Alternatively spliced ELAVL3 cryptic exon 4a causes ELAVL3 downregulation in ALS TDP-43 proteinopathy (2024)

Associated Data

Supplementary Materials

Data Availability Statement

Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disease characterized clinically by progressive weakness and neuropathologically by nuclear loss and cytoplasmic aggregation of the RNA binding protein TDP-43 in neurons of the brain and spinal cord in > 97% of cases [6]. TDP-43 is mainly a nuclear protein and plays a major role in RNA splicing. Loss of nuclear functions may lead to splice defects, such as insertion of noncanonical sequences called cryptic exons (CEs) that may cause premature stops, polyadenylation, truncation or non-functional proteins [4]. Two of the most recently identified and best characterized examples of cryptic exon inclusion are STMN2 CE 2a, leading to premature polyadenylation and a premature stop sequence, and UNC13A CE 21a, leading to a premature stop sequence [1, 3, 5, 7]. We previously reported downregulation and nuclear loss of the RNA binding protein ELAVL3 at both the transcript and protein levels in ALS nervous systems [2]. Loss of ELAVL3 in mice results in axonal deformity, loss of neuronal polarity, and synapse formation deficits [8]. Here we use reverse transcription quantitative PCR (qRT-PCR) and chromogenic in situ hybridization (CISH) applied to the spinal cord and motor cortex and show that the mechanism involved in this downregulation is also related to expression of a CE.

Within intron 3 (between exons 3 and 4) of ELAVL3 there is an intronic binding domain for TDP-43 (Fig.1a). Expression of a cryptic exon upstream of the TDP-43 binding site, ELAVL3 CE 4a, is seen when TDP-43 is reduced in a TDP-43 knock-down cellular model and in TDP-43 negative neuronal nuclei from frontotemporal lobar degeneration frontal cortex (Fig.1b) [3, 10]. Using RT-PCR and Sanger sequencing, we confirmed the existence of a 166-nucleotide sequence in mRNA transcripts, ELAVL3 CE 4a (hg38 chr19:11,463,662–11,463,496) in ALS tissues and note that this creates a frameshift leading to multiple premature stop sequences (Fig.1c). We re-confirmed our previous data demonstrating downregulation of total ELAVL3 in ALS spinal cord, and we expanded these findings into motor cortex to show reduced ELAVL3 mRNA in ALS relative to controls (Fig.1d). We detected ELAVL3 CE 4a using a short segment containing CE 4a and exon 4 by qPCR in 43% of ALS spinal cord and 77% of ALS motor cortex samples (Fig.1e, Table1). Expression of ELAVL3 transcripts containing CE 4a was extremely low relative to total ELAVL3. We also detected ELAVL3 CE 4a in 13% (2/15) spinal control cases, perhaps related to age (82 and 76years compared to control mean of 62years).

In order to visualize cell specificity of ELAVL3 expression, we performed CISH in formalin-fixed, paraffin embedded tissue sections using a BaseScope™ assay. We selected regions in ALS cases with preserved numbers of motor neurons, reasoning that early molecular events at the neuronal level were more likely to be identified in regions of the nervous system with the highest numbers of residual motor neurons [9]. For total ELAVL3 mRNA, we used a probe targeting a 300 base-pair region containing exons 1, 2, and 3. In ALS spinal cord, there were significantly fewer motor neurons positive for total ELAVL3 loci compared to control–68% of motor neurons in control contain 4 or more loci compared to 41% in ALS (Fig.1f–g). This is consistent with ELAVL3 mRNA downregulation in ALS.

To visualize cell specificity of ELAVL3 CE 4a expression by CISH, we used a probe that targets a 50-nucleotide RNA sequence containing the junction of exon 4a and exon 4 (Fig.1a). We found 46% of ALS spinal cord and 67% of motor cortex cases were positive for expression of ELAVL3 exon 4a loci (Fig.1h–i, Supplementary Fig.1a–d′, Table1). The majority of ALS motor neurons expressing ELAVL3 CE 4a contained between 1–3 loci (mean: 15.4% of total spinal cord motor neurons). In the spinal cord, ELAVL3 CE 4a were seen in anterior horn motor neurons and smaller neurons in the posterior regions of the anterior horn. We did not detect ELAVL3 exon 4a loci in glia of white matter tracts or in posterior horn neurons. In the motor cortex, we rarely detected expression of ELAVL3 CE 4a in motor neurons (Betz cells). Rather, we detected expression predominantly in smaller neurons of layers 3, 5, and 6 (Fig.1j–k). We did not detect ELAVL3 CE 4a loci above assay threshold in control nervous systems.

For comparison, we measured STMN2 CE 2a expression in our cohort by qPCR and CISH. By qPCR, full-length STMN2 was reduced in both ALS spinal cord and motor cortex, and STMN2 CE 2a was detected in 93% of ALS spinal cords and motor cortices (Supplementary Fig.2a–b). Similarly, using a CISH probe targeting 300 nucleotides of the 3′ untranslated region and a probe targeting a 50-nucleotide sequence containing the junction of canonical exon 1 and CE 2a, respectively, we detected downregulation of STMN2 and expression of STMN2 CE 2a in spinal cord motor neurons (Supplementary Fig.2c–d). Approximately a quarter of spinal cord motor neurons contained STMN2 exon 2a (Supplementary Fig.2e–f). In the motor cortex, we detected a pattern of expression of STMN2 CE 2a similar to ELAVL3 CE 4a. We did not detect STMN2 CE 2a in control nervous systems.

Using a duplex CISH assay to detect ELAVL3 CE 4a and STMN2 CE 2a simultaneously, there was not a strong concordance between expression of the CEs (Fig.1j–l, Supplementary Fig.1e–i). This suggests that neuronal subtypes have desynchronization of molecular changes, although we are at the technical limits to detect differential sensitivity of TDP-43 disruptions. The higher expression of transcripts containing STMN2 CE 2a relative to those containing ELAVL3 CE 4a fits with the pattern of the hypothesized destinations of these transcripts. STMN2 is a highly expressed transcript in anterior horn spinal cord, and the inclusion of the CE causes premature polyadenylation and inclusion of a premature stop sequence. ELAVL3 is moderately expressed relative to STMN2 and inclusion of ELAVL3 exon 4a leads to premature stop codons.

In summary, we confirm there is a CE in intron 3 of ELAVL3 that is expressed in ALS spinal cord and motor cortex using multiple modalities (Table1). We consider the overall reduction of ELAVL3 expression is by way of this splice dysregulation and CE expression, generating mRNA with premature stop sequences, similar to what has been reported for UNC13A and STMN2. Interestingly, although we presume the ELAVL3 CE expression is related to loss of TDP-43, we cannot entirely exclude the possibility that expression reduction relates to the dysfunction of ELAVL protein itself, since ELAVL2, ELAVL3, and ELAVL4 also bind intron 4 and thus could modulate its own splicing (Fig.1a).

Acknowledgements

This work wassupported by theNational Instituteof Aging andDepartment ofHealth and HumanServices [T32AG066596], theNational Instituteof NeurologicalDisorders andStroke [R21NS121805 and P30 NS047101], FightMND Foundation, and Target ALS Foundation.

Data availability

Images are available upon request.

Footnotes

References