Characterization of a novel zebrafish model of SPEG-related centronuclear myopathy

Congenital myopathies are neuromuscular disorders that typically present at birth with hypotonia and weakness (Louis et al., 2022). The prevalence of congenital myopathies is at least 1 in 26,000 (Amburgey et al., 2011). A common congenital myopathy subtype is centronuclear myopathy (CNM), which is defined pathologically by perinuclear organelle disorganization and increased centralized myofiber nuclei, and is clinically associated with respiratory failure, wheelchair dependence and early mortality (Nance et al., 2012; Jungbluth et al., 2008). CNM is a genetic disease, caused by pathogenic variants in at least five genes: X-linked recessive CNM caused by myotubularin 1 (MTM1) (Laporte et al., 1996), autosomal-dominant CNM caused by dynamin 2 (DNM2) (Bitoun et al., 2005), and autosomal-recessive forms of CNM caused by bridging integrator 1 (BIN1) (Nicot et al., 2007), ryanodine receptor 1 (RYR1) (Wilmshurst et al., 2010) and striated muscle enriched protein kinase (SPEG) (Agrawal et al., 2014).

Most mutations reported in CNM patients are associated with primary or secondary defects of the excitation–contraction coupling (ECC) machinery (Gonorazky et al., 2018; Tasfaout et al., 2018; Jungbluth et al., 2008; Dowling et al., 2021). In skeletal muscle, ECC occurs at triads, which consist of centrally located sarcolemmal invaginations called transverse tubules (or T-tubules) flanked on either side by terminal cisternae of the sarcoplasmic membranes (tSR) (Fig. 1A). These junctional membranes are connected by a mechanical interaction between the CACNA1S subunit of the dihydropyridine receptor (DHPR) in the T-tubule membrane (with additional essential proteins including STAC3 and the DHPR β1a subunit) and the ryanodine receptor 1 (RyR1) Ca²⁺ release channels in the tSR. ECC is initiated when an action potential propagates down the T-tubule to cause voltage-driven conformational changes in DHPR, which then trigger activation of RyR1 to release Ca²⁺ stored in the tSR. The resulting surge in myoplasmic Ca²⁺ promotes actin–myosin cross bridging and sarcomere shortening (i.e. muscle contraction). Notably, although the DHPR/RyR1 interaction is essential for ECC function, the junctional membranes of the triad develop independently to the recruitment of DHPR or RyR1 to the region (as reviewed by Al-Qusairi and Laporte, 2011). The tSR membrane also contains structural/regulatory proteins such as triadin, junctin and calsequestrin (Zhang et al., 1997; Park et al., 2004), as well as other regulatory proteins that modulate RyR1 channel activity and maintain RyR1 integrity (Zhang et al., 1997; Costello et al., 1986; Treves et al., 2009; Ríos and Györke, 2009; Guo and Campbell, 1995; Wium et al., 2016; Caswell et al., 1999; Groh et al., 1999; Goonasekera et al., 2007). Although many of the molecular components of the triad are known, the precise molecular mechanisms that underlie triad formation and maintenance, as well as triad disruptions in CNM, remain largely elusive.

SPEG encodes a serine/threonine-specific protein kinase that belongs to the Obscurin/MLCK (also known as MYLK) family (Fleming et al., 2021). SPEG contains two serine/threonine kinase domains, two Fibronectin-Type III (Fn) domains and nine Immunoglobulin (Ig) domains (Fig. 1B). SPEG is predominantly expressed in striated muscles, but also found in the brain (Quick et al., 2017). Previous research suggests that SPEG interacts with proteins involved in the ECC pathway (Agrawal et al., 2014; Quick et al., 2017; Quan et al., 2019; Huntoon et al., 2018). SPEG is colocalized with RyR1 (Agrawal et al., 2014), and interacts with MTM1 (Agrawal et al., 2014) and desmin (DES) (Luo et al., 2020) (Fig. 1B). Bi-allelic variants in SPEG have been identified to cause CNM in a small number of families (Agrawal et al., 2014; Wang et al., 2017, 2018; Lornage et al., 2018; Qualls et al., 2019; Tang et al., 2019; Conlon et al., 2020; Almannai et al., 2021; Zhang et al., 2021) (Fig. 1B). Mutations span different regions of the SPEG gene and most are nonsense, resulting in decreased levels of the SPEG protein. Patients with SPEG mutations have skeletal muscle weakness as well as dilated cardiomyopathy (DCM). In severe cases, neonatal mortality has been reported (Agrawal et al., 2014; Liu et al., 2009; Wang et al., 2018; Qualls et al., 2019; Tang et al., 2019). At present, it is unclear how SPEG regulates muscle development, particularly triad formation and/or function, and why mutations in SPEG lead to CNM.

