Since titin has been implicated in regulating sarcomere length and plays a prominent role in striated muscle force generation, we asked whether titin protein changes following Bmal1 knockout in skeletal muscle. Using sodium dodecyl sulfate-vertical agarose gel electrophoresis (SDS-VAGE) to analyze gross size changes in titin protein of iMSBmal1^-/- tibialis anterior (TA) muscle, we found a shift in titin isoform composition (p<0.01; Figure 1A and B) with no change in total titin expression. The shift in titin isoforms changes the predominant form from a short isoform to a mixture of both short and long titin proteins.

Titin mRNA is well known to undergo significant number of splicing events to generate its many isoforms (Guo et al., 2010). To gain insight into the changes in titin isoform following loss of Bmal1 in adult skeletal muscle, we performed RNA sequencing to identify exon-specific changes in iMSBmal1^-/- muscle (Guo et al., 2012; Schafer et al., 2015; Maatz et al., 2014). iMSBmal1^-/- TA muscle showed increased percent sliced in (PSI) of exons 52–88 of the titin transcript compared to iMSBmal1^+/+ control samples (Figure 1C). These exons code for a portion of titin’s proximal Ig segment, one of titin’s extensible segments within its spring region. Though the ∆PSI across this splicing event was stable, the variability in PSI across this event made statistical analysis difficult. To remedy this, we split this large splicing event into three, smaller splicing events (exons 52–69, 70–79, and 80–88) based on the similarity of PSI values within each sample (Figure 1C). In each of these smaller events, inclusion of exons was significantly greater in iMSBmal1^-/- muscle compared to iMSBmal1^+/+ muscle (Figure 1D; p<0.05). Thus, Bmal1 knockout in skeletal muscle altered splicing of titin’s I band localized spring region.

We next asked whether this difference in Ttn mRNA splicing resulted in detectable differences in titin protein. To date, no paper has provided side-by-side comparisons of RNA splicing and exon inclusion of titin protein. To address this gap in the field, we homogenized skeletal muscle from iMSBmal1^+/+ and iMSBmal1^-/- mice using a previously published method for analyzing MYBPC isoforms (O’Leary et al., 2019). Proteins were digested with trypsin, and peptide abundances were quantified by label-free LC-MS analyses. A total of 1218 peptides corresponding to the Ensembl ENSMUST00000099981.10 titin transcript were identified in the LC-MS analysis of which 1009 met our minimum threshold for quantification and were used to calculate exon-specific peptide abundance. These peptides corresponded to 225 of the 363 known exons in the Ttn gene (Figure 1—source data 2). We then analyzed differential expression of peptides and identified that those representing exons 52–69 were not detected in iMSBmal1^+/+ muscle and only identified in the iMSBmal1^-/- muscle samples. The abundance of peptides from the translation of exons 70–79 and 80–88 were 18 and 16% greater in the iMSBmal1^-/- compared to the iMSBmal1^+/+ muscle samples, respectively (p<0.05; Figure 1E). Notably, the only other region that displayed significant differences at the protein level is encoded by exons 112–217. This region is lowly expressed in both iMSBmal1^+/+ and iMSBmal1^-/- muscle, thus the difference in our measurement may be a result of low read numbers across this region in the RNAseq data. These results demonstrate that the difference in Ttn mRNA splicing encoding exons within the proximal Ig region in the iMSBmal1^-/- muscle is translated to titin protein. Our findings from these experiments also provide a unique analytical reference for the titin research community.

We have previously reported that the global Bmal1 knockout mouse displays sarcomere abnormalities in skeletal muscle (Andrews et al., 2010); however, these sarcomeric changes could be a result of a loss of skeletal muscle Bmal1, a loss of behavioral rhythms, and/or other systemic interactions. Since titin acts as a sarcomeric ruler and our observed titin splicing change is present in titin protein, we measured sarcomere length in iMSBmal1^-/- skeletal muscle. To determine whether iMSBmal1^-/- skeletal muscle presents with abnormal sarcomere morphology to accompany the weakness, we longitudinally cryosectioned TA muscles and measured sarcomere lengths as the distance between peak fluorescence following α-actinin immunostaining (Figure 2A). Using this approach, mean sarcomere length did not significantly change following skeletal muscle-specific loss of Bmal1 (p=0.86; Figure 2B). However, sarcomere length homogeneity is a feature of healthy striated muscle, and we noted that sarcomere length was significantly more variable in iMSBmal1^-/- muscle compared to iMSBmal1^+/+ muscle (F = 22.12, p<0.05; Figure 2B).

