|Year : 2018 | Volume
| Issue : 12 | Page : 1057-1062
Network strategy to investigate differential pathways in sporadic amyotrophic lateral sclerosis
Li-Hong Han1, Xiao-Yan Fan2, Hong Guo2, Wei Wei3, Mao-Meng Chen4, Shu-Fang Yan4
1 Department of Neurology, Zhangqiu People's Hospital, Jinan, Shandong Province, China
2 Department of Obstetrics, Zhangqiu People's Hospital, Jinan, Shandong Province, China
3 Department of Oncology, Zhangqiu People's Hospital, Jinan, Shandong Province, China
4 Department of Internal Medicine of TCM, Zhonglu Hospital, Shandong University of TCM, Jinan, Shandong Province, China
|Date of Web Publication||11-Dec-2018|
Department of Neurology, Zhangqiu People's Hospital, No. 1920, Huiquan Road, Mingshui Town, Zhangqiu, Jinan, 250014, Shandong Province
Source of Support: None, Conflict of Interest: None
Objective: The objective of this paper was to investigate differential pathways in sporadic amyotrophic lateral sclerosis (SALS) based on pathway network analysis.
Materials and Methods: To achieve this goal, first, differentially expressed genes (DEGs) between SALS and normal controls were identified, and a target network was defined as DEGs correlated interactions from the search tool for the retrieval of interacting genes/proteins (STRING). Second, topological centrality analysis was conducted on the target network to identify hub genes and hub network. Third, pathway network was constructed by taking intersections of Reactome database and STRING protein-protein interaction network. Finally, based on extracting the common interactions between target network, hub network and pathway network, we built randomized network, performed randomization test, and denoted differential pathways and hub differential pathways with P < 0.05.
Results: There were 485 DEGs and 627 interactions in the target network. The pathway network was comprised 117,370 interactions. What was more, we found that 217 pathways had intersections with the target network. By accessing randomization test and removing the intersected count <10, 21 differential pathways with P values were nearly to be 0 were obtained, of which 6 rightly were the hub differential pathways, such as gene expression, mRNA Splicing, and mRNA splicing-major pathway.
Conclusion: We have investigated 217 differential pathways and 21 significant differential pathways between SALS and normal controls based on network strategy. The findings might provide potential biomarkers for detection and therapy of SALS clinically and give great insights to reveal molecular mechanism underlying this disease. However, how these pathways cooperated with each other is still not clear, and future study should focus on this aspect.
Keywords: Differential, interaction, network, pathway, sporadic amyotrophic lateral sclerosis
|How to cite this article:|
Han LH, Fan XY, Guo H, Wei W, Chen MM, Yan SF. Network strategy to investigate differential pathways in sporadic amyotrophic lateral sclerosis. J Can Res Ther 2018;14, Suppl S5:1057-62
|How to cite this URL:|
Han LH, Fan XY, Guo H, Wei W, Chen MM, Yan SF. Network strategy to investigate differential pathways in sporadic amyotrophic lateral sclerosis. J Can Res Ther [serial online] 2018 [cited 2020 Feb 19];14:1057-62. Available from: http://www.cancerjournal.net/text.asp?2018/14/12/1057/199453
| > Introduction|| |
Amyotrophic lateral sclerosis (ALS), an idiopathic and lethal neurodegenerative disease, leads to weakness and atrophy of skeletal muscles by affecting motor neurons in the cortex, brainstem, and spinal cord. ALS patients mainly die from respiratory insufficiency after 3–5 years from symptom onset, of which 85% are sporadic ALS (SALS) and 15% are familial ALS (FALS). However, compared to extensive studies of FALS, the investigations focused on pathological and molecular mechanism underlying SALS remain a few. What is more, numerous hypotheses about the mechanisms of SALS have been proposed, for instance, gene mutation, inflammation, immune dysfunction, protein-processing, and degradation defects., However, the direct evidence to validate these assumptions are absent, and the molecular mechanism for SALS is still unclear.
