Next Article in Journal
Angiotensin II Stimulation of DPP4 Activity Regulates Megalin in the Proximal Tubules
Previous Article in Journal
The Natural Course of Non-Alcoholic Fatty Liver Disease
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Integrated Analysis of Expression Profile Based on Differentially Expressed Genes in Middle Cerebral Artery Occlusion Animal Models

1
Department of Anesthesiology, the First Affiliated Hospital of Sun Yat-sen University, Guangzhou 510080, China
2
Sun Yat-sen University School of Medicine, Sun Yat-sen University, Guangzhou 510080, China
*
Author to whom correspondence should be addressed.
These authors contributed equally to this work.
Int. J. Mol. Sci. 2016, 17(5), 776; https://0-doi-org.brum.beds.ac.uk/10.3390/ijms17050776
Submission received: 5 April 2016 / Revised: 10 May 2016 / Accepted: 16 May 2016 / Published: 20 May 2016
(This article belongs to the Section Molecular Pathology, Diagnostics, and Therapeutics)

Abstract

:
Stroke is one of the most common causes of death, only second to heart disease. Molecular investigations about stroke are in acute shortage nowadays. This study is intended to explore a gene expression profile after brain ischemia reperfusion. Meta-analysis, differential expression analysis, and integrated analysis were employed on an eight microarray series. We explored the functions and pathways of target genes in gene ontology (GO) enrichment analysis and constructed a protein-protein interaction network. Meta-analysis identified 360 differentially expressed genes (DEGs) for Mus musculus and 255 for Rattus norvegicus. Differential expression analysis identified 44 DEGs for Mus musculus and 21 for Rattus norvegicus. Timp1 and Lcn2 were overexpressed in both species. The cytokine-cytokine receptor interaction and chemokine signaling pathway were highly enriched for the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. We have exhibited a global view of the potential molecular differences between middle cerebral artery occlusion (MCAO) animal model and sham for Mus musculus or Rattus norvegicus, including the biological process and enriched pathways in DEGs. This research helps contribute to a clearer understanding of the inflammation process and accurate identification of ischemic infarction stages, which might be transformed into a therapeutic approach.

Graphical Abstract

1. Introduction

Stroke is the second most common cause of death after heart disease. Reduced brain blood flow leads to the loss of the oxygen, glucose, and ATP in cerebral tissue and the depolarization of neurons [1]. Stroke also elevates extracellular potassium concentration and increases oxidative stress. Eccentrically, reperfusion produces the excessive reactive oxygen species (ROS) and leads to additional cerebral injuries, after inflammatory changes and neurotic and apoptotic cell death pathways [2]. Inflammation is the key factor in ischemic stroke, yet the current molecular research concerning the pathways is scarce [3,4].
High throughput technologies can detect the unusual genomic changes among the mass gene expression profiling with microarray. However, previous studies have failed to provide ample samples, and results of the studies are often conflicting and contradictory, which has left their credibility in question. In recent studies, meta-analyses have been employed to detect the differentially expressed genes (DEGs) statistically with an assessment of heterogeneity [5]. In order to overcome the limitations above and generate convincing results, a careful meta-analysis must combine gene expression datasets from multiple sources [6,7].
Recently it was shown that meta-analyses of gene expression data from plenty of sources increased statistical power by pooling effect size with assessment of heterogeneity, and would lead to more powerful, reproducible, and accurate results. Several meta-analysis of microarray data have been reported for a number of central neural system (CNS) conditions. For brain ischemia, single experiments are available, but no coherent meta-analysis has been reported so far. We described an integrated analysis of available data from rodent middle cerebral artery occlusion (MCAO) models and identified molecular signatures and pathways involved in the pathophysiology of stroke in order to overcome the restriction of individual studies and solve inconsistencies among various studies.

2. Results

2.1. Microarray Data Search and Data Preprocessing

2.1.1. Microarray Data Search

After a search in the GEO and the ArrayExpress microarray data repositories, we obtained 128 records without duplications and learned that the goals of relevant studies vary according to different learning paradigms, test conditions, subjects, and tissue types. After filtration according to the inclusion and exclusion criteria, eight datasets were finally identified with four Mus musculus studies and four Rattus norvegicus studies, including a total 36 (Mus musculus: 19; Rattus norvegicus: 17) sham models and 65 (Mus musculus: 39; Rattus norvegicus: 26) MCAO models (Figure 1). The table below presents the basic information about the dataset included. Considering there are two species (Mus musculus and Rattus norvegicus), meta-analysis was performed separately. Characteristics of each study in meta-analysis are presented in Table 1.

2.1.2. Data Preprocessing

Only sham and MCAO arrays were used for further analyses. We carried out probe-to-gene mapping to convert probe-set expression levels into gene expression levels according to the specific chip dataset. After data integration among different chips, there were 13,590 common genes in Mus musculus and 11,880 common genes in Rattus norvegicus.

2.2. Meta-Analysis

We applied five methods to select the genes at the significance level of 0.05. The DEGs identified by at least four methods were selected for further studies. 360 genes (A1: Mus musculus DEGs set) were regarded as significantly differentially expressed for four methods while 255 genes (A2: Rattus norvegicus DEGs set) in Rattus norvegicus fell under the threshold of p-value <0.05 for four methods. The following function annotation and pathway enrichment were operated on Mus musculus and Rattus norvegicus separately and the two species bore some remarkable resemblances.

2.2.1. Mus Function Annotation and Pathway Enrichment Analysis

The response to the molecule of bacterial origin (GO:0002237, p = 2.10 × 10−3) and regulation of interleukin-12 production (GO:0032655, p = 2.46 × 10−3) were significantly enriched for biological processes. Calcium ion binding (GO:0005509, p = 1.03 × 10−2) and growth factor binding (GO:0019838, p = 1.21 × 10−2) were for molecular function, while cytosolic part (GO:0044445, p = 5.75 × 10−3) and membrane-enclosed lumen (GO:0031974, p = 2.27 × 10−2) were for cellular component (Table 2). The most significant pathway was the MAPK signaling pathway (p = 8.14 × 10−5) and cytokine-cytokine receptor interaction (p = 2.65 × 10−2) was highly enriched (Table 3).