After defining the region of titin mRNA that changed in iMSBmal1^-/- muscle, we asked whether splicing in this region would be sufficient to account for the increased sarcomere length variability. To test this, we performed immunohistochemistry using epitope-specific antibodies against titin’s Z1Z2 and N2A domains because these epitopes flank the region with increased exon inclusion, the proximal Ig segment of titin. The positional difference between the two domains provides a proxy for proximal Ig domain length (Figure 3A). Representative images of iMSBmal1^+/+ and iMSBmal1^-/- muscle can be seen in Figure 3B and C, respectively. The significant difference in sarcomere length variability persisted as compared through an F-test of group variance though the sarcomere lengths themselves were longer than in our previous measurements. iMSBmal1^+/+ muscle had sarcomere lengths of 2.12 ± 0.01 µm while iMSBmal1^-/- muscle had sarcomere lengths of 2.28 ± 0.02 µm (F = 3.62, p<0.001; Figure 3D). Once we determined that sarcomere length variability was maintained, we asked whether proximal Ig domain length was significantly altered following Bmal1 knockout in adult skeletal muscle. iMSBmal1^-/- had significantly longer proximal Ig domain lengths (229.5 ± 1.357 nm) than the proximal Ig domain from iMSBmal1^+/+ samples (182.4 ± 1.388 nm; p<0.0001; Figure 3E). This difference in lengths accounts for the difference we see in sarcomere lengths in these samples. However, the variation around the mean for proximal Ig domain length was not significantly different (F = 1.338, p=0.12). These changes confirm that the proximal Ig domain length in adult skeletal muscle is correlated with a corresponding change in sarcomere length.

While the changes to sarcomere length of iMSBmal1^-/- TA muscle are associated with titin splicing and length, this model does not causally link titin splicing within the proximal Ig region to sarcomeric structure. Thus, we thought it necessary to manipulate titin splicing without the influence of the clock. To directly test whether the region of titin that is altered in iMSBmal1^-/- muscle could contribute to changes in sarcomere length independent of clock changes, we used U7 snRNPs to direct titin splicing from the 5′ splice site of titin exon 51 to the 3′ splice site of titin exon 89, resulting in shorter titin protein using an in vitro C2C12 myotube model. To visualize sarcomere length, we used a stable cell line in which α-actinin is labeled with GFP (eGFP-ACTN2-C2C12). These labeled myocytes were transfected with the titin-specific U7 snRNPs and plated on micropatterned gelatin hydrogels to support maturation of the myotube culture and sarcomere formation and alignment (Denes et al., 2019; Bettadapur et al., 2016). At day 10 of differentiation, eGFP-ACTN2-C2C12 myotubes were fixed then stained with a titin N2A epitope-specific antibody similarly to the tissue samples from iMSBmal1^-/- muscle (Figure 4A). Representative images of control eGFP-ACTN2-C2C12 myotubes and Ttn-U7 transfected eGFP-ACTN2-C2C12 myotubes can be seen in Figure 4B and C, respectively. While control myotubes had an average sarcomere length of 2.756 ± 0.0495 µm, Ttn-U7 transfected myotubes had significantly shorter sarcomeres with an average sarcomere length of 2.387 ± 0.0347 µm (p<0.0001, n = 100–200 sarcomeres/two biological replicates; Figure 4D). Because ACTN2 is a key component of the Z-line, the distance between the N2A-epitope and eGFP-ACTN2 was used to calculate proximal Ig length in this experiment. Proximal Ig domain lengths were also significantly different. Ttn-U7 transfected myotubes had a proximal Ig length of 285.1 ± 3.131 nm compared to 448.0 ± 3.368 nm of control myotubes (p<0.0001; Figure 4E). These results confirm that altering the length of titin’s proximal Ig domain causes a change in sarcomere length in skeletal muscle myotubes.

Having established that iMSBmal1^-/- skeletal muscle has an altered titin isoform composition resulting in sarcomere length heterogeneity, we looked to define the mechanism linking the skeletal muscle molecular clock to titin splicing. RBM20 is a well-characterized splicing factor known to target titin pre-mRNA (Guo et al., 2012; Li et al., 2013; Maatz et al., 2014). RBM20 is an RNA-binding protein that acts as a splicing repressor such that when present at high levels it binds along titin pre-mRNA preventing normal intronic splicing. When bound by many RBM20 proteins, titin pre-mRNA undergoes large exon-skipping events whereby the resultant, processed mRNA is shorter than if RBM20 was not present. We tested whether RBM20 levels change following loss of skeletal muscle Bmal1. Rbm20 mRNA expression was reduced by 34% in iMSBmal1^-/- muscle compared to iMSBmal1^+/+ muscle (p<0.05; Figure 5A). This reduced expression was also present at the protein level where RBM20 protein content decreased by 46% compared to iMSBmal1^+/+ muscle (p<0.05; Figure 5B).

