Alternative Splicing of Titin Restores Diastolic Function in a HFpEF-like Genetic Murine Model (Ttn^ΔIAjxn)

Abstract

Rationale

HFpEF patients experience elevated filling pressures and reduced ventricular compliance. The splicing factor RBM20 regulates the contour length of titin’s spring region and thereby determines the passive stiffness of cardiomyocytes. Inhibition of RBM20 leads to super compliant titin isoforms (N2BAsc) that reduce passive stiffness.

Objective

To determine the therapeutic potential of upregulating compliant titin isoforms in a HFpEF-like state in the mouse.

Methods and Results

Constitutive and inducible cardiomyocyte specific RBM20 inhibited mice were produced on a Ttn^ΔIAjxn background to assess the effect of upregulating compliant titin at the cellular and organ levels. Genetic deletion of the I-band – A-band junction (IAjxn) in titin increases strain on the spring region and causes a HFpEF-like syndrome in the mouse without pharmacological or surgical intervention. The increased strain represents a mechanical analogue of deranged post-translational modification of titin that results in increased passive myocardial stiffness in HFpEF patients. Upon inhibition of RBM20 in Ttn^ΔIAjxn mice, compliant titin isoforms were expressed and diastolic function was normalized, exercise performance was improved and pathologic hypertrophy was attenuated.

Conclusions

We report for the first time a benefit from upregulating compliant titin isoforms in a murine model with HFpEF-like symptoms. Constitutive and inducible RBM20 inhibition improves diastolic function resulting in greater tolerance to exercise. No effective therapies exists for treating this pervasive syndrome, therefore our data on RBM20 inhibition are clinically significant.

INTRODUCTION

Greater than half of all heart failure (HF) patients suffer from increased diastolic stiffness and impaired relaxation of the left ventricle while their ejection fraction (EF) is preserved (HFpEF)¹. HFpEF is a complex syndrome that includes diastolic dysfunction, exercise intolerance, and concentric hypertrophic remodeling^{2, 3}. No effective therapies exist for treating this pervasive syndrome due in part to the limited understanding of the underlying pathophysiology. HFpEF is likely due to a range of pathomechanisms that include dysfunction in sarcomeric proteins and slowing of the early phase of diastole that might be caused by changes in calcium handling/response proteins. Titin, the largest known protein and molecular spring in the heart, has emerged as a prime therapeutic target aimed at restoring compliance to the sarcomere and thereby improving diastolic function⁴. The titin gene (TTN) contains 364 exons that encode the third myofilament of the sarcomere that spans from Z-disk to M-band⁵. Titin’s I-band region acts as a molecular spring that is responsible for maintaining the structural integrity of the sarcomere⁶. Additionally, titin’s molecular spring is the chief contributor of passive stiffness in the heart, and functions as a mechano-sensor for stress and strain in the myocyte⁷. Alternative splicing produces two main cardiac isoforms; the smaller N2B isoform (~3.0 MDa) and the larger more compliant N2BA isoform (~3.3 MDa). Diastolic stiffness of the left ventricle is in large part dependent on the N2BA:N2B isoform ratio⁸. Stiffness can also be tuned through post-translational modification (PTM) of these primary isoforms, particularly through phosphorylation of titin’s spring region^9–11.

PTM of titin’s spring region is deranged in HFpEF patients^{4, 12}, which contributes to increased diastolic stiffness. Since protein kinase G (PKG) phosphorylation of titin’s N2B element enhances spring compliance, targeted treatment of HFpEF by enhancing PKG activity with the phosphodiesterase(PDE)5 inhibitor sildenafil was performed (summarized in the recent RELAX trial)¹³, however, exercise capacity was unimproved and clinical outcomes were similar to placebo¹³. Another potential therapeutic approach is upregulation of compliant titin isoforms. Alternative splicing of messenger RNA occurs primarily in transcripts encoding the spring region of titin and therefore regulates the elastic properties of titin through changes in contour length of the spring. Since the discovery of RBM20 as a titin splicing factor¹⁰, modification of titin size through RBM20 inhibition has become possible.

