Long non‐coding RNAs in the spinal cord injury: Novel spotlight

Abstract Spinal cord injury (SCI) may lead to persistent locomotor dysfunction and somatosensory disorders, which adversely affect the quality of life of patients and cause a significant economic burden to the society. The efficacies of current therapeutic interventions are still far from satisfaction as the secondary damages resulting from the complex and progressive molecular alterations after SCI are not properly addressed. Recent studies revealed that long non‐coding RNAs (lncRNAs) are abundant in the brain and might play critical roles in several nervous system disorders. At the cellular level, lncRNAs have been shown to regulate the expression of protein‐coding RNAs and hence participate in neuronal death, demyelination and glia activation. Notably, SCI is characterized by these biological processes, suggesting that lncRNAs could be novel modulators in the pathogenesis of SCI. This review describes recent progresses in the lncRNA transcriptome analyses and their molecular functions in regulating SCI progression.

cancers), 1 resulting in immediate haemorrhage and rapid neuronal cell death. This is followed by secondary injury mechanisms, including glutaminergic excitotoxicity, oxidative stress, increased adaptive immune responses, Wallerian degeneration and scar tissue formation, leading to further structural and functional disturbances that spread spatially from the site of initial injury. [2][3][4] These biochemical alternations can be further divided into acute, subchronic and chronic phases, which require tailored therapeutic strategies. 5 An essential problem is that adult spinal cord has a low regenerative capacity. This results in paralysis or movement dysfunctions, sensation deficits and autonomic dysfunctions, such as loss of urinary and bowel functions. Unfortunately, current treatments are insufficient due to multiple and complex aetiologies of SCI. Further understandings of cellular and molecular mechanisms of primary and secondary injuries are necessary for finding a new therapeutic strategy to promote functional recovery of patients with SCI. In this regard, long non-coding RNAs may provide hints for novel treatment strategies for SCI.
Long non-coding RNAs (lncRNAs) were identified as non-protein-coding transcripts that consist of more than 200 nucleotides. 6 lncRNAs were initially considered as transcriptional noise that was transcribed by the RNA polymerase II complex. 7 However, recent studies demonstrated that lncRNA retained limited protein-coding capacity to encode short peptides. 8 LNCipedia (https ://lncip edia. org/) is a public and active database that aims to record and annotate lncRNA sequences and structures. Since its establishment in 2013, 9 five updated databases have been published. 10 Currently, a total of 127 802 long non-coding transcripts and 56 946 lncRNAs are curated by LNCipedia. Accumulating evidences revealed that many lncRNAs may functionally interact with proteins, adding a new dimension to the physiological and pathological roles of genes coding. lncRNAs play multiple roles in gene expression ( Figure 1). Firstly, lncRNAs may locally (in cis) or non-locally (in trans) modulate gene transcription. Polycomb Repressive Complex 2 (PRC2) is required for the initiation of histone modifications and subsequent chromatin compaction. 11 By interacting with PRC2, lncRNA transcripts, such as X-inactive specific transcript (XIST) and HOX Antisense Intergenic RNA (HOTAIR), were shown to regulate chromatin structure and silence gene transcription. [12][13][14] Meanwhile, lncRNAs may complementarily hybridize to promoter regions of gene loci, leading to either repression or activation of gene transcription. An example of this scenario is the transcription of PDCD4 whose promoter interacts with lncRNA CASC9, thereby recruiting the transcription repressor EZH2. 15 For its co-activator role, lncRNA Evf-2 promoted the recruitment of chromatin-binding protein Dlx-2 and then enhanced the transcription of Dlx-6 gene. 16  Given the enormous amounts of lncRNAs and their profound effects on coding RNAs, it is notsurprising that lncRNAs may play important roles in pathogenesis. It is worthwhile to note that up to 40% of all the known lncRNAs are specifically expressed in the brain,or possibly other parts of the central nervous system. 23 This finding F I G U R E 1 Summary of lncRNA functions at different regulatory levels. (1) lncRNA transcripts recruit chromatin modifiers as a coactivator to regulate subsequent chromatin compaction for gene silencing. (2) Transcriptionally, lncRNA leads to either gene repression or activation by hybridization to promoter regions of gene loci. (3) Post-transcriptionally, lncRNAs act as ceRNAs to regulate RNA transcripts by miRNAs sponging. (4) Short peptides encoded by lncRNAs interact with RNA binding proteins to modulate protein function. (5) Alternatively, short peptides interact with calcium channels for protein function modulation at the Post-translational level. (6) lncRNAs bind directly to receptors to regulate its ion channel activity indicated that lncRNAs, especially differentially expressed lncRNAs, might provide additional regulations in the pathogenesis of nervous system diseases. In this regard, identifications of differentially expressed lncRNAs using genome-wide approaches were a good start to understand lncRNA-mediated disease development. In this review, differentially expressed lncRNAs profiling in murine models of SCI will be summarized. Emerging evidence of the interplay between lncRNA function and SCI will also be highlighted.  (Table 1). Surprisingly, two studies, which were conducted by the same team, presented no overlap between the top 20 differentially expressed lncRNAs in immediate (2 hours) 24 and acute (2 days) 25 SCI stages. These results suggested that lncRNA expression was highly dynamic across different stages. Nonetheless, these studies have aimed to reveal the lncRNA profiles in different post-injury time-points, from immediate, 24 acute, 25,26 subchronic 27 to chronic 27 phases after SCI ( Figure 2). These studies provided preliminary views on stage-specific lncRNA modulation.

