Cell cycle and complement inhibitors may be specific for treatment of spinal cord injury in aged and young mice: transcriptomic analyses

Abstract

Previous studies have reported age-specific pathological and functional outcomes in young and aged patients suffering spinal cord injury, but the mechanisms remain poorly understood. In this study, we examined mice with spinal cord injury. Gene expression profiles from the Gene Expression Omnibus database (accession number GSE93561) were used, including spinal cord samples from 3 young injured mice (2–3-months old, induced by Impactor at Th9 level) and 3 control mice (2–3-months old, no treatment), as well as 2 aged injured mice (15–18-months old, induced by Impactor at Th9 level) and 2 control mice (15–18-months old, no treatment). Differentially expressed genes (DEGs) in spinal cord tissue from injured and control mice were identified using the Linear Models for Microarray data method, with a threshold of adjusted P < 0.05 and |logFC(fold change)| > 1.5. Protein–protein interaction networks were constructed using data from the STRING database, followed by module analysis by Cytoscape software to screen crucial genes. Kyoto encyclopedia of genes and genomes pathway and Gene Ontology enrichment analyses were performed to investigate the underlying functions of DEGs using Database for Annotation, Visualization and Integrated Discovery. Consequently, 1,604 and 1,153 DEGs were identified between injured and normal control mice in spinal cord tissue of aged and young mice, respectively. Furthermore, a Venn diagram showed that 960 DEGs were shared among aged and young mice, while 644 and 193 DEGs were specific to aged and young mice, respectively. Functional enrichment indicates that shared DEGs are involved in osteoclast differentiation, extracellular matrix–receptor interaction, nuclear factor-kappa B signaling pathway, and focal adhesion. Unique genes for aged and young injured groups were involved in the cell cycle (upregulation of PLK1) and complement (upregulation of C3) activation, respectively. These findings were confirmed by functional analysis of genes in modules (common, 4; aged, 2; young, 1) screened from protein–protein interaction networks. Accordingly, cell cycle and complement inhibitors may be specific treatments for spinal cord injury in aged and young mice, respectively.

Introduction

Spinal cord injury (SCI) is a common traumatic event in orthopedic clinics due to rapid industrial and economic development in China, with an estimated incidence of 23.7 per million cases in Tianjin, 25 in Shanghai, and 60 in Beijing (Hua et al., 2013). SCI results in severe or permanent motor, sensory and autonomic dysfunction, which affects a patient’s quality of life and imposes a huge economic burden on family and society (Krueger et al., 2013; Ravensbergen et al., 2016; Zhang et al., 2016b; Rabchevsky et al., 2017). More importantly, recent studies have suggested that the pathological and behavioral outcomes after SCI may be age-dependent, with elderly patients exhibiting markedly less remyelination compared with younger patients, which consequently leads to worsened functional recovery and a higher mortality rate (Siegenthaler et al., 2008; Wilson et al., 2014). Thus, distinguishing the cellular and molecular response mechanisms in aged and young people is necessary to develop targeted treatments.

Recently, the role of aging following SCI was investigated (Geoffroy et al., 2016). Accordingly, the number of M1 macrophages at the injury epicenter was increased by 50% in aged compared with young rats (Hooshmand et al., 2014), while M2 macrophages were reduced (Zhang et al., 2015), thereby inducing apoptotic cell death and greater locomotor deficits. Similarly, a lower number of infiltrating neutrophils and secreted pro-inflammatory cytokines/chemokines (e.g., interleukin 6; tumor necrosis factor α; and C-X-C motif chemokine ligand 1) were detected in microglia from young compared with adult mice (Kumamaru et al., 2012). Further studies suggest that inflammatory activation may be NADPH oxidase (NOX)- (Zhang et al., 2016) or adipokine-mediated (Bigford et al., 2012) in chronic SCI and advanced age, with high expression of NOX2 and the leptin signaling inhibitor, suppressor of cytokine signaling 3 (SOCS3), as well as lower long-form leptin receptor (LepRb) and Janus kinase 2/signal transducer and activator of transcription 3 Jak2/Stat3 signaling. High throughput analysis of gene expression profiles between aged and young rats following SCI has also been performed. Cortical transcriptome analysis of the left hemisphere suggests that genes enriched in biological processes such as apoptosis (1 day post-operation), activation of immune responses (7 days post-operation), and cell cycle and cell adhesion (35 days post-operation) may be specific to aged animals (Jaerve et al., 2012). However, specific treatments for aged and young SCI patients are not fully understood.