The current study evaluates whether manipulating post-transcriptional modification of titin isoforms through reduction of the titin splicing factor RBM20 has a beneficial effect on the diastolic function of mice with restrictive cardiomyopathy. A previously published mouse model¹⁴ deficient in titin’s IA junction (Ttn^ΔIAjxn) places increased strain on the spring region of titin. The Ttn^ΔIAjxn mouse is a choice tool for studying diastolic dysfunction as this model displays similar phenotypic characteristics found in the HFpEF patients, notably diastolic dysfunction, reduced exercise tolerance and concentric remodeling of the left ventricular chamber in response to stress. Furthermore, all known signaling elements in titin remain intact and therefore the pathology can be attributed solely to increases in titin-based strain. We aimed to determine if inhibition of RBM20 could recover function in this model back to a wild-type (WT) state. Our data establish that inhibiting titin splicing factor RBM20 in the Ttn^ΔIAjxn mouse improves diastolic function, restores exercise tolerance, and attenuates afterload induced pathologic concentric remodeling.

METHODS

Generation of mice homozygous for Ttn^ΔIAjxn and heterozygous for cardiac specific cRbm20^ΔRRM

We bred two murine models the Ttn^ΔIAjxn14 and the cardiac specific αMHC Cre cRbm20^ΔRRM14 in order to compare effects of increased or decreased strain in titin, respectively. To study the effect of RBM20 inhibition in Ttn^ΔIAjxn mice we crossed these two models to create hybrid Ttn^ΔIAjxn mice with an intermediate expression level of WT RBM20, abbreviated ΔIAjxn/ΔRRM; details in Online Figure I. Additionally, we used a Tg(Myh6-cre/Esr1) mouse that has the Myh6 promoter (cardiac muscle αMHC) directing expression of a selective estrogen receptor modulator (SERM) inducible Cre recombinase (Mer-Cre-Mer, MCM) to cardiac myocytes. Mer-Cre-Mer ΔIAjxn/ΔRRM mice were injected intraperitoneal with raloxifene (40 mg/kg) or vehicle for 8 days to induce reduction of WT RBM20. After day 28 (from start of injections), mice were examined by echo and catheterization and then euthanized. Hearts were dissected, weighed and flash frozen at −80°C for further titin isoform analysis using standard procedures (see Online Methods). Western blot studies with an antibody to RBM20’s C-terminus¹¹ revealed that the LV of the αMHC-Cre ΔIAjxn/ΔRRM expresses 40% of the total RBM20 as mutant protein; for the MCM ΔIAjxn/ΔRRM mice this value is 45% mutant RBM20 (see Fig. 3 B). Experiments were approved by the University of Arizona Institutional Animal Care and Use Committee and followed the U.S. National Institutes of Health Using

Figure 3. Inducible expression of super compliant titin isoforms recovers diastolic function in Ttn^ΔIAjxn mice.

A) Representative image of 1% agarose gel for titin analysis of LV myocardium. Raloxifine results in expression of two large N2BA isoforms (scN2BA1 and 2). B) RBM20 expression in the constitutive and inducible ΔIAjxn/ΔRRM mouse. The top shows example Western blots and the bottom the ratio of mutant RBM20 to WT RBM20 in constitutive and inducible ΔIAjxn/ΔRRM LV samples. C) E/A ratio; D) MVDT: mitral valve deceleration time; E) EDPVR (β): end diastolic pressure volume relationship. Data shown as mean ± SEM. Age at day 0 = 100 ± 5 days; Age at day 28 = 128 ± 5 days for vehicle and treatment groups. Statistical significance calculated with two-way ANOVA and Bonferroni correction (D and E) or t-test (F), (* p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001 as indicated). (Vehicle: DMSO/Saline; Treatment: raloxifene (40 mg/kg), both administered IP daily for 8 days.)

Animals in Intramural Research guidelines for animal use

Protein expression analysis

Protein expression analysis was performed by using standard titin gel electrophoresis methods¹⁵.

Cell mechanics

Skinned cardiac myocyte mechanics were performed as previously reported¹⁶. (See online Methods for details.)

Hemodynamics

Anesthetized and conscious echocardiography was performed on mice and pressure volume relations were evaluated using cardiac catheterization. Cardiac function in diastole was assessed in mice using continuous flow two percent isoflurane inhaled anesthetic with a target heart rate of 450 ± 25 bpm. (See Online Methods for details.)

