Neuronal-specific deficiency of the splicing factor Tra2b causes apoptosis in neurogenic areas of the developing mouse brain

Abstract

Alternative splicing (AS) increases the informational content of the genome and is more prevalent in the brain than in any other tissue. The splicing factor Tra2b (Sfrs10) can modulate splicing inclusion of exons by specifically detecting GAA-rich binding motifs and its absence causes early embryonic lethality in mice. TRA2B has been shown to be involved in splicing processes of Nasp (nuclear autoantigenic sperm protein), MAPT (microtubule associated protein tau) and SMN (survival motor neuron), and is therefore implicated in spermatogenesis and neurological diseases like Alzheimer's disease, dementia, Parkinson's disease and spinal muscular atrophy. Here we generated a neuronal-specific Tra2b knock-out mouse that lacks Tra2b expression in neuronal and glial precursor cells by using the Nestin-Cre. Neuronal-specific Tra2b knock-out mice die immediately after birth and show severe abnormalities in cortical development, which are caused by massive apoptotic events in the ventricular layers of the cortex, demonstrating a pivotal role of Tra2b for the developing central nervous system. Using whole brain RNA on exon arrays we identified differentially expressed alternative exons of Tubulinδ1 and Shugoshin-like2 as in vivo targets of Tra2b. Most interestingly, we found increased expression of the cyclin dependent kinase inhibitor 1a (p21) which we could functionally link to neuronal precursor cells in the affected brain regions. We provide further evidence that the absence of Tra2b causes p21 upregulation and ultimately cell death in NSC34 neuronal-like cells. These findings demonstrate that Tra2b regulates splicing events essential for maintaining neuronal viability during development. Apoptotic events triggered via p21 might not be restricted to the developing brain but could possibly be generalized to the whole organism and explain early embryonic lethality in Tra2b-depleted mice.

Figure 1. Conditional ablation of Tra2b causes perinatal lethality and disturbed cortical patterning in mice.

(A) Cross breading of Tra2b^fl/fl with Tra2b^fl/+; Nestin-Cre^tg/0 mice allowed generation of neuronal-specific knock-out (KO) animals as well as controls (CRTL) and heterozygous knock-out animals (HET) in one litter. (B) KO animals are born alive and all possible genotypes were detected according to Mendelian law (N = 113). (C) General development of conditional knock-out mice is not impaired as there are no gross morphological differences in embryo appearance. (D) Hematoxylin/Eosin staining of paraffin-embedded coronal sections at indicated developmental stages. KO animals but not controls or HET animals show ventriculomegaly of the third and lateral ventricles starting at around 14.5 dpc. Cortical layers are largely distinguishable at 14.5 dpc but cortical patterning and the ependymal lining of the lateral ventricle appears highly disturbed (black arrowheads) at 16.5 dpc in knock-out brains. (E) Immunostaining of Tra2b on paraffin-embedded coronal sections shows efficient downregulation of Tra2b protein in knock-out brains compared to controls and heterozygote animals. Cells of the ventricular and subventricular zones of the cortex show strongest decrease in staining intensity (black arrowhead). Scale bar equals 400 µm; ctx, cortex; th, thalamus; cpt, caudoputamen; cp, cortical plate; iz, intermediate zone; svz, subventricular zone; vz, ventricular zone; sn, septal nuclei; 3v, third ventricle; lv, lateral ventricle; pc, choroid plexus.

Figure 2. Brain malformations are initiated by massive apoptosis in the cortex.

(A) Immunostaining for Caspase-3 on paraffin-embedded coronal sections indicates prominent apoptosis in the proximal cortical layers and in the thalamic area of 14.5 dpc and 15.5 dpc KO embryos (black arrowheads). Remaining cortical tissue does not show apoptosis at 16.5 dpc and later stages (light arrowhead). (B) Immunostaining for Ki-67 shows initial decrease of proliferation at 14.5 dpc which is fully lost at 16.5 dpc in KO animals (black arrowheads). Control and HET animals retain strong Ki-67 signals in the proximal cortical layers at all indicated developmental stages. Scale bar equals 400 µm; ctx, cortex; th, thalamus; cpt, caudoputamen; sn, septal nuclei; 3v, third ventricle; lv, lateral ventricle.

Figure 3. Bona fide Tra2b-related splicing events can be detected in vivo in conditional Tra2b knock-out mice.

(A–G) Quantitative real-time PCR using whole brain RNA of control, HET and KO embryos at 16.5 dpc. All mRNA levels have been normalized to Gapdh. (A) Total Tra2b expression levels are effectively reduced in KO animals at 16.5 dpc, 18.5 dpc and PND1. (B) Expression levels of both the functional and non-functional Tra2b isoforms are reduced at 16.5 dpc. (C) Individual expression levels of total Tra2b largely behave accordingly to the respective genotype but are subject of variation within each group. Horizontal lines indicate the mean of each genotype group. The dashed line indicates a threshold value of 0.4. (D) Expression levels of the 3R and 4R isoforms of the Mapt transcript. Transcript levels of the 4R isoform are reduced in KO brains. (E) Expression levels of the LCbII and LCbIII isoforms of the Cltb transcript. Transcript levels of the LCbII isoform are reduced in KO brains. (F) Expression levels of the somatic and the testis-related isoform of the Nasp transcript. Transcript levels of the testis-specific T-isoform are drastically reduced by −5.5-fold in KO brains. (G) Expression levels of the functional and non-functional isoform of the Tra2a transcript. Transcript levels of the non-functional isoform are drastically reduced by −4.0-fold and levels of the functional isoform are increased by +1.4-fold. Mapt, microtubule associated protein tau; Cltb, Clathrin light chain b; Nasp, Nuclear autoantigenic sperm protein; Tra2a, Transformer 2 alpha; a.u., arbitrary units; error bars indicate the s.e.m.; significance levels are *p<0.05, **p<0.01, ***p<0.001 (Student’s t-test); N is given by numbers within bars.