2.2.2. Rat Function Annotation and Pathway Enrichment Analysis

Response to wounding (GO:0009611, p = 1.91 × 10−7) and regulation of cell morphogenesis (GO:0022604, p = 5.44 × 10−6) were significantly enriched for biological processes. Calcium ion binding (GO:0005509, p = 5.33 × 10−5) and ion channel activity (GO:0005216, p = 6.73 × 10−5) were for molecular functions, while synapse part (GO:0044456, p = 1.44 × 10−6) and plasma membrane (GO:0005886, p = 1.52 × 10−6) (Table 4) were for cellular components. The most significant pathway was in Long-term depression (p = 5.93 × 10−4). MAPK signaling pathway (p = 8.48 × 10−4) was also highly enriched.

2.3. Differential Expression Analysis in Each Study

We performed differentially expressed analysis between MCAO and sham samples in individual series. Genes were regarded as significantly differentially expressed under the threshold of p-value <0.01 and |log2 fold change (FC)| ≥ 1.5. The results were summarized in Table 5. Further study aimed at genes recognized as DEGs by at least three studies. Forty-four DEGs (B1: Mus musculus DEGs set) were identified in Mus musculus, 21 DEGs (B2: Rattus norvegicus DEGs set) in Rattus norvegicus. We inputted B1 into the Database for Annotation, Visualization, and Integrated Discovery (DAVID). Chemotaxis (GO:0006935, p = 6.10 × 10−7) and taxis (GO:0042330, p = 6.10 × 10−7) were significantly enriched for biological processes. For KEGG pathway detection, the most significant pathway was found in the cytokine-cytokine receptor interaction (p = 1.30 × 10−4). Furthermore, the Chemokine signaling pathway (p = 3.20 × 10−3) was highly enriched. However, B2 could not be well described by DAVID for the limitation of the number of B2.

2.4. Integrated Analysis

In integrated analysis, there were 12 common DEGs identified by both individual differential expression analysis and meta-analysis for MCAO Mus musculus species, and eight for MCAO Rattus norvegicus species. A protein-protein interaction (PPI) network was constructed separately centering on potential hub proteins for Mus musculus or Rattus norvegicus (Figure 2). Confidence score represented probability of predicted association for both adjacent proteins against a common reference set of functional grouping of proteins maintained in the Kyoto Encyclopedia of Genes and Genomes (KEGG) [16]. We found the common DEGs across species and identified that Timp1 and Lcn2 were both overexpressed in the MCAO Mus musculus or Rattus norvegicus.

3. Discussion

Many genes are expressed differently after cerebral ischemia. A molecular and cellular analysis during stroke is of great importance for clinical management. In the paper, we have performed an integrated analysis using eight publicly available GEO datasets to identify the mechanisms in a stroke. We identified significantly up- or down-regulated genes. We also explored their biological function based on GO enrichment analysis, KEGG pathway analysis, and PPI networks.
Five approaches were applied during meta-analysis. Three hundred and sixty genes were identified as DEGs in Mus musculus, and 255 genes in Rattus norvegicus. The function annotation and pathway enrichment were operated on Mus musculus and Rattus norvegicus separately and the two species bore some remarkable resemblances. Calcium ion binding was significantly enriched for molecular functions, while MARK signaling pathway was for KEGG pathway analysis. The activation of the p38 MAPK pathway affects the production of pro-inflammatory cytokines such as TNFα, IL-1β, and IL-6 [17]. Roy observed p38 MAPK activation in the glia scar area in a MCAO mouse and suggested that the p38 MAPK signaling pathway might have an impact on the ischemic stroke [18]. Moreover, Chang found that the MAPK signal pathway modulated MMP activity after stroke [19].
We performed differentially expressed analysis between MCAO and sham samples in each study, and obtained the intersection sets of DEG results. Chemotaxis was significantly enriched for biological processes. The cytokine-cytokine receptor interaction and chemokine signaling pathway were highly enriched. Inflammation is important in an ischemic stroke. During a stroke, activation of astrocytes and microglia rapidly strengthen, promote the production of cytokines (including IL-1β, IL-6, and TNF-α and other proinflammatory cytokines) through different pathways such as MAPK signal pathway, induce neutrophils and peripheral lymphocyte infiltration in the focal, then further aggravate the central nervous system injury [20,21,22]. In reports, chemokines were relative to stroke and more researchers regarded them as a new therapeutic target or prognostic indicators. Arac described that interleukin 6 and chemokine (C-C motif) ligand 7, contributed to stroke pathology [23]. A six month follow up study in Chinese stroke patients indicated the association between the high levels of CXC chemokine ligand 12 and unfavorable outcomes and mortality [24]. A blockade of CXCR1/2 chemokine receptors and ELR CXC chemokine antagonism protected mice in ischemic stroke [25,26].
Timp1 and Lcn2 were regarded as common significant DEGs across species when we performed integrated analysis, either of which was observed to be over-expressed in MCAO mice (Figure 3). Timp1, the TIMP metallopeptidase inhibitor 1, is a glycoprotein against most of the known MMPs, and may have an anti-apoptotic function [27]. Wang utilized a subtractive cDNA library strategy in focal strokes and identified Timp1 with increased gene expression [28]. The relationship between Timp1 and MMPs is intriguing. Specifically, the MMP9/ Timp1 was elevated on the seventh day of stroke, which might provide a biomarker of prognosis [29,30]. Considering what we mentioned before, MAPK signal pathway modulates MMP activity after a stroke. In recent years, more studies are engaged in the relationships between Timp1 and MAPK pathways [19]. Overexpression of endogenous Timp1 activates ERK1/2 and JNK1/2 of the MAPK signaling pathway [31]. p38 MAPK inhibitors block all natural haemozoin effects on Timp1 and proinflammatory molecules [32]. Lcn2, neutrophil gelatinase-associated lipocalin (NGAL) sequestrates iron and hinders bacterial growth [33,34]. Recently, rapidly elevated Lcn2 levels have been reported in the plasma of ischemic stroke patients, but its role in strokes is unknown [35,36]. Jin indicated that Lcn2 might contribute to the death of neuronal cells in the ischemic brain by promoting neurotoxic glial activation, neuroinflammation, and blood-brain barrier disruption [36,37]. It suggests that Lcn2 is a potential early stroke biomarker and a novel therapeutic target to reduce injury. In order to illustrate its molecular function in ischemia and reperfusion, more research needs to be conducted.
Except Timp1, CXCL1 is the hub of the PPI network. CXCL1 belongs to the CXC chemokine family and was previously called GRO1 oncogene, GROα [38]. CXCL1 is expressed by macrophages, neutrophils, and epithelial cells in neutrophil chemoattractant activity, inflammation, wound healing, and angiogenesis [39,40]. It elicits its effects by signaling CXCR2 [41]. In 2005, Losy observed that there was higher CXCL1 in cerebrospinal fluid level than the control group in stroke patients, and suggested the potential neutrophil chemoattractant function in the pathophysiology of stroke [42]. No new evidence had surfaced to illustrate it, until our study confirmed the finding of Losy at the gene expression level.
Undeniably, our current study has many limitations. First, different studies might be heterogeneous due to the variation of microarray platforms, labs, technicians, and so on. Heterogeneity and confounding factors may be interfering with the analysis, even under the circumstance that we applied strict criteria to the studies included in order to minimize the potential for errors; Second, there are differences in gene expression between species such as Mus musculus and Rattus norvegicus. Attempting to minimize the influence, we analyzed them separately before we integrated the analyses to detect common important genes in the species; Third, a number of samples did not provide details of different periods. As a result, we were unable to do a meta-analysis of expression profiles along with the time of developing; Fourth, meta-analysis, differential expression analysis, and integrated analysis were performed to reduce heterogeneity owing to various microarray platforms instead of putting all the expression matrix together and then performing differential expression analysis [43].

