2.1. Various Splicing of Transcripts in HuD KO Cortex
There are a number of types of various splicing mechanisms, together with exon skipping, use of mutually unique exons, various 5′ or 3′ splice websites and intron retention. Every of those occasions can lead to mRNA isoforms with totally different exons from the identical gene or intron-including isoforms of the identical gene. Differential splicing occasions in cortical tissue of HuD KO vs. wildtype (management) littermates had been analyzed utilizing the rMATS software program. As proven in Determine 1A, 310 vital splicing occasions had been recognized. Exon skipping represented the biggest proportion of different splicing between teams at 77.74%, whereas each intron retention and mutually unique exons represented 3.55% (Determine 1B). Full rMATS output for all 5 various splicing occasions together with all of the statistical analyses of the information from replicate samples of the 2 genotypes is proven in Tables S1–S5. Occasions with learn protection ≥ 5 (i.e., aligned reads counts larger than 5), ∣Δψ∣ > 5% (change in splicing larger than 5%) and FDR < 0.05 had been thought of vital (Determine 1A).
Differences in exon or intron inclusion levels between the two mouse genotypes are represented as follows: positive inclusion levels equal greater inclusion in KOs and negative inclusion levels equal decreased inclusion (more exclusion) in KOs. Altogether, we found 144 statistically significant alternative splicing events with increased inclusion levels in KOs and 166 events with decreased inclusion levels in KOs (Figure 1C). Specifically, transcripts from 114 genes exhibited increased exon inclusion in KOs, also indicating a decrease in exon skipping events. Exon inclusion levels in mRNAs from 127 genes were decreased in the KOs, indicating an increase in exon skipping events.
Ingenuity Pathway Analysis (IPA) software was used to determine the biological systems impacted by differential splicing events in HuD KOs (Figure 1D,E). The most affected biological pathways concerned cell death and survival, neurological disease, organismal injury and abnormalities, and nervous system development and function (Figure 1D). Examples of major neuronal functions associated with those categories include loss (p-value = 0.0273) and viability (p = 0.00204) of neurons and synaptic transmission of nervous tissue and pyramidal neurons (p = 0.0152 and p = 0.0406, respectively) (Figure 1E).
Since HuD KO was found to have the largest effect on exon skipping, we examined these events in more detail. To identify exons involved in each event, rMATS and Maser outputs were used to determine chromosomal locations of exon start sites in Integrative Genomics Viewer (IGV; Broad Institute, Cambridge, MA, USA). Alternative splicing was then visualized using the rmats2sashimiplot software. Not surprisingly, the exon skipping event with the greatest inclusion level difference occurred in exon 2 of the Elavl4 gene itself, which is the exon deleted in HuD KO mice. While control mice had 100% inclusion of exon 2 (Figure 2A), HuD KO cortex exhibited 0% inclusion, and reads were shifted to the intron immediately following the exon (Figure 2B). The inclusion levels and statistical analysis of this and other significantly skipped exons are shown in Table S3.
Other genes where exon skipping was greatly impacted by HuD KO were Ap4e1 and Rapgef4. In this case, both genes were found to have a positive inclusion level difference, indicating that skipping of exon 10 in Ap4e1 and exon 7 in Rapgef4 occurs less frequently in KOs than in controls (Figure 2C,D). Alternative splicing of Ap4e1 at exon 10 has not been reported before, so this may be a novel isoform. The gene encodes the AP-4 complex subunit epsilon-1, which is involved in intracellular trafficking and sorting of AMPA receptors to axons . Rapgef4 encodes the exchange protein directly activated by cAMP 2 (Epac2), which has been shown to regulate the release of excitatory neurotransmitters . Alternative splicing of exon 7 in Rapgef4 has been reported previously, with Epac2A being the major splice variant expressed in the brain .
