Functional analysis of a gene-edited mouse model to gain insights into the disease mechanisms of a titin missense variant

Titin truncating variants are a well-established cause of cardiomyopathy; however, the role of titin missense variants is less well understood. Here we describe the generation of a mouse model to investigate the underlying disease mechanism of a previously reported titin A178D missense variant identified in a family with non-compaction and dilated cardiomyopathy. Heterozygous and homozygous mice carrying the titin A178D missense variant were characterised in vivo by echocardiography. Heterozygous mice had no detectable phenotype at any time point investigated (up to 1 year). By contrast, homozygous mice developed dilated cardiomyopathy from 3 months. Chronic adrenergic stimulation aggravated the phenotype. Targeted transcript profiling revealed induction of the foetal gene programme and hypertrophic signalling pathways in homozygous mice, and these were confirmed at the protein level. Unsupervised proteomics identified downregulation of telethonin and four-and-a-half LIM domain 2, as well as the upregulation of heat shock proteins and myeloid leukaemia factor 1. Loss of telethonin from the cardiac Z-disc was accompanied by proteasomal degradation; however, unfolded telethonin accumulated in the cytoplasm, leading to a proteo-toxic response in the mice.We show that the titin A178D missense variant is pathogenic in homozygous mice, resulting in cardiomyopathy. We also provide evidence of the disease mechanism: because the titin A178D variant abolishes binding of telethonin, this leads to its abnormal cytoplasmic accumulation. Subsequent degradation of telethonin by the proteasome results in proteasomal overload, and activation of a proteo-toxic response. The latter appears to be a driving factor for the cardiomyopathy observed in the mouse model. Supplementary Information The online version contains supplementary material available at 10.1007/s00395-021-00853-z.


Cardiomyocyte contractility and calcium transients
For measurements on the IonOptix µstep apparatus, isolated cardiomyocytes were allowed to settle on a 1.5 mm coverslip in the perfusion chamber for 5 minutes before being perfused with 37 °C tyrode solution supplemented with 1.4 mM Ca 2+ and electrically paced with 40 volts at a frequency of 1 Hz. Cardiomyocyte contraction was assessed using phase-contrast microscopy. Cells with poor morphology (e.g. excessive blebbing or asynchronous contraction) were not measured. Cells with basal sarcomere lengths or contraction times more than two standard deviations from the mean were excluded as an artefact of the deteriorating function of primary cells with time in culture.
Ratiometric measurement of intracellular calcium ([Ca 2+ ]i) transients were done in fura-2 loaded LV cardiomyocytes isolated from the hearts of WT or TG mice. After isolation, cardiomyocytes were loaded for 8 mins with Fura2-AM (3 µM) in a low-Ca 2+ Tyrode solution (500 µM Ca 2+ + Pluronic) followed by two washes in low-Ca 2+ Tyrode to remove any unloaded

Cardiomyocyte size and immunofluorescence
Isolated cardiomyocytes were deposited on poly-D-lysine coated slides using a StatSpin cytofuge at 700 rpm for 2 min, then fixed for 10 minutes with 4 % paraformaldehyde in PBS.
100x magnification bright-field images of 27-39 rod-shaped cardiomyocytes per mouse were obtained. ImageJ was used to measure the area, maximum width and maximum length of each cardiomyocyte. Immunofluorescence on fixed adult mouse cardiomyocytes on poly-Dlysine coated slides (see above) was performed as described [3,10] using the primary antibodies indicated in Table S11.
For T-tubular visualisation, 1 µM of di-4-ANEPPS and 2 µM Pluronic® F-127 (Life Technologies) were directly added to freshly isolated cardiomyocytes for 20 minutes, cells were washed with storage buffer. They could then be visualised immediately using a Leica TCS SP5 X confocal microscope equipped with a 433 nm argon laser and a 63X oil lens.
Images were acquired using a photomultiplier set between 500-600 nm for optimum image resolution and fluorescence intensity.

High Resolution Episcopic Microscopy (HREM)
Adult hearts were collected and flushed with PBS to remove blood. Hearts were fixed overnight in 4% paraformaldehyde in PBS prior to extensive washing with PBS. A methanol series was performed (10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 100%, 100%) for 2 hours each prior to infiltration overnight in a 50:50 mixture of 100% methanol and JB-4 resin (18570-1, Polysciences), including 0.275g/100ml eosinB (Sigma Aldrich) and 0.055g/100ml acridine orange (Sigma Aldrich). Hearts were washed in the JB-4 infiltration solution for 1 hour prior to incubation for 1 week in fresh JB-4 solution. Hearts were sectioned at 3 µm using an optical HREM microscope (Indigo Scientific). Sections were reconstructed in 3D using Horos DICOM viewer (The Horos Project) and Amira (ThermoFisher Scientific) software. Horos was used to calculate ventricular length and width, expressed as a ratio of length:width. Average myocardial wall thickness was calculated from 5 separate measurements for each heart along the ventricular wall. To assess the fractal dimensions of left ventricle trabeculae, 20 evenly spaced sections from base to apex of the heart were assessed using the LV Fractals Analysis plugin for Osirix and Horos [2].

