Allelic heterogeneity of TTNtv dilated cardiomyopathy can be modeled in adult zebrafish

Dilated cardiomyopathy (DCM) is a heart muscle disorder that is characterized by dilation of the ventricle and systolic contractile impairment (1). It is a common cause of heart failure and associated with substantial morbidity, mortality, and health care expenditure (2–4). More than 50 genes have been linked to DCM, and the most predominant genetic factor is truncational variants in the TITIN gene (TTNtvs) (5–7), accounting for approximately 20% of cases (8–13). However, TTNtvs have also been found in about 1% of the reference population, who do not exhibit phenotypes (11, 14, 15). Studies in human patients found that TTNtvs affecting exons encoding the C-terminal A-band region (TTNtv-As) are more likely to be pathogenic than those encoding the N-terminal Z-band region (TTNtv-Zs) (10, 16, 17). This position-dependent feature of TTNtv, termed allelic heterogeneity (AH), is a major hurdle for the implementation of genotype-guided diagnosis and therapy, and the underlying mechanisms remain largely elusive (18–20).

Although TTNtv DCM has been successfully modeled in rat, mouse, zebrafish, and induced pluripotent stem cells (iPSCs), AH has only been recapitulated in iPSCs (16, 17). In one iPSC study, it was shown that cardiac microtissue expressing A-band TTNtv exhibited less than half the contractile force of those expressing I-band TTNtv or WT (16). Similarly, another iPSC study showed that TTNtv-Z-cardiomyocytes had sarcomeres and were able to contract whereas TTNtv-A cardiomyocytes were not able to (17). In contrast, rat models of Ttntv-Z and Ttntv-A manifest similar cardiac phenotypes — normal under resting conditions while higher strain rates and higher left ventricle pressures develop under transaortic constriction stress (21). In mice, a Ttntv-A model harboring a 2 bp insertion in an A-band exon was established to mimic a human pathogenic variant (22, 23). Heterozygous mutants showed left ventricular dilation, impaired fractional shortening, and diffused myocardial fibrosis when stressed with angiotensin II or isoproterenol (24). No mouse Ttntv-Z model has yet been reported. Several other mouse mutants that contain internal deletions of important Ttn domains, such as N2B, PEVK, I-band/A-band junction, and kinase domains, or a missense mutation, exhibit cardiomyopathy-like phenotypes (25–30), but they are not TTNtvs. In zebrafish, the first ttntv mutant, pickwick, contains a nonsense mutation in the N2B domain and is considered an embryonic ttntv model (31); an adult zebrafish ttntv-A mutant was recently reported to develop a DCM-like phenotype (32). We previously generated a panel of ttntv-A and ttntv-Z mutants and revealed AH during somitogenesis (33); however, whether these mutants recapitulate AH of TTNtv DCM is unknown.

Three hypotheses have been proposed to explain the AH of TTNtv DCM. First, the haploinsufficiency/exon usage hypothesis argues that TTNtv-Zs have less effect on the protein levels of full-length TTN than TTNtv-As. This hypothesis is supported by studies directly correlating exon usage with pathogenic mutations in patients with DCM (13) and engineered iPSCs (16) and indirectly by the exon-dependent penetrance of ttntvs in zebrafish somitogenesis (33). Second, the Cronos hypothesis predicts that disruption of the Cronos isoform, a shorter TTN isoform that shares the C-terminal 124 exons with the full-length TTN (34), predisposes animals to the development of TTNtv DCM. This hypothesis was initially proposed based on findings in zebrafish somitogenesis, in which TTNtv-As but not TTNtv-Zs disrupted the Cronos isoform (34, 35), and was later supported by a study in iPSCs (17). Third, the toxic peptide hypothesis proposes that TTNtvs in the A-band result in truncated TTN peptides that exert dominant toxic effects (10). This idea has attracted more attention recently because truncated TTN peptides have been detected in human patients with TTNtv DCM (36, 37). Whether these hypotheses are mutually exclusive or coexist remains to be elucidated.

The establishment of TTNtv DCM animal models laid the foundation for searching for dysregulated signaling pathways (38–41). A recent study in a rat Ttntv model uncovered autophagy dysregulation, i.e., increased expression of LC3-II and p62 proteins at steady state and failure to further accumulate LC3-II upon administration of the lysosome inhibitor bafilomycin A1 (BafA1) in Ttn-knockdown cells (21). It remains unclear whether autophagy can be leveraged for therapeutic benefits (42).

Here, using a collection of zebrafish ttntv-As and ttntv-Zs mutants, we showed that AH can be modeled in ttntv homozygous embryos as well as heterozygous adult fish. We generated null mutants with deletion of the full-length ttn genes and revealed a loss-of-function nature of TTNtv DCM. Finally, we demonstrated that autophagy is differentially dysregulated in ttntv-A versus ttntv-Z mutants and that targeting ulk1 could be a viable therapeutic avenue.

