Introduction

Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative adult disease characterized by fatal paralysis in both brain and spinal cord motor neurons. ALS can be induced by inherited mutations in the gene encoding the ubiquitously expressed enzyme Cu/Zn superoxide dismutase 1 (SOD1) [1]. The role of the normal SOD1 function—conversion of toxic superoxide into less damaging hydrogen peroxide—in ALS pathogenesis remains unclear. So we choose lumbar spinal cord of an amyotrophic lateral sclerosis mouse model to do this research for the purpose of taking a general look at transcriptional level of hSOD1-G93A transgenic mice.

Alternative splicing (AS) is known to be prominent in many important physiological processes, such as cell differentiation, apoptosis, and development, and is especially prevalent in the nervous system [2]. Pre-mRNA splicing, a step in mRNA transcription, is the process by which intervening sequences (introns) are precisely removed and the flanking functional sequences (exons) are joined together. Following splicing, the exons are always joined together in the same order in which they lie in DNA. The colinearity of gene and protein is maintained between the individual exons and the corresponding parts of the protein chain. The absence or presence of exons can directly affect the distinct proteins structurally and functionally [3]. In human cells, analysis of expressed sequence tag (EST) databases and DNA microarray experiments indicate that over 60 % of all genes is alternatively spliced [4]. Though alternative splicing is prevalent in the nervous system, little is known in ALS.

A high density exon-centric microarray, Affymetrix GeneChip® Mouse Exon 1.0 ST Array, which covers over 1 million exon clusters with approximately four probes per exon and roughly 40 probes per gene was used in this study. It enables two complementary levels of analysis—gene expression and alternative splicing. One is “gene-level” expression analysis in which multiple probes on different exons are summarized into an expression value of all transcripts from the same gene. The other is “exon-level” analysis on a whole-genome scale, which opens the door to detect specific alterations in exons that may play a central role in disease mechanism and etiology. So the information is more creditable [5].

In this study, we applied exon array to investigate the alternative splicing events and gene expression profiling in hSOD1-G93A transgenic mice.

Materials and Methods

Animals

Transgenic female mice expressing mutant human CuZn-SOD1 with a Gly93 → Ala substitution and their nontransgenic normal female littermates were generated by breeding male hemizygous carriers (B6SJL-Tg(SOD1G93A)1Gur/J) to female B6SJLF1 hybrids, both of which were purchased from the Jackson Laboratories (Bar Harbor, ME, USA). Transgenic progeny and their normal littermates were identified by polymerase chain reaction (PCR) amplification of tail DNA [6]. We collected 9 female mice, of which 6 were derived from transgenic mice (3 30-day-old in asymptomatic stage and 3 approximately 120 days of age in symptomatic stage) and 3 were derived from littermates (30 days of age) without hSOD1-G93A, namely 30A, 120A, and 30C, to be used in this study. Animals were killed at two time points to obtain fresh lumbar spinal cord tissues (L3 to L5 segment) for analysis. To monitor disease progression, the animals were visually inspected daily. Behavioral testing was measured twice a week starting at 90 days of age (Supplementary Material 1). We selected hSOD1-G93A mouse whose performance began to decline steadily in the rotarod test and any hindlimb extension was absent as symptomatic stage (approximately 120 days old).

The animals were housed in an asepsis room with 12 h light cycle and free access to water and diet. All experiments were carried out in accordance with the regulation of laboratory animal management promulgated by the Ministry of Science and Technology of the People’s Republic of China (1988, No. 134).

