Supplementary Materials for

Transcription

1 Supplementary Materials for Integrated allelic, transcriptional, and phenomic dissection of the cardiac effects of titin truncations in health and disease Angharad M. Roberts, James S. Ware, Daniel S. Herman, Sebastian Schafer, John Baksi, Alexander G. Bick, Rachel J. Buchan, Roddy Walsh, Shibu John, Samuel Wilkinson, Francesco Mazzarotto, Leanne E. Felkin, Sungsam Gong, Jacqueline A. L. MacArthur, Fiona Cunningham, Jason Flannick, Stacey B. Gabriel, David M. Altshuler, Peter S. Macdonald, Matthias Heinig, Anne M. Keogh, Christopher S. Hayward, Nicholas R. Banner, Dudley J. Pennell, Declan P. O Regan, Tan Ru San, Antonio de Marvao, Timothy J. W. Dawes, Ankur Gulati, Emma J. Birks, Magdi H. Yacoub, Michael Radke, Michael Gotthardt, James G. Wilson, Christopher J. O Donnell, Sanjay K. Prasad, Paul J. R. Barton, Diane Fatkin, Norbert Hubner, Jonathan G. Seidman, Christine E. Seidman,* Stuart A. Cook* *Corresponding author. stuart.cook@nhcs.com.sg (S.A.C.); cseidman@genetics.med.harvard.edu (C.E.S.) The PDF file includes: Published 14 January 2015, Sci. Transl. Med. 7, 270ra6 (2015) DOI: /scitranslmed Phenotype ascertainment methods for study cohorts Fig. S1. Schematic representation of the unselected DCM cohort recruitment pathway and analyses. Fig. S2. TTN sequencing coverage for each cohort. Fig. S3. Sites susceptible to truncating events are non-uniformly distributed within the TTN gene but do not influence clustering effects in the A-band. Fig. S4. Alternative splicing of TTN in the human heart. Fig. S5. Time to events in FHS individuals grouped by TTNtv presence. Fig. S6. Truncated transcript length is correlated with indices of cardiac impairment severity in DCM. Fig. S7. FHS exam 7 CMR. Fig. S8. FHS and JHS additional CMR and echocardiography exams.

2 Fig. S9. mrna transcripts encoding truncated TTN proteins are expressed in human LV. Table S1. TTNtv identified in UK prospective DCM cohort. Table S2. TTNtv identified in the FHS offspring cohort. Table S3. TTNtv identified in the JHS cohort. Table S4. TTNtv identified in end-stage DCM. Table S5. TTNtv identified in healthy volunteers. Table S6. Titin reference transcript and protein identifiers. Table S7. Overview of TTN transcripts and exon usage. Table S8. TTNtv in publicly available control populations. Table S9. Burden, type, and distribution of TTNtv in publicly available control populations. Table S10. FHS exam 7 CMR phenotype grouped by TTNtv presence. Table S11. Prevalence of DCM in FHS and JHS participants, grouped by TTNtv presence. Table S12. Time to event empirical Cox proportional hazard models for the FHS cohort. Table S13. TTNtv identified in replication cohorts. Table S14. Linear modeling of the relationship between TTN genotype and phenotype for 14 continuous variables in the unselected DCM cohort. Table S15. Full linear model describes impact of multivariate TTN genotype on phenotype for 14 continuous variables in the unselected DCM cohort. Table S16. Linear models for FHS exam 7 CMR. Table S17. Linear models for additional FHS and JHS exams. Table S18. Allele-specific expression of exons containing TTNtv using RNA sequencing data. References (65 67)