Cardiac function and sarcomere structure are more severely affected in homozygous ttntv-A embryos than in ttntv-Z mutants. Zebrafish have 2 homologs of mammalian TTN, namely, ttna (also designated ttn.2) and ttnb (also designated ttn.1). They are in tandem on chromosome 9, with ttna being predominantly expressed in the heart (43). Prompted by the success of modeling AH during somitogenesis (32), we decided to analyze the cardiac phenotypes of these ttntv-A and ttntv-Z embryos (Figure 1A). Comparing aZ/aZ with aA/aA mutants that affect ttna, we noted that approximately 48% of aZ/aZ fish maintained weak heart contraction while 98% of aA/aA mutants manifested complete or near-complete cessation of contraction at 48 hpf (Table 1 and Figure 1B). The residual contraction of the ventricle in the aA/aA mutants appeared to be a passive movement initiated from the contraction in the atrium (Supplemental Video 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.175501DS1). Moreover, ttnb mutants were generally normal until 9 dpf, when reduced cardiac contraction was noted in bA/bA but not bZ/bZ (Figure 1C).

Figure 1

AH in cardiac function and sarcomere structure is noted in homozygous ttntv embryos. (A) Schematics of the ttntv mutants. These mutants have been generated previously (33). * indicates the location of truncation mutations. (B) Quantification of fractional shortening in the embryonic hearts at 2 dpf. *P < 0.0001. (C) Quantification of fractional shortening in the embryonic hearts at 9 dpf. (D) Comparison of cardiac sarcomere structure in 2 dpf embryos via F59 antibody staining. Scale bar = 20 μm. Insets show enlarged images of the area enclosed with dashed lines. Scale bar = 5 μm. Arrowheads indicate sarcomeres with striated patterns. Arrows indicate immature sarcomeres with less striated patterns.

Table 1

Comparison of the heart contraction of homozygous ttntv embryos at 2 dpf

Visual inspection of the ttntv embryos indicated pericardial effusion, defective somite structure (Supplemental Figure 1, A and C), and abnormal ventricle size (Supplemental Figure 1, D and F) in aA/aA, aZ/aZ, dA/dA, and dZ/dZ but not bA/bA or bZ/bZ. We compared the sarcomere structure between ttntv-Z and ttntv-A mutants by immunostaining with F59, an anti–myosin heavy chain antibody that labels the thick filament (44). In agreement with the lack of contraction, the striation of the myofibrils was severely disrupted, and the lateral growth of sarcomeres was halted in homozygous aA/aA mutants (Figure 1D). In contrast, striated sarcomeric structures were still detectable in approximately 50% of aZ/aZ mutants, consistent with half of the aZ/aZ mutants retaining certain contractions (Figure 1, B and D). Although bZ/bZ exhibited sarcomeric defects in somites at 9 dpf, as reported previously (Supplemental Figure 2A) (33), we did not observe any sarcomere defects in their hearts (Supplemental Figure 2B). Furthermore, we compared dZ/dZ and dA/dA double mutants that affected both ttna and ttnb in the corresponding exons; however, this comparison was not informative for AH because both mutants had silent hearts and lacked striated myofibrils (Table 1 and Figure 1, B and D).

Cardiomyopathy-like phenotypes are noted in heterozygous zebrafish ttntv-A adult mutants but not in ttntv-Z mutants. To better understand patients with TTNtv, we examined heterozygous TTNtv-A and TTNtv-Z mutants at the adult stage. Using high-frequency echo to quantify cardiac pump function, we found a significant reduction in ejection fraction (EF) in aA/+ but not aZ/+ fish starting from the age of 3 months (Figure 2, A–C). Similar cardiac dysfunction was also noted in dA/+ but not dZ/+ mutants (Figure 2, A–C). We did not detect significant changes in markers for mitochondrial function and mass (Drp1 and Tom20) in the hearts of 3-month dA/+ fish (Supplemental Figure 3), suggesting that mitochondria dysfunction is not prominent at this early pathological stage. While cardiac dysfunction was not detected in bA/+ fish until 12 months, bZ/+ fish remained normal (Figure 2, A–C). Moreover, mirroring a recent report, cardiac function was more significantly reduced in female fish than in male fish (Supplemental Figure 4) (32).