Gene expression data have been applied to explore diagnostic gene signatures and biological processes of human diseases, which provide novel insights into the underlying biological mechanisms of SALS. Currently, a variety of methods have been proposed to analyze gene expression data, but few methods pay attention to quantify the interrelated behavior among genes within gene interaction network. Even though the incidence is thought to be closely related to the abnormal expression of genes, the study on differentially expressed genes (DEGs) is inadequate, and there is a long distance beyond only identifying DEG for the complex mechanism of disease. Therefore, gene interactions related studies play significant roles in biological processes, such as pathway. Pathway analysis is imperative to gain deep insight underlying individual genes, which could reduce complexity and increase explanatory power. Meanwhile, Barter et al. performed a comparative analysis and found that the network-based method was more stable than single gene and gene set method. However, there are few studies about identifying differential pathways dependent on network-based approaches.
Therefore, in the present study, we proposed a novel method to explore differential pathways between SALS patients and normal controls. The method, a network-based method, integrated target network, hub network, pathway network, and randomized network-related analyses. In detail, target network was constructed for DEGs and their correlated interactions which extracted from the search tool for the retrieval of interacting genes/proteins (STRING). Hub network was extracted from the target network via capturing hub gene-related interactions. Pathway network was built by taking intersections of Reactome database and STRING protein-protein interaction (PPI) network. Randomized network was implemented to investigate differential pathways between SALS and normal controls. These pathways might give hand to reveal mechanism underlying SALS, and provide potential biomarkers for detection and therapy of this disease.
| > Materials and Methods|| |
Target network construction
The gene expression profile for SALS with accessing number E-MTAB-2325 was recruited from ArrayExpress database. E-MTAB-2325 was consisted of 31 SALS samples and 10 normal controls and presented on A-AGIL-28 - Agilent Whole Human Genome Microarray 4×44K 014850 G4112F (85 columns × 532 rows) platform. Subsequently, standard procedures were conducted to control the quality of the data, background correction carried out based on robust multiarray average algorithm; normalization preformed according to quantile-based algorithm; probe correction implemented by microarray suite algorithm, and expression summarization proceeded through median polish method. By converting the preprocessed data on probe level into gene symbol measure, a total of 18,411 genes were obtained in the gene expression data.
Differentially expressed genes identification
DEGs between SALS and normal controls were identified for the gene expression data using significance analysis of microarrays (SAM) approach SAM assigns a score to each gene on the basis of change in gene expression relative to the standard deviation of repeated measurements. By conducting a set of gene specific t- test, genes with statistically significant changes in expression were identified. To identify significant different expressed genes further, n genes were ranked in descending order of the relative difference d(i) values, so that d(1) was the largest relative difference, d(2) was the second largest relative difference, and d(i) was the ith largest relative difference. Meanwhile dt(i) was the ith largest relative difference for permutation t. The expected relative difference, dE(i), was defined as the average over all permutations,
For the vast majority of genes, d(i) ≌ dE(i), but some genes were represented by points displaced from the d(i) = dE(i) line by a distance greater than a threshold Δ. As Δ decreased, the number of genes called significant by SAM increased. The Δ value for SALS was 0.234.
Target network construction
Due to certain significant genes may not be identifiable through their own behavior, but their changes are quantifiable when consider in conjunction with other genes (e.g., as a network); hence, we utilized a human PPI dataset from STRING to capture interactions among DEGs. Subsequently, these interactions with the edge score >0.7 were visualized by Cytoscape and formed a PPI network which was defined as the target network. Here, Cytoscape is a free software package for visualizing, modeling, and analyzing the integration of bimolecular interaction networks with high-throughput expression data and other molecular states.
Hub network extraction
For the purpose of evaluating the biological significance and functions of genes in target network, we employed the index of topological centrality, degree. Degree quantifies the local topology of each gene by summing up the number of its adjacent genes and gives a simple count of the number of interactions of a given node. The genes with degree distribution ≥15 in the significantly perturbed networks were defined as hub genes. In addition, the network, which was composed of hub genes and their interactions, was denoted as hub network.