3 PHENOTYPE ASCERTAINMENT METHODS FOR STUDY COHORTS DCM subjects: prospectively recruited unselected cohort Clinical data were collected for risk factors for non-ischemic DCM including history of alcohol excess, myocarditis, chemotherapy, iron overload and a peripartum presentation. A family history of a minimum of 3 generations was taken with direct query for DCM, other cardiomyopathy, skeletal myopathy/dystrophy and sudden cardiac death. Significant CAD was excluded as described above. Electrocardiographic data collected included conduction disease and ventricular arrhythmia. Ventricular tachycardia (VT) was defined as 3 or more consecutive ventricular extra-systoles at a rate of >120bpm, with non-sustained VT lasting < 30 seconds, and sustained VT > 30 seconds. Patient clinical data were obtained by questionnaire at time of enrolment and from clinical notes. Quantitative cardiac imaging: CMR was performed on Siemens Sonata 1.5-T or Avanto 1.5-T. Cine images were acquired using a steady-state free-precession sequence in standard 2-, 3- and 4- chamber long-axis views (TE [echo time]/tr [repetition time] 1.6/3.2 ms, flip angle 60 ), with subsequent sequential 8-mm short-axis slices (2-mm gap) from the atrioventricular ring to the apex. Ventricular volumes and function were measured for both ventricles using standard techniques (65). Left atrial (LA) area and length were recorded manually from 2-chamber and 4- chamber images. The biplane area-length method was used to calculate LA volume (LAV) (66). In addition to maximum wall thickness measured in the short axis, mean wall thickness was measured at mid-ventricular (papillary muscle) level in the septum and lateral wall. LAV, LV and right ventricular (RV) volumes, wall thicknesses and LV mass were indexed to body surface area (BSA). Image analysis was performed using semi-automated software (CMRtools, Cardiovascular Imaging Solutions). Late gadolinium enhancement (LGE) images were acquired 10 min after intravenous gadolinium-dtpa (Schering AG; 0.1 mmol/kg) in identical short-axis

4 planes using an inversion-recovery gradient echo sequence, as previously described (34). Midwall LGE was recorded as present when the area of signal enhancement could be seen in both phase-swapped images and in a cross-section long-axis image, as previously described (34). DCM subjects: end-stage A diagnosis of non-ischemic DCM was confirmed from patients medical records, however detailed quantitative clinical and phenotype data were not available for this cohort. Healthy volunteers Healthy volunteer CMR studies were performed on a 1.5T Philips Achieva system. Cine images were acquired using a balanced-steady-state free-precession sequence in standard 2-, 3- and 4- chamber long-axis views (TE/TR 1.5/3.0 ms, flip angle 60 ), with subsequent sequential 8-mm short-axis slices (2-mm gap) from the atrioventricular ring to the apex. Volumes were measured and indexed as above for the DCM cohort. Community-based cohorts Quantitative cardiac imaging: Subjects in the FHS Offspring cohort were studied by echocardiography during exams 2, 4, 5, 6, and 8 (range of sample size 2,898 to 3,787) and primary analyses were performed using M-mode measurements. Cardiac morphology was also assessed in 1,781 subjects of the FHS Offspring cohort by CMR imaging. JHS subjects were studied by echocardiography (exam 1, n=5,076; exam 2, n=820) and primary analyses were performed using 2D-mode measurements. All relevant measurements were indexed to BSA, as for the DCM and healthy volunteer cohorts.

5 DCM was diagnosed based on echo or CMR parameters following adjustment for clinical covariates. Adjustments were independently derived for each dataset, including all individuals regardless of whether they were sequenced (see below) (33, 51). Studies with an echocardiographic LV end-diastolic dimension (EDD) or CMR derived indexed LV end-diastolic volume (EDVi) residual >2sd above the mean, and LV ejection fraction (EF, CMR and echo) or fractional shortening (FS, echo only) residual at >2sd below the mean were considered consistent with DCM (67). Subjects were considered to have DCM morphology if any study met DCM imaging criteria. Non-ischemic DCM was defined as meeting DCM imaging criteria in the absence of CAD, defined as a history of a myocardial infarction, death attributed to coronary artery disease, percutaneous coronary intervention, or coronary artery bypass surgery. We identified nonischemic DCM in 40 (0.9%) of 4,305 FHS and 35 (0.7%) of 4,979 JHS participants. Clinical parameters used for linear regression models and outcome analyses included age, sex, BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), hypertension status (HTN), HTN treatment status (HTN_tx), diabetes status (DM), and hyperlipidemia. Hypertension was defined as SBP > 140 mmhg or DBP >90 mmhg or anti-hypertensive therapy. Anti-hypertensive therapy was defined as use of known anti-hypertensive medication either for an unknown indication or for the treatment of hypertension. Diabetes was defined as a fasting blood glucose >= 126 mg/dl or treatment for diabetes. Diabetes status in FHS exams 1 and 2 were subjected to clinical review. Cardiac disease was defined as CAD, heart failure, or coronary insufficiency. Hyperlipidemia was defined in FHS as total cholesterol > 240 mg/dl or high-density cholesterol < 40 mg/dl or triglycerides > 150 mg/dl in a fasting sample or use of an anti-dyslipidemia