SPEG shares high sequence conservation across vertebrates. In humans, a single SPEG locus produces a ∼10.8 kb transcript that encodes a 3267 amino acid (a.a.) protein (NP_005867.3). In mice, the Speg locus includes two alternative transcription start sites that produce four distinct transcripts: aortic preferentially expressed gene (Apeg), brain preferentially expressed gene (Bpeg), Speg alpha and Speg beta (Hsieh et al., 2000). Of these, the two longest transcripts (Speg beta, 3262 a.a., NP_031489.4; Speg alpha, 2527 a.a., NP_001078839.1) encode the striated muscle isoforms and are well conserved with human SPEG. The striated muscle isoforms are involved in the maturation and differentiation of neonatal cardiomyocytes, while SPEGα is the predominant isoform involved in skeletal muscle differentiation (Liu et al., 2009; Hsieh et al., 2000). Zebrafish, however, possess two separate Speg genes, spega (Chromosome 6, NP_001007110.1) and spegb (Chromosome 9, XP_021334681), each encoding a single Speg transcript. The two zebrafish Speg proteins (Spega and Spegb) share high sequence conservation with human SPEG, and with both mouse SPEGα and SPEGβ isoforms.

In this study, we investigated the role of SPEG in skeletal muscle development by generating and characterizing multiple zebrafish Speg knockout (KO) models [spega-KO, spegb-KO and spega;spegb double KO (speg-DKO) zebrafish]. A main advantage of the zebrafish model is that zebrafish larvae can survive up to 7 days post-fertilization (dpf) without a functional heart (Pelster and Burggren, 1996; Stainier, 2001), allowing for skeletal muscle studies even in the setting of impaired cardiac development. We established zebrafish Speg KOs using CRISPR/Cas9-directed mutagenesis and confirmed that the resulting mutants exhibited reduced SPEG expression. Phenotypic characterization revealed that speg-DKO mutants effectively recapitulate phenotypes reported in SPEG-related CNM patients, including abnormal triad structure, disrupted calcium dynamics and impaired skeletal muscle function. Furthermore, similar to other CNM subtypes, we found that SPEG deficiency results in DES accumulation and DNM2 upregulation. In total, we established a new zebrafish model suitable for defining the biological role of SPEG in skeletal muscle and for identifying therapies for SPEG-related CNM.

There are two Speg genes in zebrafish, spega and spegb. To study tissue expression of spega and spegb during development, we performed quantitative real-time PCR (RT-qPCR) on total RNAs extracted from the head region (brain and predominantly cardiac muscle) or tail region (skeletal muscle) of 2 dpf and 7 dpf zebrafish wild-type (WT) embryos. We detected similar temporal expression pattern of spega (Fig. 2A) and spegb (Fig. 2B), with significant upregulation observed between 2 dpf and 7 dpf in the head region. We performed whole-mount in situ hybridization to compare spatial expression patterns. We observed spega expression in the brain and developing neural tube, but not in the notochord or somites. In contrast, spegb expression was predominantly detected in the chevron-shaped somites (Fig. 2C). These data show that both spega and spegb are expressed from early development, and that zebrafish skeletal muscle predominantly expresses spegb.

To characterize the subcellular localization of SPEG in muscle during development, we stained isolated zebrafish myofibers with an anti-Speg antibody that recognizes both zebrafish Spega and Spegb proteins. We observed, by confocal imaging, that Speg is localized to sarcolemmal and perinuclear regions at 2 dpf, with little transverse striations or colocalization with the tSR marker RyR1 (Fig. 2D,D′). However, at 5 dpf, Speg loses its early sarcolemmal/perinuclear localization, and instead shows robust transverse colocalization with RyR1 at the triads (Fig. 2E,E′). These results indicate that Speg translocates to the triad junction during muscle development shortly after initial triad formation.

To study SPEG function during development, we used CRISPR/Cas9 to target the two zebrafish Speg genes, generating spega or spegb knockout (KO) lines. We designed guide RNAs (gRNAs) and tested their in vivo cutting efficiency using high-resolution melting (HRM) analysis (Fig. S1). The most efficient gRNAs were chosen to generate single mutant lines, with double mutant lines later generated by intercrossing F2 single mutants (Fig. 3A). Sanger sequencing confirmed the genotype of one spega mutant as carrying a 10 bp deletion in exon 27 (i.e. spegaΔ10), and the genotype of one spegb mutant as carrying a 17 bp deletion in exon 26 (i.e. spegbΔ17), both of which are predicted to introduce premature stop codons in the inter-kinase region (Fig. 3B). To examine the level of Speg transcripts, we performed RT-qPCR on total RNA extracted from single mutants, double mutants and their WT siblings. Although no significant transcript reductions were detected in the single mutants (Fig. S2), we observed significantly decreased spega (∼50% of WT) and spegb (∼40% of WT) in spegaΔ10;spegbΔ17 double mutants (Fig. 3C). Owing to lack of anti-Speg antibodies suitable for western blot analysis in zebrafish, we quantified relative protein levels in skeletal muscle by measuring the intensity (gray values) of immunostaining in isolated myofibers (Fig. 3D). We observed that anti-Speg staining intensity was significantly reduced by ∼70% in spegaΔ10;spegbΔ17 myofibers compared to WT myofibers, confirming the double mutant zebrafish to be a Speg-deficient zebrafish line, or speg-DKO.

We generated single and double knockouts from carrier crosses, and analyzed embryos at the F3 generation and beyond. With the exception of deflated swim bladders in the speg-DKO zebrafish, no obvious morphological abnormalities were noted, with embryos and larvae appearing of normal size and without clear bend or curvature to the body shape (Fig. 3E). However, survival was significantly reduced. Using Kaplan–Meier based methodology, we observed that spegbΔ17 and spegaΔ10;spegbΔ17 (or speg-DKO) died at a median age of 10 dpf, but spegaΔ10 survived to adulthood and could successfully reproduce (Fig. 3F).

