A total of 99 mice were used in this study; 15 mice were used for the functional assessment of sciatic functional index (SFI), 72 mice were used for microarray analysis and 12 mice were used for further PCR verification. All the mice were obtained from the Laboratory Animal Research Center, Academy of Military Medical Sciences, of the Chinese People's Liberation Army, Beijing, China.

The right sciatic nerves of 15 C57Bl6 mice were crushed using an ultra-fine, smooth, straight hemostat (tip width, 0.6 mm, Fine Science Tools) for 20 sec, as previously descrbied (16). After this procedure, the SFI was measured in each mouse daily for 28 days. The hind paws of the mice were immersed in non-toxic ink. The mice moved without assistance along a walking track and generated footprints were recorded. As previously described (17), several parameters were calculated for the SFI: print length (PL), toe spread (TS) and intermediate toe spread (ITS). All parameters were measured for normal (N) and experimental (E) animals. The SFI was determined according to the formula described by Inserra et al (17) as follows: SFI, 118.8 (ETS − NTS/NTS) − 51.2 (EPL − NPL/NPL) − 7.5, where NTS is normal toe spread, ETS is experimental toe spread, EPL is experimental print length, and NPL is normal print length.

A total of 72 C57Bl6 mice, approximately 2 months of age were selected and randomly classified into 4 groups according to the time points of 0, 3, 7 and 14 days post-surgery. The 0 day group was used as a control, while all the other groups were the experimental groups. Due to the insufficient volume of a single sciatic nerve, 6 mice were pooled into one sample, and 3 samples for each time point. After the mice were anaesthetized via an intraperitoneal injection of chloral hydrate, an incision was created on the right lateral thigh to expose and lift the sciatic nerve. As described in a previous study (18), right sciatic nerve crush was performed at the upper thigh level using an acutenaculum. The nerve was compressed for 30 sec to ensure that the axon was disconnected while the epineurium remained intact. The crush site was then ligated to generate a marker for the damage site. The incision was then closed. To relieve painful mechanical stimulation and discomfort, the mice were housed in clean cages with sawdust bedding. We provided free access to food and water. These experiments were performed according to the NIH Guidelines for the Care and Use of Laboratory Animals (http://oacu.od.nih.gov/regs/index.htm). All procedures were approved by the Ethics Committee of Tianjin Medical University, Tianjin, China.

The microarray experiment was conducted using an Agilent-074622 Mouse lncRNA micro-array (Agilent Technologies, Santa Clara, CA, USA). The mice were sacrificed by cervical dislocation. Tissue perfusion was performed to remove the blood in the sciatic nerve. Distal segments of crushed sciatic nerves (0.5 cm) were isolated and harvested from the mice of each group 0, 3, 7 and 14 days post-surgery. Total RNA was extracted using a Takara RNAiso Plus kit (#9109) according to the manufacturer's instructions. The quality of the total RNA was assessed by the RNA integrity number (RIN) on an Agilent Bioanalyzer 2100 (Agilent Technologies). The total RNA was purified using an RNase-Free DNase kit and an RNeasy micro kit (both Qiagen GmBH, Hilden, Germany). The total RNA samples had a RIN of ≥7.0 and a 28S/18S ratio of ≥0.7.

The preparation of One-Color Spike Mix was performed using an Agilent One-Color RNA Spike-In kit. We amplified and labeled total RNA using a Low Input Quick Amp WT Labeling kit (Agilent Technologies) and purified cDNA using an RNeasy mini kit (Qiagen GmBH).

A total of 1.65 µg of Cy3-labeled cDNA was hybridized onto each microarray using a Gene Expression Hybridization kit in a hybridization oven (both from Agilent Technologies). Microarray slides were washed after a 17-h hybridization using a Gene Expression Wash Buffer kit (Agilent Technologies). All aforementioned procedures were performed in accordance with the manufacturer's instructions.

After scanning the microarrays with an Agilent Microarray Scanner we used Feature Extraction software v10.7 (both from Agilent Technologies) to extract the data. Raw data were normalized by a quantile algorithm in GeneSpring software v11.0 (Agilent Technologies).

