Immunoprecipitation (IP) was performed by using custom ELC antibodies covalently coupled to magnetic beads. Liquid chromatography followed by mass spectrometry (LC-MS) was used for identification. A Validation of ELC precipitation by immunoblot (IB) before performing LC-MS (see B and C). B Classification of ELC enriched proteins identified by LC-MS. From 3452 spectra 110 proteins were identified in total. 106 proteins were found to interact with ELC. 3 different kinases were found. (Peptide FDR: 5%, minimum of two identified peptides per protein) C Ranking of ELC enriched proteins by foldchange of unique peptide counts enriched by ELC compared to unique peptide counts of the same protein captured by the isotype control (IgG Ctrl). Data of the screening experiment are shown without statistical analysis. Candidates were validated using independent methods. The string intersection score was calculated for known ELC interacting proteins⁶⁷. (NA: not available) D Two representative mass spectrometry spectra of unique peptides accepted to identify NIMA related kinase 9 (NEK9) (left panel) or calcium/calmodulin-dependent protein kinase type II subunit gamma (CamK2G) (right panel). The protein coverage is shown above (red boxes). E Validation of ELC-kinase interaction. Human myc-tagged ELC protein was overexpressed in human cells followed by specific immunoprecipitation (IP) and immunoblot (IB) analysis. Ca²⁺-influx was obtained by ionophore stimulation. One representative experiment is shown (see also Fig. 2D). F Schematic illustration of ascorbate peroxidase (APEX) catalyzed proximity labeling. Overexpression of ELC fused to APEX enables biotin-labeling of proteins in close proximity. The reaction is catalyzed by hydrogen peroxide (H₂O₂). Streptavidin precipitation followed by LC-MS analysis identified NEK9 and CamK2G. The foldchange of unique peptide counts of biotin-labeled proteins in the proximity of ELC (APEX-ELC) compared to unique peptide counts of the same protein in the APEX-Ctrl sample is shown. Data of the validation experiment are shown without statistical analysis.

A NEK9 and CamK2G RNA expression in human total RNA tissue panel. Ribosomal protein large subunit P0 (RPLP0) was used as housekeeping gene. B NEK9 (left panel) and CamK2G (right panel) protein expression in piscine and human heart tissue to detect cross-species applicability. C mRNA expression profile of NEK9 and CamK2G in different human tissues from 4 weeks post-conception to adulthood⁶⁸. D, E Relationship of NEK9-ELC binding efficiency on intracellular Ca²⁺-concentration. NEK9-ELC interaction was analyzed after overexpression of myc-tagged ELC protein in human cells followed by specific immunoprecipitation (IP) and immunoblot (IB) analysis (D). Ca²⁺-influx was obtained by ionophore stimulation. Quantification of NEK9-ELC interaction dependent on Ca²⁺-influx illustrated as the averaged intensity of precipitated NEK9 protein normalized to actin protein expression of input protein lysate (E). Data are mean ± SD (n = 3). **P < 0.01 by the analysis of ordinary one-way ANOVA followed by Bonferroni’s multiple comparisons test.

A Validation of siRNA mediated knockdown (KD) of NEK9 in HEK293 cells by quantitative real time PCR. mRNA expression shown as 2^(−ΔCt) normalized to RPLP0 (ribosomal protein large subunit P0). Ca²⁺-influx was obtained by ionophore stimulation. Data are mean ± SD (n = 3). ****P < 0.0001 by analysis of ordinary one-way ANOVA followed by Bonferroni’s multiple comparisons test. B Representative immunoblot (IB) of NEK9 protein expression after specific knockdown (KD). C 2D immunoblot of ELC-phosphorylation pattern after myc-tagged ELC overexpression and Ca²⁺ -influx using myc-specific antibody. After protein isolation samples were split and half of the sample was treated with phosphatase inhibitor (Native) and the other half was incubated with phosphatase (Dephospho). D Representative 2D immunoblot of myc-tagged ELC-phosphorylation pattern using myc-specific antibody (above) and ELC-phosphorylation illustrated as averaged intensity normalized to total ELC protein (down). ELC-phosphorylation was analyzed after myc-tagged ELC overexpression and siRNA knockdown (KD) of NEK9 or after overexpression (Ox) of NEK9 in human cells. Ca²⁺-influx was obtained by ionophore stimulation. Data are mean ± SD (n = 3, n = 5 for Ctrl group). *P < 0.05, **P < 0.01 by the analysis of two-tailed paired Student’s t-test. E Relative ELC protein phosphorylation (sum of phosphorylated (+1P + 2P) ELC forms) illustrated as averaged intensity normalized to total ELC protein. Data are mean ± SD (n = 3, n = 5 for Ctrl group). **P < 0.01 by the analysis of ordinary one-way ANOVA followed by Bonferroni’s multiple comparisons test.