We next used an environmental model to chronically disrupt the expression of the muscle clock to ask whether this was sufficient to change Rbm20 expression in muscle. C57BL/6J mice were subjected to 8 weeks of repeated phase advances, a well-established model of circadian disruption most commonly referred to as chronic jet lag. Prior studies using this model demonstrated significant disruption of circadian clock gene expression in skeletal muscle (Wolff et al., 2013). For this analysis, we included collections at two time points so that we could capture potential changes during either the rest (CT30) or active (CT42) parts of the day. In Figure 5D, we found that Rbm20 mRNA did exhibit a significant difference in expression in the quadriceps muscles under normal light:dark conditions. In contrast, muscles from the CPA mice exhibit depressed expression of Rbm20, mostly notably at CT30, during the subjective rest phase for the mice. The results for the TA muscle from the same mice follow a similar trend with reduced Rbm20 expression at both time points in the CPA group, but this change was not significant (p=0.09). The observation that Rbm20 mRNA expression was differentially expressed in the rest vs. active period of the quadriceps muscles was not expected but is consistent with the muscle clock regulating Rbm20 expression. In summary, these data indicate that Rbm20 expression is regulated downstream of the skeletal muscle circadian clock; this regulation is evident in two different muscles, and it provides further evidence that altering the muscle clock, either genetically or physiologically, leads to altered Rbm20 expression.

Since skeletal muscle Bmal1 knockout and physiological disruption of circadian rhythms reduced Rbm20 expression, we asked whether Rbm20 is a direct transcriptional target of the core clock factors in skeletal muscle. Our first approach was to interrogate our skeletal muscle ChIP-seq data sets for BMAL1 and CLOCK as these proteins comprise the positive limb of the skeletal muscle circadian clock (Gabriel et al., 2021). We also looked into our skeletal muscle ChIP-seq data set for MYOD1 as we have previously shown with Tcap that the clock factors can function synergistically with MYOD1 to transcriptionally regulate gene expression (Hodge et al., 2019; Cao et al., 2010). As seen in Figure 5C, we detected significant binding of BMAL1, CLOCK, and MYOD1 within a common region of intron 1 of Rbm20. Analysis of this region identified a putative regulatory element as defined by the third phase of the ENCODE project (Moore et al., 2020). To confirm binding of these proteins, we performed targeted ChIP-PCR using primers flanking this 88 bp site. We identified significant binding of BMAL1 (p<0.01), CLOCK (p<0.001), and MYOD1 (p<0.001; Figure 5F). To test whether this site served as an enhancer for Rbm20 transcription in muscle cells, we generated luciferase reporter constructs with either the Rbm20 proximal promoter alone (Rbm20-Luc) or the ~400 bp enhancer region from intron 1 cloned upstream of the Rbm20 promoter (E-Rbm20-Luc). We found that overexpression of the clock factors in E-Rbm20-Luc transfected myotubes resulted in a 255-fold increase in luciferase activity compared to empty vector control transfected myotubes (n = 3, p<0.0001, Figure 5G). Importantly, this increase is 9.8-fold higher than the increase seen in myotubes transfected with Rbm20-P-Luc, which lacks the enhancer element (p<0.0001). These data have identified a novel circadian clock-sensitive enhancer in the Rbm20 gene and support a model in which the skeletal muscle molecular clock modulates Rbm20 transcription through this intronic enhancer.

Since Rbm20 expression is regulated by the skeletal muscle molecular clock, we tested whether overexpression of RBM20 expression, using direct muscle injection of Rbm20-AAV, in the iMSBmal1^-/- muscle would be sufficient to restore titin splicing and protein uniformity (Figure 6A). Consistent with our prior measures, RBM20 protein levels decreased by 38% in iMSBmal1^-/- muscle compared to iMSBmal1^+/+ muscle injected with GFP-AAV (p<0.05). Injection of RBM20-AAV into iMSBmal1^-/- muscle increased Rbm20 gene expression by 2.8-fold (p_adj<0.05) and RBM20 protein expression almost 2-fold above iMSBmal1^+/+-GFP-AAV levels (p<0.05; Figure 6B). We performed SDS-VAGE with these muscles to determine titin isoform distribution. Consistent with earlier work, loss of muscle Bmal1 resulted in a significant increase in the long form of titin (p<0.01; Figure 6C). AAV expression of RBM20 in the iMSBmal1^-/- muscle was sufficient to significantly reduce the amount of the long titin isoform to levels closer to wildtype (p<0.05; Figure 6C). These findings demonstrate that the decrease in RBM20 expression downstream of loss of muscle Bmal1 contributes significantly to titin isoform heterogeneity in skeletal muscle.