4. Materials and Methods

4.1. Microarray Data Search

We searched the Gene Expression Omnibus (GEO, available at: http://0-www-ncbi-nlm-nih-gov.brum.beds.ac.uk/geo/) and the ArrayExpress (available at:http://www.ebi.ac.uk/arrayexpress/) microarray data repositories for microarray gene expression datasets by using the keywords “brain ischemia reperfusion”, “stroke”, and “MCAO”. We also used the literature database PubMed and Embase to search for relevant studies. The available raw data were downloaded from the GEO and ArrayExpress public databases.

4.2. Included and Excluded Criteria

The eligible criteria were as follows: (1) the examined samples were tissue section of ipsilateral cerebral cortex; (2) the data were on gene expression level; (3) the species were Rattus norvegicus or mice; (4) the microarray platform was limited to single channel; (5) the occlusion in the trial group must be in the middle cerebral artery and sham groups must undergo the same surgery but without real occlusion. The exclusion criteria were as follows: (1) samples were from other tissues or vitro assays; (2) datasets were not the original gene expression level data files; (3) the data were with redundant subdatasets; (4) samples were without at least three repetitions in an experiment; (5) the samples had permanent ischemia reperfusion; (6) the data had only miRNA profiles; (7) the series had low quality, such as containing more than 20% missing values on arrays.

4.3. Data Preprocessing

The data were analyzed by R software (available at: http://www.R-project.org/) and packages in Bioconductor (available at: http://www.bioconductor.org/). We used the visual inspection for image quality, relative log expression (RLE), and normalized unscaled standard errors (NUSE) for data quality and affyPLM package for RNA degradation. We only included the samples with high quality and the independent decision by the original authors [44]. The expression data were processed using the limma package or affy package in R software, including background correction, quantile normalization, and final probe summarization [45,46]. Missing values were filled by using the k-nearest neighbors (KNN) method based on the average of non-missing neighboring values of its neighbor. For each probe-set, intensity values were log2-transformed and normalized to 0 mean and unit variance [47]. Lastly, we carried out probe-to-gene mapping to convert probe-set expression levels into gene expression levels according to the specific chip dataset [48]. If multiple probes mapped to a gene, the mean of the probe effect size was selected.

4.4. Meta-Analysis

For either Rattus norvegicus or Mus musculus, we merged the common gene-set among the studies before meta-analysis. The original data sets have been preprocessed and subsampled to reduce the computational complexity of the meta-analysis. Considering there were cross species (Mus musculus and Rattus norvegicus), we performed meta-analysis separately. Using MAMA package, five approaches had been utilized to select the genes at the significance level of 0.05, including (1) combine p-values in sense of sum of logs by Fisher′s S-statistic; (2) combine effect sizes by classical and moderated t-test and random-effect model [49]; (3) non-parametric statistic of RankProduct that detects up-regulated and down-regulated genes under one condition against another condition [50]; (4) Z-statistic utilizing posterior mean differential expression weighted by variances; (5) combine effect sizes by Hedge’s and Olkin’s g method. The DEGs were recorded by at least four of five approaches, which were further researched in function annotation and pathway enrichment analysis.