Samples of Exon Array Preparation and Array Hybridization

All mice were killed under deep anesthesia with 10 % chloral hydrate and then fresh specimens of lumbar spinal cord tissues were rapidly dissected. The specimens were immediately put into RNAlater (RNAlater RNA Stabilization Reagent, Qiagen), and additional specimens were instantly frozen in liquid nitrogen and then stored at −80 °C until use. Total RNA stored in RNAlater was extracted for exon array using RNeasy Mini kit (Qiagen, Valencia, CA, USA) following the manufacturer’s protocol. Purity and integrity of the RNA were assessed on the Agilent 2100 bioanalyzer with the RNA 6000 Nano LabChip® reagent set (Agilent Technologies, Santa Clara, CA, USA). The RNA specimens were quantified spectrophotometrically. For each experimental condition, three independent samples were used for exon array analysis

This exon array experiment was performed in CapitalBio Corp. (Beijing, China) using Affymetrix GeneChip® Mouse Exon 1.0ST arrays. Manufacturer’s instructions (Affymetrix, Inc.) were followed for biotinylation and synthesis of single-stranded DNA targets, hybridization, washing, and scanning steps [5].

Data Analysis Algorithms

The Affymetrix Expression Console Software V1.0 was used to process gene- and exon-level expression signal estimates. Signal estimates were derived from the CEL files of the nine samples by quantile sketch normalization using the “Probe Logarithmic Intensity Error Estimation” (PLIER) algorithm for probeset (exon-level) intensities and Iter-PLIER (a variant algorithm) for gene-level intensities. Candidate exons for alternative splicing were detected using a Splicing Index. Presence/absence of exons was determined by “Detection Above Background (DABG),” using surrogate GC mismatch intensities [5]. Details were shown in Supplementary Material 2.

Filtering

At the gene level, after variance stabilization and log transformation, the fold change and P value were calculated using control samples as base values. We set fold change >2 or <0.5 and P value < 0.05 as cutoff values for significantly differentially expressed genes.

At the exon level, the standard by which we filtered the exons as alternative splicing events is shown below: (a) SI (Splicing Index) >2 means this exon expression in sample 1 is higher than in sample 2, indicating the “exon inclusion” event; SI < 0.5 means this exon expression in sample 2 is higher than in sample 1, indicating the “exon skipping” event. (b) P value < 0.05. (c) Focus on exons that NI (Normalized Intensity) is near 1.0 in the group predicted to have a higher inclusion rate and near 0 in the group predicted to have a lower inclusion rate. (d) Limit search to genes that have very large differential expression. (e) Limit search to high-confidence exons. (f) Filter probes with unusually low variance.

The formulas are shown below:

Exon Level = Transcription × Splicing

$$ Gene-level\ Normalized\ Intensity \left( {NI} \right)=\frac{\mathrm{Probe}\ \mathrm{set}\ \mathrm{intensity}}{{\mathrm{Expression}\ \mathrm{level}\ \mathrm{of}\ \mathrm{the}\ ``\mathrm{gene}''}} $$
$$ \mathrm{Splicing}\ \mathrm{index}\ \mathrm{value}\left( {\mathrm{SI}} \right) = \mathrm{lo}{{\mathrm{g}}_2}\frac{{\mathrm{NI}\ \left( {\mathrm{Sample}1} \right)}}{{\mathrm{NI}\ \left( {\mathrm{Sample}\ 2} \right)}} $$

GO and Pathway Analysis of the Filtered Genes

The functions and pathways of these significant change genes were analyzed using CapitalBio® Molecule Annotation System (CB-MAS) V4.0 available through CapitalBio Corp. (http://bioinfo.capitalbio.com/mas), which included the gene ontology (GO) hierarchy analysis to group genes functionally and pathway analysis to group genes into respective pathway. Three GO terms which are biological process, cell component, and molecular function were evaluated. Pathways analysis also includes three sheets: KEGG, BioCarta, and GenMAPP. In this research, we chose molecular function and KEGG to demonstrate.

Validation by Reverse Transcription-Polymerase Chain Reaction (RT-PCR)

RT-PCR was used to verify the levels of phosphoinositide-3-kinase gamma polypeptide (PIK3CG, NM_020272) and NADPH oxidase 2 (NOX2, NM_007807) mRNA and to check whether Fyn-binding protein (FYB, AK030905) has alternative splicing. The RNA was extracted from the lumbar spinal cord tissues stored at −80 °C using Trizol reagent (Invitrogen, Carlsbad, CA), following the manufacturer’s instructions. RNA concentration and purity were determined spectrophotometrically at 260 and 280 nm Reverse transcription used reagents all from Promega following the manufacturer’s instructions.