Figure 2

AH in cardiac contraction is noted in adult zebrafish heterozygous ttntv mutants. (A–C) Quantification of the ejection fraction in adult zebrafish at 3–12 months via high-frequency echocardiography. One-way ANOVA was applied. (D and E) Quantification of end-systolic volume/body weight (ESV/BW) in adult zebrafish at 6–12 months. One-way ANOVA was applied. (F) Comparison of dissected ventricular size in 6-month fish. Independent samples t test was applied to compare WT with dZ/+ and WT with dA/+. (G) Expression of nppa, nppb, vmhc, and vmhcl gene transcripts by quantitative real-time PCR in 6-month WT, dZ/+, and dA/+ fish. One-way ANOVA was applied. (H) The swimming challenge test indicated a reduced swimming ability in ttntv fish compared with WT fish at 15 months of age. Two-way ANOVA was applied to compare WT versus dZ/+, WT versus dA/+, and dZ/+ versus dA/+ in each day group. Data are presented as mean ± SD. VSA, ventricular surface area; BL, body length.

To further define DCM-like phenotypes in ttntv-As, we carried out detailed analysis of the echo data and found that the end-systolic volume (ESV) was significantly increased in dA/+ and aA/+ fish (Figure 2, D and E). At 6 months of age, standardized ventricle surface area was enlarged in dA/+ but not dZ/+ fish (Figure 2F). In contrast with the myocardium of rodents and humans, the myocardium of the adult zebrafish is composed of a compact layer and a trabeculated layer. Because a clearly defined wall is absent, the density of the trabeculated myocardial muscles has been utilized as a metric to assess cardiomyopathy-like changes in a zebrafish heart (45, 46). No significant changes of the muscle density were found in the zebrafish ttntv model at 3 months of age (Supplemental Figure 5, A and B). At the mRNA level, the cardiac remodeling marker nppa was greatly activated in 6-month-old dA/+ mutants and, to a lesser extent, dZ/+ mutants (Figure 2G), suggesting some abnormality in dZ/+ as well. Given that patients with DCM perform poorly on physical activity assessments, we assessed the exercise capacity of the ttntvs via a continuous 3-day swimming test. As widely accepted, Ucrit, instead of total distance, is a recognized index to measure swimming capacity (47, 48). The assay is always carried under the same light condition (49). We found lower Ucrit in both dA/+ and dZ/+ at day 2 but only in dA/+ at day 3 (Figure 2H), supporting the notion that dZ/+ is more resilient but not entirely normal.

To assess striated sarcomere structure in myofibrils, we developed a tissue-clearing whole-mount staining protocol that enables better resolution of the sarcomeric structure in isolated adult fish whole hearts. While the striated pattern of sarcomeres was not affected, we noted aberrant trabecular muscle organization in the hearts of dA/+ but not dZ/+ fish at 15 months. In contrast with well-aligned myofibrils, some trabecular muscles in dA/+ formed unique circular structures (Figure 3). Together, these data strongly suggested that ttntvs in zebrafish recapitulate the AH of TTNtv DCM in humans.

Figure 3

AH in myofibril organization is noted in adult zebrafish ttntv mutants. Shown is anti–Troponin T immunostaining of a whole heart isolated from a 15-month-old zebrafish using tissue clearance technology. Distorted myofibrils were noted in dA/+ but not in dZ/+ and WT. n = 3 in each group. Scale bar = 100 μm. Images in the bottom panel are enlarged images of the marked area in the upper panel. Scale bar = 50 μm.

Loss of function of ttn mutants manifests DCM-like phenotypes. To investigate the molecular mechanisms underlying AH in TTNtv DCM, we examined Titin protein expression using agarose-based SDS protein gels (50). Consistent with previous results (33), dZ/dZ mutants lost the higher–molecular weight Titin band but maintained the lower–molecular weight Titin band (Figure 4A), suggesting again the affected exons are unnecessary for some Titin isoforms. Surprisingly, unlike previous F₂ generation mutants (33), dA/dA fish in the current F₈ generation no longer expressed any full-length Titin proteins (Figure 4A). To assess whether nonsense-mediated RNA decay was aggravated from the F₂ through the F₈ generation, we examined mRNA expression. We found that the changes in transcript levels in the F₈ generation were the same as those in the F₂ generation (33), i.e., fewer ttna transcripts in dA/dA than in dZ/dZ and undisturbed ttnb transcripts in dA/dA (Figure 4B), suggesting that the depletion of the full-length Titin protein in the F₈ generation occurs at the protein level rather than the transcription level. Of note, truncated Titin proteins were not detected in either mutant (data not shown).