In this study, we aimed to further investigate gene expression differences in the injured spinal cord between aged and young mice using microarray data downloaded from the Gene Expression Omnibus (GEO) database (Takano et al., 2017). A total of 364 differential genes between aged and young mice were identified in the study by Takano et al. (2017), among which 169 down-regulated genes were involved in regulation of synapse-, ion transport-, or axon-related functions, while 195 up-regulated genes were involved in the cell cycle, cell stress responses, or maintenance of extracellular matrix (Takano et al., 2017). Shared or unique differentially expressed genes (DEGs) for aged and young mice were not identified. Thus, our study focused on screening crucial genes and pathways for aged and young mice, and as a result, is able to suggest targeted treatments.

Materials and Methods

Animals

Ten female C57BL/6J mice (young, 2–3 months old, n = 6; aged, 15–18-months old, n = 4) were housed in groups under 12-hour light/dark cycles with free access to food and water. All protocols were approved by the Institutional Animal Care and Use Committee of Keio University School of Medicine, Japan, and performed in accordance with the Guidelines for the Care and Use of Laboratory Animals of Keio University School of Medicine, Japan.

Young and aged mice were randomly assigned to undergo SCI or control treatment: young injured mice (n = 3), young normal mice (n = 3), aged injured mice (n = 3), and aged normal mice (n = 3). SCI model was induced using a commercially available SCI device (Infinite Horizon Impactor, 70-kdyn; Precision Systems & Instrumentation, Fairfax Station, VA, USA) at the thoracic level, Th9. Spinal cord samples were collected nine days after injury (Takano et al., 2017). Injured mice exhibiting low Basso Mouse Scale scores indicate successful model establishment (Takano et al., 2017). Normal mice underwent no treatment.

Microarray data

SCI microarray data were extracted from the GEO database (http://www.ncbi.nlm.nih.gov/geo/) using the accession number, GSE93561 (Takano et al., 2017). This contains spinal cord samples from three young injured mice (GSM2454721_AG1408, GSM2454722_AG1409, and GSM2454723_AG1410), three young normal mice (GSM2454718_AG1405, GSM2454719_AG1406, and GSM2454720_AG1407), three aged injured mice (GSM2454727_AG1414, GSM2454728_AG1415, and GSM2454729_AG1416), and three aged normal mice (GSM2454724_AG1411, GSM2454725_AG1412, and GSM2454726_AG1413). Because of its expression, sample GSM2454729_AG1416 was deemed “not available”. Consequently, this aged injured sample and the corresponding aged normal sample (GSM2454726_AG1413) were removed from the study. As a result, the study ultimately included: young injured mice (n = 3), young normal mice (n = 3), aged injured mice (n = 2), and aged normal mice (n = 2).

Data normalization and DEG identification

Raw CEL files were preprocessed and normalized using the Robust Multichip Average algorithm (Irizarry et al., 2003) as implemented in the Bioconductor R package (http://www.bioconductor.org/packages/release/bioc/html/affy.html). DEGs between injured and control samples were screened using the Linear Models for Microarray data method (Ritchie et al., 2015), also in the Bioconductor R package (http://www.bioconductor.org/packages/release/bioc/html/limma.html). After performing t-tests, P-values were adjusted by the Benjamini-Hochberg algorithm (Thissen, 2002). Adjusted P < 0.05 and |logFC(fold change)| > 1.5 were set as threshold values. A Venn diagram was constructed to show unique or shared genes in aged and young injured mice using an online tool (http://bioinformatics.psb.ugent.be/webtools/Venn/).