A Lateral view of zebrafish embryo hearts 72 h post fertilization (hpf) injected with control oligonucleotides or anti-sense oligonucleotides blocking the translational start site of nek9 (KD NEK9) (left). Immunostaining of atrium-specific (A) and ventricle-specific (V) myosin heavy chains (green: antibody against atrial-specific myosin (S46), red: antibody against ventricular and atrial myosin (MF20)) (right). B Phenotype rescue of KD NEK9. Zebrafish eggs were injected with nek9 mRNA (Ox + KD NEK9) followed by KD NEK9 in one-cell stage. Analysis was performed at 72 hpf. Data are mean ± SD (Ctrl, n = 143; KD NEK9, n = 192; Ox + KD NEK9, n = 155 fish embryos out of 3 experiments). *P < 0.05; **P < 0.01; ***P < 0.001 by the analysis of two-tailed paired Student’s t-test. C Fractional shortening (FS) of zebrafish morphants at 72 hpf. Data are mean ± SD (Ctrl, n = 25; KD NEK9, n = 21; Ox + KD NEK9, n = 21 fish embryos out of 3 experiments). *P < 0.05; **P < 0.01 by the analysis of two-way ANOVA followed by Bonferroni’s multiple comparisons test. The respective heart rate is shown in Supplementary Fig. 4A. D–F Two heterozygous transgenic zebrafish lines were generated by CRISPR/Cas9. Heterozygous transgenic zebrafish nek9^78del/+ and nek9^+/500del (F1 parents) were incrossed and F2 generation was analyzed at 72 hpf regarding heart morphology (D), penetrance of heart failure phenotype (E) and FS (F). Recordings were performed for all embryos showing a heart failure phenotype compared to randomly chosen wild type looking littermates. After phenotype analysis all embryos were genotyped as shown in Supplementary Fig. 5B and genotypes were associated to phenotypes (total number of genotype (number of phenotype: wild type looking/heart failure/severe heart failure): nek9^+/+, n = 77 (71/4/2); nek9^78del/+, n = 64 (40/20/4); nek9^+/500del, n = 19 (6/10/3); nek9^78del/500del, n = 19 (0/16/3)). Statistical significance in F was calculated by a mixed effect model followed by Bonferroni’s multiple comparisons test. *P < 0.05; **P < 0.01; ***P < 0.001 are shown for the corresponding chamber of Ctrl embryos.

Two heterozygous transgenic zebrafish lines were generated by CRISPR/Cas9. A NEK9^78del protein is lacking the ATP-binding domain shown in the 3D structure of NEK9 (A). NEK9^500del is harboring a frame shift mutation leading to a premature stop codon. Sanger sequencing results of the affected allele are shown in Supplementary Fig. 5C. B, C Piscine ELC and mutated piscine myc-tagged NEK9 protein either lacking the ATP-binding domain (NEK9^78del) or harboring a premature stop codon (NEK9^500del) were overexpressed in HEK293 cells. The ELC-NEK9 interaction was analyzed by specific immunoprecipitation (IP) and immunoblot (IB) (B). Quantification of ELC-NEK9 interaction illustrated as averaged intensity of precipitated ELC protein normalized to ELC expression of input protein lysate (C). Data are mean ± SD (n = 3). *P < 0.05; ***P < 0.001 by the analysis of ordinary one-way ANOVA followed by Bonferroni’s multiple comparisons test. D, E Validation of ELC-NEK9 interaction. Human myc-tagged ELC protein and different deletion variants of human flag-tagged NEK9 protein were overexpressed and interaction intensity was quantified after IP and IB. Data are mean ± SD (n = 3). **P < 0.01 by the analysis of ordinary one-way ANOVA followed by Bonferroni’s multiple comparisons test.