Pathway network construction
Network can provide significant instructions for mining unknown connections in incomplete networks. Although the data of large-scale protein interactions are keeping accumulated with the development of high-throughput testing technology, a certain number of significant interactions are not tested. This type of difficulty might be resolved to some extent by utilizing subnetworks of the complex network. Therefore, we identified pathway networks by exploring interactions of pathway-enriched genes with the global human PPI network from STRING database. The pathway-enriched genes were originated from Reactome database, which is a manually curated open-data resource of human pathways and reactions.
Differential pathway identification
We had constructed the target network, hub network, and pathway network, but how to select differential pathways based on the three kinds of networks was a great challenge. To overcome the problem, we first took the intersection of interactions between pathway network and target network, hub network, respectively, and the quantity of intersected interactions was denoted as count. Subsequently, we employed randomization test to determine P value of each pathway from the intersected interactions. Randomization test provides a general means of constructing tests that control size in finite samples whenever the distribution of the observed data exhibits symmetry under the null hypothesis. Before conducting this test, we regarded the number of DEGs as A, all the possible interactions among DEGs were B, so B was equal to A × (A − 1)/2. We selected N (N was the number of interactions in target network) interactions randomly from B to construct a random network for 1000 times. Note that, to make them more confident and feasible, only pathways with P < 0.05 and count >10 were considered to be significant differential pathways.
| > Results|| |
A total of 485 DEGs were identified between SALS samples and normal controls utilizing SAM method with the threshold of Δ = 0.234. When inputting them into the STRING database and discarding the interactions with edge score <0.7, we obtained 627 interactions. Using Cytoscape, 485 DEGs and 627 interactions were mapped to the target network as shown in [Figure 1].
|Figure 1: Target network for sporadic amyotrophic lateral sclerosis. Nodes represented genes, and edges stood for gene-gene interactions. The purple nodes were hub genes. There were 485 differentially expressed genes and 627 interactions|
Click here to view
In this paper, by accessing degree centrality analysis, we detected 22 hub genes in total, HSP90AB1 (degree = 23), RNPS1 (degree = 22), ACTB (degree = 20), EEF2 (degree = 19), PRPF8 (degree = 19), RPL3 (degree = 18), PRPF6 (degree = 18), GNB2L1 (degree = 17), RPL4 (degree = 16), PPP2R1A (degree = 16), SF3B2 (degree = 16), HNRNPD (degree = 16), HSPA5 (degree = 16), SRSF5 (degree = 16), POLR2E (degree = 16), CD2BP2 (degree = 16), FUS (degree = 15), SEC61A1 (degree = 15), PABPN1 (degree = 15), SUGP1 (degree = 15), and HNRNPA2B1 (degree = 15). Further, the hub network, which was composed of hub genes and their interactions, was extracted from the target network. All of the hub genes were mapped to the hub network [Figure 2]. There were 270 edges and 111 nodes in the hub network.
|Figure 2: Hub network for sporadic amyotrophic lateral sclerosis. Nodes represented genes, and edges stood for gene-gene interactions, the purple nodes were hub genes. A total of 21 hub genes were identified|
Click here to view
The construction of pathway network was on the basis of STRING PPI network which was composed of 787,896 interactions, and the Reactome database which included 1675 human pathways. The STRING PPIs of pathway-enriched genes were captured, and thus, we obtained 117,370 interactions to form pathway network. The 117,370 interactions might contain reduplicative interactions which resulting from repeated enrichments of one interaction, that was to say, one interaction probably enriched in two or more pathways.