Protein–protein interaction (PPI) network construction

To screen crucial genes associated with SCI (aged or young), DEGs were mapped onto PPI data collected from the Search Tool for the Retrieval of Interacting Genes (STRING) 10.0 database (http://string db.org/) (Szklarczyk et al., 2015). Combined scores > 800 were set as cut-off values for identifying significant protein pairs for constructing PPI networks. These were then visualized using Cytoscape software 2.8 (www.cytoscape.org/) (Kohl, 2011). To identify functionally related and highly interconnected clusters from PPI networks, module analysis was performed using the Molecular Complex Detection plugin of Cytoscape software, with a degree cutoff of 5, node score cutoff of 0.5, k-core of 5, and maximum depth of 100 (ftp://ftp.mshri.on.ca/pub/BIND/Tools/MCODE) (Bader and Hogue, 2003). Significant modules were identified with Molecular Complex Detection scores ≥ 4 and nodes ≥ 6.

Functional enrichment analysis

Kyoto encyclopedia of genes and genomes (KEGG) pathway and Gene Ontology (GO) enrichment analyses were performed to investigate the potential function of all DEGs (shared or unique DEGs), or genes in modules using The Database for Annotation, Visualization and Integrated Discovery (DAVID) 6.8 online tool (http://david.abcc.ncifcrf.gov). False discovery rate < 0.05 was chosen as the cut-off point for GO and KEGG analyses.

Results

Identification of DEGs in aged and young SCI mice

Based on a threshold of adjusted P < 0.05 and |logFC| > 1.5, a relatively higher number of DEGs were identified after SCI in aged mice (1,604: 952 up-regulated and 652 down-regulated) compared with young mice (1,153: 721 up-regulated and 432 down-regulated). These genes clearly differentiated the samples (Figure 1). Further, Venn diagram showed 960 shared DEGs between young and aged injured groups (640 up-regulated and 320 down-regulated), suggesting these genes are important for development of SCI. Additionally, 644 (312 up-regulated and 332 down-regulated) and 193 (81 up-regulated and 112 down-regulated) DEGs were unique for the aged and young injured groups, respectively, suggesting these genes are age-dependent (Figure 2).

Figure 1.

Heat map of differentially expressed genes between young/aged spinal cord injury and normal control mice.

Expression was Z-score normalized within samples. High expression levels are indicated in red and low levels in green. The horizontal coordinate represents the sample list, while the vertical coordinate represents the Z-score value. (A) AG1405, AG1406, and AG1407 are young normal control mice, and AG1408, AG1409, and AG1410 are young injured mice. (B) AG1411 and AG1412 are aged normal control mice, and AG1414 and AG1415 are aged injured mice. The results show that these differentially expressed genes can clearly differentiate the samples.

Figure 2.

Venn diagram of differentially expressed genes between young/aged spinal cord injury and normal control mice.

(A) Overall, (B) up-regulated, and (C) down-regulated genes. Overall, there are 960 shared differentially expressed genes between young and aged injured groups (640 up-regulated and 320 down-regulated). Additionally, 644 (312 up-regulated and 332 down-regulated) and 193 (81 up-regulated and 112 down-regulated) differentially expressed genes were unique for aged and young injured mice, respectively. NC: Normal control.

Functional enrichment analysis of shared and unique DEGs

Shared and unique DEGs were subjected to functional enrichment analysis using the online tool DAVID, with the mouse genome as background and false discovery rate < 0.05 as the cut-off point. For shared up-regulated DEGs, 33 KEGG pathways were enriched including osteoclast differentiation, phagosome, extracellular matrix (ECM)–receptor interaction, nuclear factor-kappa B (NF-κB) signaling pathway, cytokine-cytokine receptor interaction, and focal adhesion. Further, nine pathways were identified for shared down-regulated DEGs, including synaptic vesicle cycle and glutamatergic synapse (Table 1). Furthermore, three pathways showed enrichment in unique genes of the aged injured group: up-regulated (cell cycle and lysosome) and down-regulated (cholinergic synapse). While four pathways were enriched in unique genes of the young injured group: up-regulated (pertussis, and complement and coagulation cascades) and down-regulated (nicotine addiction and retrograde endocannabinoid signaling).

Table 1.