A Schematic illustration of genetic sensitizing by nek9 in an ELC phospho-deficient genetic background. The concept is based on the progressive amplification of a response by functionally coupled factors. B Penetrance of NEK9 KD in zebrafish embryos at 72 hpf. Different concentrations of antisense oligonucleotides against zebrafish nek9 were injected in a genetic wild-type background (KD NEK9: knockdown of NEK9 resulting in a significant heart failure phenotype; LD NEK9: low-dose knockdown showing no phenotypical effect). Data are mean ± SD (Ctrl, n = 517; LD NEK9, n = 162; KD NEK9, n = 461 fish embryos out of 4 (Ctrl), 3 (LD NEK9) and 5 (KD NEK9) experiments). For statistical analysis a mixed effect model followed by Bonferroni’s multiple comparisons test was used. **P < 0.01 shown for the corresponding embryos injected with Ctrl oligonucleotides. C Validation of NEK9 KD or LD NEK9 by immunoblot (IB). D Lateral view of heterozygous lazy susan (laz^m647/+) embryo hearts 72 hpf injected with control oligonucleotides or low-dose antisense oligonucleotides against zebrafish ilk (LD ILK) or nek9 (LD NEK9). E Genotype-phenotype-association of laz^m647 zebrafish injected with control oligonucleotides, LD ILK or LD NEK9. Individual genotyping was performed after blinded phenotyping at 72 hpf. Data are mean ± SD (Ctrl: laz^+/+n = 41, laz^m647/+n = 81; LD ILK: laz^+/+n = 21, laz^m647/+n = 46; LD NEK9: laz^+/+n = 49, laz^m647/+n = 97 fish embryos out of 3 experiments). *P < 0.05, **P < 0.01 by the analysis of two-tailed paired Student’s t-test. F Fractional shortening (FS) of genetic sensitized lazy susan (laz^m647) embryos. After analyzing FS, individual genotyping was performed and associated with the phenotype. Data are mean ± SD (Ctrl: laz^+/+n = 41, laz^m647/+n = 82; LD ILK: laz^+/+n = 21, laz^m647/+n = 46; LD NEK9: laz^+/+n = 49, laz^m647/+n = 97 fish embryos out of 3 experiments). ***P < 0.001 by the analysis of ordinary one-way ANOVA followed by Bonferroni’s multiple comparisons test.

A Representative 2D silver stain of human LV tissue. B Selected spot areas 1 to 6 were analyzed by liquid chromatography followed by mass spectrometry (LC-MS). Peptide counts uniquely assigned to ventricular (v) ELC or vRLC are shown for 6 independent samples. The mean percentage of the protein coverage of vELC and vRLC are given above (table inset). Data are mean ± SD. C Schematic illustration of identified ELC or RLC-phosphorylation sites and deamidated amino acid residues detected in human LV tissue (n = 6; DCM: n = 4; Ctrl: n = 2). D Validation of polyclonal rabbit antibody against human ELC (Biozol, GeneTex, ZF127578) by immunoblot (IB) of different amounts of human LV protein. No cross reactivity with vRLC was detected. E Quantification of relative ELC protein expression illustrated as averaged intensity normalized to total ELC protein. ELC expression intensities were plotted against the amount of human heart protein. Exposure time was varied from 0.25 to 5 min. For 2 min exposure R = 0.9383 and P < 0.0001 was calculated by two-tailed Spearman Rank Correlation (n = 5). F Validation of in-vitro dephosphorylation of human LV tissue. After protein isolation samples were split and half of each sample was treated with phosphatase inhibitor (Native), while the other half was incubated with phosphatase (Dephospho). Proteins of the same aliquot were analyzed either by ELC or RLC 2D immunoblot (IB) (left panel) or IB against phospho-cardiac troponin I (TnI) (right panel). The expression of total cardiac TnI was used as loading control. The same aliquots were used for Pro-Q® Diamond Phosphoprotein Stain to detect phosphorylated protein species in 2D IB (Supplementary Fig. 9).

A Ratio of ventricular (v) MLC and atrial (a) MLC mRNA expression in human left ventricular tissue of DCM patients and HTX controls (HTX: n = 28; DCM: n = 44). Samples were analyzed by deep mRNA sequencing (mRNAseq)³⁸. Data are mean ± SEM. Statistical significance was tested by two-tailed unpaired Student’s t-test. B Quantification of relative basal (Basal) and phosphorylated (+1P, + 2P) ELC protein illustrated as averaged intensity normalized to total ELC protein (Control: n = 9; DCM: n = 11). 6 HTX controls (Supplementary Table 1) and 3 healthy controls were included. Data are mean ± SD. **P < 0.01 by the analysis of two-tailed unpaired Student’s t-test. C Representative ELC 2D immunoblot (IB) of human LV heart tissue of a DCM patient and a healthy control. Native ELC protein forms (left panel) and ELC protein forms after in-vitro dephosphorylation (right panel). Exposure times were varied from 1 to 5 min. D Correlation analysis of phosphorylated ELC expression (+2P) and left ventricular ejection fraction (LVEF). R = −0.6193 and P = 0.0036 by the analysis of two-tailed Pearson Correlation (Control: n = 9; DCM: n = 11). E Sum of phospho peptides detected in spot arrays assigned to ventricular (v) ELC (see Fig. 7A, B) normalized to total ELC peptide counts. Proteins were separated by 2D electrophoresis and selected spot areas were analyzed by liquid chromatography followed by mass spectrometry (LC-MS) (DCM: n = 4; Ctrl: n = 2). Data are mean ± SD and are shown without statistical analysis. F Distribution of ELC-phosphorylation in DCM patients compared to healthy controls (Ctrl). Normalized amount of ELC phospho-peptides shown for the respective ELC-phosphorylation site (DCM: n = 4; Ctrl: n = 2). Data are mean ± SEM and are shown without statistical analysis.