Figure 4

ttn-null mutants manifest similar phenotypes to ttntv-A but not ttntv-Z mutants. (A) Protein analysis of 2 dpf larvae of WT, dZ/dZ, dA/dA, aA/aA, a-null/a-null, and d-null/d-null mutants. The upper lane shows silver staining of a 2% SDS-agarose gel. The same protein lysates were then analyzed by Coomassie blue staining with a 12.5% SDS-PAGE system, as shown in the bottom lane. The expression of Actin was used as a loading control. (B) ttna and ttnb mRNA expression levels in dA/dA and dZ/dZ embryos, as revealed by quantitative PCR using primer pairs that recognize the M-line exons of ttna and ttnb, respectively. (C) Schematics of the location of sgRNAs that were used to generate the large deletions in ttn-a-null, ttn-b-null, and ttn-d-null. The red and blue sequences are sgRNAs on either side. Arrowheads indicate the cutting sites, and the resulting genome sequence is listed below. (D) High-frequency echocardiography was performed in adult zebrafish at 6 months to quantify the ejection fraction. One-way ANOVA was used to compare multiple groups for each mutation. Data are presented as mean ± SD.

To ensure that AH indeed results from loss of function of full-length Titin, we generated null mutants with the entire genomic sequences of ttna, ttnb, or both removed via CRISPR-mediated genome editing (Figure 4C). As expected, both Titin bands were reduced in a-null/a-null fish and absent in d-null/d-null mutants (Figure 4D). In the somites, sarcomere assembly was halted at the premyofibril stage, as manifested by the greatly reduced width of the Z-disc and length of the sarcomere in d-null/d-null mutants (Supplemental Figure 6, A–C). Similar sarcomeric defects were noted in the heart (Supplemental Figure 6D), pericardial effusion and abnormal ventricle could be detected (Supplemental Figure 1, B and E), and cardiac contraction was nearly completely halted in d-null/d-null mutants (Supplemental Video 1). We raised the heterozygous null mutants to adulthood and found that cardiac functions in d-null/+ and a-null/+ were significantly impaired at 6 months, comparable to that seen in dA/+ and aA/+ (Figure 4E). Interestingly, b-null/+ fish exhibited an earlier decline in cardiac function (6 months) (Figure 4D) than bA/+ fish (12 months) (Figure 2, B and C). Together, these data strongly suggested a loss-of-function nature of ttntv DCM in adult zebrafish models.

Defective autophagic flux is noted in ttntv-As but not ttntv-Zs. Because dysregulated autophagy has been reported to be a pathological event in a rat Ttntv model (21), we asked whether dysregulated autophagy also occurs in zebrafish ttntv models. Similar to the rat model, we noted significantly increased basal LC3-II expression in 12-month-old dA/+ mutants that failed to further accumulate upon treatment with BafA1, an inhibitor of autophagosome-lysosome fusion (Figure 5, A and B), suggesting a defect in autophagic flux. This autophagic dysregulation can be detected in dA/+ mutants as early as 3 months (P < 0.01, see Figure 6, G and H). In contrast, dZ/+ hearts manifested a different mode of autophagic dysregulation: a normal response to BafA1 treatment despite reduced basal LC3-II levels (Figure 5, A and B).

Figure 5

AH in autophagy dysregulation is noted in ttntv models, and autophagic flux is affected by ttntv-A. (A) Representative Western blot of LC3-II from 12-month-old zebrafish hearts. (B) Quantification of A; autophagy is differentially dysregulated in dA/+ vs. dZ/+ fish, with dA/+ showing increased basal LC3-II and impaired autophagic flux, while dZ/+ exhibits reduced basal LC3-II. (C) Quantification of autophagosome puncta in 2 dpf BafA1-treated embryos (Supplemental Figure 7A). (D) Quantification of the autolysosome/autophagosome ratio of 2 dpf non-BafA1-treated embryos (Supplemental Figure 7A). (E) Ubiquitin-conjugated (Ub-conjugated) proteins were moderately increased in detergent-insoluble fractions of dA/dA embryo lysates. The dA/dA fish and WT siblings at 5 dpf were subjected to cell fractionation, and the resulting total lysates (T), soluble fractions (S), and insoluble fractions (P) were analyzed by Western blot using an Ub antibody. The blot is representative of 3 independent experiments. (F) Representative image and quantification of LC3-GFP and lysosome marker LAMP1 colocalization compared with the total LAMP1-stained lysosome. The arrow indicates a yellow puncta with both GFP and LAMP1 that represents colocalization, and the arrowhead indicates a red puncta with only LAMP1. Scale bar: 5 μm. (G) Quantification of F. In this figure, independent samples t test is used. Data are presented as mean ± SD.

Figure 6

ulk1a mutants repair defective autophagic flux in both embryonic and adult stages of ttntv-A mutants. (A) Schematics of ulk1a^e6 generated by MMEJ sgRNA. Red indicates deleted nucleotides. * indicates the nonsense codons. (B and C) Quantification of the autophagosome puncta in 2 dpf embryos either with or without BafA1 treatment and the autolysosome/autophagosome ratio of BafA1-untreated groups. ulk1a^+/– can reduce the basal autophagosome level of dA/dA and can recover the autophagic flux affected by dA/dA. One-way ANOVA was applied. (D–F) Representative Western blot and quantification of ulk1a and atg7 from 3-month zebrafish hearts. One-way ANOVA was applied. (G and H) Representative Western blot and quantification of LC3-II from zebrafish hearts either with or without BafA1 treatment at 3 months. Independent samples t test was applied to compare 2 groups. Data are presented as mean ± SD.