4.5. Differential Expression Analysis in Each Study

Using the limma package, false discovery rate correction was performed and we selected DEGs with the criteria of both p-value < 0.05 and |log2fold change (FC)| ≥ 1.5. Considering there are cross species (Mus musculus and Rattus norvegicus), we performed differential expression analysis separately and identified the common DEGs in at least three of four series for further exploration. Venn diagrams were constructed to display the intersection of DEGs results.

4.6. Integrated Analysis

The individual datasets are useful for validation to show that genes identified from meta-analysis show up in the individual datasets and are concordant in expression change. To elevate the positive predictive power, we entered DEGs from meta-analysis into each microarray. Thus, more specific DEGs were filtered when they were significantly expressed in at least three individual datasets (there were four datasets in total). Next, the potential hub proteins were sought through the protein-protein interaction network based on the DEGs above. The overall analysis workflow is depicted in Figure 4.

4.7. Functional Annotation

Functional and pathway analyses were performed mainly using the Database for Annotation, Visualization, and Integrated Discovery (DAVID, available at: https://david.ncifcrf.gov/). To understand the significance of genes, we performed the Gene Ontology (GO) classification, making use of the following categories: BP_Fat (biological process), CC_Fat (cellular component), and MF_Fat (molecular function) [51]. We also performed the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis to detect the potential pathway of target genes and had the hypergeometric test with p value <0.05 [52]. STRING version 10.0 (available at: http://string-db.org/) comprises of >1100 completely sequenced organisms [53]. To identify the interactive associations among genes, the DEGs were input into STRING. The Cytoscape v2.8.0 software was used to visualize these associations and the mined modules [16].
Data were integrated across studies and across array platforms after data preprocessing and analyzed in several ways: meta-analysis and individual differential expression analysis and integrated analysis.

5. Conclusions

In conclusion, through multiple analyses based on gene expression, we have exhibited a global view of the potential molecular differences between MCAO model and sham for Mus musculus or Rattus norvegicus, including DEGs and their enriched pathways. This research helps contribute to the further understanding of the pathophysiology of stroke involving inflammation process and accurate identification of different stages of ischemic infarction, which might be transformed into a therapeutic approach.

Acknowledgments

This work was funded in part by the Fundamental scientific research program of Sun Yat-sen University (No. 16ykpy36).

Author Contributions

Zhongxing Wang conceived and designed the study; Huaqiang Zhou and Zeting Qiu analyzed the data; Shaowei Gao and Qinchang Chen performed literature search. Si Li, Wulin Tan, and Xiaochen Liu worked with the figures and table. Huaqiang Zhou and Zeting Qiu wrote the paper.

Conflicts of Interest

The authors declare no conflict of interest.

Abbreviations

GOGene ontology
DEGsDifferentially expressed genes
ROSReactive oxygen species
CNSCentral neural system
RLERelative log expression
NUSENormalized unscaled standard errors
KNNk-nearest neighbors
DAVIDDatabase for Annotation, Visualization and Integrated Discovery
KEGGKyoto Encyclopedia of Genes and Genomes
PPIProtein-protein interaction
NGALNeutrophil gelatinase-associated lipocalin
MCAOMiddle cerebral artery occlusion