Target sequences were obtained from Affymetrix. Forward and reverse primers were 5′-g c t t a g a g g a c g a t g a c g t t-3′ and 5′-g g a t a g g a c t g t g g g a t c a g-3′ for PIK3CG (524 bp product), 5′-t g a a c g a a t t g t a c g t g g a c-3′ and 5′-a c t t g a g a a t g g a g g c a a a g-3′ for NOX2 (593 bp product), and 5′-a a t g g g t g c c t t g t t g c c-3′ and 5′-g c t t c t g a t c c t c t t g c c t t c-3′ for the exon 19 of FYB (539 and 226 bp product when aberrant splicing). Forward and reverse primers of exon 19 were designed adjacent to constitutive exons, and these primers amplified specific bands of differentially expressed transcripts. The densitometric values were normalized with respect to the values of glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

Results

Differentially Expressed Genes

In 30A vs.30C, 202 differentially expressed genes (DEGs; P < 0.05) passed our signal threshold and were identified for subsequent analysis among 14039 transcripts on the chip. There were four genes changed more than 1.5-fold, including Fc receptor-like S scavenger receptor (Fcrls, NM_030707). Only 1 gene without gene annotation was upregulated more than 2-fold (definitely 2.8-fold). In 120A vs. 30A, 2869 DEGs (P < 0.05) were identified in 14314 transcripts, consisting of 263 (9.17 %) filtered genes upregulated more than 2-fold and 71 (2.47 %) filtered genes downregulated more than 2-fold. Four genes regulated more than 1.5-fold in 30A vs. 30C were detected in 120A vs. 30A, but they had no statistical significance (P > 0.05). It can be seen that at least 330 significantly changed genes were emerging with the progression of disease process. Take comparison of 120A and 30A groups as an example, the significant DEGs was visualized using Cluster 3.0 software with TreeView tools after hierarchical, average linkage clustering (Fig. 1a). The DEGs in 120A vs. 30A were described using volcano plot (Fig. 1b). One hundred DEGs including top 90 upregulated genes and bottom 10 downregulated genes in 120A vs. 30A were shown (Tables 1 and 2). In addition, 16.2 % (54/334) of the transcripts which were changed significantly contained alternatively spliced exons in 120A vs. 30A, indicating the possibility of combinatorial regulation of transcription and splicing.

Fig. 1
figure 1

a Clustering display of exon array data of 120A vs. 30A groups. The differentially expressed genes, consisting of 263 genes upregulated more than 2-fold and 71 genes downregulated more than 0.5-fold in 120A vs. 30A groups, were determined using Cluster 3.0 software and visualized with Tree View tools after hierarchical clustering. b Volcano Plot reflected the comparison of the 120A and 30A groups. Each dot represents the log2 values for the transcript cluster from the Exon Array (x axis). Green and red dot signs indicate that the transcript cluster was significantly downregulated or upregulated

Full size image
Table 1 Upregulated genes in lumbar spinal cords of 120A vs. 30A mice (top 90)
Full size table
Table 2 Downregulated genes in lumbar spinal cords of 120A vs. 30A mice (bottom 10)
Full size table