To validate the specificity of the phenotypes that we observed, we generated additional spega and spegb mutant lines, and examined the survival of spegaΔ5;spegbΔ8, another Speg double mutant line. Specifically, we created lines that induce premature stop codons in spega (carrying a 5-bp deletion in exon 5, spegaΔ5) and spegb (carrying an 8-bp deletion in exon 8, spegbΔ8). These lines were identical in appearance to spegaΔ10 and spegbΔ17, and we observed similar reduced survival rates in spegbΔ8 and spegaΔ5;spegbΔ8 (Fig. S3). These data confirm that the phenotypes we describe are due to spega/b mutations and not to off-target effects. We focused the majority of the remaining analyses on speg-DKO zebrafish, and specifically spegaΔ10;spegbΔ17, as they best genetically and phenotypically model the human disease.

Patients with SPEG mutations manifest both skeletal muscle and cardiac (e.g. dilated cardiomyopathy) abnormalities, and heart dysfunction could explain the early lethality seen in our spega/b zebrafish. Therefore, we examined speg-DKO embryos and larvae for signs of cardiac abnormalities. One change associated with cardiomyopathy in zebrafish is pericardial edema, caused by disrupted blood flow and decreased contractile function (Huttner et al., 2018). No evidence of pericardial edema was observed in spega/b zebrafish. We also did not detect any obvious alterations in heart rate or cardiac rhythm in speg-DKO zebrafish. Therefore, no clear cardiac abnormalities were present in speg-DKO zebrafish (possibly due to residual Speg protein), suggesting that the early lethality is due to skeletal muscle dysfunction.

To determine whether spega/b deficiency in zebrafish leads to skeletal muscle phenotypes that mirror human CNM, we first evaluated the expression pattern of key triad proteins using immunofluorescence (IF) in myofibers isolated from 2 dpf (Fig. 4A,B,C,D) and 5 dpf (Fig. 4A′,B′,C′,D′) WT versus speg-DKOs. In WT myofibers, both RyR1 and DHPR formed transverse striations (i.e. triads), as did the sarcomeric Z-disk protein α-Actinin, while SERCA1 formed both transverse and longitudinal striations reflective of the sarcoplasmic reticulum (SR) membrane network. In speg-DKO myofibers, however, we observed fragmentation of this striated RyR1 and DHPR pattern, with occasional RyR1 mislocalization to the sarcolemma (Fig. 4A), consistent with abnormal early triad development and/or RyR1/DHPR decoupling. To further dissect this, we performed electron microscopy (EM) analyses and found that triad density (the number of triads per 60 µm²) decreased significantly in myofibers from speg-DKO zebrafish (Fig. 4F,F′,G), with many triads appearing structurally abnormal (Fig. 4E′). These findings indicate that spega/b deficiency leads to triad loss/abnormality. Interestingly, transverse, but not longitudinal, SERCA staining is depleted in speg-DKO myofibers from 5 dpf speg-DKO zebrafish (Fig. 4C,C′), while α-Actinin staining remained normally striated (Fig. 4D,D′), consistent with the normal overall sarcomeric organization observed under EM (Fig. 4E,F). Together, these results indicate that SPEG is essential for triad formation/stability in skeletal muscle.

We then set out to examine the effects of speg-DKO on ECC and RyR1 function. ECC was quantified in single myofibers by measuring intracellular calcium dynamics during electrical stimulation and ligand-induced RyR1 Ca²⁺ release was assessed following local caffeine application. In brief, single myofibers were isolated from genotyped larvae at 7 dpf, loaded with Ca²⁺-sensitive dye (fluo-4-AM) (Fig. 5A) and then exposed to a series of electric stimulations (1 Hz, Fig. 5B; 10 Hz, Fig. 5C,D), followed by 30 s application of 10 mM caffeine (Fig. 5E). Peak changes of fluo-4-AM fluorescence (myoplasmic-free Ca²⁺) were recorded and normalized to background (ΔRatio). Myofibers from both spegb-KO (bKO) and spega/b-DKO (DKO) zebrafish exhibited significantly reduced electrically evoked and caffeine-induced Ca²⁺ release (∼50% reduction in ΔRatio) compared to that observed for myofibers from WT zebrafish (Fig. 5B-E). These results indicate that spega/b deficiency results in reduced ECC and RyR1 function in skeletal muscle, consistent with the observed alterations in T-tubules and triads.

To examine whether speg-DKO alters motor performance, we conducted whole-zebrafish swimming assays using ZebraBox (Viewpoint, France) (Fig. 5F). As early as 3 dpf, speg-DKO zebrafish traveled significantly less distance (29-49% reduction) compared to their WT siblings (Fig. 5G). By 7 dpf, speg-DKO zebrafish had a nearly complete absence of movement. Of note, spegb mutants also demonstrated abnormal swim behavior, consistent with spegb representing the primary functionally relevant SPEG paralog in skeletal muscle.