6 medication and in JHS as total cholesterol >= 240 mg/dl or LDL >= 160 mg/dl or antidyslipidemia medication

7 Figure S1. Schematic representation of the unselected DCM cohort recruitment pathway and analyses. Referral for CMR with possible DCM and biobanked (n=960) Not DCM HCM 3 IHD 42 Infiltrative disease 5 LV Hypertrophy 18 Myocarditis 5 No CMR abnormality 104 LVNC 8 Other CV disease 107 Valvular disease 6 Aortic disease 2 Arrhythmia syndromes 8 ARVC 4 No CMR abnormality on 130 family screening Patients diagnosed with DCM by CMR scan following referral with another possible diagnosis Arrhythmia syndromes 4 ARVC 4 IHD 4 LV Hypertrophy 3 Other CV disease 3 Valvular disease 1 Clinical diagnosis of DCM but not meeting strict CMR criteria (n=55) Extended phenotype analyses (n=374) Patients meeting strict CMR criteria for DCM (n=319) TTNtv positive vs. TTNtv negative genotype / phenotype studies Positional effect of TTNtv on cardiac structure and function

8 Patients with a presumptive diagnosis of DCM (n=960) or an alternative cardiac diagnosis (n=19) were recruited to the biobank and demonstrated to have DCM based on clinical history and supportive CMR characteristics (n=55) or confirmed with DCM by strict CMR characteristics(33, 51) (n=319) by two experienced, Level 3 CMR accredited cardiologists. We excluded 442 patients referred with a possible diagnosis of DCM as not DCM based on CMR characteristics. Borderline EDVi and EF (EDVi>1.5SD above & EF>1.5SD below the mean for age and sex) were considered supportive CMR characteristics in the presence of corroborating clinical and/or echocardiographic evidence. Abbreviations: HCM, hypertrophic cardiomyopathy; IHD, ischemic heart disease; LVNC, LV non compaction; ARVC, arrhythmogenic right ventricular cardiomyopathy; TTNtv, TTN truncating variant.

9 Figure S2. TTN sequencing coverage for each cohort. A % callable B Exon number (size not to scale) % callable Exon number (size not to scale) Points for each cohort show the percentage of each exon sequenced with sufficient depth and quality to call variants, calculated as a mean across samples. Error bars show the 95%CI. Exon numbering is according to LRG_391. Vertical grey lines mark the domain boundaries between Z-disc, I-band, A-band, and M-band (from left to right). Points denote exon number and are not representative of exon size. Figure S2a shows all coding exons (2-364). Figure S2b shows the zoomed region from (A) within the I-band (exons ; 6315 bp; 2105aa; ~5% of the gene) that exhibited the poorest callability in all cohorts. The exons in this region are small and contain repetitive sequence affecting capture and mapping of reads in all cohorts.

10 Red = Healthy Volunteers, Orange = JHS, Magenta = FHS, Dark blue= Unselected DCM, Pale blue = Endstage DCM.

11 Figure S3. Sites susceptible to truncating events are non-uniformly distributed within the TTN gene but do not influence clustering effects in the A-band. splice susceptibility nonsense susceptibility codon position nonsense/splice susceptibility = number of potential nonsense/splice sites in a fixed window window size = 100bp for nonsense, 400bp for splice sites In the top track a cartoon of the TTN metatranscript is shown, with regions of the sarcomere demarcated as follows: Z-disk (red), I-band (blue), A-band (green), and M-band (purple). The PEVK domain of Titin is shown in yellow, the N2B sequence in dark green and the N2A sequence in dark blue. As exon lengths are non-uniform the number of exon boundaries (and hence sites at which splicing may be disrupted) are shown using a 400bp sliding window along the TTNtranscript. Short exons in the PEVK region lead to an excess of exon boundaries and potential splice site variants, with a relative depletion of splice sites in the A-band (P= 8.7x10-16 ; Fisher test for number of potential splice sites in the A-band vs. rest of the gene by length), where some exons are longer. Despite this, variants affecting splice donor / acceptor sites in DCM cases cluster in the A-band. Nonsense susceptibility is also non-uniform, the PEVK region is relatively invulnerable to nonsense variation (P= 1.3x10-8 (Fisher test for number of potential nonsense variants in the A-band vs. rest of the gene by length)), as the codons representing these four amino acids can not be converted to a stop codon through a single nucleotide substitution. While this may explain some of the observed clustering in nonsense variants, an enrichment of variants in the A-band is still observed when controlling for this effect (OR=3.6( ), P= 6x10-3 ; Fisher test for number of nonsense TTNtv in A-band vs. rest of gene by number of potential nonsense sites).