Figure 4. Mouse whole exon array analysis reveals Tubd1 exon4 and Sgol2 exon4 as in vivo targets of Tra2b.

Whole brain RNA of 4 CTRL animals and 4 KO animals was analyzed on mouse exon array. (A) The inclusion ratio (PSI, percent splicing inclusion) of each identified exon is defined as [PSI_KO]/[PSI_CTRL]. PSI distribution reached from ∼0.2 until ∼4.0. (B) Initial filtering strategies comprised exclusion of PSIs between 0.66 and 1.5 (grey bars) as well as restriction to p-values smaller than 0.05 which yielded a total of 1,006 exons. Exons associated with transcripts identified as being transcriptionally up- or downregulated were excluded from analysis. Ranking of those was further refined using large PSI values and considering presence of putative Tra2b binding sites (AGAA-motifs). Thereby, exons had to contain at least a single AGAA-site and a AGAA-frequency higher than 1.5. (C,D,G,H) Semi-quantitative RT-PCT on whole brain RNA was carried out using isoform specific primers for Sgol2 FL (C), Sgol2 Δ4 (D), Tubd1 FL (G) and Tubd1 Δ4 (H) confirming splicing events identified on the microarray. All isoform expression levels were densitometrically measured and normalized against Hprt (E,F,I). The Tubd1 Δ4 isoform could not be detected using whole brain RNA, as skipping of exon4 introduces numerous premature termination codons leading to nonsense-mediated decay of the transcript (H). Treatment of wt and Tra2b-depleted murine embryonic fibroblasts with emetine successfully inhibited NMD and the Tubd1 Δ4 isoform was detectable in Tra2b-depleted cells only (J). FL, full length; Δ4, transcript lacking exon 4, (−) PCR negative control; a. u., arbitrary units; error bars show the s.e.m.; significance levels are *p<0.05, **p<0.01, ***p<0.001 (Student’s t-test).

Figure 5. Splicing of Sgol2 and Tubd1 is responsive to changes in Tra2b concentration.

(A) The murine and human versions of the genomic regions comprising the identified exons of Sgol2 and Tubd1 were cloned into the pSPL3 exon trapping vector. (B) HEK293T cells were co-transfected with the pSPL3 minigene vector and siRNA specific for Tra2b or a TRA2B-GFP expression vector. Western Blot analysis shows efficiently reduced Tra2b protein levels and solid overexpression of TRA2B-GFP. (C) RNA was analyzed for exon inclusion after 48 h by semi-quantitative RT-PCR. (D–G) RT-PCR results were densitometrically quantified. Exon 4 of murine but not human Sgol2 is responsive to increased concentrations of Tra2b as splicing inclusion significantly increased from 43% to 89%. Knock-down of Tra2b is insufficient to reduce Sgol2 exon4 splicing inclusion. Exon 4 of human but not murine Tubd1 is responsive to increased concentrations of Tra2b as splicing inclusion significantly increased from 79% to 94%. Knock-down of Tra2b decreased inclusion of exon 4 from 79% to 71%. Nt, non-treated; scr, scrambled siRNA; si, siRNA against Tra2b; n.s., non significant; error bars show the s.e.m.; significance levels are *p<0.05, **p<0.01, ***p<0.001 (Student’s t-test).

Figure 6. p21 is upregulated as a response to Tra2b depletion in the mouse brain and in neural stem cells.

(A,B) Semi-quantitative RT-PCR using whole brain RNA of neuronal specific Tra2b KO mice and controls. p21 is significantly upregulated by 1.5-fold in KO mice as compared to HET or control mice. p21 expression is indifferent between controls and HET mice. p21 expression was normalized to Hprt. (C–G) NSC34 neural stem cells were transfected with siRNAs specific to Tra2b or scrambled siRNAs. siRNA treatment but not scr-treatment effectively reduced Tra2b protein and mRNA levels after 24 h, 48 h and 72 h after transfection (C–E). Tra2b function was strongly reduced as the Nasp transcript showed a significantly lower inclusion of the T-exon at 24 h, 48 h and 72 hours after transfection (F). 48 hours after transfection p21 expression was found slightly but significantly increased on RNA level (G) but not on protein level (D). 72 hours after transfection p21 was massively and highly significantly upregulated on RNA and protein level by +2.2-fold (D,G). a.u., arbitrary units; nt, non-treated; scr, scrambled siRNA; si, siRNA against Tra2b; error bars show the s.e.m.; significance levels are *p<0.05, **p<0.01, ***p<0.001 (Student’s t-test).