Previous work identified DES (Luo et al., 2020) and MTM1 (Agrawal et al., 2014) as SPEG-binding partners in skeletal muscle, and aberrant DES aggregation in Mtm1 KO mice (Hnia et al., 2011; Luo et al., 2020). We thus compared Desmin expression in a series of CNM zebrafish models at stages at which muscle phenotypes are fully penetrant: (1) speg-DKO at 5-7 dpf (Figs 3–5), (2) mtm1-KO at 7 dpf (Sabha et al., 2016) and (3) DNM2-S619L-eGFP transgenics at 3 dpf (Zhao et al., 2019). To examine Desmin localization, we stained isolated myofibers with an anti-Desmin antibody that picks up both Desmin-a and Desmin-b in zebrafish (Kayman Kürekçi et al., 2021). In myofibers from WT zebrafish, Desmin formed transverse striations marking the sarcomeric Z-disks, and was also localized around the nucleus and below the sarcolemma (Fig. 6A,D). In both speg-DKO (Fig. 6B) and DNM2-S619L (Fig. 6E), this clear striated pattern was lost, and abnormal accumulation was observed longitudinally within the middle of the myofiber and below the sarcolemma. In mtm1-KO, mild accumulation of Desmin was observed longitudinally within the middle of the myofiber without obvious loss of transverse striations (Fig. 6C).

To determine whether overall DES expression levels are altered, we performed western blot analysis on whole-zebrafish protein lysates. In WT sibling controls (5 dpf, Fig. 6F; 7 dpf, Fig. 6G), we observed two Desmin bands (an upper band ∼55 kDa and a lower band ∼50 kDa), consistent with a previous report in adult zebrafish muscle (Kayman Kürekçi et al., 2021). In the mutants, we detected significant Desmin upregulation in both 5 dpf speg-DKO zebrafish (2- to 2.5-fold, Fig. 6F′) and 7 dpf mtm1-KO zebrafish (9- to 12-fold, Fig. 6G′) compared to WT controls. Conversely, no difference in Desmin expression was detected between 3 dpf DNM2-WT and DNM2-S619L zebrafish (Fig. 6H,H′). Of note, only a single Desmin band at ∼55 kDa was observed in the DNM2 experiments, most likely due to the earlier developmental time point (3 dpf) used in these studies. In summary, increased longitudinal Desmin accumulation was observed across all three models, while increased Desmin expression was observed only with 5 dpf speg-DKO and 7 dpf mtm1-KO zebrafish.

Upregulation of endogenous DNM2 protein is a common phenotype observed in non-DNM2-related CNM models including mice lacking Mtm1 (Cowling et al., 2014) and Bin1 (Cowling et al., 2017). Thus, we evaluated Dnm2 protein expression in our speg-DKO zebrafish and mtm1-KO zebrafish. We first studied Dnm2 localization by performing IF on isolated myofibers using an antibody specifically against zebrafish Dnm2. In WT myofibers (5 dpf), Dnm2 staining appeared in transverse striations (Fig. 7A). In speg-DKO (Fig. 7B) and mtm1-KO (Fig. 7C), although Dnm2 staining formed similar striated patterns, occasional Dnm2 aggregates could be observed along the striations. However, these patterns were distinct in appearance and localization from the large sarcolemmal DNM2-EGFP aggregates previously observed with DNM2-S619L overexpression (Zhao et al., 2019). To examine Dnm2 protein levels, we performed western blot analysis in 5 dpf speg-DKO (Fig. 7D) and 7 dpf mtm1-KO (Fig. 7E) zebrafish using whole-zebrafish protein lysates. We detected two Dnm2 bands at ∼100 kDa in WT siblings, with a significant (∼2- to 3-fold) increase in the upper band in both speg-DKO (Fig. 7D,D′) and mtm1-KO zebrafish (Fig. 7E,E′), consistent with DNM2 upregulation in SPEG-CNM and MTM1-CNM. Together, our results demonstrate that DNM2 is upregulated in SPEG-CNM and MTM1-CNM in the absence of the detectable sarcolemmal DNM2 aggregation seen with DNM2-CNM.

SPEG is a novel CNM-causing gene and the only known kinase associated with CNM. The function of SPEG in skeletal muscle remains unclear. In this study, we provide the first characterization of spega/b in zebrafish. We define the developmental expression of spega/b mRNA, as well as the subcellular localization of Speg protein in skeletal muscle fibers. We generated single and double knockouts of zebrafish Speg paralogs, and show that the double knockout zebrafish exhibit molecular, pathologic and behavioral phenotypes consistent with the human disease. We also examined potential SPEG-CNM pathomechanisms and therapeutic targets, and found that, as with other forms of CNM, SPEG deficiency is associated with both DES accumulation and DNM2 upregulation. In all, we establish a new model of SPEG-related CNM that will be ideal for studying disease mechanisms and identifying therapies.

There are two Speg genes in zebrafish, spega and spegb. Although both share high sequence conservation with human SPEG and are expressed from early zebrafish development, our in situ hybridization data demonstrate that the zebrafish skeletal muscles predominantly express spegb. This is further supported by our loss-of-function data, whereby spegb single KOs showed similar triad and Ca²⁺ transient defects as spega/b double knockouts, and also had reduced survival. However, we only detected significantly reduced swimming distance in spega/b double knockouts, suggesting that the lack of both paralogs is required to induce the full muscle phenotype in zebrafish that mirrors that of SPEG-CNM patients.