In this paper, randomization test was implemented to identify differential pathways between SALS and normal controls. The first step was taking intersections among pathway network and target network, and pathway network and hub network, respectively. We found that 217 pathways had interactions with the target network, but numbers of interactions for different pathways were considerable difference. What was more, differential pathways must meet to two conditions; one was P value for the pathway was <0.05, and the other was the count which meant the quantity of intersected interactions between pathway network and target network or hub network was ≥10. The result showed that a total of 21 differential pathways were gained dependent on pathway network and target network [Table 1]. The top five differential pathways were gene expression (count = 105), mRNA Splicing (count = 55), mRNA splicing - Major pathway (count = 55), processing of capped intron-containing pre-mRNA (count = 55), and metabolism of proteins (count = 29).
|Table 1: Differential pathways based on pathway network and target network|
Click here to view
Meanwhile, we explored 57 pathways had intersections with hub network. After satisfying two conditions described above, 6 differential pathways were identified; interestingly, all of them belonged to the 21 differential pathways based on pathway network and target network, which suggested that the 6 pathways were more significant than others, termed with hub differential pathways. They were displayed in [Table 2], gene expression (count = 76), mRNA splicing (count = 54), mRNA splicing - Major pathway (count = 54), processing of capped intron-containing pre-mRNA (count = 54), metabolism of proteins (count = 16), and disease (count = 11).
|Table 2: Hub differential pathways based on pathway network and hub network|
Click here to view
If P < 0.05, we considered the pathway was differential pathway. Interestingly, most of the P was close to 0 or equal to 0, which suggested that these pathways were significantly differential. Due to the similar P of differential pathways, count might be another measure to evaluate the significance of pathways.
| > Discussion|| |
To gain a system-level understanding, it is necessary to examine how genes interact on a large-scale level; meanwhile, genes do not work in isolation, they are connected in highly structured networks., Besides the understanding of the molecular process that led to SALS might facilitate the discovery of novel therapeutic targets. Therefore, in this paper, we proposed a novel method to identify differential pathways which might be potential molecular biomarkers between SALS and normal controls. The main components of the novel method were target network construction, pathway network construction, and differential pathway identification according to randomization tests.
There were 485 DEGs and 627 interactions in the target network, 270 edges and 111 nodes in the hub network, and 117,370 interactions in the pathway network. What was more, we found that 217 and 57 pathways had intersections with the target network and hub network, respectively. By accessing randomization test and removing the intersected count <10, 21 differential pathways and 6 hub differential pathways were obtained. The P values for these pathways were all nearly to be 0, which suggested that they were significantly different between SALS and normal controls. In addition, we discovered that the 6 hub differential pathways were the part of the 21 differential pathways, and thus concluded that the 6 were significant differential pathways for SALS. They were gene expression, mRNA splicing, mRNA splicing - Major pathway, processing of capped intron-containing pre-mRNA, metabolism of proteins, and disease.
Gene expression is exactly regulated to ensure the complement of RNA and proteins, and mainly includes transcription, processing of capped intron-containing pre-mRNA, mRNA splicing, translation, and posttranslation. Transcription and translation occurred in separate compartments, nucleus, and cytoplasm, respectively. This allows the extensive posttranscriptional process of pre-mRNA and produced more diverse assortment of mRNAs. It is well known that RNA has crucial roles in cell biology due to the diversity of its sequence and structure. Significantly, the well-studied ALS-related gene FUS was demonstrated to have a function on ALS by regulating the process of transcription, RNA splicing, and transport. Besides TAR DNA-binding protein of 43 kDa (TDP-43), a nuclear DNA/RNA binding protein that involved in the regulation of transcription, alternative splicing, miRNA processing, and mRNA stabilization were proved to function an important role in ALS by abnormal splicing event. Furthermore, mutations in ALS-genes have a modulatory effect on miRNAs to regulate gene products.,,, All previous studies described above demonstrated that the differential pathways related to SALS isolated by a novel method are reliable. Therefore we might infer that significant differential pathways were correlated to SALS closely, and their dysregulation possibly led to the progression of this disease.
There were other investigations demonstrated that SALS were related to other metabolisms, such as a study by Wang et al. proved that a pathway named Notch which controls cell fate decisions, migration, growth, synaptic plasticity and neuronal survival played a critical role in ALS. And particularly, Xie et al. isolated 10 pathways associated with SALS based on the Gene Set Analysis Toolkit V2. Different with the pathways isolated in this investigation, These pathways were Wnt signaling pathway, T-cell receptor signaling pathway, MAPK signaling pathway, axon guidance, phosphatidylinositol signaling system, arachidonic acid metabolism and neurotrophin signaling pathway. Considering the complicated mechanisms under the disease, it is possible that various pathways were involved in SALS, thus further detailed investigations are indispensible to find effective treatment for ALS.
| > Conclusion|| |
A complex interaction among genes, impaired pathways, and environmental factors contribute to the pathology of SALS. In our investigation, we have successfully identified differential pathways and hub differential pathways associated with SALS. Most of these pathways focused on gene expression. This provides us a new therapy direction to SALS and gives great insight to reveal molecular mechanism underlying this disease. However, how these pathways interact with each other is still not clear, and our future study aims to work out the challenge.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| > References|| |
Robberecht W, Philips T. The changing scene of amyotrophic lateral sclerosis. Nat Rev Neurosci 2013;14:248-64.
Mitchell JD, Borasio GD. Amyotrophic lateral sclerosis. Lancet 2007;369:2031-41.
Renton AE, Chiò A, Traynor BJ. State of play in amyotrophic lateral sclerosis genetics. Nat Neurosci 2014;17:17-23.
Zhang R, Gascon R, Miller RG, Gelinas DF, Mass J, Hadlock K, et al.
Evidence for systemic immune system alterations in sporadic amyotrophic lateral sclerosis (sALS). J Neuroimmunol 2005;159:215-24.
Bruijn LI, Miller TM, Cleveland DW. Unraveling the mechanisms involved in motor neuron degeneration in ALS. Annu Rev Neurosci 2004;27:723-49.
Cleveland DW, Rothstein JD. From charcot to lou gehrig: Deciphering selective motor neuron death in ALS. Nat Rev Neurosci 2001;2:806-19.
Jordán F, Nguyen TP, Liu WC. Studying protein-protein interaction networks: A systems view on diseases. Brief Funct Genomics 2012;11:497-504.
Toyoshiba H, Yamanaka T, Sone H, Parham FM, Walker NJ, Martinez J, et al.
Gene interaction network suggests dioxin induces a significant linkage between aryl hydrocarbon receptor and retinoic acid receptor beta. Environ Health Perspect 2004;112:1217-24.
Toyoshiba H, Sone H, Yamanaka T, Parham FM, Irwin RD, Boorman GA, et al.
Gene interaction network analysis suggests differences between high and low doses of acetaminophen. Toxicol Appl Pharmacol 2006;215:306-16.
Khatri P, Sirota M, Butte AJ. Ten years of pathway analysis: Current approaches and outstanding challenges. PLoS Comput Biol 2012;8:e1002375.
Barter RL, Schramm SJ, Mann GJ, Yang YH. Network-based biomarkers enhance classical approaches to prognostic gene expression signatures. BMC Syst Biol 2014;8 Suppl 4:S5.
Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. Summaries of affymetrix GeneChip probe level data. Nucleic Acids Res 2003;31:e15.
Bolstad BM, Irizarry RA, Astrand M, Speed TP. A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 2003;19:185-93.
Li J, Tibshirani R. Finding consistent patterns: A nonparametric approach for identifying differential expression in RNA-Seq data. Stat Methods Med Res 2013;22:519-36.
Zhuang DY, Jiang L, He QQ, Zhou P, Yue T. Identification of hub subnetwork based on topological features of genes in breast cancer. Int J Mol Med 2015;35:664-74.
Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T. Cytoscape 2.8: New features for data integration and network visualization. Bioinformatics 2011;27:431-2.
Haythornthwaite C. Social network analysis: An approach and technique for the study of information exchange. Libr Inf Sci Res 1996;18:323-42.
Nibbe RK, Chowdhury SA, Koyutürk M, Ewing R, Chance MR. Protein-protein interaction networks and subnetworks in the biology of disease. Wiley Interdiscip Rev Syst Biol Med 2011;3:357-67.
Wu Y, Jing R, Jiang L, Jiang Y, Kuang Q, Ye L, et al.
Combination use of protein-protein interaction network topological features improves the predictive scores of deleterious non-synonymous single-nucleotide polymorphisms. Amino Acids 2014;46:2025-35.
Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, et al.
STRING v10: Protein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res 2015;43:D447-52.
Croft D, Mundo AF, Haw R, Milacic M, Weiser J, Wu G, et al.
The reactome pathway knowledgebase. Nucleic Acids Res 2014;42:D472-7.
Canay IA, Romano JP, Shaikh AM. Randomization Tests Under an Approximate Symmetry Assumption. Technical Report No. 2014-13, Stanford University, USA.
Liu ZP, Wang Y, Zhang XS, Chen L. Network-based analysis of complex diseases. IET Syst Biol 2012;6:22-33.
Chen L, Wang RS, Zhang XS. Reconstruction of gene regulatory networks, in biomolecular networks. John Wiley & Sons, Inc.; 2009. p. 47-87.
Glisovic T, Bachorik JL, Yong J, Dreyfuss G. RNA-binding proteins and post-transcriptional gene regulation. FEBS Lett 2008;582:1977-86.
Licatalosi DD, Darnell RB. RNA processing and its regulation: Global insights into biological networks. Nat Rev Genet 2010;11:75-87.
Vance C, Rogelj B, Hortobágyi T, De Vos KJ, Nishimura AL, Sreedharan J, et al.
Mutations in FUS, an RNA processing protein, cause familial amyotrophic lateral sclerosis type 6. Science 2009;323:1208-11.
Baralle M, Buratti E, Baralle FE. The role of TDP-43 in the pathogenesis of ALS and FTLD. Biochem Soc Trans 2013;41:1536-40.
Xiao S, Sanelli T, Chiang H, Sun Y, Chakrabartty A, Keith J, et al.
Low molecular weight species of TDP-43 generated by abnormal splicing form inclusions in amyotrophic lateral sclerosis and result in motor neuron death. Acta Neuropathol 2015;130:49-61.
Dini Modigliani S, Morlando M, Errichelli L, Sabatelli M, Bozzoni I. An ALS-associated mutation in the FUS 3'-UTR disrupts a microRNA-FUS regulatory circuitry. Nat Commun 2014;5:4335.
Kabashi E, Bercier V, Lissouba A, Liao M, Brustein E, Rouleau GA, et al.
FUS and TARDBP but not SOD1 interact in genetic models of amyotrophic lateral sclerosis. PLoS Genet 2011;7:e1002214.
Gascon E, Gao FB. The emerging roles of microRNAs in the pathogenesis of frontotemporal dementia-amyotrophic lateral sclerosis (FTD-ALS) spectrum disorders. J Neurogenet 2014;28:30-40.
Jiao J, Herl LD, Farese RV, Gao FB. MicroRNA-29b regulates the expression level of human progranulin, a secreted glycoprotein implicated in frontotemporal dementia. PLoS One 2010;5:e10551.
Wang SY, Ren M, Jiang HZ, Wang J, Jiang HQ, Yin X, et al.
Notch pathway is activated in cell culture and mouse models of mutant SOD1-related familial amyotrophic lateral sclerosis, with suppression of its activation as an additional mechanism of neuroprotection for lithium and valproate. Neuroscience 2015;301:276-88.
Ables JL, Breunig JJ, Eisch AJ, Rakic P. Not(ch) just development: Notch signalling in the adult brain. Nat Rev Neurosci 2011;12:269-83.
Xie T, Deng L, Mei P, Zhou Y, Wang B, Zhang J, et al.
Genome-wide association study combining pathway analysis for typical sporadic amyotrophic lateral sclerosis in Chinese Han populations. Neurobiol Aging 2014;35:1778.
[Figure 1], [Figure 2]
[Table 1], [Table 2]