KEGG pathway enrichment for differentially expressed genes in spinal cord tissue of aged and young injured mice

PPI network construction and module analysis for shared and unique DEGs

PPI networks were constructed after mapping shared or unique DEGs onto PPI data. For shared DEGs, four significant modules were screened from the PPI network (Figure 3 and Table 2). Module 1 was involved in neuroactive ligand–receptor interaction, module 2 in ECM-receptor interaction, focal adhesion, and phosphoinositide 3-kinase (PI3K)-Akt signaling pathway-related, and module 4 in osteoclast differentiation and NF-κB signaling pathway-associated (Table 3). In the aged injured group, two significant modules were screened from the PPI network for unique DEGs (Figure 4), with module 1 involved in the cell cycle and module 2 in the chemokine signaling pathway. No pathways or significant pathways were enriched in module 3 of shared and unique DEGs from the young injured group (Figure 5). Moreover, GO analysis indicated that unique DEGs in the young injured group exert effects on SCI via inflammatory processes (Table 2).

Figure 3.

Modules obtained from protein–protein interaction networks of shared differentially expressed genes between young/aged spinal cord injury and normal control mice.

(A–D) Modules 1–4. Red: Up-regulated genes; and green: down-regulated genes. Abbreviations are shown in Additional file 1 ^{(90.1KB, pdf)} .

Table 2.

Significant functional modules from protein–protein interaction networks constructed by shared or unique differentially expressed genes in spinal cord tissue of aged and young injured mice

Table 3.

KEGG pathway enrichment for functional modules screened from protein–protein interaction networks constructed by shared or unique differentially expressed genes in spinal cord tissue of aged and young injured mice

Figure 4.

Modules obtained from protein–protein interaction networks of unique differentially expressed genes for aged spinal cord injury mice.

(A) Module 1 and (B) module 2. Red: Up-regulated genes; and green: down-regulated genes. Abbreviations are shown in Additional file 1 ^{(90.1KB, pdf)} .

Figure 5.

Module obtained from protein–protein interaction network of unique differentially expressed genes for young spinal cord injury mice.

Red: Up-regulated genes; S1pr3: sphingosine-1-phosphate receptor 3; C3: complement C3; Cxcl10: C-X-C motif chemokine ligand 10; Cxcl1: C-X-C motif chemokine ligand 1; Cxcl2: C-X-C motif chemokine ligand 2; Cxcl5: C-X-C motif chemokine ligand 5; Hcar2: hydroxycarboxylic acid receptor 2; P2ry13: purinergic receptor P2Y, G-protein coupled 13.

Additional files

Abbreviations of Table 1, 2, 3 and Figure 3, 4.

Discussion

By integrating functional analyses of all DEGs and module genes, our present study preliminarily demonstrates that cell cycle (including polo like kinase 1 [PLK1], cell division cycle 6 [CDC6]; cell division cycle 20 [CDC20], and BUB1 mitotic checkpoint serine/threonine kinase [BUB1]) and complement-related genes (including complement C3 [C3]) may be specifically altered in spinal cord of aged and young injured mice, respectively. All DEGs were up-regulated, consequently use of cell cycle and complement pathway inhibitors may be potential treatment measures for aged and young SCI patients. Indeed, our hypothesis has been indirectly demonstrated by previous studies.