To visually assess autophagy dysregulation at the cellular level, we injected the mCherry-EGFP-LC3-II tandem repeat plasmid into embryos at the 1-cell stage for fluorescence analysis of puncta. The hearts were dissected at 2 dpf and treated with either BafA1 or DMSO solution in a culture dish, and then autophagosomes (yellow dot) and autolysosomes (red dot) were quantified (Supplemental Figure 7A). We found significantly more autophagosomes in dA/dA hearts that could not be further increased in response to BafA1 treatment, as well as fewer (though not significantly, P = 0.0916) autophagosomes in dZ/dZ fish that were responsive to BafA1 (Figure 5, C and D). Consistent with the flux defect in dA/dA fish, ubiquitin-tagged protein aggregates in the detergent-insoluble portion, which is usually degraded via the autophagy pathway, accumulated (Figure 5E). Next, we costained LC3-GFP (autophagosome marker) and LAMP1 (lysosome marker) and found a reduced colocalization in dA/dA hearts (Figure 5, F and G) (51), indicating an impairment in autophagosome-lysosome fusion. In d-null/d-null mutants, we also noted aberrant accumulation of autophagosome and defective autophagic flux (Supplemental Figure 7, B and C), as has been noted in dA/dA, supporting a loss-of-function nature underlying how ttntv mutants result in autophagic defects. Together, these data strongly suggested that AH in autophagy dysregulation could occur in both zebrafish heterozygous adults and homozygous ttntv embryos.

ulk1a can be targeted to repair defects in autophagic flux and ameliorates cardiomyopathy-like phenotypes in ttntv-As. To determine whether defective autophagy can be repaired for therapeutic benefits, we elected to knock down atg7 that is involved in autophagosome formation and ulk1a that functions in both autophagosome formation and autophagosome-lysosome fusion (52–54) via microhomology-mediated end-joining (MMEJ) technology (55, 56). At the average knockout (KO) efficiency approximately 80% (Supplemental Table 2), injection of MMEJ-inducing sgRNAs against ulk1a or atg7 did not cause significant autophagic changes in WT fish (Supplemental Figure 7A). However, knockdown of Ulk1a but not Atg7 was able to normalize steady-state autophagosome numbers and restore response to BafA1 treatment in dA/dA hearts (Supplemental Figure 7A). At the cellular level, injection of ulk1a-sgRNA improved LC3-GFP and LAMP1 colocalization in the dA/dA fish (Supplemental Figure 7, B and C), indicating a repair in the autophagosome-lysosome fusion.

To confirm this observation from transient genetic analysis, we generated a stable ulk1a mutant that contains a 5 bp deletion in exon 6 of the ulk1a gene (Figure 6A). ulk1a transcripts were significantly reduced in the ulk1a^+/– mutant (Supplemental Figure 9), presumably due to nonsense-mediated RNA decay. Consistent with our observation from transient genetic assay, both ulk1a^+/– and ulk1a^–/– were able to rescue defects in the autophagic flux in the dA/dA mutant (Figure 6, B and C, and Supplemental Figure 10, A and B). In adult dA/+ mutant hearts, expression levels of both Atg7 and Ulk1a were increased, which could be attenuated by ulk1a^+/– (Figure 6, D–F). Moreover, ulk1a^+/– could restore autophagic flux in adult dA/+ hearts, as indicated by the normalized LC3-II levels in the basal and BafA1-treated states (Figure 6, G and H).

Encouraged by the rescuing effects in autophagy, we examined whether ulk1a mutants can ameliorate cardiomyopathy-like phenotypes in dA/dA embryos and dA/+ adults. Indeed, we found that cardiac defects, including the distorted shape of the ventricle (Figure 7A), reduced ventricular size (Figure 7B), and reduced cardiac function (Figure 7C) in dA/dA, could be effectively rescued in the dA/dA ulk1a^–/– double mutants. Consistent with the rescued cardiac contractility, the disrupted sarcomere structure was partially rescued (Figure 7A). At the cellular level, we noted unchanged cardiomyocyte size in the outer curvature and increased nucleus size in dA/dA, the latter of which could be partially rescued in dA/dA ulk1a^–/– double mutants (Supplemental Figure 11). Moreover, cardiomyopathy-like phenotypes in dA/+ could be rescued in dA/+ ulk1a^–/– double mutants at 3 months of age (Supplemental Figure 12) and in dA/+ ulk1a^+/– double mutants at 6 months of age (Figure 7D), as indicated by the improved cardiac function. Consistent with these data, aberrant activation of nppa could be rescued in dA/+ ulk1a^+/– double mutants (Figure 7E). Together, our data suggested Ulk1a as a therapeutic target gene for TTNtv DCM.