Electron microscopy
Specimens for electron microscopy were prepared from frozen samples which were brought to 0 °C in a buffer containing PBS, 1 mM MgCl 2, 5 mM EGTA and 20 mM 2,4, butanedione monoxime (BDM). The pieces were then fixed in 2.5 % glutaraldehyde, 2 % PFA in the same buffer for 90 min on ice. After washing, small pieces were subsequently fixed in 1 % osmium tetroxide, dehydrated in alcohol and embedded in Araldite. Thin sections were stained with UranyLess stain followed by lead citrate. Sections were viewed in a JEM 1400 transmission electron microscope in the Centre for Ultrastructure Imaging, King's College London.
For assessment of Z-disc width, data was obtained from scans of Z-discs in Image J (v.1.53c) where it was judged there was little or no overlap of the ends of the thick filaments with the Z-disc. Width limits were taken as positions at half the density difference between the central Z-disc and the neighbouring I-band.

Passive tension measurements
Left ventricular endocardial septal trabeculae were dissected in dissecting solution and demembranated over night at 4 °C in skinning solution. Before the experiment, preparations were further thinned out, and aluminium "T-clips" were attached to each end. The Aurora Scientific (Aurora, ON, Canada) 1400A permeabilised fibre apparatus, with 403A force transducer and 315C-I length controller, was used for tension mesurements and length control. After mounting, the fibre the baseline sarcomere length (SL) was set at 2.0 µm (1.0 L0) using laser diffraction, and allowed to stabilise for ten minutes in relaxing solution. The preparation length and thickness in two dimension was measured. The latter were used to approximate cross sectional area to an elliptical form. A series of four incremental stretches of 50 ms duration, increasing the fibre length on 5 % increments, length clamped for 1 s at each increment, to a maximum of 1.2 L0 was performed followed by a return to SL 2.0.
Incubation in relaxing solution supplemented with 0.9 M KCl and then relaxing solution with 0.6 M KI as described elsewhere [12] removed the titin derived passive tension after which the stretch procedure was repeated. To derive titin generated tension, post-treatment values were subtracted from pre-treatment values. All tension values were normalised to cross sectional area (mN/mm2) and titin fraction of total tissue derived by dividing titin derived tension by total tension. All experiments were carried out at 15 °C.

Immunofluorescence on cryo-sections
Cryo-section were done on O.C.T. (VWR) embedded tissue on a ryotome Cryotome FSE (Thermo Scientific; 10 µm thickness). Cryo-sections from tibialis anterior (cross-sections) were performed as described [1], using laminin to visualise the extracellular matrix [11]. Cryosections on cardiac tissue were performed as above and unfixed sections stained and imaged as described for isolated cardiomyocytes. Antibodies and their dilutions are given in Table   S11.

Histology
Histology on paraffin embedded samples (7 µm sections) was performed with hematoxylin and eosin or with Sirius red, using standard protocols.             Table S4. Heart weight normalised to tibial length on the same animals is also shown (Student's test). All animals in this experiment were male.    derived fraction of total passive tension was assessed, C -After extraction of titin derived tension, the remaining tension was assumed ascribable to extracellular collagen. D -Titin derived fraction of total passive tension was assessed. All parameter were found to be normal (Student's t-test; n = 6, age WT 132 ± 2 d/A178D 129 ± 1 d, 4M/2F).

Figure S9
The plateau in titin derived tension (B), and reduced titin derived fraction of tension (D) at longer stretches reflect the known reduced dependency of titin in passive tension regulation at the higher sarcomere lengths [4].
For values at each stretch length, see Table S8. All measurements are normalised to Gapdh; (n = 6; Kruskal-Wallis followed by Wilcoxon rank sum test with Bonferroni correction). All mice in this experiment were male, age 120 ± 2 d.    Figure S17: Analysis for Fhl1 as in described in Figure S16. Please note, in the 3 months cohort (A), Fhl1 (arrow) was not detectable in WT or A178D samples, but clearly detectable in a positive control ('C' = Csrp3 null heart). * Gapdh signal bleed through on the membrane. D -3 months cohort not quantified.    Ankrd1 50 not quantified Figure S19: Analysis for Rcan1 as in Figure S16. Arrow indicates lane excluded from quantification due to high background.            Figure S16.
All measurements are normalised to Gapdh; Kruskal-Wallis followed by Dunn's multiple comparison's test; n = 6, all males, age WT 412 ± 2 d , HET 412 ± 2 d, A178D 423 ± 3 d. C -Assessment in the adrenergic challenge cohort. All measurements are normalised to Gapdh; (n = 6; age 120 ± 2 d, all males; Kruskal-Wallis followed by Wilcoxon rank sum test with Bonferroni correction). Hspa1a is shown separately on the right with a different scaling of the y-axis.  Figure S29: Analysis for Bag3 as in described in Figure S16. Figure S30: Analysis for Hsp70 as in described in Figure S16. Figure S31: Analysis for Hsc70 as in described in Figure S16. Figure S32: Analysis for αβ-crystallin as in described in Figure S16. Figure S33: Analysis for Hsp27 as in described in Figure S16.