Alternative Splicing Events

Using the “Splicing Index” algorithm described in the methods, 630 probe select regions (PSRs) were identified as alternatively spliced exons that belonged to 563 alternatively spliced transcripts (ASTs) in 120A vs. 30C, and only 85 PSRs were identified as alternatively spliced exons that belonged to 85 ASTs in 30A vs. 30C. In 120A vs. 30A group, 537 PSRs with higher expression were classified as “exon inclusion” events, while the other 93 PSRs with lower expression were considered as “exon skipping” events. In 30A vs. 30C group, compared with normal mice, 61 PSRs were classified as “exon inclusion” events and 24 PSRs were considered as “exon skipping” events. In 120A vs. 30A, we filtered 46 genes whose exon’s NI near 1.0 in one group predicted to have a higher inclusion rate and near 0 in the other group predicted to have a lower inclusion rate to show (Table 3). The same went for 30A vs. 30C, 13 genes were filtered, including lengsin (Lgsn, NM_153601), RNA-binding motif protein 15B (Rbm15b, BC052180), tryptophan hydroxylase 2 (Tph2, NM_173391), colony stimulating factor 2 receptor beta 2 (Csf2rb2, NM_007781), prolactin (Prl, NM_011164), polycystic kidney disease 1-like 3 (Pkd1l3, NM_001039700), and other 7 unnamed genes. Furthermore, 12.06 % of alternatively spliced genes was changed at least 2-fold in 120A vs. 30A. In addition, only 7 of 85 ASTs in 30A vs. 30C were detected in 120A vs. 30A; that is, 556 ASTs were emerging with the progression of disease process.

Table 3 Alternatively spliced transcripts in lumbar spinal cords of 120A vs. 30A mice
Full size table

GO and Pathway Analysis of Transcription and Splicing

GO and pathway analysis were performed on 4 DEGs (fold change ≥ 1.5, P < 0.05) and 13 ASTs in 30A vs. 30C using MAS V4.0 with P value cutoff of less than 10−3. Only Fcrls in four DEGs was shown to be related to receptor activity and no pathway result was found. Six in 13 ASTs showed GO analysis results, including Rbm15b, Pkd1l3, Csf2rb2, Prl, Tph2, and gluld1, which were related to nucleotide binding, ion channel activity, receptor activity, hormone activity, monooxygenase activity, and catalytic activity, respectively. Furthermore, Csf2rb2 and Prl belonged to cytokine–cytokine receptor interaction and Jak-STAT signaling pathway based on KEGG library. Likewise, 100 DEGs and 46 ASTs in 120A vs. 30A were analyzed, which were shown in details (DEGs in Supplementary Table 1, 2 and ASTs in Table 4).

Table 4 GO and pathway analysis of 46 alternatively spliced transcripts in lumbar spinal cords of 120A vs. 30A mice
Full size table

Taking into account the molecular function of the 100 DEGs in 120A vs. 30A, there is a GO mapping to describe (Fig. 2a). With regard to the GO analysis, it is of note that the DEGs are significantly enriched in “protein binding, oxidoreductase activity and hydrolase activity” functional annotation. The 100 DEGs in 120A vs. 30A participate in 34 pathways. We chose KEGG data to show the 24 top enrichment pathways of these differentially expressed genes in 120A vs. 30A (Fig. 2b). Take “Leukocyte transendothelial migration” pathway for example, it includes NOX2 and PIK3CG (Supplementary Fig. 1). We found that this pathway contained not only 9 significantly upregulated genes, but also 2 exons inclusion events. Their change and change tendency have been validated by RT-PCR or supported by literature to be congruous with the exon array system [7–11]. Genes affected at both transcriptional and splicing levels appeared in this pathway. Otherwise, NOX2 appears in 5 pathways and PIK3CG appears in 28 pathways based on the three databases. Simultaneously, we can obtain the homologous gene’s correlation graph which was generated to illustrate the correlativity of two or more genes within networks (Supplementary Fig. 2). The number in this graph represents the number of pathways. This graph contains 14 genes and PIK3CG is the core, indicating that PIK3CG may function synergistically in this module in response to development of ALS. Based on this exon array system, we can focus on these genes and pathways to explore the pathogenesis of ALS.

