Although there is ample evidence for extensive regulation of AS in metazoans, there are conflicting views as to the extent of its biological significance (18–21
). By amassing a large validated data set of AS transitions that occur in response to myogenic differentiation, we characterized global splicing regulation directly and in a biologically relevant context. We found nearly 100 splicing transitions that were robustly regulated during differentiation, most AS events detected in both mouse and quail myoblasts underwent conserved transitions during differentiation, were dependent on the myogenic program, and can be grouped into four temporal patterns, demonstrating coordinated regulation. In addition, AS transitions initiate prior to the appearance of the earliest transcriptional markers of differentiation, indicative of links to myogenic differentiation program within proliferating myoblasts. These findings strongly support a biologically significant role for AS as an integral component of myogenic differentiation. This conclusion is further supported by results from mRNA-Seq analysis published while this manuscript was in revision, demonstrating extensive splicing changes during C2C12 differentiation (48
We found that more than half of the splicing transitions detectable in the QM7 quail cell line were conserved with mouse. This high level of conservation is particularly striking given the 300 million years of evolution that separates mammals and birds and the fact that not only is AS conserved, but also a regulated transition of splicing patterns during a shared developmental process. Genome-wide analysis using EST and full-length cDNA databases have detected low levels of AS conservation across species (21
). However, here we find that a high proportion of splicing transitions in myogenic differentiation are conserved, consistent with previous results (13
). Previously, we identified a conserved network of splicing transitions that occur during postnatal murine heart development (13
). It is interesting to note that 49 out of 78 (62.8%) of the splicing transitions that occur during postnatal heart development are regulated during C2C12 differentiation. Additionally, in both systems splicing transitions occurred in tightly grouped temporal waves, suggesting that the splicing regulatory programs in myoblast differentiation and heart development at least partially overlap. GO analysis of the regulated transcripts in C2C12 showed a strong enrichment in molecular function terms relating to muscle development and function, even when compared against a muscle specific background (D), consistent with other muscle splicing data sets (23
). Additionally, several of the regulated AS transitions we observed during C2C12 differentiation have already been experimentally shown to be important for normal muscle function (52–56
), further supporting the view that AS regulation is affecting processes that are biologically significant to muscle. A substantial subset of alternative exons that have been previously shown to be enriched in human (42
) and mouse (23
) muscle were found to undergo transitions in exon inclusion during C2C12 differentiation. Specifically, of the 56 muscle-enriched human exons described by Das et al
., 21 (37.5%) were orthologous to exons we observed to undergo splicing transitions during C2C12 differentiation. Consistent with our results, this study identified enrichment of Fox and CELF-binding sites within 200-nt downstream of muscle enriched exons. Sugnet et al.
reported 28 exons enriched in adult mouse skeletal muscle. Of these 28 exons, 12 (42.9%) were found to undergo regulated splicing transitions during C2C12 differentiation. However, it should be noted that both of these studies examined adult tissues, thus it is likely that many of the events they identified were excluded from our study because they do not undergo regulated changes in the level of alternative exon inclusion during C2C12 differentiation (despite being enriched in adult muscle). It is also important to note that we did not attempt to exhaustively identify all exons that undergo changes in inclusion levels during C2C12 differentiation, but instead our goal was to identify a large set of robust splicing transitions which we could then use for computational and molecular analysis, and it is likely that many splicing transitions that occur during C2C12 differentiation were not detected in our study.
To determine whether the splicing transitions are specific to the myogenic differentiation pathway or are a general response to cell-cycle withdrawal, we included BDM in the differentiation medium which allows induction of cell-cycle arrest but prevents progression of the differentiation program (46
). The results indicated that most of the splicing transitions tested are dependent on induction of the myogenic program and cannot be induced through serum deprivation, cellular confluence and/or cell-cycle exit alone. Time course experiments showed that ≥10 splicing transitions occur largely before the addition of differentiation media (Hour 0), suggesting that these events are regulated by signals associated with increasing cell density or cell-cycle exit. It is particularly interesting that for the majority of the early events tested, the transition initiated in proliferative cells was reversed when myogenic differentiation was blocked by using BDM treatment (Cii). These results indicate the presence of a signaling mechanism that links early differentiation-dependent splicing transitions with components of the myogenic program that initiate as proliferating myoblasts become confluent.
Two independent computational analyses to identify motifs associated with specific temporal patterns showed that specific sequence motifs are associated with distinct temporal splicing transition patterns. Both analyses identified hnRNP-L motifs (ACACA) within the proximal upstream intronic region of alternative regions as strongly associated with an early increase in alternative region inclusion ( and B). FOX-binding motifs were also identified in both analyses as strongly associated with increased inclusion when located within the proximal downstream intronic region of alternative regions. Such concordance of two independent computational analyses is strongly supports a position-dependent role for these two proteins in AS regulation.
Computational analysis demonstrated that motifs associated with known regulators of AS, including the FOX, CELF, PTB, MBNL and hnRNP-L families of splicing regulators, are enriched within the intronic regions surrounding the variably spliced regions that undergo regulation during differentiation. Many of these enriched binding motifs are also conserved with regard to sequence and relative location among eight mammalian species. Regression analysis predicted that FOX-binding upstream from alternative regions promotes skipping, and downstream promotes inclusion, which is in agreement with experimental evidence from other groups (57–59
). While FOX-binding sites are highly enriched near differentiation-induced splicing events, steady-state nuclear FOX2 levels remain relatively constant, while FOX1 levels modestly increase very early in C2C12 differentiation but remain constant thereafter (A). Regression analysis showed that FOX sites are most enriched in early AS changes, suggesting that the early increase in FOX1 steady-state levels could account for early AS transitions observed. However, the FOX2 transcript contains two exons which undergo regulated splicing changes of >20 percentage points during C2C12 differentiation, which could alter its activity independent of its steady-state levels, leading to another mechanism for the regulation of FOX activity. This possibility is supported by reports that show AS in the FOX1 and FOX2 transcripts result in alterations in splicing regulatory activity (11
MBNL family members show dynamic regulation in steady-state nuclear protein levels early in myogenic differentiation. Considering the large proportion of validated splicing transitions that show dramatic changes within the first 24
h of differentiation, these data indicate that MBNL family members are likely contributors to developmentally regulated myogenic AS. While MBNL-binding motifs were detected in our computational analysis, the high stringency of the analysis combined with the relatively high variability of MBNL-binding motifs likely led to their under-representation. Nevertheless, MBNL-binding motifs were still found to be significantly enriched (corrected P
) in the downstream distal intronic regions of regulated alternative region. Furthermore, the well characterized role that MBNL family members play in the pathogenesis of myotonic dystrophy suggest they are important for normal muscle development (29
). Overall, these data demonstrate that AS during myogenic differentiation is highly conserved, extensively regulated and suggests that splicing regulation is influenced by multiple regulatory factors associated with distinct temporal clusters of splicing transitions.