| LN CRNA E XPRE SS I ON PROFILING IN SCI
The first study focusing on lncRNAs was performed to identify differentially expressed lncRNAs in a period of 1-21 days (acute to subchronic phase) after a moderate contusion SCI 26 . Hundreds of lncRNAs were up-regulated or down-regulated at all time-points (ie 1, 3, 7 and 21 days) after injury. This suggested that lncRNAs expression was sensitive to the pathological changes across acute and subchronic phases of SCI. It is not surprising to find that some of these lncRNAs might participate in the pathological changes. Similar to coding RNAs, the number of differentially expressed lncRNAs at the epicentre of injury gradually increased on day 1 and day 3, peaked on day 7 and then recovered on day 21. Interestingly, co-expression analysis using gene quantification values of different time-points demonstrated that many lncRNAs expression levels were highly correlated with differentially expressed coding RNAs (coefficients of correlation > 0.997). The lncRNA-mRNA co-expression network was then constructed with these coefficients of correlation. This further revealed that several lncRNAs had high degrees and K-core values and belong to the 'hub' nodes of co-expression network.
In graph theory, higher degrees and K-core values indicated that a node (gene) is connecting with higher number of other nodes in the network or subnetwork (K-core). 28 .The network analysis highlighted that these hub lncRNAs correlated with a substantial amount of coding genes, in terms of expression levels. Further experiments, such as effects of overexpressing or knockdown of these hub ln-cRNAs on SCI-associated transcriptome in relation to histological and functional recovery, will be required. Neither were functional enrichments of co-expressed coding mRNAs annotated. The functional insights of differentially expressed lncRNAs in the progression of SCI were lacking.
Recently, a systematic analysis focusing on differentially expressed lncRNAs in subchronic (1-3 months) and chronic (6 months) phases of moderate (150-kdyn) contusive SCI in rats was done. 27 Compared to sham control, a total of 277 lncRNAs were identified to be differentially expressed at all the time-points of the study, ie 1, 3 and 6 months after SCI. A co-expression network was then constructed using the 277 lncRNAs and the mRNAs that were significantly correlated. Gene Set Enrichment Analysis (GSEA) revealed  nih.gov/gds). Although these previous studies were not concentrated in lncRNAs, many of them did contain expression data of lncRNAs.
Retrieving these datasets is an efficient way to study lncRNAs. On the other hand, owing to the limited knowledge of lncRNAs, it is not easy to functionally annotate lncRNAs. The study performed by Duran and his college represented a good example of systematic analysis. 27 LncRNA-mRNA network reconstructions based on co-expression coefficients or the identification of coding neighbours appear to be two essential analyses for subsequent pathways (eg KEGG or GSEA) and gene ontology annotations. Alternatively, weighted correlation network analysis (WGCNA) would be a good option. 35 In addition to gene modules (clusters) that contain highly correlated lncRNAs and coding RNAs, WGCNA may determine particular phenotypes (eg time-points of injury or gliogenesis) that are associated with these gene modules.
Weighted correlation network analysis therefore may provide more clues from bioinformatic annotations. In this connection, cell-type specific transcriptome analysis will be of great interest to researchers. By