While titin protein size is partially rescued in RBM20-overexpressing iMSBmal1^-/- muscle, whether or not the proximal Ig domain length of titin is restored remained undetermined. We performed RNAseq on iMSBmal1^-/--GFP and iMSBmal1^-/--RBM20 TA muscle to analyze the splicing of the titin transcript. iMSBmal1^-/--RBM20 muscle showed decreased expression of exons 70–88 of titin compared to iMSBmal1^-/--GFP muscle (Figure 6D). Importantly, these exons are increased in iMSBmal1^-/- muscle compared to iMSBmal1^+/+ muscle, suggesting that these exons are regulated downstream of the molecular clock through Rbm20 expression. We did note, however, that exons 52–69 showed only a modest increase in expression in muscle-overexpressing RBM20. This change was unexpected and is likely a result of other splice factors or mechanisms regulating inclusion/exclusion of those exons. These changes were confirmed using LC-MS with a significant decrease in exons 70–79 (p<0.001) and 80–88 (p<0.01) inclusion in iMSBmal1^-/--RBM20-AAV TA muscle with no significant different in inclusion of exons 52–69 in titin protein (Figure 6E). We used IHC to probe whether rescue of Rbm20 expression in the iMSBmal1^-/- muscle results in a change in the distance from the Z-disc to the PEVK region of titin. Longitudinal sections were labeled for these two epitopes (α-actinin/TTN-PEVK) demonstrated that RBM20 rescue with rescue of TTN-splicing resulted in a reduced distance of the PEVK domain from the Z-disc, consistent with a smaller titin isoform within the sarcomere (Figure 6—figure supplement 1). Unfortunately, we only had a small amount of tissue remaining, so our sample size is small, but these results are consistent with the levels of the level of exon inclusion in both the RNAseq and LC-MS data,.thus, demonstrating that splicing of titin is modulated downstream of the circadian clock via regulation of Rbm20 expression.

Why the molecular clock would regulate splicing of a protein as large as titin is puzzling. Estimations of titin turnover have been measured on the order of days, much longer than processes that would predictably be regulated daily by the circadian clock (Isaacs et al., 1989). However, studies in cardiomyocytes suggest that sarcomeric titin exhibits dynamic exchange within ~14 hr independent of protein turnover (da Silva Lopes et al., 2011). If such a mechanism exists in skeletal muscle, this would allow the fiber to modulate titin length on a more frequent, potentially daily basis. Previous studies have identified titin-cap (Tcap) as a direct clock-controlled gene that exhibits time-of-day variations in protein amount with highest levels during the active phase (Gregorio et al., 1998; Hammouda et al., 2012). TCAP functions to anchor the titin filament at the Z-line by binding titin’s Z1Z2 domain (Lee et al., 2006; Gregorio et al., 1998). These observations suggest a model in which titin protein could be exchanged daily, likely in sync with times at which TCAP protein is lowest. This daily replacement would serve to provide flexibility in the muscle fibers to adjust to patterns of use and disuse. Further, this daily replacement could lead to a vulnerable period for skeletal muscle damage (i.e., when TCAP is low and titin is being replaced). The potential for this period of vulnerability is somewhat supported by studies that measured diurnal variation in muscle damage following maximum performance tests in athletes. In these studies, muscle creatine kinase (Hammouda et al., 2012) and leukocyte infiltration (Hammouda et al., 2011) increased more substantially following evening exertion compared to morning exertion. In our proposed scenario, sarcomeric replacement during the inactive period would leave skeletal muscle susceptible to damage in the evening (i.e., leading into the inactive period), thus damage markers should increase at this stage. Studies that measure sarcomeric replacement over time of day with differing time of exercise bouts should be performed to directly test this scenario as they could prove useful for prescribing exercise as a therapeutic in cases of circadian rhythm disruption.

This study uniquely links the field of circadian rhythms with well-established properties of skeletal muscle structure and function through the RNA-binding protein RBM20. Since maintenance of circadian rhythms and skeletal muscle strength is gaining recognition for their roles in human health, understanding these processes is needed to promote health. Our work provides important new insights into understanding fundamental mechanisms regulating muscle homeostasis and potential new targets through which diseases that modulate the circadian clock could lead to skeletal muscle weakness.

The iMSBmal1 mouse model has been described (Schroder et al., 2015; Hodge et al., 2015). The Bmal1^lox mouse was purchased from Jackson Laboratories (stock number 7668) and crossed with the skeletal muscle-specific Cre-recombinase mouse (HSA-Cre) generated in-house (McCarthy et al., 2012). All experiments were performed in male mice housed in an AAALAC-regulated facility under 12 hr light–12 hr dark cycle with ad libitum food and water availability. Breeding for iMSBmal1^fl/fl mice was performed as described in Hodge et al., 2015, and recombination was induced in 12–16-week-old male mice through intraperitoneal vehicle (15% ethanol in sunflower seed oil) or tamoxifen injections (2 mg day^–1) for 5 days (Hodge et al., 2015). Five weeks after injections were completed, mice were sacrificed through cervical dislocation and tissues were collected. Tissues were flash-frozen in liquid nitrogen and stored at –80°C for RNA and protein analysis. All animal procedures were approved by the University of Kentucky’s and University of Florida’s Institutional Animal Care and Use Committees.