Although HuD KO significantly affected alternative splicing of several genes, it was unclear whether this was a direct effect of HuD or a result of indirect compensatory mechanisms in the KO mice. To identify genes that were directly affected by KO of this protein, we focused only on transcripts that had been shown to directly bind to HuD by RNA immunoprecipitation (RIP) assays. Significant splicing events were then compared with our previously identified 738 HuD targets (Table S6). These included common HuD targets from RIP-Chip and GST-HuD pulldowns of mouse forebrain  and RIP-seq from mouse striatum (Gardiner et al., manuscript in preparation). Comparison of this list with significant rMATS events (Tables S1–S5) identified 17 genes with transcripts known to bind to HuD that also displayed alternative splicing changes: Cspp1, Whsc1l1, Atp2c1, Whsc1l1, Ap3s1, Cbx3, Sbno1, Per3, Gria2, Atp2c1, Derl2, Ttc3, Clint1, Ube2w, Ube2w, Snap25, Stau2, Snap23, Ptpn12, Dram2, Ube2w and Cbx3. Multiple alternative splicing events were found in some of these genes (Atp2c1, Cbx3, Ube2w and Whsc1l1), and the majority of events were found to be exon skipping (Figure 3A).
The gene that exhibited the greatest inclusion level difference (41.3%) in KOs relative to controls was Cbx3, which encodes chromobox 3 (CBX3), a protein involved in transcriptional repression through the binding of histone H3 tails at methylated sites . The highest inclusion level difference occurred at exon 3. However, we found that the downstream exon for this event was exon 5, indicating that exon 4 is coregulated with exon 3 (Figure 3B). In contrast, the overall read counts of exon 3 were greater in control mice (Figure 3B), stressing the importance of using appropriate methods for identifying alternative splicing events instead of individual exon reads. The lowest inclusion level difference in KOs relative to controls (−39.1%) was found in Cspp1. This gene encodes the centrosome and spindle pole associated protein 1 (CSPP1), which functions in spindle organization and is required for primary cilia formation [40,41]. Primary cilia are known to be critical for neuronal development . Alternative splicing of exon 17 has been documented in Cspp1, resulting in a long isoform that is more physiologically relevant during mitosis . In contrast, we found an exon skipping event at exon 12, which is excluded more frequently in HuD KOs than in controls (Figure 3C). In this case, read coverage also indicated greater number of reads in controls relative to KOs.
From the 17 genes with mRNAs that were alternatively spliced and targeted by HuD, splicing has been shown to functionally impact two genes, Snap25 and Gria2, which encode proteins primarily involved in synaptic transmission and plasticity. In HuD KO cortices, there was a significant increase in Snap25 exon 5 inclusion (Figure 4A).
Furthermore, there are two Snap25 exon 5 isoforms: exon 5a and exon 5b (Figure 4B). SNAP-25 a and b isoforms differ in their ability to promote vesicle priming and release, with the SNAP-25b isoform primarily expressed in mature neurons [44,45]. In this mRNA, skipping of exon 5b was decreased in HuD KOs compared to controls (Figure 4C). Although the overall reduction of this exon was 8%, since we used bulk RNA-seq for the analyses, it is possible that only a low percentage of neurons was affected by this change.
For Gria2, inclusion of exon 14 was decreased in HuD KOs, indicating exon skipping occurred more frequently in these mice (Figure 4D). This gene encodes the glutamate receptor 2 (GluR2) protein, an AMPA receptor subunit involved in excitatory neurotransmission . Alternative splicing is known to occur between exons 14 and 15, which are specified as “flip” and “flop” exons (Figure 4E). GRIA2 subunits with inclusion of exon 14 are considered the “flop” isoforms, while those with inclusion of exon 15 constitute the “flip” isoforms . Visualization of exons 14 and 15 showed lower read coverage in HuD KOs at exon 14 compared to controls, indicating that KOs contained decreased levels of Gria2 flop isoforms (Figure 4F). In contrast, there were no changes in alternative splicing at exon 15. Given that the flop GRIA2 shows more rapid AMPA channel opening and faster glutamate desensitization than flip GRIA2 [48,49], our data suggest that HuD may be important in regulating sensitivity of this glutamate receptor through alternative splicing.
Finally, as shown in Tables S1–S5, none of the alternative splicing differences in HuD KOs resulted in alterations in the overall expression of these genes, including Ap4e1, Rapgef4, Cbx3, Cspp1, Snap25 and Gria2 (Figure 5). In comparison to this set, 432 genes exhibited significant differences in mRNA levels in HuD KOs (Table S7).