Insights into the roles of SPEG during muscle development can be inferred from known SPEG-interacting partners identified by yeast two-hybrid and validated in subsequent co-immunoprecipitation experiments, such as DES (Luo et al., 2020) and MTM1 (Agrawal et al., 2014). DES is a major intermediate filament protein in muscle that promotes muscle integrity by forming dynamic filamentous networks crosslinking the sarcolemma, sarcomeres, and mitochondria (Paulin and Li, 2004). Emerging evidence has shown that DES is an important CNM disease marker. Abnormal DES accumulation is observed in muscle biopsies from MTM1-CNM and DNM2-CNM patients (Romero and Bitoun, 2011). SPEG-deficient mice (Luo et al., 2020), MTM1-deficient mice (Hnia et al., 2011), and our speg-DKO, mtm1-KO and DNM2-S619L zebrafish all show abnormal DES accumulation and/or overall DES upregulation. Although the cause of DES accumulation remains unknown, reduced/altered DES phosphorylation disrupts intermediate filament disassembly and promotes DES accumulation (Winter et al., 2013). Thus, it is tempting to speculate that SPEG regulates DES phosphorylation and that lack of SPEG leads to DES accumulation because of reduced phosphorylation. Future experimentation will be required to test this potential link, and more generally to define substrates of the kinase activity of SPEG.

The biological consequences of DES accumulation in SPEG-CNM remain unclear. Notably, DES accumulation has been associated with mitochondrial defects in Mtm1 mouse muscles (Hnia et al., 2011), and mitochondrial defects along with triad disruption are a key aspect of the CNM disease process (Lawlor et al., 2016; Muñoz et al., 2020; Zanoteli et al., 2009). It will thus be of interest in the future to examine mitochondrial dynamics in SPEG knockouts. Notably, DES accumulation is also present in desminopathies, a subset of myofibrillar myopathies caused by autosomal-dominant DES mutations, which have neither triad defects nor other histopathologic features of CNM, indicating that DES accumulation is unlikely a direct cause for the triad defects observed in CNM. Furthermore, in contrast to desminopathies, we did not observe myofibril dissolution (as indicated by normal α-Actinin striations) in our CNM zebrafish models, further supporting distinct disease pathways of the two myopathies, consistent with the clinical and histopathological differences between CNM and desminopathy patients.

The triad defects observed in SPEG-CNM are potentially due to disrupted SPEG/MTM1 interaction. MTM1, DNM2 and BIN1 all encode components of the endocytic machinery. Because mutations in each of these genes cause CNM with defective triads, membrane trafficking is considered a key aspect of triadogenesis (Dowling et al., 2008). One theory on triad biogenesis is that T-tubule invagination is first initiated by membrane deformation promoted by BIN1 and then completed by DNM2-mediated (Nicot et al., 2007) membrane fission (Cowling et al., 2017; Picas et al., 2014). However, the precise roles of SPEG and MTM1 {a phosphatase that acts to dephosphorylate 3-position phosphoinositides [PI(3)P and PI(3,5)P2]} in this pathway are unclear. In Mtm1-KO and Bin1-KO mice, the level of Dnm2 protein is upregulated (Cowling et al., 2014, 2017). In our mtm1-KO and speg-DKO zebrafish, we observed similar increases in Dnm2 protein expression, demonstrating that DNM2 is also upregulated in SPEG-CNM. SPEG/MTM1 interaction therefore may regulate triad biogenesis by regulating DNM2 levels and/or activity. The mechanisms leading to DNM2 upregulation in SPEG-CNM (this study), BIN1-CNM (Cowling et al., 2017) and MTM1-CNM (Cowling et al., 2014) remain unclear, although lack of BIN1 inhibition (Cowling et al., 2017), dysregulated microRNA (Liu et al., 2011; Chen et al., 2020) and altered post-translational modifications (e.g. SPEG phosphorylation) (Kusić et al., 2020) have been proposed. Moreover, abnormal DNM2 aggregates are observed (Chin et al., 2015; Zhao et al., 2019) or suggested (Muñoz et al., 2020) in DNM2-CNM caused by hyperactive DNM2 mutations. However, unlike DNM2-S619L zebrafish (which model DNM2-CNM), we did not observe sarcolemmal Dnm2 aggregates in myofibers from speg-DKO or mtm1-KO zebrafish, indicating a level of variability in DNM2 abnormalities across different CNM subtypes. Interestingly, DNM2 reduction has been used as an effective strategy in treating muscle phenotypes of MTM1-CNM and BIN1-CNM mouse models (Tasfaout et al., 2018). It will thus be important to test reducing DNM2 levels (e.g. via genetic ablation, antisense oligonucleotides or manipulating upstream interactions/events) in models of SPEG-CNM.

Given that the subcellular localization of SPEG changes rapidly from being perinuclear/sarcolemmal at 2 dpf to located within the triad junction at 5 dpf, SPEG may play distinct roles in developing versus mature muscles. The colocalization of SPEG/RyR1 in mature myofibers suggests that SPEG and RyR1 may interact (similar to SPEG/RyR2 interaction in cardiac muscles) (Quick et al., 2017), such that SPEG promotes phosphorylation events that modulate RyR1 channel function/stability (Jungbluth et al., 2018; Witherspoon and Meilleur, 2016). Of note, it remains unclear how SPEG is recruited to each cellular compartment. Some of these questions could be addressed in future studies that compare RyR1 phosphorylation/interactomes in developing versus mature muscles from WT and speg-DKO zebrafish. Lastly, in the absence of any effective treatments for SPEG-CNM, our speg-DKO zebrafish will enable large-scale chemical screens to identify novel small-molecules/pathways to treat the devastating disease.