References

  1. Donnan, G.A.; Fisher, M.; Macleod, M.; Davis, S.M. Stroke. Lancet 2008, 371, 1612–1623. [Google Scholar] [CrossRef]
  2. Higgins, G.C.; Beart, P.M.; Shin, Y.S.; Chen, M.J.; Cheung, N.S.; Nagley, P. Oxidative stress: Emerging mitochondrial and cellular themes and variations in neuronal injury. J. Alzheimers Dis. 2010, 20, S453–S473. [Google Scholar] [PubMed]
  3. Ferrer, I.; Planas, A.M. Signaling of cell death and cell survival following focal cerebral ischemia: Life and death struggle in the penumbra. J. Neuropathol. Exp. Neurol. 2003, 62, 329–339. [Google Scholar] [CrossRef] [PubMed]
  4. Wong, C.H.Y.; Crack, P.J. Modulation of neuro-inflammation and vascular response by oxidative stress following cerebral ischemia-reperfusion injury. Curr. Med. Chem. 2008, 15, 1–14. [Google Scholar] [PubMed]
  5. Siddiqui, A.S.; Delaney, A.D.; Schnerch, A.; Griffith, O.L.; Jones, S.J.M.; Marra, M.A. Sequence biases in large scale gene expression profiling data. Nucleic Acids Res. 2006, 34. [Google Scholar] [CrossRef] [PubMed]
  6. Ramasamy, A.; Mondry, A.; Holmes, C.C.; Altman, D.G. Key issues in conducting a meta-analysis of gene expression microarray datasets. PLoS Med. 2008, 5, e184. [Google Scholar] [CrossRef] [PubMed]
  7. Feichtinger, J.; Julia, F.; Thallinger, G.G.; McFarlane, R.J.; Larcombe, L.D. Microarray meta-analysis: From data to expression to biological relationships. In Computational Medicine; Springer-Vienna: New York, NY, USA, 2012; pp. 59–77. [Google Scholar]
  8. Gertz, K.; Kronenberg, G.; Kälin, R.E.; Baldinger, T.; Werner, C.; Balkaya, M.; Eom, G.D.; Hellmann-Regen, J.; Kröber, J.; Miller, K.R. Essential role of interleukin-6 in post-stroke angiogenesis. Brain 2012, 135, 1964–1980. [Google Scholar] [CrossRef] [PubMed]
  9. Chen, M.J.; Wong, C.H.; Peng, Z.F.; Manikandan, J.; Melendez, A.J.; Tan, T.M.; Crack, P.J.; Cheung, N.S. A global transcriptomic view of the multifaceted role of glutathione peroxidase-1 in cerebral ischemic—Reperfusion injury. Free Radic. Biol. Med. 2011, 50, 736–748. [Google Scholar] [CrossRef] [PubMed]
  10. White, R.E.; Palm, C.; Xu, L.; Ling, E.; Ginsburg, M.; Bernie, J.D.; Han, R.; Patterson, A.; Altman, R.B.; Giffard, R.G. Mice lacking the β2 adrenergic receptor have a unique genetic profile before and after focal brain ischaemia. ASN Neuro 2012, 4. [Google Scholar] [CrossRef] [PubMed]
  11. Wu, L.; Jiang, Y.; Zhu, J.; Wen, Z.; Xu, X.; Xu, X.; Xie, Y.; Yang, L.; Xu, L.; Lan, W.; et al. Orosomucoid1: Involved in vascular endothelial growth factor-induced blood-brain barrier leakage after ischemic stroke in mouse. Brain Res. Bull. 2014, 109, 88–98. [Google Scholar] [CrossRef] [PubMed]
  12. Szydlowska, K.; Gozdz, A.; Dabrowski, M.; Zawadzka, M.; Kaminska, B. Prolonged activation of ERK triggers glutamate-induced apoptosis of astrocytes: Neuroprotective effect of FK506. J. Neurochem. 2010, 113, 904–918. [Google Scholar] [CrossRef] [PubMed]
  13. Mengozzi, M.; Cervellini, I.; Villa, P.; Erbayraktar, Z.; Gökmen, N.; Yilmaz, O.; Erbayraktar, S.; Manohasandra, M.; van Hummelen, P.; Vandenabeele, P. Erythropoietin-induced changes in brain gene expression reveal induction of synaptic plasticity genes in experimental stroke. Proc. Natl. Acad. Sci. USA 2012, 109, 9617–9622. [Google Scholar] [CrossRef] [PubMed]
  14. Lai, W.; Zheng, Z.; Zhang, X.; Wei, Y.; Chu, K.; Brown, J.; Hong, G.; Chen, L. Salidroside-mediated neuroprotection is associated with induction of early growth response genes (EGRS) across a wide therapeutic window. Neurotox. Res. 2015, 28, 108–121. [Google Scholar] [CrossRef] [PubMed]
  15. Wang, L.; Yu, Y.; Yang, J.; Zhao, X.; Li, Z. Dissecting Xuesaitong’s mechanisms on preventing stroke based on the microarray and connectivity map. Mol. BioSyst. 2015, 11, 3033–3039. [Google Scholar] [CrossRef] [PubMed]
  16. Franceschini, A.; Szklarczyk, D.; Frankild, S.; Kuhn, M.; Simonovic, M.; Roth, A.; Lin, J.; Minguez, P.; Bork, P.; von Mering, C.; et al. STRING v9.1: Protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Res. 2012, 41, D808–D815. [Google Scholar] [CrossRef] [PubMed]
  17. Wang, C.X.; Shuaib, A. Involvement of inflammatory cytokines in central nervous system injury. Prog. Neurobiol. 2002, 67, 161–172. [Google Scholar] [CrossRef]
  18. Roy Choudhury, G.; Ryou, M.-G.; Poteet, E.; Wen, Y.; He, R.; Sun, F.; Yuan, F.; Jin, K.; Yang, S.-H. Involvement of p38 MAPK in reactive astrogliosis induced by ischemic stroke. Brain Res. 2014, 1551, 45–58. [Google Scholar] [CrossRef] [PubMed]
  19. Chang, D.; Wang, Y.-C.; Bai, Y.-Y.; Lu, C.-Q.; Xu, T.-T.; Zhu, L.; Ju, S. Role of P38 MAPK on MMP activity in photothrombotic stroke mice as measured using an ultrafast MMP activatable probe. Sci. Rep. 2015, 5. [Google Scholar] [CrossRef] [PubMed]
  20. Guillemin, G.J.; Brew, B.J. Microglia, macrophages, perivascular macrophages, and pericytes: A review of function and identification. J. Leukoc. Biol. 2004, 75, 388–397. [Google Scholar] [CrossRef] [PubMed]
  21. Waje-Andreassen, U.; Krakenes, J.; Ulvestad, E.; Thomassen, L.; K.-M, M.; Aarseth, J.; Vedeler, C.A. IL-6: An early marker for outcome in acute ischemic stroke. Acta Neurol. Scand. 2005, 111, 360–365. [Google Scholar] [CrossRef] [PubMed]
  22. Graeber, M.B.; Streit, W.J. Microglia: Biology and pathology. Acta Neuropathol. 2009, 119, 89–105. [Google Scholar] [CrossRef] [PubMed]
  23. Arac, A.; Grimbaldeston, M.A.; Nepomuceno, A.R.B.; Olayiwola, O.; Pereira, M.P.; Nishiyama, Y.; Tsykin, A.; Goodall, G.J.; Schlecht, U.; Vogel, H.; et al. Evidence that meningeal mast cells can worsen stroke pathology in mice. Am. J. Pathol. 2014, 184, 2493–2504. [Google Scholar] [CrossRef] [PubMed]
  24. Cheng, X.; Lian, Y.-J.; Ma, Y.-Q.; Xie, N.-C.; Wu, C.-J. Elevated serum levels of CXC chemokine ligand-12 are associated with unfavorable functional outcome and mortality at 6-month follow-up in Chinese patients with acute ischemic stroke. Mol. Neurobiol. 2016, in press. [Google Scholar] [CrossRef] [PubMed]
  25. Sousa, L.F.d.C.; Coelho, F.M.; Rodrigues, D.H.; Campos, A.C.; Barcelos, L.d.S.; Teixeira, M.M.; Rachid, M.A.; Teixeira, A.L. Blockade of CXCR1/2 chemokine receptors protects against brain damage in ischemic stroke in mice. Clinics 2013, 68, 391–394. [Google Scholar] [CrossRef]
  26. Connell, B.J.; Gordon, J.R.; Saleh, T.M. ELR-CXC chemokine antagonism is neuroprotective in a rat model of ischemic stroke. Neurosci. Lett. 2015, 606, 117–122. [Google Scholar] [CrossRef] [PubMed]