Figure 7

ulk1a mutants repair cardiac dysfunction in both adult and embryonic stages of ttntv-A mutants. (A) Immunostaining and differential interference contrast (DIC) images of 2 dpf embryos. Sarcomere structures were obtained by immunostaining with an anti-F59 (green) antibody in WT, dA/dA, ulk1a^–/–, and dA/dA ulk1a^–/–. Higher magnifications of the boxed areas are shown in the insets. White arrowheads indicate striated sarcomere structures. The embryonic ventricle is outlined by dashed red lines in the DIC images. The scale bar of the upper panel is 100 μm, the lower panel is 10 μm, and the inset is 2.5 μm. (B) Quantification of the ventricle area based on the area outlined by the dashed red lines in the DIC panels of Figure 7A. (C) Quantification of fractional shortening in embryonic hearts. (D) High-frequency echocardiography was performed on 6-month-old zebrafish to quantify EF%. (E) Evaluation of nppa gene transcript expression by quantitative real-time PCR in hearts from 3-month-old adult zebrafish. One-way ANOVA was used to compare multiple groups for each mutation. Data are presented as mean ± SD.

AH in TTNtv DCM can be recapitulated in adult zebrafish. One of the novelties of this manuscript is that we present zebrafish as an in vivo model that is able to recapitulate AH in TTNtv DCM. Cardiomyopathy-like phenotypes occur only in adult heterozygous ttntv-A mutants such as dA/+ but not in ttntv-Z mutants such as dZ/+. AH in cardiac dysfunction was also noted in aA/+ or bA/+ but not aZ/+ or bZ/+. Consistent with the lower cardiac expression level of ttnb in zebrafish heart, bA/+ manifests phenotypes at an older age than either aA/+ or dA/+. A limitation of our present study is that we only assessed a single exon that affects either the Z-disc or A-band. Investigations of more ttntvs affecting other ttn exons are justified in the future, which could potentially establish zebrafish as a faithful model for assessing the pathogenicity of TTNtvs suggested from human patients.

We then defined the pathogenesis of ttntv DCM in adult zebrafish by detailed phenotyping of the dA/+ allele. Similar to a previous study of a ttntv-A affecting zebrafish ttna exon 202 that corresponds to human exon 327 in TTN (32), we noted decreased cardiac function, aberrant expression of molecular markers for heart remodeling, myofibril abnormality, and reduced exercise capacity, which are all phenotypic traits of DCM (Figures 2 and 3). The decreased EF percentage could be attributed to increased ESV and ventricular cavity enlargement. Moreover, we detected additional phenotypes in our dA/+ allele than in the reported ttna^tv allele (32). First, we were able to detect cardiac dysfunction as early as 3 months in dA/+, probably because both ttn alleles have been mutated and because Ttn proteins are nearly completely eliminated in dA/dA at the F₈ generation (Figure 4A). Second, we used a tissue-clearing technique that allowed us to better define structural changes in the trabecular muscle. We noted normal sarcomere structure at 15 months, when trabecular muscle disarrangement appeared in dA/+ but not dZ/+ fish. Given that reduced EF percentage is already detectable as early as 3 months, when mitochondria dysfunction does not occur, and long before trabecular muscle disorganization, our longitudinal observation strongly suggested that neither defective sarcomere structure nor mitochondria dysfunction is a primary pathological event but rather sequential abnormalities of ttntv DCM.

Loss of function of ttn is sufficient to trigger DCM-like phenotypes. The establishment of zebrafish as an in vivo animal model for AH in ttntv DCM enabled us to further pinpoint the underlying mechanisms. We generated null mutants with complete removal of all zebrafish Ttn proteins. Because genetic compensation should have been effectively eliminated, our observation that heterozygous d-null/+ manifests DCM-like phenotypes provided a strong piece of genetic evidence supporting that ttn loss of function is sufficient to trigger DCM-like phenotypes. This genetic evidence is in line with both the haploinsufficiency hypothesis and Cronos hypothesis (11, 17) because both full-length Titin protein and the shorter Titin Cronos isoform are presumably depleted in our null mutants. However, our current data cannot distinguish between the haploinsufficiency hypothesis and the Cronos hypothesis, which would require future genetic studies that target the Cronos isoform.