For the chronic phase advance (CPA) model, 12–16-week-old C57/Bl6J mice were individually housed in 12L:12D conditions in temperature-, humidity-, and light-controlled circadian cabinets (Actimetrics, Wilmette, IL). After 2 weeks of acclimation to individual housing, mice were randomly assigned to either control (12L:12D) conditions or a CPA regimen. The CPA paradigm consisted of a 6 hr advance in lights-on every 4 days for a total of 8 weeks. This protocol was chosen as our lab has previously shown that this protocol disrupts the core molecular clock rhythm in skeletal muscle (Wolff et al., 2013). Disruptions in activity rhythms were confirmed using wheel-running activity. At the end of the 8-week experimental protocol, mice were put into constant darkness then sacrificed under red light at either circadian time (CT) 30 or CT42. For reference, CT30 reflects a sample collected in the middle of the rest phase for the mice and CT42 is the middle of the active phase. Muscles were collected and snap-frozen in liquid nitrogen prior to processing.

TA muscles were pinned at slack length on a cork block, flash-frozen in OCT compound (Tissue-Tek) in liquid nitrogen-cooled isopentane, and stored at –80°C. Muscles were longitudinally sectioned at 6 μm and fixed with 2% paraformaldehyde (PFA). Sections were permeabilized using 0.1% Tween 20 in phosphate-buffered saline (PBS), blocked with bovine serum albumin (BSA), and incubated overnight at 4°C with primary antibodies. Tissue sections were washed with PBS, then incubated for 1 hr at room temperature with Alexa Fluor conjugated secondary antibodies (Thermo Fisher) prior to mounting with ProLong Gold (Molecular Probes).

Sections with only α-actinin were imaged using Leitz DMRBE microscope with a Leica DCF480 digital camera. Sarcomere lengths were measured using the 1D line tool within ImageJ as the distance between peak fluorescence signals. For titin-epitope-specific immunohistochemistry, sections were permeabilized using 0.1% Tween 20 in PBS, blocked with 10% BSA 5% goat serum in permeabilization buffer, and incubated overnight at 4°C for each of the primary antibodies. Secondary antibodies were added for 1 hr at room temperature between primary antibodies. Sections with titin-epitope-specific antibody staining were imaged using a Leica TCS-SP8 confocal microscope equipped with HyVolution deconvolution software. Measurements were made using Leica LASX software.

For RBM20-AAV rescue experiments, tissue sections were permeabilized with 0.5% Triton for 10 min, washed 3 × 5 min with TBS, incubated with Image-iT FX Signal Enhancer (Thermo Fisher) for 30 min, and blocked for 1 hr at room temperature with 5% BSA, 5% normal goat serum, 1% glycine, 0.1% Triton X-100 in TBS. Following blocking, sections were incubated overnight at 4°C with TTN-PEVK (rabbit, Myomedix, 1:500) and myomesin (mouse, DSHB, 1:1000). The following day, sections were washed with TBS-Tween (0.1%) before incubation with secondary antibodies (goat anti-mouse Alexa Fluor 488 1:500, goat anti-rabbit Alexa Fluor Plus 405 1:500) in blocking buffer. Following post-secondary wash (2 × 5 TBS-T, 1 × 5 TBS), sections were post-blocked with goat anti-rabbit unconjugated fab fragment antibodies (1:200, Jackson ImmunoResearch), before incubation overnight with α-actinin antibody (Abcam, rabbit, 1:1000) in blocking buffer. The following day, sections were washed before incubation with secondary antibody (goat anti-rabbit Alexa Fluor 647). After secondary incubation, sections were washed (2 × 5 TBS-T, 1 × 5 TBS, 1 × 5 Ultrapure Water) before being mounted with ProLong Glass Antifade. Sections were imaged using a Nikon Spinning Disk SR microscope CSU-X1 using ×100 oil-objective lens, standard zoom. Z-stacks (0.3 µm steps) were obtained and reconstructed in ImageJ for analysis.