Increasing evidence, including gene expression profiles in spinal cord (Di, 2003), indicate that cell cycle activation plays an important role in the pathophysiology of SCI (Wu et al., 2011). First, cell cycle activation contributes to neuronal and oligodendroglial apoptosis after SCI (postmitotic cells) (Byrnes et al., 2007). Further, it also promotes microglial proliferation (mitotic cells), which produce pro-inflammatory cytokines and cause functional deficits (Tian et al., 2007a, b). Cell cycle-related proteins, such as cyclin D1, cyclin dependent kinase 4 (CDK4), and proliferating cell nuclear antigen are all significantly up-regulated following SCI (Wu et al., 2012, 2014). Moreover, systemic administration of CDK inhibitors, such as olomoucine, flavopiridol, or CR8, suppresses these processes and improves neurodegeneration and neuropathic pain (Ren et al., 2014; Wu et al., 2016). However, whether cell cycle activation is specific for aged SCI (Jaerve et al., 2012), and whether there are treatment differences in CDK inhibitors for aged and young mice is unclear and needs further confirmation. Human PLK1 is an evolutionarily conserved serine/threonine kinase that regulates cell division at the M phase. PLK1 can phosphatase CDC6 (Yim and Erikson, 2010), Cdc25C (Toyoshimamorimoto et al., 2002), CDC20 (Jia et al., 2016), CCD14B (Bassermann et al., 2008), BUB1 (Qi et al., 2006), BubR1 (BUB1-related) (Elowe et al., 2007), and CDK5 regulatory subunit associated protein 2 (CDK5RAP2) (Hanafusa et al., 2015) to promote spindle checkpoint signaling. Use of PLK1 inhibitors, such as RO3280 (Wang et al., 2015), GSK461364 (Chou et al., 2016; Pajtler et al., 2017), and BI2536 (Frost et al., 2012; Kumar et al., 2015), induces cell cycle arrest and growth inhibition, enabling treatment of various diseases. In our PPI network, we found that PLK1 interacts with 44 DEGs, including CDC20, CDC6, and BUB1. These findings suggest a possible crucial role of PLK1 in SCI and an underlying therapeutic effect for PLK1 inhibitors. Unfortunately, there are no experimental studies that confirm our conclusion, but this may be a new direction for our future studies.

Classical (C1q and C4), alternative (Factor B), and terminal (C5b-9) complement pathways in neurons and oligodendrocytes are suggested to initiate an inflammatory cascade and induce secondary injury and functional deficits following traumatic SCI (Anderson et al., 2004). Mice with a deficiency in the complement component C1q (Galvan et al., 2008), C3 (Qiao et al., 2006), complement receptor 2, and complement receptor C5aR (Li et al., 2014; Brennan et al., 2015), or treated with complement antagonists (Qiao et al., 2006; Li et al., 2009; Brennan et al., 2016; Biggins et al., 2017) exhibit improved functional outcomes. As expected, complement activation was detected in SCI mice in our study. More importantly, our study reveals that this pathway may be specific to young injured mice, although this is not consistent with a previous study (Jaerve et al., 2012). We believe this may be due to the following reasons: (1) Jaerve et al., (2012) investigated cortical samples from the left hemisphere, which was the resulting site induced by SCI. Thus, a delayed effect may be present; (2) our sample size is small; and (3) C1qb, C3, and C4 are early complement proteins after SCI, while C5, C6, C7, and C9 are terminal complement proteins after SCI (Nguyen et al., 2008). Therefore, we speculate that different gene expression profiles may explain the different phenomena. As anticipated, C3 (as a complement activation pathway gene) was significantly up-regulated in young SCI mice in our study. Accordingly, the C3 inhibitor, CR2-Crry (Qiao et al., 2006), may be an effective treatment for young SCI patients. Nevertheless, further confirmation is still needed.

In addition to unique DEGs, we also found several shared between aged and young injured mice. These DEGs are involved in the NF-κB signaling pathway, indicating that these genes and pathway may be important for SCI, regardless of age. These findings are in accordance with previous studies. For example, NF-κB and related inflammatory cytokines were up-regulated in the injured rat spinal cord (Ni et al., 2015; Yarar-Fisher et al., 2016). Treatment with hyperbaric oxygen (Yang et al., 2013; Kang et al., 2015), curcumin (Ni et al., 2015), and butein (Ming et al., 2013) ameliorated SCI-induced hindlimb locomotion deficits, spinal cord edema, and apoptosis by down-regulating the toll-like receptor 4 (TLR4)/NF-κB inflammatory signaling pathway.

In conclusion, our present study reveals preliminarily findings showing differences in specific genes in aged and young injured mice. Cell cycle- (PLK1) and complement (C3)-related gene inhibitors may be more effective for treatment of SCI in aged and young mice, respectively. However, further in vivo experimental studies are needed to confirm our findings due to small sample size, which is a limitation of our study.

Additional file:

Additional file 1 ^{(90.1KB, pdf)} : Abbreviations in Tables 1–3 and Figures 3, 4.