Exercise performance

Exercise tolerance was evaluated by measuring time and distance ran using a treadmill running test. (See Online Methods for details.)

TAC surgery

Male mice 3–4 months old were subjected to minimally invasive transverse aortic constriction (TAC)¹⁷ performed under Ketamine/Xylazine (120/12 mg/kg) anesthesia; details are explained in the Online Methods. Briefly, a 27-gauge needle was used for banding and mice were studied by echocardiography at 28 days and then euthanized. Hearts were immediately dissected, weighed and frozen for further analysis.

Myocyte cross section area

Cross-sectional area (CSA) of cardiomyocytes was performed as described¹⁸ with minor modifications. Hearts in diastole were removed from sham and TAC mice and the LV tissue was covered with OCT (Tissue-Tek) and frozen with liquid nitrogen–cooled isopentane and stored at −80°C. 8-μm thick sections were collected on VWR glass microscope slides and stored at −20°C overnight. The cross-sections were skinned and stained with anti-Laminin (Sigma) to demarcate cell borders and DAPI (Vector Laboratories) to stain nuclei. Images were collected on an Axio Imager M.1 microscope (Carl Zeiss) using an Axio Cam MRC (Carl Zeiss). CSAs were measured using the ImageJ program (National Institutes of Health) and their borders of stained cells that included nuclei were traced manually. The CSA of 100 myocytes from the myocardium of the left ventricle (6 mice per group) were collected from each sample using 4 randomly selected sections.

Statistics

Statistical analysis was performed in Graphpad Prism (GraphPad Software, Inc). Group significance was defined using one or two-way ANOVA followed by multiple testing correction, as appropriate. Results are shown as mean ± standard error of the mean. P<0.05 was considered significant.

RESULTS

Inhibition of Rbm20 upregulates super compliant titin isoforms (N2BAsc) and normalizes passive stiffness in cardiomyocytes of Ttn^ΔIAjxn mice

Ttn^ΔIAjxn mice are deficient in the IA junction of titin (see Figure 1 A for schematic) but have normal expression of the adult N2B and N2BA isoforms. However, both isoforms are reduced in size as revealed by their slight increase in mobility in gel electrophoresis (compare Fig. 1 B left and middle lanes). The larger and more compliant N2BA isoform (~3.3 MDa in WT and ~3.15 MDa in Ttn^ΔIAjxn) comprises ~ 20 percent of total titin in both WT and Ttn^ΔIAjxn mice (Fig. 1C, left and middle bar, Online Table I). It was recently shown that as a consequence of deleting the IA junction of titin, the attachment point of titin’s spring region is moved away from the Z-disk, resulting in greater titin strain in Ttn^ΔIAjxn mice¹⁴.

Figure 1.

A) Half sarcomere models of the 3 genotypes used in this work: WT, ΔIAjxn and ΔIAjxn/ΔRRM. The domain composition of titin’s spring region (N2B isoform) is shown at the top. B) Representative image of 1% agarose gel for titin analysis of LV myocardium. WT myocardium expresses N2BA and N2B titin (T2 is a minor degradation product). The ΔIAjxn mouse expresses titin with a slightly higher mobility consistent with the deletion of the IA junction. Crossing the ΔIAjxn with the ΔRRM mouse (αMHC-Cre cRbm20^ΔRRM+/−) results in expression of two very large N2BA isoforms that we refer to as super compliant (N2BAsc). C) Ratio of N2BA isoforms over total titin. D) Representative skinned cardiomyocyte with passive stress in cardiomyocytes. E) Passive stiffness in cardiomyocytes. Abbreviations: ΔIAjxn: Ttn^ΔIAjxn; ΔRRM: αMHC-Cre cRbm20^ΔRRM); N2BAsc: super compliant N2BA titin. Data shown as mean ± SEM. Statistical significance calculated by one-way ANOVA with Bonferroni correction. (*symbol versus WT; ^† symbol versus Ttn^ΔIAjxn; 3 symbols: p<0.001; 4 symbols: p<0.0001).