Our genetic data do not absolutely disprove the toxic peptide hypothesis because we have not purposed to study truncational TTN peptides. In fact, the 3 hypotheses on AH in TTNtv DCM are not mutually exclusive because each of these hypotheses concerns different versions of TTN proteins. The haploinsufficiency hypothesis emphasizes the full-length TTN proteins, the toxic peptide hypothesis relates to the truncated TTN peptides, and the Cronos hypothesis emphasizes a particular shorter TTN isoform. While our current evidence favors the haploinsufficiency hypothesis and the Cronos hypothesis, future studies in the in vivo zebrafish model are still warranted to directly test the toxic peptide hypothesis.

ulk1a is a therapeutic target gene for ttntv DCM in zebrafish. The discovery of autophagy dysregulation in the rat Ttntv model prompted us to test whether this pathological signaling is conserved in our zebrafish ttntv models. Indeed, we observed similar autophagic flux defects, manifesting as an increased accumulation of LC3-II and a lack of response to BafA1 (21). Using dA/dA homozygous embryos, our data further suggested that the dysregulated autophagy in ttntv DCM is characterized by defective degradation, likely owing to reduced fusion between autophagosome and lysosome. Precisely how ttntv leads to autophagosome-lysosome fusion defects warrants future studies. Nevertheless, the fact that these autophagy defects are detectable as early as 3 months of age suggests that autophagy dysregulation is a proximal pathological event triggered by ttntv, which is different from myofibril structural changes that occur at 15 months of age, a much later pathological stage. Our data in the d-null/d-null mutant further suggested that autophagy defects can be triggered by depletion of Ttn proteins, without any contribution from truncated Ttn as a toxic peptide (Supplemental Figure 7, B and C).

Another potentially novel discovery of the present study is to demonstrate that repairing autophagy could be a therapeutic avenue for TTNtv DCM. ulk1a was identified as a candidate gene that can be manipulated to repair autophagic defects and to rescue cardiac dysfunction. This rescuing effect is manifested at the autophagosome-lysosome fusion step, and whether the more classic function of Ulk1 in autophagy initiation also contributes to the therapeutic benefits remains to be investigated. While Atg7 expression level is aberrantly increased in dA/+ that is attenuated in ulk1^+/–, the seemingly negative data from transient genetic studies of atg7 argues against that changes in Atg7 expression might contribute significantly to the therapeutic effect of ulk1a inhibition (Figure 6F and Supplemental Figure 8A).

Recent studies showed complex functions of ULK1 in the autophagosome-lysosome fusion via phosphorylation of its downstream targets STX17 and YKT6, 2 components of the SNARE complex essential for autophagosome-lysosome fusion (57–59). While ULK1-dependent phosphorylation of STX17 promotes the fusion process, phosphorylation of YKT6 impairs autolysosome formation. How STX17 and YKT6 are regulated in our ulk1a fish warrants further analysis. Moreover, a mouse KO study showed that perinatal inhibition of either Ulk1 or Ulk2 alone in the heart unexpectedly activates autophagy while coinhibition of Ulk1 and Ulk2 suppresses autophagy, suggesting an Ulk dose-dependent impact on autophagy (60). Whether this dose-dependent response also occurs in zebrafish models (given the existence of ulk1a, ulk1b, and ulk2 in zebrafish) needs to be considered in the future when translating Ulk1-based therapy for ttntv DCM.

We noted that knocking down of either ulk1 or atg7 did not cause significant autophagy changes (Supplemental Figure 7A). It remains to be determined whether this is owing to the haploinsufficient nature of these CRISPants, compensation by family members, or species difference between zebrafish and mammals. Future studies are warranted to further test these possibilities, including studies of stable atg7-null fish and ulk1 ulk2 compound mutants.

Zebrafish is a promising animal model for developing genotype-guided therapy for TTNtv DCM. The present study established zebrafish as an important alternative animal model for studying TTNtv DCM. Compared with existing rodent models (16, 17, 22, 25–28, 61), zebrafish models of ttntv DCM possess remarkable advantages, including notable cardiac phenotypes without the need for any biomechanical stresses and the capacity to recapitulate AH. Our data highlighted zebrafish as an increasingly valuable animal model for studying cardiomyopathies of different etiologies. In the last decade, a panel of cardiomyopathy models have been generated (62), and novel phenotyping tools have been developed in this vertebrate model with a tiny heart, enabling the definition of both common and unique phenotypes in cardiomyopathies of different etiology (63, 64). In searching therapeutic avenues using these disease models, it appears that different models benefit from different therapeutic strategies. For example, although bag3 DCM (56), lamp2 hypertrophic cardiomyopathy (65), and anthracycline-induced cardiotoxicity models (66) benefit from mtor inhibition, the RRAGC^S75Y knockin model does not but rather benefits from tfeb activation in cardiomyocytes (67). Here, we reported ulk1a as a candidate therapeutic target gene for the ttntv DCM model. Whether ulk1a would benefit other genetic types of cardiomyopathies and whether therapeutic avenues discovered from other cardiomyopathies are extendable to ttntv DCM remain to be investigated. Findings from zebrafish need to be tested in mammalian models before further translational efforts. It is possible that systematic studies in the efficient zebrafish model would accelerate the implementation of precision medicine for cardiomyopathies of different etiologies.