TA muscles were pulverized under liquid nitrogen using a mortar and pestle to prevent fiber type and muscle region bias. Muscle powder was dissolved in sample buffer consisting of 8 M urea, 2 M thiourea, 0.05 M Tris base, 75 mM dithiothreitol (DTT), and 3% SDS at a 60:1 volume–muscle ratio to ensure titin solubilization (Warren et al., 2003). Protein concentrations were measured using the RC-DC protein assay (Bio-Rad) and stored at –80°C until use. Samples were resolved on a vertical agarose gel consisting of 1.0% w/v Seakem Gold Agarose (Lonza), 30% glycerol, 50 mM Tris base, 0.384 M glycine, 0.1% SDS run at 5 mA/gel for 30 min, then 7.5 mA/gel for 4 hr at 4°C (Warren et al., 2003). Titin protein was visualized using Sypro Ruby (Invitrogen) and imaged using a ChemiDoc MP System (Bio-Rad) at 450 nm. Bands were densitometrically analyzed using Bio-Rad ImageLab software as long T1 titin (upper), short T1 titin (lower), T2 titin (a major titin degradation product), and total titin (all bands combined).

RNA was isolated using procedures common to our lab for next-generation RNA sequencing (Schroder et al., 2015; Hodge et al., 2015; Terry et al., 2018). Briefly, muscle powder was homogenized in TRIzol (Invitrogen) according to the manufacturer’s directions, then isolated using QIAGEN RNeasy Mini Kit. RNA samples were treated with TURBO DNase (Ambion) to remove genomic DNA. RNA integrity numbers (RIN) were determined using the Agilent 2100 Bioanalyzer. RNA samples with RIN values >8 were used for sequencing library preparation. Then, 250 ng of total RNA was used to prepare barcoded RNAseq libraries using Stranded RNA-Seq Kit with RiboErase (KAPA Biosystems). Samples were sequenced on the HiSeq 2500 (Figure 2) or NovaSeq 6000 (Figure 6; Illumina) using a 2 × 100 kit to a read depth >45 million reads/sample. Reads were checked using FastQC before mapped to mm10 with HISAT2 (Kim et al., 2015). Gene expression data were collected using HTSeq and DESeq2 analysis packages (Love et al., 2014), and splicing analysis was performed using the protocol outlined by Schafer et al., 2015. Exon usage data is reported as PSI and change in PSI (∆PSI; iMSBmal1^-/- - iMSBmal1^+/+). Data have been deposited in NCBI’s Gene Expression Omnibus and are accessible using the GEO accession number GSE189865.

Our generation of eGFP-labeled α-actinin-2 C2C12 cells has been previously described (Denes et al., 2019). Briefly, C2C12 myoblasts obtained from ATCC were stably transfected with ACTN2-pEGFP gene fusion using the PiggyBac Transposon system and selected using puromycin and flow cytometry. Myoblasts were maintained on plastic cell culture dishes in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin in a humidified incubator kept at 37°C and 5% CO₂. When cells reached 70% confluency, they were separated and plated on micropatterned gelatin hydrogels. Gelatin hydrogels and PDMS stamps used for patterning were prepared as previously described (Denes et al., 2019; Bettadapur et al., 2016). Once 100% confluent, cells were serum-restricted with differentiation medium (DMEM, 2% horse serum, 1% penicillin-streptomycin) and left to differentiate for 10 days. Media were replenished each day, and all myoblasts were used at passage 8 or earlier. This cell line was validated using RNA deep sequencing (Denes et al., 2019) as well as through manufacturer certification and has tested mycoplasma negative.

Titin-specific, modified U7 snRNAs were generated using methods previously published to direct dystrophin splicing (Brun et al., 2003; Echevarría et al., 2018; Goyenvalle and Davies, 2011). The donor splice site on exon 51 and the acceptor splice site on exon 89 were targeted to induce the exon-skipping event studied through RNA sequencing of iMSBmal1^-/- and iMSBmal1^+/+ muscle. Sequences were generated to block these splice sites according to published standards and cloned into pAAVsc-CB6-PI-mCherry (Goyenvalle and Davies, 2011). Briefly, a base U7smOpt vector was generated from the wildtype U7 snRNA gene using published primers and QuikChange II Site-Directed Mutagenesis Kit. Forward- and reverse-antisense primers were then used to insert titin-specific sequences into the pAAV-U7smOpt vector with a free tail harboring canonical binding sites for hnRNP A1/A2 to increase exon-skipping efficiency (Goyenvalle and Davies, 2011). Each cloned plasmid was confirmed by Sanger sequencing. Then, 150 ng of each plasmid were transfected with X-tremeGENE 9 (Roche) in C2C12 myoblasts immediately after trypsinization (Escobedo and Koh, 2003). Day 10 myotubes were fixed in 2% PFA for 4 min, then immunostained and imaged using the same conditions as with cryosectioned TA muscle; however, the antibody against the Z1Z2 region of titin was not used due to insufficient penetrance into myotubes.