  27. Gardner, J.; Ghorpade, A. Tissue inhibitor of metalloproteinase (TIMP)-1: The TIMPed balance of matrix metalloproteinases in the central nervous system. J. Neurosci. Res. 2003, 74, 801–806. [Google Scholar] [CrossRef] [PubMed]
  28. Wang, X.; Barone, F.C.; White, R.F.; Feuerstein, G.Z. Subtractive cloning identifies tissue inhibitor of matrix metalloproteinase-1 (TIMP-1) increased gene expression following focal stroke. Stroke 1998, 29, 516–520. [Google Scholar] [CrossRef] [PubMed]
  29. Vukasovic, I.; Tesija-Kuna, A.; Topic, E.; Supanc, V.; Demarin, V.; Petrovcic, M. Matrix metalloproteinases and their inhibitors in different acute stroke subtypes. Clin. Chem. Lab. Med. 2006, 44, 428–434. [Google Scholar] [CrossRef] [PubMed]
  30. Kurzepa, J.; Szczepanska-Szerej, A.; Stryjecka-Zimmer, M.; Malecka-Massalska, T.; Stelmasiak, Z. Simvastatin could prevent increase of the serum MMP-9/TIMP-1 ratio in acute ischaemic stroke. Folia Biol. 2006, 52, 181–183. [Google Scholar]
  31. Thang, N.M.; Kumasawa, K.; Tsutsui, T.; Nakamura, H.; Masaki, H.; Ono, T.; Kimura, T. Overexpression of endogenous TIMP-2 increases the proliferation of BeWo choriocarcinoma cells through the MAPK-signaling pathway. Reprod. Sci. 2013, 20, 1184–1192. [Google Scholar] [CrossRef] [PubMed]
  32. Polimeni, M.; Valente, E.; Ulliers, D.; Opdenakker, G.; van den Steen, P.E.; Giribaldi, G.; Prato, M. Natural haemozoin induces expression and release of human monocyte tissue inhibitor of metalloproteinase-1. PLoS ONE 2013, 8, e71468. [Google Scholar] [CrossRef] [PubMed]
  33. Kjeldsen, L.; Johnsen, A.H.; Sengeløv, H.; Borregaard, N. Isolation and primary structure of NGAL, a novel protein associated with human neutrophil gelatinase. J. Biol. Chem. 1993, 268, 10425–10432. [Google Scholar] [PubMed]
  34. Yang, J.; Goetz, D.; Li, J.Y.; Wang, W.; Mori, K.; Setlik, D.; Du, T.; Erdjument-Bromage, H.; Tempst, P.; Strong, R.; et al. An iron delivery pathway mediated by a lipocalin. Mol. Cell 2002, 10, 1045–1056. [Google Scholar] [CrossRef]
  35. Hasegawa, M.; Ishii, J.; Kitagawa, F.; Takahashi, K.; Hayashi, H.; Koide, S.; Tomita, M.; Takahashi, H.; Ozaki, Y.; Yuzawa, Y. Urinary neutrophil gelatinase-associated lipocalin as a predictor of cardiovascular events in patients with chronic kidney disease. Heart Vessels 2015, 30, 81–88. [Google Scholar] [CrossRef] [PubMed]
  36. Wang, G.; Weng, Y.-C.; Han, X.; Whaley, J.D.; McCrae, K.R.; Chou, W.-H. Lipocalin-2 released in response to cerebral ischaemia mediates reperfusion injury in mice. J. Cell. Mol. Med. 2015, 19, 1637–1645. [Google Scholar] [CrossRef] [PubMed]
  37. Jin, M.; Kim, J.-H.; Jang, E.; Lee, Y.M.; Soo Han, H.; Woo, D.K.; Park, D.H.; Kook, H.; Suk, K. Lipocalin-2 deficiency attenuates neuroinflammation and brain injury after transient middle cerebral artery occlusion in mice. J. Cereb. Blood Flow Metab. 2014, 34, 1306–1314. [Google Scholar] [CrossRef] [PubMed]
  38. Haskill, S.; Peace, A.; Morris, J.; Sporn, S.A.; Anisowicz, A.; Lee, S.W.; Smith, T.; Martin, G.; Ralph, P.; Sager, R. Identification of three related human GRO genes encoding cytokine functions. Proc. Natl. Acad. Sci. USA 1990, 87, 7732–7736. [Google Scholar] [CrossRef] [PubMed]
  39. Iida, N.; Grotendorst, G.R. Cloning and sequencing of a new gro transcript from activated human monocytes: Expression in leukocytes and wound tissue. Mol. Cell. Biol. 1990, 10, 5596–5599. [Google Scholar] [CrossRef] [PubMed]
  40. Schumacher, C.; Clark-Lewis, I.; Baggiolini, M.; Moser, B. High- and low-affinity binding of GROα and neutrophil-activating peptide 2 to interleukin 8 receptors on human neutrophils. Proc. Natl. Acad. Sci. USA 1992, 89, 10542–10546. [Google Scholar] [CrossRef] [PubMed]
  41. Tsai, H.-H.; Frost, E.; To, V.; Robinson, S.; Ffrench-Constant, C.; Geertman, R.; Ransohoff, R.M.; Miller, R.H. The chemokine receptor CXCR2 controls positioning of oligodendrocyte precursors in developing spinal cord by arresting their migration. Cell 2002, 110, 373–383. [Google Scholar] [CrossRef]
  42. Losy, J.; Zaremba, J.; Skrobański, P. CXCL1 (GRO-α) chemokine in acute ischaemic stroke patients. Folia Neuropathol. 2005, 43, 97–102. [Google Scholar] [PubMed]
  43. Yang, Z.; Chen, Y.; Fu, Y.; Yang, Y.; Zhang, Y.; Chen, Y.; Li, D. Meta-analysis of differentially expressed genes in osteosarcoma based on gene expression data. BMC Med. Genet. 2014, 15. [Google Scholar] [CrossRef] [PubMed]
  44. Gentleman, R.; Carey, V.; Huber, W.; Irizarry, R.; Dudoit, S. Bioinformatics and Computational Biology Solutions Using R and Bioconductor; Springer Science & Business Media: New York, NY, USA, 2006; p. 494. [Google Scholar]
  45. Gautier, L.; Cope, L.; Bolstad, B.M.; Irizarry, R.A. affy—Analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 2004, 20, 307–315. [Google Scholar] [CrossRef] [PubMed]
  46. Ritchie, M.E.; Phipson, B.; Wu, D.; Hu, Y.; Law, C.W.; Shi, W.; Smyth, G.K. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015, 43. [Google Scholar] [CrossRef] [PubMed]
  47. Troyanskaya, O.; Cantor, M.; Sherlock, G.; Brown, P.; Hastie, T.; Tibshirani, R.; Botstein, D.; Altman, R.B. Missing value estimation methods for DNA microarrays. Bioinformatics 2001, 17, 520–525. [Google Scholar] [CrossRef] [PubMed]
  48. Irizarry, R.A. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003, 4, 249–264. [Google Scholar] [CrossRef] [PubMed]
  49. Marot, G.; Foulley, J.-L.; Mayer, C.-D.; Jaffrézic, F. Moderated effect size and P-value combinations for microarray meta-analyses. Bioinformatics 2009, 25, 2692–2699. [Google Scholar] [CrossRef] [PubMed]
  50. Choi, J.K.; Yu, U.; Kim, S.; Yoo, O.J. Combining multiple microarray studies and modeling interstudy variation. Bioinformatics 2003, 19, i84–i90. [Google Scholar] [CrossRef] [PubMed]
  51. Huang, D.W.; Sherman, B.T.; Lempicki, R.A. Bioinformatics enrichment tools: Paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009, 37, 1–13. [Google Scholar] [CrossRef] [PubMed]
  52. Ashburner, M.; Michael, A.; Ball, C.A.; Blake, J.A.; David, B.; Heather, B.; Michael Cherry, J.; Davis, A.P.; Kara, D.; Dwight, S.S.; et al. Gene Ontology: Tool for the unification of biology. Nat. Genet. 2000, 25, 25–29. [Google Scholar] [CrossRef] [PubMed]
  53. Kanehisa, M. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res. 2000, 28, 27–30. [Google Scholar] [CrossRef] [PubMed]
Figure 1. Search in the public microarray data repositories identified 128 microarray datasets. We finally selected eight studies for studies according to the included and excluded criteria.
Figure 1. Search in the public microarray data repositories identified 128 microarray datasets. We finally selected eight studies for studies according to the included and excluded criteria.