The microarray data generated in our study were deposited into the NCBI Gene Expression Omnibus (GEO) under accession no. GSE74087 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE74087).

Raw data were extracted using Feature Extraction 10.7 software, and the quantiles were normalized using GeneSpring GX 11.0 software (Agilent Technologies). The global distribution characteristics of the sample data were normalized and visualized using a box plot. To comprehensively and clearly depict the associations and differences among the samples, clustering into expressed genes or differentially expressed genes was performed. Samples with similar characteristics may present in the same cluster after processing, and genes presenting in the same cluster may have similar biological functions. Following the normalization of the raw data, the fold change was determined, and multiple hypothesis testing was performed to identify differentially expressed genes. Genes with a fold change (linear) ≤0.5 or ≥2 and a P-value <0.01 in a t-test were selected. Differentially expressed genes that were identified were visually presented using scatter plots and heat maps.

To visualize the time dependence of the differential lncRNA expression, we performed a time series analysis in which expression levels were summed to obtain relative expression patterns over time. lncRNAs were clustered into several distinct profiles using k-means clustering with a distance matrix based on Pearson's correlation.

lncRNAs evidently influence the expression of protein-coding genes; thus, we predicted the target genes of differentially expressed lncRNAs to further investigate their functions in biological processes. The possible target genes for cis- or trans-regulating lncRNAs were predicted by applying two independent algorithms.

For cis-regulating prediction, the genomic positional associations between the lncRNAs and their potential paired target genes were determined from the RefSeq and UCSC Known Genes databases (19). Based on the distance between each lncRNA gene and its neighboring known protein-coding gene, algorithms, including an ORF-Predictor and BLASTP pipeline were used to identify cis-regulating lncRNA-mRNA potential pairs. We selected 10 kb as the cut-off of the distance between lncRNAs and mRNAs based on a previous study (19).

For trans-regulating prediction, the RNAplex method (http://www.tbi.univie.ac.at/RNA/RNAplex.1.html) was applied to identify possible target genes for lncRNAs based on a previous study (20). RNAplex, which is specifically designed to rapidly identify possible hybridization sites for an RNA query in large RNA databases, adopts a slightly different energy model that reduces the computational time significantly.

To predict the function of differentially expressed lncRNAs, we selected predicted target genes that overlapped with the differentially expressed mRNAs for further GO analysis. GO analysis was performed to analyze the primary function of the target coding genes according to the GO database, which defines the terms 'cellular components', 'molecular functions', and 'biological process' for each gene. These analyses were used to determine the functional enrichment of the target mRNA to analyze the functions of the lncRNAs involved in the response to nerve injury.

To validate the microarray results, the expression level of three randomly selected lncRNAs, ENSMUSG00000087366, NONMMUG018386 and NONMMUG018381, were profiled at different time points after injury. For the preliminary validation of the lncRNA-mRNA pairs, the NONMMUG014387-Cthrc1, NONMMUG042364-Ntm and ENSMUSG00000097535-Icam1 pairs were detected. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a control. In total, 12 mice were used for PCR verification. Total RNA was extracted from the sciatic nerve as described above. Reverse-transcribed cDNAs were synthesized using a PrimeScript RT Reagent kit according to the manufacturer's instructions (Shanghai Biotechnology Corp., Shanghai, China). PCR reactions were performed using SYBR Premix Ex Taq according to the manufacturer's instructions in a Rotor-Gene 6000 instrument. The PCR reaction used the following cycling conditions: 10 min at 95°C; 95°C (10 sec), 60°C (60 sec) and 95°C (10 sec) for a total of 40 cycles; and a final temperature increase from 60 to 99°C. The relative quantity of RNA was calculated and analyzed using the 2ΔΔCq method. All experiments were performed in triplicate. The primer sequences are presented in Table I.