cRbm20^ΔRRM mice (αMHC-Cre) are deficient in titin splicing factor RBM20 and, as previously described¹¹, they upregulate large N2BA titin isoforms that we refer to as super compliant titin (N2BAsc). To determine the therapeutic potential of RBM20 inhibition, we bred Ttn^ΔIAjxn mice with αMHC-Cre cRbm20^ΔRRM mice in order to produce Ttn^ΔIAjxn mice that are heterozygous deficient in splicing factor RBM20 (see online Fig. I, left for more detail on the breeding scheme and Online Table II for tissue morphometry). Ttn^ΔIAjxn mice constitutively inhibiting RBM20 express super compliant N2BA titin isoforms estimated ~3.35 and ~3.45 MDa (Fig. 1 B, right lane) encompassing ~80% of total titin in the murine left ventricle (Fig. 1 C, Online Table I). Thus, heterozygous inhibition of the titin splicing factor RBM20 is highly effective in upregulating compliant titin.

Previously we have shown that altering titin size and therefore strain on titin’s spring region directly affects passive stress in skinned left ventricular (LV) cardiomyocytes¹⁴. We tested whether inhibiting RBM20 in Ttn^ΔIAjxn mice reduces the elevated diastolic stress in skinned myocytes to WT levels. Cardiomyocytes were skinned and isolated in relaxing solution and cells were attached at one end to a force transducer and at the other end to a servomotor and subjected to a stretch–hold–release protocol using standard procedures (see Online Methods for details). Passive stress was significantly increased in Ttn^ΔIAjxn mice but was normalized in Ttn^ΔIAjxn mice expressing super compliant titin (Fig. 1 D). Passive stiffness (slope of stress–sarcomere length (SL) relation within physiologic SL range) was similarly reduced to WT levels (Fig. 1 E).

Reduction of RBM20 expression restores diastolic function to Ttn^ΔIAjxn mice

Results at the cellular level were further evaluated at the organ level to assess cardiac function in diastole (Fig. 2). Pulse wave Doppler in the 4-chamber apical view revealed significant differences in mitral valve deceleration time (MVDT) and E/A ratios of Ttn^ΔIAjxn mice compared to WT control, indicating diastolic dysfunction (Fig. 2C and D and Table 1). The MVDT is significantly reduced in Ttn^ΔIAjxn mice (21.9 ± 1.0 vs. 25.9 ± 0.7 ms in WT), a parameter inversely related to diastolic stiffness¹⁹. The MVDT was found to be at WT levels (26.5 ± 1.5 ms) in Ttn^ΔIAjxn mice expressing N2BAsc titin, indicating restored LV chamber compliance. Additionally, a greater E/A ratio in Ttn^ΔIAjxn mice (1.5 ± 0.1 vs. 1.3 ± 0.03 in WT) indicates restrictive LV filling²⁰, and this ratio was improved in Ttn^ΔIAjxn mice expressing super compliant titin (1.2 ± 0.03).

Figure 2. Diastolic function of Wildtype, ΔIAjxn and ΔIAjxn/ΔRRM mice.

A) representative pulse-wave Doppler E and A waves of the 3 genotypes that were used; B) cardiac catheterization PV loops; C) E/A ratio; D) MVDT: mitral valve deceleration time; E) EDPVR: end-diastolic pressure volume relationship. Data shown as mean ± SEM. Statistical significance calculated with one-way ANOVA and Bonferroni correction. (*symbol versus WT; ^† symbol versus Ttn^ΔIAjxn. (2 symbols: p<0.01; 3 symbols: p<0.001; 4 symbols: p<0.0001). * p<0.05 compared to WT; ^† p<0.05 compared to ΔIAjxn.)

Table 1.