| FUN C TI ONAL ROLE S OF LN CRNA S IN SCI
Recently, the biofunctional characterizations of differentially expressed lncRNAs in SCI were increasing. In particular, the modulation of glial activation and neuronal apoptosis by lncRNAs have become areas of intense investigation (Table 2; Figure 3).
TA B L E 2 Functional characterization of the lncRNAs in spinal cord injury F I G U R E 3 Known functional roles of lncRNAs on glial activation and neuronal apoptosis. Green box: Glial activation induced by lncRNAs in spinal cord injury (SCI). Down-regulation of lncSCR1 in acute contusive SCI model led to up-regulation of Bmp7 and Adm, resulting in astrogliosis. Up-regulation of metastasis associated lung adenocarcinoma transcript 1 (MALAT1) in the same model sponged miR199b, leading to pro-inflammatory cytokine production and microgliosis. Blue Box: Neuronal apoptosis regulated by lncRNAs in SCI. In the spinal cord ischaemic/reperfusion injury (SCIRI) model, suppression of MALAT1 sponged miR-204-dependent apoptosis, whereas that of CasC7 sponged miR-30c-dependent apoptosis. In the contusive SCI model, X-inactive specific transcript (XIST) was up-regulated, followed by miR-494 down-regulation and phosphatase and tensin homolog deletion on chromosome 10 (PTEN) activation-induced PI3K/AKT pathway, resulting in neuronal protection against apoptosis and hence promote the production of pro-inflammatory cytokines.
Moreover, knockdown of spinal MALAT1 reduced the expression of Iba-1 (microglial marker) and pro-inflammatory cytokines in contusion epicentre, and improved locomotor function of hindlimb in the same model of SCI. However, the roles of MALAT1 in microglial polarization were not explored in this study. In another report, 45 the expression of lncSCIR1 was found to be constantly down-regulated on the 1st, 4th and 7th day after moderate contusive SCI. lncSCIR1 (long non-coding spinal cord injury related 1) was inversely correlated to the expression of Bmp7 (bone morphogenetic protein 7) and Adm (Adrenomedullin), both of which have been reported to promote astrogliosis in the spinal cord, 46,47 indicating that lncSCIR1 might have a role in regulating astrocytes. Indeed, knockdown of lncSCIR1 was sufficient to promote the migration and proliferation of cultured astrocytes. 45 However, most of the functional studies were performed in vitro. The distribution of lncSCIR1, ie scar or perilesional area, was unclear. The evidence that whether lncSCIR1 replenishment was beneficial for SCI was also lacking. Nonetheless, these studies provided preliminary evidence that lncRNAs might participate in gliogenesis after SCI.   51

| CON CLUS ION
Spinal cord injury is still a tough clinical issue that needs intensive research. Previous studies have gradually disclosed the pathological changes of SCI and their underlying mechanism. Current investigations on the molecular properties of SCI are focusing on protein-coding genes, yet the clinical translation is still not satisfactory. lncRNAs, via interacting with the coding gene network, participate in various cellular and tissue alterations in all stages of SCI. lncRNA deregulations therefore represent a new dimension of molecular mechanisms of SCI.

CO N FLI C T O F I NTE R E S T
There is no conflict of interest.