Fig. 2
figure 2

a Go mapping classification of the molecular function of the 100 differentially expressed genes in 120A vs. 30A with MAS software (http://bioinfo.capitalbio.com/mas). b The top 24 statistically significant (P < 0.001) KEGG pathway are shown and the numbers indicate the numbers of the genes involved in this pathway

Full size image

RT-PCR Validation

Two DEGs exhibiting significant differences in expression were detected verified by RT-PCR. Exon array showed that NOX2 gene was upregulated 9.83-fold (P = 0.0130) and PIK3CG gene was upregulated 2.4-fold (P = 0.0136) in 120A vs. 30A, and no change in 30A vs. 30C. The expression change and changed tendency of NOX2 and PIK3CG genes validated by RT-PCR were consistent with the results acquired by the exon array (Fig. 3). One alternative splicing exon was selected for validation by RT-PCR. From the exon array, the exon 19 of FYB gene (SI = 5.25, P = 0.042) was highly included in 120A (NI = 0.9704) and lower expressed in 30A (NI = 0.185), which is consistent with the results of the RT-PCR (Fig. 4). All these results suggest that the exon array system maybe reliable and effective enough to detect differential expression at both transcriptional and splicing levels.

Fig. 3
figure 3

RT-PCR validation of differentially expressed genes in 30A (n = 3), 30C (n = 3), and 120A (n = 4) samples from lumbar spinal cord. Representative bands of NOX2 and PIK3CG mRNA expression in lumbar spinal cord. GAPDH was used as an internal control

Full size image
Fig. 4
figure 4

RT-PCR validation of alternatively spliced genes in 30A (n = 3), 30C (n = 3), and 120A (n = 4) samples from lumbar spinal cord. RT-PCR was performed with primers hybridizing to neighboring constitutively expressed exons. Gene symbols are listed to the left gel, and sequence-validated exon structures are shown to the right with primer postions indicated by arrows. This is a “exon inclusion” event

Full size image

Discussion

Alternative splicing is a common strategy for creating functional diversities of proteins that have cell and developmentally specific functions. Given the important role for splicing, it is not surprising that a recent estimate has proposed that 50 % to 60 % of mutations linked to disease affect splicing [12]. Some previous researches support the critical morbigenous contribution [13] of exon splicing to ALS in particular, such as TDP-43, FUS/TLS, and EAAT2 [14–16]. But they only focused on single gene. Exon arrays provide the most comprehensive coverage of the genome, including empirically supported and predicted transcribed sequences, enabling the discovery of previously unidentified novel events. Most bioinformatics studies [17] rely on identifying ESTs that come from the same gene and looking for differences between them that are consistent with alternative splicing. But given that only a few ESTs have been sequenced for most genes, it seems possible that even more alternative splicing exists that is not yet detectable in the available ESTs. Moreover, there are many researches on gene expression profiling of ALS not only in hSOD1-G93A mouse but also in patients [18]; however, application of GeneChip® Mouse Exon 1.0 ST array on lumber spinal cord of ALS mice at different stages that allows genome-wide identification of differential splice variation, and concurrently provides a flexible and inclusive analysis of gene expression was few. While conventional identification of splice variants generally targets individual genes, we present this exon array to take a more extensive and deeper look in ALS caused by hSOD1-G93A mutation.

Complex Regulation of Transcription and Splicing in hSOD1-G93A Mouse

In 120A vs. 30A, 12.06 % of alternatively spliced genes was upregulated or downregulated at least 2-fold, and 25.24 % of alternatively spliced genes was upregulated or downregulated by at least 1.5-fold. Alternative splicing can expand the protein repertoire and influence protein function by altering protein domains. The affected domains in the coding regions of alternatively spliced exons confirmed the existence of changes in the transcriptome and proteome resulting from alterations in the domain architecture of biological networks [19]. It was hypothesized that splicing may modify transcription activity [20] and can change RNA stability which in turn may affect gene expression [21], while transcription may change the splicing efficiency [22]. It is also possible that different expression levels of the upstream genes such as splicing factors facilitate or inhibit splicing machinery by influencing spliceosome assembly or the cis-elements during the splicing process. These two aspects of regulations may result in high degree of correlation between splicing patterns and transcriptional expression [23]. Evidence from the splicing classifications and the overlap between the two levels suggests a combinatorial regulation. Especially, the research of alternatively spliced exons was analyzed based on the coding regions, which would be more significant.