Total RNA from iMSBmal1^-/- and iMSBmal1^+/+ TA muscles was extracted using TRIzol Reagent (Ambion) and purified using the RNeasy Mini Kit (QIAGEN). Then, 500 ng total RNA was reverse-transcribed using SuperScript III first-strand cDNA synthesis system (Invitrogen). Primers were generated using previously published sequences for Rbm20 and Rpl26 (Hodge et al., 2019; van den Hoogenhof et al., 2018). qRT-PCR was performed on a QuantiStudio 3 thermal cycler (Applied Biosystems) from 10 ng of cDNA in a total volume of 20 μl containing Fast SYBR Green Master Mix (Applied Biosystems) and 400 nM of each primer. Rbm20 was normalized to corresponding Rpl26, and relative quantification was calculated using the ΔΔCT method. Gene expression is shown as the relative fold change of expression in iMSBmal1^-/- muscle compared to iMSBmal1^+/+ samples.

Small ~1–2 mg pieces of TA muscle were obtained from five iMSBmal1^+/+ and six iMSBmal1^-/- mice. The proteins were solubilized, reduced, alkylated, and digested to peptides using trypsin, as previously described (O’Leary et al., 2019). In brief, each piece of muscle was placed in a glass-bottom dissection chamber containing 150 μl 0.1% RapiGest SF Surfactant (Waters) and mechanically triturated with forceps. The solubilized proteins were reduced by addition of 0.75 μl 1 M DTT and heating (100°C, 10 min). Proteins were alkylated by addition of 22.5 μl of a 100 mM iodoacetamide (Acros Organics) in 50 mM ammonium bicarbonate (30 min, ~22°C, in the dark). The proteins were cleaved into tryptic peptides by addition of 25 μl of a 0.2 μg/μl trypsin (Promega) in 50 mM ammonium bicarbonate and incubation (18 hr, 37°C). The samples were dried down by centrifugal evaporation and reconstituted in 100 μl of a 7% formic acid in 50 mM ammonium bicarbonate solution to inactivate trypsin and degrade RapiGest (1 hr, 37°C). The samples were dried down once more and reconstituted in 100 μl 0.1% trifluoroacetic acid (TFA) for further cleavage of RapiGest (1 hr, 37°C). The samples were dried down a final time, reconstituted in 150 μl 0.1 TFA, centrifuged for 5 min at 18,800 RCF (Thermo, Sorvall Legend Micro 21R), and 125 μl of the supernatant were removed for analysis by mass spectrometry.

Tryptic peptides were separated via high-pressure liquid chromatography (LC) and analyzed by mass spectrometry (MS) as previously described (O’Leary et al., 2019). In brief, a 20 µl aliquot of each sample was injected into a Waters XSelect HSS T3 column using a Dionex UltiMate 3000 LC system. The effluent was directly infused into a Q Exactive Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo) through an electrospray ionization source. Data were collected as Xcalibur .raw files in a data-dependent MS/MS mode, with the five most abundant ions selected for fragmentation.

Titin peptides were identified from the Xcalibur .raw files and their LC-MS peak areas were quantified using Thermo Proteome Discoverer 2.2 (PD) software as previously described (O’Leary et al., 2019). The protein database used to identify the peptides was generated by translation of the Ensembl ENSMUST00000099981.10 titin transcript. This nucleotide sequence contained 347 of the possible 363 genomic exons. However, the exon numbering in the Ensembl database referred to the exon position in the transcript. To number the exons according to the genomic sequence, the Ensembl ENSMUST00000099981.10 nucleotide sequence was compared to the BN001114.1 entry in GenBank and numbered accordingly. This genomic exon numbering scheme is commonly used in the current literature (Guo et al., 2010). The peptides identified within the LC-MS data were then labeled in accordance with the genomic exon(s) from which they were translated.

To determine whether there were differences in the expression of peptides from exons 52–69, 70–79, and 80–88, peptide abundances were compared by a pairwise analysis. Peptides with minimal LC-MS peak areas, defined as having a median abundance of less than 1.5E7 units within the control group, were excluded to ensure the use of robust signal intensities. Each remaining LC-MS peak area was normalized by dividing its value by the average LC-MS peak area of the peptides coming from exons 256–363 in the sample. This region was selected for normalization because there is a lack of evidence for alternative splicing (Guo et al., 2010). The median normalized abundance of each peptide within the iMSBmal1^-/- and iMSBmal1^+/+ samples was determined. The average abundance and standard deviation of all of the peptides from each of the three exons were determined and plotted using ggplot2 in R.

Protein lysates were prepared by homogenizing approximately 20 μg of TA muscle from iMSBmal1^+/+ and iMSBmal1^-/- mice in RIPA buffer containing protease inhibitors using a Bullet Blender (Next Advance). Then, 40 μg of total protein were loaded into each well for SDS-PAGE (10%). Proteins were transferred with a semidry apparatus (Bio-Rad) onto PVDF membrane, then blotted overnight at 4°C with anti-RBM20 antibody (Myomedix) or anti-γ-tubulin (Sigma) antibodies. Secondary antibodies were applied at room temperature for 1 hr, and the blot was visualized using a ChemiDoc MP System (Bio-Rad). RBM20 bands were quantified with Image Lab software and normalized to the γ-tubulin loading control.