Ijms 17 00776 g001
Figure 2. Protein–protein interaction (PPI) network. The confidence scores are marked in the line of interaction. (White: Mus musculus; Black: Rattus norvegicus).
Figure 2. Protein–protein interaction (PPI) network. The confidence scores are marked in the line of interaction. (White: Mus musculus; Black: Rattus norvegicus).
Ijms 17 00776 g002
Figure 3. Forest plots of Timp1 and Lcn2. Forest plots of (A) Timp1 in Mus musculus; (B) Lcn2 in Mus musculus; (C) Timp1 in Rattus norvegicus; and (D) Lcn2 in Rattus norvegicus. For each forest plot, the bar at the top shows the occurrence of combining p-values and combining effect size methods in selected genes. The dark gray box means that the result of the particular method in the selected gene is present, while light gray means that it is absent. The graph at the bottom shows the combination of effect size. The point marks the effect size, the horizontal lines denote the variance of effect size. Adjusted FDR (false discovery rate) on the right side of the graph presents the statistical significance of the difference in gene expression.
Figure 3. Forest plots of Timp1 and Lcn2. Forest plots of (A) Timp1 in Mus musculus; (B) Lcn2 in Mus musculus; (C) Timp1 in Rattus norvegicus; and (D) Lcn2 in Rattus norvegicus. For each forest plot, the bar at the top shows the occurrence of combining p-values and combining effect size methods in selected genes. The dark gray box means that the result of the particular method in the selected gene is present, while light gray means that it is absent. The graph at the bottom shows the combination of effect size. The point marks the effect size, the horizontal lines denote the variance of effect size. Adjusted FDR (false discovery rate) on the right side of the graph presents the statistical significance of the difference in gene expression.
Ijms 17 00776 g003aIjms 17 00776 g003b
Figure 4. A summary of analysis workflow. RLE, relative log expression; NUSE, normalized unscaled standard errors; DEGs, differentially expressed genes. Solid line with arrow refers to directed workflow. Dash line refers to enriched gene sets of several studies.
Figure 4. A summary of analysis workflow. RLE, relative log expression; NUSE, normalized unscaled standard errors; DEGs, differentially expressed genes. Solid line with arrow refers to directed workflow. Dash line refers to enriched gene sets of several studies.
Ijms 17 00776 g004
Table 1. Characteristics of included studies analyzed in the meta-analysis.
Table 1. Characteristics of included studies analyzed in the meta-analysis.
Series DatasetIschemic Period (min)Reperfusion Period (h)No. of SamplesPlatform
MCAOSham
Mus musculus
E-MEXP-2547 [8]3024, 24055Affymetrix GeneChip Mouse Gene 1.0 ST Array
GSE23160 [9]1202, 8, 24248Illumina MouseRef-8 v2.0 expression beadchip
GSE30655 [10]602473Affymetrix Mouse Genome 430 2.0 Array
GSE58720 [11]902433Agilent-028005 SurePrint G3 Mouse GE 8 × 60 K Microarray
Rattus norvegicus
E-MEXP-2222 [12]906, 24123Affymetrix Rat Genome 230 2.0 Array
GSE33725 [13]NA2, 666Agilent-014879 Whole Rat Genome Microarray 4 × 44 K G4131F
GSE52001 [14]12014433Agilent-028282 Whole Rat Genome Microarray 4 × 44 K v3
GSE61616 [15]12016855Affymetrix Rat Genome 230 2.0 Array
GEO, gene expression omnibus; GSE, gene expression series; MCAO, middle cerebral artery occlusion; No, number; NA, not available.
Table 2. Gene ontology (GO) functional annotation for differentially expression genes in Mus musculus (Top five).
Table 2. Gene ontology (GO) functional annotation for differentially expression genes in Mus musculus (Top five).
GO IDGO Term%p-Value
Biological Process
GO:0002237response to molecule of bacterial origin1.672.10 × 10−3
GO:0032655regulation of interleukin-12 production1.112.46 × 10−3
GO:0043086negative regulation of catalytic activity2.222.78 × 10−3
GO:0010033response to organic substance5.562.92 × 10−3
GO:0044092negative regulation of molecular2.503.28 × 10−3
Cellular Component
GO:0044445cytosolic part1.675.75 × 10−3
GO:0031974membrane-enclosed lumen8.892.27 × 10−2
GO:0005829cytosol5.002.33 × 10−2
GO:0005886plasma membrane18.612.59 × 10−2
GO:0005938cell cortex1.942.78 × 10−2
Molecular Function
GO:0005509calcium ion binding7.501.03 × 10−2
GO:0019838growth factor binding1.671.21 × 10−2
GO:0043167ion binding25.282.49 × 10−2
GO:0046872metal ion binding24.722.75 × 10−2
GO:0003677DNA binding12.782.92 × 10−2
Table 3. List of the top five pathways based on Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis.
Table 3. List of the top five pathways based on Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis.
IDTerm%p-Value
Mus musculus
mmu04010MAPK signaling pathway4.728.14 × 10−5
mmu04060Cytokine-cytokine receptor interaction3.062.65 × 10−2
mmu04062Chemokine signaling pathway2.503.18 × 10−2
mmu04722Neurotrophin signaling pathway1.944.90 × 10−2
mmu05410Hypertrophic cardiomyopathy (HCM)1.399.16 × 10−2
Rattus norvegicus
rno04730Long-term depression2.771.44 × 10−6
rno04010MAPK signaling pathway5.141.52 × 10−6
rno04070Phosphatidylinositol signaling system2.371.82 × 10−6
rno04540Gap junction2.372.29 × 10−6
rno04670Leukocyte transendothelial migration2.773.29 × 10−6
Table 4. GO functional annotation for differentially expressed genes in Rattus norvegicus. (Top five).
Table 4. GO functional annotation for differentially expressed genes in Rattus norvegicus. (Top five).
GO IDGO Term%p-Value
Biological Process
GO:0009611response to wounding9.491.91 × 10−7
GO:0022604regulation of cell morphogenesis4.745.44 × 10−6
GO:0006954inflammatory response5.533.57 × 10−5
GO:0006813potassium ion transport4.357.65 × 10−5
GO:0006811ion transport9.881.16 × 10−4
Cellular Component
GO:0044456synapse part6.721.44 × 10−6
GO:0005886plasma membrane24.901.52 × 10−6
GO:0043005neuron projection8.701.82 × 10−6
GO:0005856cytoskeleton13.442.29 × 10−6
GO:0045202synapse7.913.29 × 10−6
Molecular Function
GO:0005509calcium ion binding9.495.33 × 10−5
GO:0005216ion channel activity6.326.73 × 10−5
GO:0043167ion binding25.308.37 × 10−5
GO:0022838substrate specific channel activity6.329.35 × 10−5
GO:0022836gated channel activity5.539.48 × 10−5
Table 5. Summary of differential expression analysis in each study.
Table 5. Summary of differential expression analysis in each study.
StudyCountUp-RegulatedDown-Regulated
Mus musculus
E-MEXP-25471791790
GSE2316022220
GSE30655341128213
GSE587201162841321
Rattus norvegicus
E-MEXP-222283812
GSE3372538380
GSE52001946628
GSE61616827695132
Count = the number of differentially expressed genes.