In conclusion, we describe the generation and characterization of the first zebrafish model of SPEG-CNM. We demonstrate that speg-DKO zebrafish faithfully recapitulate multiple features of SPEG-CNM, and identify changes in DES and DNM2 expression/localization as critical (and conserved) disease markers that can be dissected in future studies.

Zebrafish (AB strain) were raised and maintained at 28.5°C at the Zebrafish Facility at the Hospital for Sick Children, Toronto, ON, Canada. Experiments were performed on zebrafish embryos and larvae from the one-cell stage up to 15 dpf. All zebrafish procedures were performed in strict accordance with the Animals for Research Act of Ontario and the Guidelines of the Canadian Council on Animal Care.

Total mRNA was isolated from 2 dpf or 7 dpf zebrafish homogenates using an RNeasy Mini Kit (Qiagen, 74104) and reverse-transcribed with SuperScript VILO (Invitrogen, 1755050). Approximately 10-25 embryos were collected for each condition, in triplicates. Head and trunk dissections were done to cut at the region indicated at the diagram (Fig. S4). RT-qPCR was performed using Platinum SYBR Green reagent (Invitrogen, 11744500) and a Step-One-Plus Real-Time PCR System (Applied Biosystems). All reactions were performed in technical triplicates and the results represent biological triplicates. The zebrafish peptidylprolyl isomerase B gene, ppib, was used as the endogenous control. Primers used are as follows: spega forward 5′-ACAAAGAGATTGGCAGAGGGG-3′, reverse 5′-ACTCTCGCAATGCACAAGTC-3′; spegb forward 5′-CAACAACAAGTACGGCAGCG-3′, reverse 5′-TGCAAATCGAGGAGTCTCGC-3′; and ppib forward 5′-ACCCAAAGTCACGGCTAAGG-3′, reverse 5′-CTGTGGTTTTAGGCACGGTC-3′. Protocol conditions for RT-qPCR were the following: denaturation at 95°C for 10 min; followed by 40 cycles of 95°C for 15 s and 60°C for 1 min; and one melt curve, i.e. 95°C for 15 s, 60°C for 1 min and 95°C for 15 s.

Probes were generated via PCR amplification from 2 dpf total cDNA synthesis using SuperScript VILO (Invitrogen, 11755050). Primers used were as follows: spega forward 5′-AAGAAGCAAGCTCACCCACA-3′, reverse 5′-AAGTCAAGGTCTGTCGACGC-3′; spegb forward 5′-CGAAACTCACACGGGGAAGA-3′, reverse 5′-GACTGTGATGCTCAAGGGCT-3′. Digoxigenin (DIG)-labeled in-situ probes were synthesized using DIG RNA Labeling Kits (Roche, 11277073910). RNA in-situ hybridization was carried out as previously described (Thisse and Thisse, 2007). Briefly, embryos were fixed in 4% paraformaldehyde (PFA) and then dehydrated in 100% methanol. Embryos were then permeabilized and incubated with DIG-labeled antisense RNA probes at a final concentration of 300 ng/200 µl in hybridization solution. Hybridizations of the probe with the RNA were detected with an alkaline phosphatase-conjugated antibody (1:5000; anti-DIG-AP, Fab Fragments, Roche, 11093274910). Finally, stained embryos were cleared overnight in a 70% glycerol solution with 30% PBSTw (0.1% Tween 20 in PBS) and imaged under a Leica M205FA stereomicroscope.

The program Chopchop (http://chopchop.cbu.uib.no/) (Montague et al., 2014) was used to design each of the gRNAs used in this project. According to Chopchop, there were no predicted off-targets for the gRNAs tested. Next, 50-100 one-cell-stage WT embryos were injected with the gRNA (150 pg per embryo) and Cas9 mRNA (100 pg per embryo) with a Picopump (World Precision Instruments). Strong gRNAs were identified by isolating genomic DNA from 24 individual injected embryos and three uninjected embryos at 3-5 dpf. DNA was digested with 1 µg/µl Proteinase K, and all embryos were genotyped using HRM analysis and Sanger Sequencing. HRM analysis was performed on a Roche Lightcycler 96. Once we identified gRNAs that were cutting at the desired genomic region, potential founders (F0) were outcrossed to WT AB zebrafish. In-cross progeny from the F3 and F4 generations were used for the characterization of the spega/b mutant phenotype. The targets in this study were: spega 5′-GATGGACAACCCTGCCAAAGG-3′ at exon 5, spega 5′-CTATCGACCTGGACCTGTAGG-3′ at exon 27, spegb 5′-GCTCCATATGACTTGAGGCGG-3′ at exon 8, and spegb 5′-ATGGCTCTCGCTTAGGGGAGG-3′ at exon 26. Primers used for HRM were as follows: spegaΔ5 at exon 5 forward 5′-GAAGACTGAAGAAACTGGAAAGCA-3′, reverse 5′-GCTGTTTTGTTTTATCTGCCAGG-3′; spegaΔ10 at exon 27 forward 5′-AAGAAAGCTCACCGGTTCCC-3′, reverse 5′-TGGACAGACTTGGATTTTTCCT-3′; spegbΔ8 at exon 8 forward 5′-CGAGGGTCGTAATGCACGAT-3′, reverse 5′-TACAGTACAGCACTCTGGGGA-3′; and spegbΔ17 at exon 26 forward 5′-CCCTCCCAAAGAGCCAAGTC-3′, reverse 5′-GCGAGCTTCAAAAACCTCCT-3′. Primers used for PCR prior to Sanger sequencing were as follows: spegaΔ5 forward 5′-CGACAAGCAGTCCAGTGTCA-3′, reverse 5′-TGGGGCTTTCACCAAACCAT-3′; spegaΔ10 forward 5′-ACACCATTACCGACACCAGTT-3′, reverse 5′-TACGTCGCACTGCAAGGAC-3′; spegbΔ8 forward 5′-AGCATCCATAGAGCCCCTTT-3′, reverse 5′-CCGTGTAGAGGCCCTCATCT-3′; and spegbΔ17 forward 5′-TCTTTTCTCGGGTTGCCTCC-3′, reverse 5′-GAGAGTCGCCTCATGAACCC-3′.