MYOD1, BMAL1, and CLOCK ChIP-seq data (accession numbers GSE122082 and GSE143334) recently published by our lab were searched for binding peaks in Rbm20 using UCSC Genome Browser (Gabriel et al., 2021; Hodge et al., 2019; Cao et al., 2010). Following identification of overlapping peaks, putative regulatory regions identified by the ENCODE Registry of cis-Regulatory Elements were scanned for canonical E-box sequences 5′-CANNTG-3′ (Moore et al., 2020). Primers flanking one of these E-box containing peaks in Rbm20 were designed and used for ChIP-PCR to confirm binding.

For ChIP-PCR, approximately 800 mg of adult gastrocnemius muscle was homogenized in cell lysis buffer (10 mM HEPES, pH 7.5, 10 mM MgCl₂, 60 mM KCl, 300 mM sucrose, 0.1 mM EDTA, pH 8.0, 0.1% Triton X-100, 1 mM DTT, protease inhibitors) and prepared as previously described (Hodge et al., 2019). After pre-clearing with Protein A/G beads, supernatant was split into equal parts and incubated overnight at 4°C with MYOD1 antibody, CLOCK antibody, BMAL1 antibody, or pre-immune serum prior to DNA recovery with a PCR purification kit (QIAGEN). qRT-PCR was performed with Rbm20 intron 1 primers. Data are presented as percent of input Ct, and significance was determined by comparing experimental ChIP values to those of the IgG control.

The promoter region of Rbm20 (–500 to +138) was cloned into pGL3-basic from mouse genomic DNA to generate Rbm20-P-Luc. The putative enhancer element (389 bp) was then cloned into Rbm20-P-Luc upstream of the promoter region. All cloned fragments were confirmed via Sanger sequencing.

Luciferase assays were performed from lysates of C2C12 myotubes (Andrews et al., 2010). In each reaction, 50 ng luciferase reporter and 150 ng expression vectors were transfected into C2C12 myoblasts in 24-well plates (50,000 cells per well) using X-tremeGENE 9 (Roche). Also, 5 ng modified pRL null plasmid (Hodge et al., 2019) was used as a transfection control. C2C12 myoblasts were transfected as previously described (Hodge et al., 2019). Luciferase activity was measured using the Promega Dual Luciferase assay system. Firefly luciferase activity was normalized to Renilla luciferase activity. Normalized luciferase activity for each transfection is plotted as average fold change relative to the pGEM empty vector control. Results were analyzed using a two-way ANOVA.

Mouse Rbm20 cDNA was cloned into pAAV-CMV with a C-terminal HA-tag and confirmed by Sanger sequencing. AAV was produced using transient transfection of HEK 293T cells and CsCl sedimentation by the University of Massachusetts Medical School Viral Vector Core as previously described (Sena-Esteves and Gao, 2020). Vector preparations were determined by ddPCR, and purity was assessed by SDS-PAGE and silver staining (Invitrogen). Expression of RBM20-HA was confirmed in C2C12 myoblasts via Western blot.

iMSBmal1^fl/fl mice were treated with tamoxifen or vehicle as described above. Four weeks after treatment, mice were injected with AAV into the TA muscle. AAV was diluted in sterile PBS, then 1 × 10¹² GC of RBM20-AAV or 0.5 × 10¹² GC of GFP-AAV were injected evenly throughout the entire TA of anesthetized mice. Muscles were collected 4 weeks after AAV injection and flash-frozen in liquid nitrogen for protein analyses.

Statistical analyses were performed using GraphPad Prism and R. Student’s t-tests were used to compare two groups with F-tests to compare variance between groups. One-way ANOVA with Tukey’s post-hoc analysis was used to detect significant differences in experiments with three groups. Values are reported as mean ± SEM. Statistical differences in the average abundances of peptides in LC-MS data were determined from the individual values for each peptide. To account for differences in the intrinsic ionization efficiency of peptides, the normalized peptide abundance values were subjected to a natural log transformation to produce a normally distributed array of values. The peptide values within each region were compared by Student’s t-test. To adjust for multiple comparisons, a Bonferroni correction (0.05/2 = 0.025) was used to determine the threshold p-value of 0.025 for the two proximal Ig regions compared. Figures were created using a combination of GraphPad Prism, ggplot2, and karyoploteR packages within R, and Adobe Illustrator. *p<0.05, **p<0.01, ***p<0.005, ****p<0.0001.