Share and Cite

MDPI and ACS Style

Zhou, H.; Qiu, Z.; Gao, S.; Chen, Q.; Li, S.; Tan, W.; Liu, X.; Wang, Z. Integrated Analysis of Expression Profile Based on Differentially Expressed Genes in Middle Cerebral Artery Occlusion Animal Models. Int. J. Mol. Sci. 2016, 17, 776. https://0-doi-org.brum.beds.ac.uk/10.3390/ijms17050776

AMA Style

Zhou H, Qiu Z, Gao S, Chen Q, Li S, Tan W, Liu X, Wang Z. Integrated Analysis of Expression Profile Based on Differentially Expressed Genes in Middle Cerebral Artery Occlusion Animal Models. International Journal of Molecular Sciences. 2016; 17(5):776. https://0-doi-org.brum.beds.ac.uk/10.3390/ijms17050776

Chicago/Turabian Style

Zhou, Huaqiang, Zeting Qiu, Shaowei Gao, Qinchang Chen, Si Li, Wulin Tan, Xiaochen Liu, and Zhongxing Wang. 2016. "Integrated Analysis of Expression Profile Based on Differentially Expressed Genes in Middle Cerebral Artery Occlusion Animal Models" International Journal of Molecular Sciences 17, no. 5: 776. https://0-doi-org.brum.beds.ac.uk/10.3390/ijms17050776

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop