Titin-truncating variants (TTNtvs) are the most commonly observed genetic cause of dilated cardiomyopathy (DCM) and are found in up to 25% of DCM cases (Herman et al., 2012; Roberts et al., 2015; Yotti et al., 2019; Mazzarotto et al., 2020). Moreover, approximately 1% of the general population carry TTNtvs, necessitating the parameters for TTNtv pathogenicity prediction (Roberts et al., 2015; Schafer et al., 2017; Yotti et al., 2019). Percent spliced-in (PSI) has been reported as a crucial pathogenicity predictor of TTNtv in DCM. Clinical data demonstrated that TTNtvs located in exons that are constitutively expressed in the human left ventricle (PSI > .9) are significantly associated with DCM (Roberts et al., 2015; Schafer et al., 2017; Haggerty et al., 2019). In addition to PSI, the location of premature termination codon (PTC) has also been investigated as a parameter for pathogenicity prediction of TTNtv. As a result, among the four protein regions of titin, i.e., Z-disk, I-band, A-band, and M-band, A-band enrichment of TTNtv in patients with DCM has been repeatedly reported (Herman et al., 2012; Roberts et al., 2015; Akinrinade et al., 2016; Schafer et al., 2017; Rich et al., 2020; Fomin et al., 2021). TTNtvs from all titin regions can lead to DCM, and A-band TTNtvs are known to have a higher odds ratio (OR) than other regions (Akinrinade et al., 2016; Schafer et al., 2017). The mechanism of A-band enrichment remains unclear. Meanwhile, in few reports, exon 327 within the A-band has been briefly mentioned as a frequently mutated TTN exon (Akinrinade et al., 2016; Tharp et al., 2019; Fomin et al., 2021). The degree of TTNtv enrichment in exon 327 and other A-band exons has not been separately analyzed.

The A-band is the longest region of TTN, which comprises 49.8% (53,799 bp/1,07,976 bp) of the TTN coding region, and when limited to high-PSI exons (PSI > .9), the A-band accounts for 67.0% (53,799 bp/80,325 bp) of the coding region. Considering the high share A-band holds in the high-PSI exons, a high frequency of DCM-related TTNtv in the A-band is expected. In this study, we aimed to assess the degree of TTNtv enrichment in the titin regions when the effect of the distribution of high-PSI exons was removed. Additionally, we aimed to measure the degree of TTNtv enrichment in the exon 327 and other A-band exons separately.

As a disease mechanism of TTNtv-related DCM, in addition to haploinsufficiency caused by the shortage of full-length titin proteins, a dominant negative effect caused by truncated titin protein has been demonstrated in recent reports (Fomin et al., 2021; McAfee et al., 2021). In this scenario, evasion of nonsense-mediated mRNA decay (NMD) is necessary to create a truncated titin protein that acts as a poison peptide. Because the amount of truncated titin protein has been demonstrated to affect the outcome of DCM (Fomin et al., 2021), a reasonable assumption is that NMD efficiency may affect the pathogenesis of TTNtv in DCM.

Exon 327 is the longest exon in TTN—by far—consisting of 17,106 bp (median = 261 bp, second-longest exon = 5,609 bp; Supplementary Table S1). According to reports on the emerging parameters of NMD efficiency, PTCs in long exons tend to decrease the efficiency of NMD because of the long average distance from the PTC to the downstream intron (PTC-to-intron distance). A long PTC-to-intron distance hinders the contact between PTC-bound ribosomes and the downstream exon-junction complex, which is required for NMD initiation (Lindeboom et al., 2016; Hoek et al., 2019; Supek et al., 2021). The association of a large PTC-to-intron distance and inefficient NMD was recorded in a study that utilized single-molecule imaging (Hoek et al., 2019). In addition, TTN was reported to be one of the genes enriched for pathogenic truncating variants that do not trigger NMD (Lindeboom et al., 2019). Our third aim was to identify whether the exon 327 enrichment of TTNtv in patients with DCM is the consequence of reduced NMD efficiency caused by long PTC-to-intron distance. Lastly, we also aimed to determine whether there is a correlation between NMD efficiency and clinical outcomes of TTNtv-related DCM.

Studies have reported that TTNtvs that affect high-PSI (>.9) exons have a 93% probability of being pathogenic, and for A-band TTNtvs, the probability is higher, at 97% (Roberts et al., 2015; Akinrinade et al., 2016). From case control studies, A-band TTNtv showed an OR of 49.8–70.4, and TTNtv from other regions showed an OR of 5.3 (Z-disk), 3.7 (M-band), 19.0 (proximal I-band), and 19.5–32.0 (distal I-band), when limited to constitutive exons (Akinrinade et al., 2016; Schafer et al., 2017). In our study, the total number of patients with DCM (with or without TTNtv) was not available in three studies (Table 1) (Jansen et al., 2019; Akhtar et al., 2020; Rich et al., 2020), making it impossible to derive OR. Instead, we analyzed the degree of enrichment based on the ratio of the observed and expected numbers of TTNtv (expected based on the length of the high-PSI coding region). The degree of TTNtv enrichment among titin regions and exon 327 could be quantitatively compared.

Although exon 327 has been designated as a potential hotspot in several reports (Akinrinade et al., 2016; Tharp et al., 2019; Fomin et al., 2021), the degree of enrichment of exon 327 and other A-band exons has not been separately analyzed. In the DCM cohorts in our study, the number of TTNtv observed in exon 327 was 37% higher than the expected number based on uniform distribution (p < .001), and no other regions showed a significant increment or increment of >10%. The non-exon 327 A-band showed a minimally (2%) higher number of TTNtv than that with the uniform distribution, which was statistically non-significant (p = 0.644). Overall, except for exon 327, the number of DCM-related TTNtv was almost the same as the expected number of uniform distributions in the A-band and I-band, whereas that of the Z-disk and M-band was significantly lower. Based on this result, the widely accepted A-band enrichment appears to be mostly attributable to exon 327.

In contrast with the DCM cohorts, in which the fluctuation in the degree of TTNtv enrichment was not substantial and consistent with the existing knowledge except for the exon 327 vs. other A-band exons, in the gnomAD cohort, the degree of TTNtv enrichment showed considerable fluctuation in an unpredicted manner. Although the allele frequency (AF) of a variant in the general population is known to reflect the severity of the phenotype by the purifying selection pressure (Gorlov et al., 2011), this might not work in a precise, quantitative way for late-onset, incompletely-penetrant phenotypes, such as TTN-related DCM. A high proportion of Clinvar (likely) pathogenic variants (21.6% among high-PSI variants, 12.1% among all variants, Supplementary Table S8) among the TTNtv included in gnomAD cohort indicates that the control group may not consist of individuals with a purely normal cardiac phenotype. Although exon 327 remained the most significantly enriched area when both DCM and gnomAD cohorts were considered, we posit that the results from DCM cohorts are a more direct reflection of the status of TTNtv distribution in patients with DCM.

The pathogenic mechanism of TTNtv leading to DCM is not well understood. It was once proposed that Cronos titin isoform, which encodes the most distal region of the I-band, the A-band and the M-band, can rescue the phenotype explaining the A-band enrichment (Zou et al., 2015; Schafer et al., 2017; Santiago et al., 2021). However, TTNtvs located in pre-Cronos regions also showed high OR (Schafer et al., 2017) and the absence of compensatory changes in Cronos isoform was demonstrated (Fomin et al., 2021), contradicting the hypothesis. The debate between haploinsufficiency and a dominant negative mechanism is ongoing (Santiago et al., 2021) and despite the clues for dominant negative mechanisms, such as the tendency of higher pathogenicity of more distally located TTNtv (Roberts et al., 2015), direct evidence had not been observed until the existence of truncated titin proteins was proved in affected human heart tissue in two recent studies (Fomin et al., 2021; McAfee et al., 2021). In these two reports, truncated titin protein was quantitatively detected by methods, such as Western blotting. Truncated titin proteins were present in intracellular aggregates rather than sarcomeres (Fomin et al., 2021), and the sizes of truncated proteins corresponded to that predicted by the location of TTNtvs (McAfee et al., 2021). Disease severity did not correlate with TTNtv location, but the level of truncated titin protein negatively correlated with patient age at HT (Fomin et al., 2021). These two reports suggested a multifactorial mechanism of combined dominant-negative and haploinsufficiency.

In the RNAseq data analysis, a negative correlation between PTC-to-intron distance and AI deviation was observed, confirming the negative effect of long PTC-to-intron distance on NMD efficiency in the literature (Hoek et al., 2019; Supek et al., 2021). DCM-related TTNtv showed generally reduced NMD efficiency (median AI deviation = .10), and exon 327 TTNtv showed almost no NMD (median AI deviation = .04). Although the difference between exon 327 TTNtv and other DCM-related TTNtv did not reach statistical significance (p = .0501), the low AI deviation values generally observed for DCM-related TTNtvs demonstrated the importance of NMD evasion in the pathogenesis of TTNtv-related DCM. This result is in concordance with a report in which TTN was included in the list of genes enriched for pathogenic truncating variants that do not trigger NMD (Lindeboom et al., 2019). The extremely low NMD efficiency of exon 327 TTNtvs, presumably caused by the long PTC-to-intron distance, seems to be a cause of the high frequency of exon 327 TTNtvs observed in patients with DCM (Figure 4).

There was no significant difference in the clinical outcome parameters between patients with exon 327 TTNtvs and those with other TTNtvs. However, when limited to exon 327 TTNtvs, the age at HT showed a negative correlation with PTC-to-intron distance, indicating that extremely reduced NMD efficiency may affect the clinical outcome. However, whether predicting the severity of DCM based on the PTC-to-intron distance of TTNtv is possible remains unclear.

Although NMD efficiency seems to explain certain levels of the pathogenicity of TTNtv, especially in exon 327, other factors might contribute to its pathogenesis. The level of truncated titin protein should have a more direct effect on penetrance and pathogenicity than the mRNA level. Considering the observation that decreased NMD efficiency did not lead to a proportionally increased production of truncated proteins (Fomin et al., 2021), factors other than NMD might regulate the production and degradation of truncated proteins. Because of the small sample size used in the study that negated the direct relationship between NMD efficiency and protein levels (n = 11) (Fomin et al., 2021), further investigation of the regulation of mRNA and protein levels and its effect on clinical outcomes is warranted.

In conclusion, the accepted A-band enrichment of DCM-related TTNtv was mostly attributable to exon 327 enrichment. The extremely low NMD efficiency of exon 327 TTNtvs, presumably caused by the long PTC-to-intron distance, might cause the high frequency of exon 327 TTNtvs observed in patients with DCM. Reduced NMD efficiency may play an important role in the pathogenesis of TTNtv-related DCM, particularly in exon 327.