12 Figure S4. Alternative splicing of TTN in the human heart. PSI DCM GTEX Exon For each exon, the proportion spliced in (PSI) estimates the proportion of transcripts that incorporate that exon. Average exon usage is show for 84 human DCM hearts, and 105 cardiac samples from GTEx donors(27). Though the absolute levels of each transcript may differ, the overall pattern of exon usage in health and disease is equivalent (Pearson correlation coefficient = 0.98)

13 Figure S5. Time to events in FHS individuals grouped by TTNtv presence. TTNtrunc = TTNtv

14 Figure S6. Truncated transcript length is correlated with indices of cardiac impairment severity in DCM N LV EDVi b = 12, p = 0.59 C N LV ESVi b = 34.8, p = 0.13 C LV SVi b = 22.3, p = N C LV EF b = 18.4, p = N C RV EDVi b = 1.74, p = 0.91 N C N RV ESVi b = 21.5, p = 0.12 C RV SVi b = 22.9, p = N C N RV EF b = 21, p = C Lateral WTi b = 0.945, p = Septal WTi b = 0.47, p = 0.36 Max WTi b = 0.496, p = 0.43 LVMi b = 18.2, p = N C N C N C N C LA vol indexed b = 14.4, p = 0.36 N C Age at Dx b = 2.16, p = 0.79 N C Variant position (aa) We assessed 14 CMR sub-phenotypes in 43 prospectively recruited DCM subjects with TTNtv and strict CMR diagnostic criteria. A range of traits show a trend for more severe phenotypes associated with more distal variants (located towards the amino end (C) of the protein). These relationships were formally assessed using linear modelling, described above. LV / RV left / right ventricle; EDV end-diastolic volume; ESV end-systolic volume; SV stroke volume; EF ejection fraction; WT wall thickness; LVM LV mass; LA vol left atrial volume; i indexed to body surface area.

15 Figure S7. FHS exam 7 CMR.

16 Genotype-phenotype relationships are shown for 10 CMR phenotypes assessed in 12 individuals carrying TTNtv. Each circle or triangle represents measurements for a single individual. The molecular location of the TTNtv is represented on the x-axis, and the cardiac expression of the variant exon is represented on a color scale (PSI: 0 to 1, see legend). The blue histogram represents the phenotypic distribution in the full population. Linear models were constructed to predict each morphologic feature using empirically selected clinical covariates, TTNtv presence, and TTNtv exon expression as predictors. The P-value compares by ANOVA the fit of the model with and without TTNtv presence and expression. Circles are subjects with variants that correspond to nonsense, frameshift, or essential splice changes and triangles are subjects with non-essential splice variant predictions.

17 Figure S8. FHS and JHS additional CMR and echocardiography exams.

18

19 Each circle or triangle represents measurements for a single individual in JHS CMR exam 2 (a), FHS echocardiography exam 2 (b), exam 4 (c), exam 5 (d), exam 6 (e), exam 8 (f), or JHS echocardiography exam 1 (g). The shape color denotes TTNtv exon expression (PSI: 0 to 1, see legend). The blue histogram represents the distribution in the full population. Linear models were constructed to predict each morphologic feature using empirically selected clinical covariates, TTNtv presence, and TTNtv exon expression as predictors. The P-value compares by anova the fit of the model with and without TTNtv presence and expression.

20 Figure S9. mrna transcripts encoding truncated TTN proteins are expressed in human LV. 1 20AA FG MC AH JB PD BM JH RD Allele Balance DH JM SB DS JR SF EG JT01288 N C N C N Variant position C RNA sequencing was used to assess transcript expression in 84 LV samples, including 18 samples with TTNtv. Allelic expression of single nucleotide polymorphisms (SNPs) were assessed in RNA from each sample, and the allele balance of each SNP plotted against their location within the full-length TTN transcript (x-axis). The location of the TTNtv is shown as a vertical red line, and the allele balance of the TTNtv as a red dot. Note that the allele balance of the TTNtv allele does not deviate significantly from 0.5 (see also Main Text Figure 4). Other SNPs (heterozygous, black; homozygous, grey) found throughout the same sample are plotted. There is no consistent SNP allele imbalance throughout the length of the transcript, which would be expected if there were differential transcription or decay of the TTNtv and wild-type alleles.

21 Table S1. TTNtv identified in UK prospective DCM cohort. Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type Variants affecting meta-transcript (LRG_391_t1) 10AP c.4724_4728deltgaaa p.met1575serfsx Y Y Y Frameshift 10RW c.8307_8308deltg p.ala2770hisfsx Y Y Y Frameshift 10DW c.12643_12644delca p.gln4215valfsx Y Y Frameshift 14MW c.12757c>t p.gln4253x Y Y Nonsense 10GG c.27607g>a p.glu9203lys Y Splice variant prediction 10JF c a>c Y Canonical splice variant 12SK c.29062delg p.ala9688glnfsx Y Frameshift 12PB c g>a Y Canonical splice variant 10SB c delg Y Canonical splice variant 12MD c g>a Y Canonical splice variant 12MA c t>a Y Splice variant prediction 10RC c.41447delg p.gly13816alafsx Y Y Frameshift 10PC c.41473c>t p.arg13825x Y Y Nonsense 10JM c.43792delg p.val14598x Y Y Frameshift 10BC c.45756dupa p.tyr15253ilefsx Y Y Frameshift 10WS c.45812t>g p.leu15271x Y Y Nonsense 10VS c.47697c>a p.cys15899x Y Y Nonsense 10DR c g>a Y Y Canonical splice variant 10CH c.48527g>a p.trp16176x Y Y Nonsense 10JN c.50170c>t p.arg16724x Y Y Nonsense 10LP c.50170c>t p.arg16724x Y Y Nonsense 10RN c.50170c>t p.arg16724x Y Y Nonsense 10LB c.52035_52036instt p.leu17346phefsx Y Y Frameshift 10DH c.52223_52227dupagaaa p.asp17410argfsx Y Y Frameshift 12AM c g>t Y Y Canonical splice variant 10JS c g>a Y Y Splice variant prediction 10HH c g>a Y Y Canonical splice variant 10JF c.55525_55531delgacagga p.asp18509serfsx Y Y Frameshift 10SF c.55525_55531delgacagga p.asp18509serfsx Y Y Frameshift

22 Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type 10RN c a>g Y Splice variant prediction 10AL c t>c Y Y Canonical splice variant 12SR c.60931c>t p.arg20311x Y Y Nonsense 10TM c.63025c>t p.arg21009x Y Y Nonsense 10CS c.69630c>a p.tyr23210x Y Y Nonsense 12JL c.74306dupa p.asn24769lysfsx Y Y Frameshift 10BK c.78184g>t p.glu26062x Y Y Nonsense 10JC c.78507delt p.gly26170valfsx Y Y Frameshift 10DM c.81262_81269delcagatgct p.gln27088cysfsx Y Y Frameshift 10PP c.81262_81269delcagatgct p.gln27088cysfsx Y Y Frameshift 12NP c.81321c>g p.tyr27107x Y Y Nonsense 10CP c.82513dela p.ile27505phefsx Y Y Frameshift 10PP c.85090c>t p.arg28364x Y Y Nonsense 12MH c.86640delc p.his28881thrfsx Y Y Frameshift 12JL c.86967g>a p.trp28989x Y Y Nonsense 12ML c.87716delg p.gly29239aspfsx Y Y Frameshift 10MV c.89750dupg p.val29918serfsx Y Y Frameshift 10TD c.90086_90088delaaginsa p.glu30029aspfsx Y Y Frameshift 10KF c.92683c>t p.arg30895x Y Y Nonsense 10AG c.94849_94855delgatgccc p.ala31618tyrfsx Y Y Frameshift 10KW c.95415_ delcagt Y Y Canonical splice variant 10CS c.98265_98268dupaaca p.his32757asnfsx Y Y Frameshift 14CT c g>a p.trp33999x Y Y Nonsense 12PF c dela p.ala35544profsx Y Y Frameshift Variants affecting Novex 3 transcript only (LRG_391_t2) 12FB c.13022c>g p.ser4341x 48 na Y Nonsense

23 Table S2. TTNtv identified in the FHS offspring cohort. Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type Variants affecting meta-transcript (LRG_391_t1) c a>g Y Y Y Splice variant prediction c.3100g>a Y Y Y Splice variant prediction c.9727c>t p.q3243* Y Y Y Nonsense c.10799c>a p.s3600* Nonsense c.10852c>t p.q3618* Nonsense c.11183dupg p.g3728fs Frameshift c g>t Y Canonical splice variant c g>c Y Canonical splice variant c g>a Y Y Canonical splice variant c.52128_52128delt p.f17376fs Y Y Frameshift c.62506c>t p.r20836* Y Y Nonsense c.68824g>a Y Y Splice variant prediction c.80635c>t p.q26879* Y Y Nonsense c.87716_87716delg p.g29239fs Y Y Frameshift c.98551c>t p.r32851* Y Y Nonsense c.98551c>t p.r32851* Y Y Nonsense Variants affecting Novex 3 transcript only (LRG_391_t2) c.10443_10443dela p.k3481fs 48 na Y Frameshift c.13660_13661insa p.i4554fs 48 na Y Frameshift c.13705g>t p.e4569* 48 na Y Nonsense c.13939_13939delg p.e4647fs 48 na Y Frameshift c.15305_15305delc p.t5102fs 48 na Y Frameshift

24 Table S3. TTNtv identified in the JHS cohort. Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type Variants affecting meta-transcript (LRG_391_t1) J c.2494g>t Y Y Y Splice variant prediction J c.2841g>t Y Y Y Splice variant prediction J c.6355g>t p.e2119* Y Y Y Nonsense J c a>t Y Y Y Canonical splice variant J c.10592c>g p.s3531* Nonsense J c.11113_11113dela p.r3705fs Frameshift J c.11113_11113dela p.r3705fs Frameshift J c g>a Y Y Canonical splice variant J c.23386c>t p.r7796* Y Nonsense J c g>a Y Splice variant prediction J c.27607g>a Y Splice variant prediction J c a>c Y Canonical splice variant J c a>c Y Canonical splice variant J c a>c Y Canonical splice variant J c.31594g>t Y Splice variant prediction J c g>t Y Canonical splice variant J c g>a Y Y Splice variant prediction J c g>a Y Y Splice variant prediction J c g>a Y Y Splice variant prediction J c g>a Y Y Splice variant prediction J c.68824g>a Y Y Splice variant prediction J c.75123t>a p.y25041* Y Y Nonsense J c.85008_85011deltagt p.v28336fs Y Y Frameshift J c.91715_91716insa p.n30572fs Y Y Frameshift J c.92288_92289insaaaag p.s30763fs Y Y Frameshift J c.92288_92289insaaaag p.s30763fs Y Y Frameshift J c a>g Y Y Splice variant prediction J c _102630dela p.k34210fs Y Y Frameshift J c g>t p.e34470* Y Y Nonsense

25 Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type J c c>t p.r34698* Y Y Nonsense J c delg Y Y Canonical splice variant Variants affecting Novex 3 transcript only (LRG_391_t2) J c.13410_13411insa p.k4470fs 48 na Y Frameshift

26 Table S4. TTNtv identified in end-stage DCM. Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type Previously published Variants affecting meta-transcript (LRG_391_t1) 20HS c.3100g>a p.val1034met Y Y Y Splice variant prediction Y 20SF c.3100g>a p.val1034met Y Y Y Splice variant prediction Y 20AH c.3100g>a p.val1034met Y Y Y Splice variant prediction 20VW c.40558g>c p.val13520leu Y Y Splice variant prediction Y 20JT c.44281c>t p.pro14761ser Y Y Splice variant prediction 20MC c g>a Y Y Canonical splice variant 20SG c.46236c>a p.c15412* Y Y Nonsense Y 20JH c.46782c>a p.tyr15594* Y Y Nonsense 20JB c g>a Y Y Canonical splice variant 20RD c.50247_50247delt p.f16749fs Y Y Frameshift Y 20JL c.53488g>t p.g17830* Y Y Nonsense Y 20PD c.58172dela p.asp19391fs Y Y Frameshift 20VW c g>a Y Y Canonical splice variant Y 20EG c.63601c>t p.arg21201* Y Y Nonsense 20SW c.64453c>t p.r21485* Y Y Nonsense Y 20JD c.67495c>t p.r22499* Y Y Nonsense 20PS c g>a Y Y Canonical splice variant Y 20CW c.69843_69843dela p.k23281fs Y Y Frameshift Y 20HS c.70791_70791dela p.e23597fs Y Y Frameshift Y 20FG c.74338c>t p.arg24780* Y Y Nonsense 20SF c.76116_76117insa p.asn25372fs Y Y Frameshift 20SB c.76383_76386deltaat p.asn25462fs Y Y Frameshift 20DH c.81518delc p.pro27173fs Y Y Frameshift 20LN c.86459_86460delct p.s28820fs Y Y Frameshift Y 20BM c.87624c>a p.y29208* Y Y Nonsense Y 20DB c.89900_89903delatta p.n29967fs Y Y Frameshift Y

27 Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type Previously published Variants affecting meta-transcript (LRG_391_t1) 20AA c.90322_90323inst p.glu30108* Y Y Frameshift 20AK c.93166c>t p.r31056* Y Y Nonsense Y 20JM c.94562dupc p.thr31522fs Y Y Frameshift 2 0PA c.96460_96461insa p.t32154fs Y Y Frameshift Y 20DS c.98265_98268dupaaca p.his32757fs Y Y Frameshift 20LB c.98299_98300delag p.r32767fs Y Y Frameshift Y 20PS c.98506c>t p.r32836* Y Y Nonsense Y 20IR c c>a p.s33482* Y Y Nonsense Y 20JR c c>a p.ser33482* Y Y Nonsense 20RD c _101022delga p.arg33674fs Y Y Frameshift 20JM c g>a Y Y Canonical splice variant Variants affecting Novex 3 transcript only (LRG_391_t2) c.14470_14471inscacactcc 20JL ATA p.r4824fs 48 na Y Frameshift Y Previously published: Some full-length Titin truncations presented in the current study were previously reported (12) with reference to ENSEMBL transcript ENST Here they are noted with reference to standard LRG transcripts, and annotations are updated. Splice variant prediction criteria in this study differ from those used previously, as explained in the Methods and Supplementary Material.Both 20IR01555 and 20JR01203 were independently recruited as probands, but retrospectively identified as very likely related. Both samples have been retained in this cohort as recruitment was not biased by family history. 20JM01785 contains 2 TTNtv, a frameshift and a canonical splice variant. 20HS01530 contains 2 TTNtv, a frameshift and a splice variant prediction. Without phasing it is not known whether these variants are in cis or trans. For burden testing between cohorts we have used the number of individuals with TTNtv rather than the total number of TTNtv.

28 Table S5. TTNtv identified in healthy volunteers. Sample ID Hg19 genomic start position LRG genomic start position Transcript effect Protein effect Exon number Median exon PSI Present in N2BA Present in N2B Present in Novex-3 Variant type Variants affecting meta-transcript (LRG_391_t1) 14AG c.3100g>a p.val1034met Y Y Y Splice variant prediction 14SS c.10852c>t p.gln3618x Nonsense 14JD c.21142c>t p.arg7048x Y Nonsense 14EC c a>g Y Canonical splice variant 14KN c g>a Y Splice variant prediction 14AH c g>c Y Canonical splice variant 14MO c.45391dela p.ile15131tyrfsx Y Y Frameshift 14JM c.67159dela p.ile22387x Y Y Frameshift

29 Table S6. Titin reference transcript and protein identifiers. Isoform Description LRG Ensembl transcript Ensembl protein RefSeq transcript RefSeq protein Uniprot Length (aa) meta-transcript Inferred complete meta-transcript 391_t1 ENST ENSP NM_ NP_ N2-BA (long) Principal cardiac long isoform ENST ENSP NM_ NP_ Q8WZ N2-A Soleus / skeletal long isoform ENST ENSP NM_ NP_ Q8WZ N2-B (short) Principalcardiac short isoform ENST ENSP NM_ NP_ Q8WZ Novex-1 Minor cardiac short isoform ENST ENSP NM_ NP_ Q8WZ Novex-2 Minor cardiac short isoform ENST ENSP NM_ NP_ Q8WZ Novex-3 Minor small cardiac isoform 391_t2 ENST ENSP NM_ NP_ Q8WZ

30 Table S7. Overview of TTN transcripts and exon usage. For each of 364 TTNexons, Hg19 and LRG genomic co-ordinates are shown, along with the coordinates of the amino-acid residues encoded by that exon, using LRG 391_t1 (meta-transcript) protein coordinates. The recommended exon numbering is shown alongside the historical numbering system in use in much of the literature. The exon rank in published cardiac transcripts is shown next for seven important transcripts, followed by a description of the exon and its relationship to important protein features. Exon length and an indication of symmetry is given, where the length of symmetrical exons is a multiple of 3, so that removal of the exon does not alter the reading frame. An estimate of exon usage in the heart is shown derived from RNAseq data from both DCM and unselected donor hearts. PSI indicates proportion spliced-in the proportion of expressed transcripts that contain the exon listed (See text for details). Recommended exon number (HAVANA) Historical exon number (GenBank: AJ )(26) Hg19 genomic start position Hg19 genomic end position LRG genomic start position LRG genomic end position Metatranscript protein start position Metatranscript protein end position Metatranscript exon number N2BA exon number N2A exon number N2B exon number Novex-1 exon number Novex-2 exon number Novex-3 exon number Exon status Protein Region Protein Domain Exon coding length Symmetric (S) / Asymmetric (A) Average exon PSI (DCM) Average exon PSI (GTEx) ' utr 0 NA Z-disk Ig-like 1 91 A Z-disk Ig-like S Z-disk Ig-like S Z-disk 86 A Z-disk 245 A Z-disk 331 A Z-disk Z-repeat S Z-disk Z-repeat S Z-disk Z-repeat S Z-disk Z-repeat S Z-disk Z-repeat S Z-disk Z-repeat S Z-disk Z-repeat S Z-disk 123 S near Z-disk 282 S near Z-disk 66 S near Z-disk Ig-like A near Z-disk 64 A near Z-disk Ig-like S near Z-disk Ig-like A near Z-disk 206 A near Z-disk Ig-like S near Z-disk Ig-like A near Z-disk Ig-like A

31 near Z-disk Ig-like S near Z-disk Ig-like A near Z-disk / Ig-like 7/8/9/ A I-band I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band 261 S I-band Ig-like S I-band 264 S I-band 261 S I-band Ig-like S I-band Ig-like S I-band Ig-like A I-band Ig-like A I-band 232 A I-band Ig-like S I-band Ig-like S I-band Ig-like S Novex S Novex S I-band 57 S Novex A N2B unique I-band Ig-like 21/22/ S sequence I-band Ig-like S I-band Ig-like 25/ S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S

32 I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S I-band Ig-like S N2A unique I-band Ig-like S sequence N2A unique I-band Ig-like A sequence N2A unique I-band 184 A sequence N2A unique I-band 90 S sequence N2A unique I-band 268 A sequence N2A unique I-band Ig-like S sequence N2A unique I-band Ig-like S sequence N2A unique I-band Ig-like S sequence I-band 27 S I-band 60 S I-band 84 S I-band 72 S I-band 48 S I-band 405 S I-band 63 S I-band 78 S I-band 78 S I-band 87 S I-band 81 S I-band PEVK 1 84 S

33 I-band PEVK 2 84 S alternative exon - removed from Havana model I-band PEVK 3 84 S I-band PEVK 4 81 S I-band PEVK 5 84 S I-band PEVK 6 84 S I-band 102 S I-band 114 S I-band 81 S I-band 78 S Low cardiac I-band PEVK7 84 S expression?(12) Low cardiac I-band PEVK8 84 S expression?(12) Low cardiac I-band 84 S expression?(12) I-band PEVK9 84 S I-band 81 S I-band 207 S I-band 78 S I-band 75 S Low cardiac I-band 93 S expression?(12) Low cardiac I-band 78 S expression? (12) Low cardiac I-band 162 S expression?(12) Low cardiac I-band PEVK10 84 S expression? (12) Low cardiac I-band PEVK11 78 S expression?(12) I-band PEVK12 84 S I-band PEVK13 84 S I-band PEVK14 84 S Low cardiac I-band 297 S expression?(12) New in meta 87 S transcript Low cardiac I-band 75 S expression? (12) S I-band 75 S I-band 96 S I-band 78 S I-band 69 S Low cardiac I-band 75 S expression?(12) Low cardiac I-band 297 S expression?(12) Low cardiac I-band PEVK15 81 S expression? (12) I-band PEVK16 78 S S S