Fish were examined daily for morphological abnormalities, and imaged at 6 dpf using a Zeiss Axio Zoom V16 microscope (16×, 10 ms exposure). Survival analysis was performed using Kaplan–Meier methods. Briefly, 3-5 dpf larvae were fin clipped and then grouped by genotype after HRM analysis. Survival counts and health checks were performed daily until the fish reached adulthood. The number of live embryos was plotted across time and analyzed using Prism version 8 (GraphPad Software).

Skeletal myofibers from WT and speg-KO zebrafish were isolated as previously described (Horstick et al., 2013). Briefly, embryos at each respective time point were digested with collagenase type II (LS004176, Worthington Biochemical Corporation) and plated on 12 mm circular coverslips. Samples were then fixed with 4% PFA for 20 min at room temperature or 100% methanol for 10 min at 4°C, permeabilized with 1× PBSTw (0.1% Tween 20 in PBS), incubated with blocking solution [0.2% Triton X-100, 0.2% bovine serum albumin (BSA), 5% goat serum in PBS] for 30 min to 1 h at room temperature, and then incubated with primary antibodies overnight at 4°C. Samples were stained in the dark with secondary antibodies for 1 h at room temperature and mounted with ProLong Gold Antifade Mountant (Invitrogen). Sample slides were dried at room temperature overnight in the dark and stored at 4°C until imaging. Images were taken using a Leica SP8 Lightning Confocal microscope and analyzed using Fiji ImageJ (Schindelin et al., 2012): single stacks were used for Fig. 2, and average projections were used for Figs 3, 4, 6 and 7. The primary antibodies used were as follows: rabbit polyclonal anti-Speg (1:100; PA553875, Invitrogen), mouse monoclonal anti-RyR1 (1:100; 34C, DSHB) (Hirata et al., 2007), rabbit polyclonal anti-CACNA1S (1:100; ab203662, Abcam; which labels the DHPR protein), mouse monoclonal anti-SERCA1a (1:200; ab2819, Abcam) (Xiyuan et al., 2017), rabbit polyclonal anti-Desmin (1:100; D8281, Sigma-Aldrich) (Kayman Kürekçi et al., 2021), rabbit polyclonal anti-Dnm2 (1:100; GTX127330, GeneTex) (Eno et al., 2016) and mouse monoclonal anti-α-Actinin (1:100; A7811, Sigma-Aldrich) (Sztal et al., 2015). Secondary antibodies used were anti-rabbit Alexa Fluor^® 488 (1:300; A11034, Invitrogen) and anti-mouse Alexa Fluor^® 594 (1:300; A11005, Invitrogen).

To quantify IF intensity, all z-stack images were taken with the same microscope settings (i.e. objective, laser power, intensity, zoom, pixel size and scanning speed). At least three fibers were imaged per group (WT or mutant) per myofiber immunostaining experiment, and three independent experiments were performed. For each image, the average projection was generated using Fiji ImageJ, gray value was measured for signals (average of three measurements within the fiber) versus backgrounds (average of three measurements outside the fiber) using the square tool (area specified as 25 µm×8 µm), and background value was subtracted from signal. The means of background-subtracted signals per group per experiment were plotted in GraphPad Prism version 8 and compared using unpaired two-tailed Student's t-test.

At 7 dpf, larvae were anaesthetized using 0.1% tricaine and fixed in Karnovsky's fixative (2.5% glutaraldehyde/2% PFA in 0.1 M cacodylate buffer, pH 7.5) at room temperature for 2 h, and re-fixed in fresh fixatives overnight at 4°C. The samples were then washed 3×5 min in 0.1 M cacodylate buffer (pH 7.5), post-fixed in 1% osmium (in 0.1 M cacodylate buffer, pH 7.5) for 1.5 h at room temperature, and washed 3×5 min with 0.1 M cacodylate buffer. Samples were then dehydrated with serial ethanol washes (70%, 90%, 95% and 100%), infiltrated with Epon and embedded in Epon to polymerize in a 60°C oven for 24-48 h. Semi-thin (1 µm) and ultra-thin (90 nm) sections were cut using Leica Ultracut ultramicrotomes and transferred on 200 nm copper grids. Grids were post-stained with 2% uranyl acetate at room temperature for 20 min, washed 7×1 min with MilliQ water and stained with lead citrate for 5 min, followed by 7×1 min water wash. Samples were imaged using an FEI Tecnai 20 transmission electron microscope.