Hemodynamics. Data shown as mean ± SEM

	WT	TtnΔIAjxn	ΔIAjxn/ΔRRM
Echocardiography
n	8	8	8	Conscious
HR (bpm)	650.5 ± 10.3	650.5 ± 13.4	669.0 ± 9.3
FS (%)	65.2 ± 1.8	57.5 ± 3.2	52.4 ± 1.7**
LVID;d (mm)	2.7 ± 0.1	2.8 ± 0.2	3.0 ± 0.1
LVID;s (mm)	1.0 ± 0.1	1.2 ± 0.2	1.4 ± 0.1
LVPW;d (mm)	1.1 ± 0.03	1.2 ± 0.1	1.1 ± 0.1
LVPW;s (mm)	1.9 ± 0.1	1.8 ± 0.1	1.7 ± 0.1
LVDD/WT (ratio)	2.5 ± 0.2	2.6 ± 0.3	2.7 ± 0.1
Heart rate maintained at 450 ± 25 (bpm)
LA Diameter (mm)	2.4 ± 0.2	2.7 ± 0.1	2.9 ± 0.1	Anesthetized
MV Decel (ms)	25.9 ± 0.7	21.9 ± 1.0**	26.5 ± 1.5††
MV E (mm/s)	770.5 ± 37.2	835.4 ± 22.7	780.7 ± 28.0
MV A (mm/s)	621.6 ± 35.8	547.4 ± 14.2	650.9 ± 26.9†
MV E/A	1.3 ± 0.03	1.5 ± 0.04****	1.2 ± 0.03††††
Cardiac Catheterization
n	7	7	8
HR (BPM)	431.3 ± 7.0	457.1 ± 23.2	492.5 ± 8.1
ESP (mmHg)	94.4 ± 4.7	98.5 ± 7.0	93.5 ± 2.5
EDP (mmHg)	2.0 ± 4.4	3.7 ± 1.0	1.4 ± 0.6
dPmax (mmHg/sec)	6807.4 ± 431.4	8257.2 ± 667.2	9224.3 ± 261.0*
dPmin (mmHg/sec)	−8116.3 ± 415.4	−7073.2 ± 254.7	−8894.5 ± 205.9†††
ESV (uL)	34.5 ± 7.6	33.9 ± 3.2	25.1 ± 3.0
EDV (uL)	70.0 ± 7.6	73.6 ± 4.1	60.1 ± 3.6
SV (uL)	35.5 ± 1.3	39.7 ± 3.4	37.1 ± 1.9
EF (%)	54.0 ± 5.7	53.9 ± 3.5	62.2 ± 2.7
EA (mmHg/uL)	2.7 ± 0.2	2.3 ± 0.2	2.4 ± 0.2
Tau Glantz (ms)	12.2 ± 3.3	18.2 ± 4.5	9.8 ± 0.4
ESPVR	4.3 ± 0.6	2.9 ± 0.5	3.1 ± 0.3
EDPVR	0.03 ± 0.004	0.08 ± 0.01****	0.03 ± 0.004††††

Abbreviations: ΔIAjxn: (Ttn^ΔIAjxn); ΔRRM (αMHC-Cre cRbm20^ΔRRM+/−). For additional abbreviations, see Non-standard Abbreviations and Acronyms list. Statistical significance calculated with one-way ANOVA with Bonferroni correction.

^*p<0.05;

^**p<0.01;

^***p<0.001;

^****p<0.0001.

(*symbol versus WT; † symbol versus Ttn^ΔIAjxn. (1 symbol: p<0.05; 2 symbols: p<0.01; 3 symbols: p<0.001; 4 symbols: p<0.0001).)

Diastolic dysfunction was further demonstrated in Ttn^ΔIAjxn mice using high fidelity pressure-volume admittance catheters (Fig. 2 B and E and Table 1). Occlusion of the inferior vena cava allowed us to determine the steepness of the end-diastolic pressure-volume relationship (EDPVR), β, which is a load-independent parameter. The EDPVR (β) is the coefficient of stiffness in the LV and was found to be (0.03 ± 0.004 mmHg/μL) in WT mice. Ttn^ΔIAjxn mice had an elevated EDPVR (β) of (0.08 ± 0.01 mmHg/μL) that was normalized in Ttn^ΔIAjxn mice expressing N2BAsc titin (0.02 ± 0.002 mmHg/μL). Therefore, consistent with our skinned cell experiments, compliance in the LV chamber was recovered in Ttn^ΔIAjxn mice deficient in splicing factor RBM20.