Mouse Schwann cells (MSCs) were puchased from Shanghai Cellbio Co. (Shanghai, China). The cells were inoculated in a new culture flask, prepared medium was added, 90% HyClone Dulbecco's modified Eagle's medium (DMEM) [10% fetal bovine serum (FBS); Gibco, Grand Island, NY, USA] 1–1.5% penicillin-streptomycin. The cells were then placed in an incubator chamber at 37°C over a period of 24 h and the density was observed under a microscope. The cells were passaged when they reached 80% confluence. The medium was then removed and the culture flask was washed with 3–5 ml D-Hanks or phosphate-buffered saline (PBS). This was followed by the addition of 0.5–1 ml trypsin with 0.25% EDTA for digestion. The digestion was terminated by the addition of 3–5 ml medium, followed by centrifugaqtion at 1,000 rpm/5 min. The supernatant was then discarded and passaged according to the amount of cells.

NONMMUG014387 was inserted into the pHBAd-MCMV-GFP vector. One day prior to transfection, 293 cells were seeded in a 60 mm dish and incubated at 37°C with 5% CO₂ overnight with Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS. The cells were transfected pHBAd-MCMV-GFP-NONMMUG014387 at 70–80% confluence. Following transfection, recombinant adenoviruses carrying NONMMUG014387 were harvested. All of the viral particles were purified by cesium chloride density gradient centrifugation and tittered by the TCID50 method.

The MSCs were plated in 6-well plate (cell/well), and different adenoviruses (Ad-GFP and Ad-NONMMUG014387) were added to each well at an MOI of 20. at 48 h post-transfection, the MSCs, and the MSC-GFP- and MSC-NONMMUG014387-transfected cells were digested into a single cell suspension. Each group of cells was equipped with a single cell suspension with 3×10⁴ cells/ml concentration. Following cell adherence, 10 µl of CCK-8 (Hanheng Biotechnology Corp., Shanghai, China) were aded to the wells at 0, 24, 48, 72 and 96 h. The optical denstity (OD) was at 450 nm using a spectrophotometer.

The statistical methods employed in this study were performed using the SPSS 20.0 software packages (SPSS, Inc., Chicago, IL, USA). The statistical significance of the differential expression of various lncRNAs in the micro-array analysis and RT-qPCR validation was determined using an independent sample t-test. GO and pathway analyses were conducted using Fisher's exact test. Our data are expressed as the means ± standard deviation (SD). Statistical significance was defined as P<0.01.

To assess the degree of functional impairment and recovery of locomotion following sciatic nerve injury in mice, the SFI was calculated daily for 28 days. The SFI values were almost −100 immediately following PNI; after 7 days, there was rapid neural recovery; however, after 14 days, nerve function recovery had decelerated (Fig. 2A).

The Agilent-074622 mouse lncRNA microarray consisted of 64,221 lncRNA probes collected from public databases, including GENCODE v21/Ensembl, Noncode v4 and UCSC. A total of 26,531 mRNA probes were also present on this array. Based on the SFI results, microarray experiments were performed at 0, 3, 7 and 14 days after sciatic nerve crush to identify lncRNAs that were differentially expressed during Wallerian degeneration.

The 12 specimens shared a similar level of total characteristics, which is shown in a box plot (Fig. 2B). Sample clustering revealed that the samples from each time group were grouped into the same cluster (Fig. 2C). A total of 5,354 lncRNAs were identified as significantly differentially expressed among all groups following injury (P<0.01; fold change >2). We also compared lncRNA expression independently. A comparison between the 0 and 3 days data revealed that 3,788 lncRNAs were differentially expressed. Among these, 1,521 were upregulated, while 2,267 were downregulated. A comparison between the 0 and 7 days data revealed that 3,314 lncRNAs were differentially expressed. Among these, 1,754 were upregulated, while 1,560 were downregulated. A comparison of the data between 0 and 14 days revealed a total of 2,400 differentially expressed lncRNAs, 1,401 of which were upregulated and 999 down-regulated (Fig. 2D). The top 10 downregulated and upregulated lncRNAs between the different groups are shown in Tables II–IV. A scatter plot was generated to assess the variation in lncRNA expression among the different groups (Fig. 3A, C and E). In addition, hierarchical cluster analysis revealed lncRNA expression patterns. In hierarchical clustering analysis, we used a heatmap to indicate that differentially expressed lncRNAs at different post-injury time -points were segregated into different clusters (Fig. 3B, D and F). The results of RT-qPCR for NONMMUG018381, NONMMUG018386 and ENSMUSG00000087366 were consistent with those of the microarray analysis (Fig. 4).