Functional and Pathway Analysis in hSOD1-G93A Mouse

According to the exon array system, the gene function and pathway analysis of DEGs and ASTs revealed the direct or indirect effect of mutant SOD1 in ALS. With regard to the GO analysis, it is of note that the DEGs are significantly enriched in “protein binding, oxidoreductase activity and hydrolase activity” functional annotation. Since genes are usually functionally organized into pathways, it is necessary to explore the gene regulation in terms of the pathways involved. As shown in Supplementary Fig. 1, the “Leukocyte transendothelial migration” pathway includes NOX2/gp91phox (also known as cytochrome b-245, beta polypeptide that is Cybb), PIK3CG and FYB. The change of the PI3K/Rac1/NOX2/ROS pathway contained by the “Leukocyte transendothelial migration” pathway was detected at gene level by exon array system and RT-PCR. FYB gene undergoes alternative splicing in ALS model, which may inference that it maybe affect the function of some signaling molecules implicated in T cell receptor (TCR) signaling, including PI3K, Fyn, and so on [24, 25]. The combination of previous researches and exon array suggested that PIK3CG, NOX2, and FYB play important roles in ALS. Based on this exon array system, we can focus on these genes and pathways to explore the pathogenesis of ALS.

Cascade Amplification Reaction in hSOD1-G93A Mouse

At the gene level, there were 202 DEGs (P < 0.05) in 30A vs. 30C, consisting of only one gene upregulated more than 2-fold. In 120A vs. 30A, we detected 2869 DEGs (P < 0.05), consisting of 263 (9.17 %) genes with more than 2-fold upregulation and 71 (2.47 %) genes with downregulation. Four genes which were 1.5-fold changed in 30A vs. 30C were all detected in 120A vs. 30A, but they have no statistical significance because of P > 0.05. It can be seen from that at least 330 significantly changed genes were emerging with the progression of the disease process. Likewise, 556 ASTs were emerging. At asymptomatic stage of ALS, there is no significant change between transgenic mice and non-transgenic mice at transcriptional level, but the transgenic mice at symptomatic stage differ significantly from asymptomatic stage. This suggests that the mutant SOD1 may be involved in a number of key steps in the pathological cascade of events leading to neuronal injury and this kind of ALS may be a cumulative result of mutant SOD1. It may revoke the cascade amplification reaction. The changes of NOX2 and PIK3CG are good examples. They exist in “Leukocyte transendothelial migration” pathway, which were upregulated accompanied by time lapse. From Supplementary Fig. 1, we can see that many cytokines are induced, as measured by the exon array. These DEGs and ASTs provide a new direction for further study of possible pathogenic mechanism and treatment of ALS at transcriptional level. We cannot neglect aging-associated changes, but it is not the key point in morbility process of transgenic mice. The lifespan of littermate mice without mutant SOD1 reaches at least 600 days; however, the transgenic mice would die at approximately 150 days old. Moreover, exon array did not detect the change of several transcription factors (e.g., Ink4, LASS2, some zinc-binding proteins, IP3R, and some collagen isoforms) that might be implicated in aging.

Conclusion

In this study, hSOD1-G93A mice were used as ALS model to explore the occurrence and development processes of ALS, followed by a genome-wide expression profiling of transcription and splicing by exon array system. Functional and pathway analysis of gene level and exon level demonstrated the importance of transcriptional and splicing regulation in physiological and pathological processes. There is increasing evidence for alterations in RNA splicing as being associated with not only ALS, but also related forms of motor neuron diseases. Because very few studies have investigated splicing regulation in hSOD1-G93A mice, elucidating the underlying transcriptional mechanisms associated with these phenomena is critical for a better understanding of the pathogenesis of ALS.