For all experiments, individual myofibers were isolated by enzymatic digestions and then loaded with 5 µM fluo-4-AM (Molecular Probes) for 45 min at room temperature in a normal rodent Ringer's solution consisting of 145 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, pH 7.4. Fibers were then transferred to dye-free rodent Ringer's solution supplemented with 25 µm N-benzyl p-toluene sulfonamide for 20 min at room temperature to block contractions. Fluo-4-AM-loaded fibers were excited at 480±15 nm, and fluorescence emission detected at 535±20 nm was collected at 10 kHz using a photomultiplier system. Myoplasmic Ca²⁺ transients in fluo-4-AM-loaded myofibers were stimulated by an electrical field stimulation protocol using a glass electrode filled with 200 mM NaCl placed adjacent to the cell of interest. The stimulation protocol consisted of five twitch stimuli delivered at 1 Hz followed by a single 5 s, 10 Hz stimulation train, and then exposure to 10 mM caffeine for 30 s in the absence of electrical stimulation. Peak change in fluo-4-AM fluorescence was measured and expressed as (F_max−F₀)/F₀.

To quantify muscle performance, 3 dpf and 5 dpf zebrafish were individually transferred to a 96-well plate and incubated in an optovin analog 6b8 (10 μM in 200 μl embryo water, ChemBridge, 5707191) at 28.5°C for 5 min in the dark. Motor activity of the larvae was recorded and analyzed using ZebraBox (Viewpoint, France) as previously described (Zhao et al., 2019) with 30 s light on, 1 min light off, 30 s light on, 1 min light off and 30 s light on. Four independent experiments were conducted including WT versus speg-KO embryos, and n=24 larvae per group. Total distance traveled (mm) was plotted and analyzed using GraphPad Prism version 8. For each group, s.e.m. was calculated, and one-way ANOVA was performed to test statistical significance.

Embryos were fin clipped at 3-4 dpf and genotyped using HRM analysis as described above; 5 dpf WT or speg-KO embryos were collected (∼20 embryos per group) and immediately stored at −80°C. Samples were homogenized using a Pellet Mixer (VWR, 47747-370) in 1× RIPA buffer (Cell Signaling Technology, 9806) supplemented with Complete Mini EDTA-free Protease inhibitor tablets (Roche, 11836170001, half a tablet per 5 ml lysis buffer) and phosphatase inhibitors (Sigma-Aldrich, 524625, 1:100). Lysates were chilled at 4°C for 10 min, sonicated and centrifuged at 12,000 g for 30 min at 4°C. Supernatants were collected and protein concentration quantified using a Pierce™ BCA protein assay kit (Thermo Fisher Scientific, 23225).

Protein lysates (40 µg/lane) were mixed with LDS sample buffer X4 (Invitrogen, B0007, X1)/DTT (100 mM) and boiled at 95°C for 5 min before loading. Samples were run at 100 V, transferred using semi-dry transfer at 10 V for 70 min and resolved on PVDF membranes. Equal loading and transfer efficiency were assessed by total protein (REVERT™ 700, Li-cor) staining prior to blocking. Membranes were blocked in 1× TBST containing 3% BSA for 1-2 h at room temperature, and then incubated overnight at 4°C with primary antibody in blocking solution. Membranes were washed and probed with secondary antibody (1:5000; anti-rabbit-HRP, 1706515; anti-mouse-HRP 1706516, Bio-Rad) in blocking solution. Blots were imaged by chemiluminescence (Clarity Max™ ECL, Bio-Rad) using the Gel Doc™ XR+Gel Documentation System (Bio-Rad). Band signal intensities were determined using Fiji ImageJ software. All densitometry values were individually standardized to corresponding values of total protein stain and expressed as the fold difference from the average of the WT group of each blot. Primary antibodies used were as follows: anti-Desmin (1:1000; Sigma-Aldrich, D8281), anti-β-actin (1:1000; Abcam, 8226) and anti-Dnm2 (1:1000; GeneTex, GTX127330).

Post-capture analysis, including tests of statistical significance, was performed using Microsoft Excel 2016 (Microsoft) and GraphPad Prism version 8. The difference between three or more groups was assessed by ordinary one-way ANOVA followed by Tukey's multiple comparisons test. The difference between two groups was assessed by unpaired two-tailed Student's t-test. All survival curves were assessed by Mantel–Cox test. Differences were considered to be statistically significant if P<0.05. All data unless otherwise specified are presented as mean±s.e.m.

The authors gratefully thank Dr Yukari Endo for reviewing the manuscript; Dr Ramil R. Noche for useful discussions and insightful suggestions; Alejandro Salazar and Elyjah Schimmens for zebrafish maintenance (SickKids Zebrafish Facility); and SickKids Imaging facility and SickKids Nanoscale Biomedical Imaging Facility for help with microscopic training and analysis.