The dysregulated lncRNAs were characterized by their lengths and chromosomal distribution (Fig. 5A and B). Several cellular biological processes, such as dedifferentiation, demyelination, redifferentiation and remyelination, are involved in Wallerian degeneration, which results from a cut or crush injury (21–23). Furthermore, lncRNA expression in response to injury may be time-dependent. We therefore performed a time series analysis to classify these dysregulated lncRNAs. Consequently, 11 lncRNA profiles had significant P-values (Fig. 5C). Recent studies have demonstrated that lncRNA transcription is often initiated simultaneously with the expression of their overlapping or interspersed sequences (20,21). This finding implies that the lncRNAs that are associated with adjacent genes may be involved inregulating gene expression. To obtain a better understanding of this association, lncRNAs were crudely classified into five categories according to their association with adjacent genes: i) sense lncRNAs overlap with the exons of coding genes and are transcribed from the same strand; ii) antisense lncRNAs overlap with the exons of coding genes and are transcribed from the antisense strand; iii) bidirectional lncRNAs are transcribed with a coding transcript that is transcribed in close proximity on the opposite strand; iv) intronic lncRNAs are derived from the introns of coding genes; and v) intergenic lncRNAs are located within the interval between two genes (24). In this study, we analyzed the categorical distribution of the 5,354 lncRNAs that were differentially expressed following sciatic nerve injury (Fig. 5D).

To further explore the dysregulated lncRNAs, we predicted the target genes that would be regulated using cis and trans mechanisms. The target genes overlapped with mRNAs that were differentially expressed at different time-points. We chose the overlapping mRNAs for further GO analysis. For further analysis, we selected 9 classes of biological processes that are closely related to nerve regeneration following PNI, including stimulation responses, inflammatory responses, immune responses, cell proliferation, cell migration, axon guidance, myelination, extracellular matrix processes, protein kinase activity and growth factor activity. The numbers of dysregulated lncRNAs that were predicted to participate in these processes are shown in Fig. 6.

Based on the classification of lncRNAs in relation to different biological processes, we predicted the lncRNA-mRNA pairs in the 9 biological processes that may be important for peripheral nerve regeneration following injury, according to the following criteria: i) the lncRNA was predicted to be involved in the 9 biological processes according to the results described above; ii) the lncRNA was predicted to use a cis or trans mechanism to target mRNAs that have been reported to be involved in peripheral nerve regeneration; iii) the mRNA has been reported to be involved in peripheral nerve regeneration; iv) both the lncRNA and mRNA were differentially expressed in the microarray experiment. A list of these lncRNA-mRNA pairs is provided in Table V.

For the preliminary validation of the predicted lncRNA-mRNA pairs, we selected three lncRNA-mRNA pairs for RT-qPCR verification. NONMMUG014387 was predicted to have cis-regulating potential with Cthrc1, NONMMUG042364 was predicted to have cis-regulating potential with Ntm, and ENSMUST00000180870 was predicted to have cis-regulating potential with Icam1. The results of RT-qPCR and microarray analysis revealed that NONMMUG014387 and NONMMUG042364 were upregulated following PNI, and Cthrc1 and Ntm were also upregulated following PNI. The expression trends of the two lncRNAs at different time points were consistent with the separate trends of the two mRNAs. ENSMUSG00000097535 was downregulated following PNI, while Icam1 was upregulated following PNI, although in a different direction; the expression level of ENSMUSG00000097535 was mostly consistent with that of Icam (Fig. 7).

NONMMUG014387 was overexpressed through recombinant adenoviruses, the relative level of NONMMUG014387 was detected by RT-qPCR, and the value of the MSCs transfected with NONMMUG014387 was higher than that of the control MSCs or the MSCs + GFP group (Fig. 8A). CCK8 reavealed that the overexpression of NONMMUG014387 in the MSCs increased cell proliferation compared with the control MSCs and the MSCs + GFP group d (Fig. 8B).