A detailed version of the Methods section is presented in the online Supplemental Methods.

Sex as a biological variable. In Supplemental Figure 4 we examined male and female zebrafish, and similar findings are reported for both sexes (32). In other results, sex was not considered as a biological variable.

Generation of ttn-null mutants. Large deletions in ttn-null mutants were generated by coinjecting 2 MMEJ-based sgRNAs (68). The sgRNA sequences are listed in Figure 4C. Tail fins from F₁ generation animals were collected for DNA extraction (HotSHOT method) (69), and the DNA lysates were used as templates for PCR to identify fish harboring the desired large deletions. The primer pairs for ttna-null are F: 5′ CGCACCAGTTGTTACTGTC 3′, R: 5′ CATAGTCAGTCTGAACACAAGG 3′; the primer pairs for ttnb-null are F: 5′ CAGCAAAAATCACTTTATTCTG 3′, R: 5′ CAAAATGGTGCAGAACTTATGG 3′; and the primer pairs for ttnd-null are F: 5′ CGCACCAGTTGTTACTGTC 3′, R: 5′ CAAAATGGTGCAGAACTTATGG 3′. Founders containing large deletions in ttna-null were identified by searching embryos with a 134 bp PCR product that contains 5′ TCAAGTGACATCCTTCTACTAGGCCTTGT 3; founders containing large deletions in ttnb-null were identified by searching embryos with a 304 bp PCR product that contains 5′ CATTTCTAACATATTCATTGATAAAGACCTGGATTTCTGT 3′; and founders containing large deletions in ttnd-null were identified by searching embryos with a 256 bp PCR product that contains 5′ TCAAGTGACATCCTTCTTTGATAAAGACCTGGATTTCTGT 3′.

LC3-II plasmid injection and puncta quantification. mCherry-EGFP-LC3-II fragments were amplified from the pBABE-puro-mCherry-EGFP-LC3-II plasmid (70) using a primer pair (F: 5′ GGGGATCCGCCGGCCGGATCTGCCACCATGG 3′, R: 5′ CGCTCGAGCCACAGGGTCGACTTACACTGAC 3′) and were inserted into the BamHI-XhoI sites of a pENTR1A vector (Addgene plasmid 11813-011), a middle clone plasmid from the Tol2Kit (71). The attL-to-attR recombination reaction was carried out using a 5′ clone, p5E-βactin (5.3 kb β-actin ubiquitous promoter); a middle clone, pENTR1A-mCherry-EGFP-LC3-II; and a 3′ clone, p3E-polyA (SV40 late polyA signal), to generate the final β-actin: tandem plasmid.

To generate the β-actin: tandem plasmid with a final concentration of 33 ng/μL, 33 ng/μL transpose, 0.2 M KCl, and 0.01% phenol red were freshly mixed. Three nanoliters of the mixture solution was injected into 1-cell–stage embryos. Hearts at 2 dpf were dissected and transferred to a slide with Leibovitz’s L-15 medium. The dissected hearts were treated with 167 nM BafA1 (MilliporeSigma, 19148) or DMSO solution for 2 hours at 28°C. Hearts were then fixed in 4% paraformaldehyde for 7 minutes, washed with PBS for 10 minutes, and mounted with mounting medium containing DAPI (Vector Laboratories, H-1200). Images were documented using a Zeiss fluorescence microscope using a 63× lens. To quantify autophagic puncta, red and yellow dots around the DAPI-stained nuclei were counted manually.

Statistics. Data were analyzed by Prism 9 (GraphPad Software by Dotmatics). Independent samples t test was used to compare differences between 2 groups, while 1-way ANOVA followed by Tukey’s post hoc test was performed to compare the differences among multiple groups. All quantitative data are presented as the mean ± SD. The detailed P values and sample size (N) represents the animal number (unless otherwise specifically designated as biological replicates) as listed in the graphs. Unless otherwise specifically mentioned, P values less than 0.05 were considered significant.

Study approval. Zebrafish (Danio rerio; WIK strain) were maintained on a 14-hour light/10-hour dark cycle at 28.5°C. All animals were handled with care following the Mayo Clinic Institutional Animal Care and Use Committee (IACUC) and the Guide for the Care and Use of Laboratory Animals (National Academies Press, 2011). The Mayo Clinic IACUC–approved protocol number is A3511.

Data availability. Values for all data points in graphs are reported in the Supporting Data Values file. Supplemental Methods and Supplemental Figures are presented in the supplemental materials. More detailed information on data that support the findings of this study is available from the corresponding author.