We further examined diastolic recovery of Ttn^ΔIAjxn mice with inducible (α-MHC-MCM) cRbm20^ΔRRM mice that express N2BAsc on a Ttn^ΔIAjxn background (see breeding scheme in Fig. S1, right for further details). This evaluation is important because it allows us to test whether diastolic dysfunction, once it has occurred, is recoverable therapeutically by upregulating compliant titin. Inducible mice were echoed at day 0 and then injected intraperitoneal for 8 days with either vehicle or selective estrogen receptor modulator (SERM) raloxifene (see Methods for additional details, and see Online Table III for raloxifene control experiments). After 8 days of raloxifene injections, greater than 60% N2BAsc titin expression occurred at 28 days (Fig. 3 A and Online Figure II). To evaluate the expression levels of mutant and wildtype RBM20, we used an antibody raised against the C-terminus of RBM20 that labels a single band in wildtype control mice and a doublet in heterozygous RBM20^ΔRRM mice (Fig. 3 B, top). The top band reflects the level of wildtype RBM20 and the bottom band the level of mutant RBM20 (i.e. missing the RNA Recognition Motif (RRM)). We used this antibody on LV samples of constitutive (α-MHC ΔIAjxn/ΔRRM) and inducible (α-MHC-MCM ΔIAjxn/ΔRRM) models. Since the mice were heterozygous for the ΔRRM allele, mutant RBM20 is expected to make up 50% of total protein. However, there is slightly less than 50% mutant RBM20 (Fig. 3 B, bottom) indicating upregulation of wildtype RBM20 in the heterozygous mice. Furthermore, the samples from inducible ΔIAjxn/ΔRRM mice have a higher level of mutant RBM20 (45% of total) than the constitutive mice (40%), suggesting that the degree of upregulation of wildtype RBM20 is higher in the inducible mice. These results combined with the protein and functional data suggest that inactivating ~40–45% of total RBM20 results in robust upregulation of super compliant titin.

Inducible (α-MHC-MCM ΔIAjxn/ΔRRM) mice were echoed at day 28 to assess diastolic function and then evaluated with a terminal cardiac catheterization procedure for studying the effect of expressing N2BAsc titin. Similar to our constitutive model (α-MHC ΔIAjxn/ΔRRM), diastolic function was improved as shown by pulse-wave Doppler revealing a greater duration of the mitral valve deceleration time and a significant reduction in the E/A ratio (Fig. 3 C and D). Moreover, pressure-volume analysis showed significant improvement of the EDPVR (β) upon induction of N2BAsc expression (Fig. 3 E). Systolic function was unaltered between vehicle and treatment groups (Online Table IV). Additionally, WT controls undergoing the same protocol of raloxifene treatment were evaluated by echo and systolic function was unaffected (see Online Table III). Thus, acutely reducing strain in titin has a beneficial effect on the diastolic function of the stiff left ventricle.

Assessment of associated HFpEF symptoms

Similar to HFpEF patients^{21, 22}, Ttn^ΔIAjxn mice display exercise intolerance¹⁴. We tested whether upregulating compliant titin has a beneficial effect on exercise tolerance by submitting mice to incremental speed running on a treadmill. Mice with compliant titin ran longer and for a greater distance in both the constitutive (Fig. 4 A and B) and the inducible N2BAsc expressing models (Fig. 4 C and D). Overall these findings suggest that increasing titin compliance in the heart has a beneficial effect on exercise performance in the Ttn^ΔIAjxn murine model.

Figure 4. Mice with upregulated compliant titin display enhanced exercise capacity in a treadmill test.

A–B) Time and distance of Ttn^ΔIAjxn mice and Ttn^ΔIAjxn mice constitutively expressing N2BAsc; C–D) Time and distance of Ttn^ΔIAjxn mice and Ttn^ΔIAjxn mice with inducible expression of N2BAsc. Data shown as mean ± SEM. (Constitutive: αMHC-Cre cRbm20^ΔRRM+/−; inducible: αMHC–MerCreMer cRbm20^ΔRRM+/−). (Vehicle: DMSO; Treatment: raloxifene, both administered IP daily for 8 days with exercise testing 28 days after last injection) Note that the IAjxn is also deleted from the skeletal muscles of these mice and that therefore future control experiments are needed on cardiac-specific Ttn^IAjxn deletion mice.

In addition to exercise intolerance, concentric remodeling in the heart is a common clinical manifestation in HFpEF patients²³. We therefore tested the responsiveness of mice with differing strain in titin to afterload stress. Mice were subjected to minimally invasive transverse aortic constriction (TAC) surgery, a procedure widely used to elicit afterload-dependent hypertrophic remodeling in the left ventricle. Compared to WT mice, Ttn^ΔIAjxn mice displayed an exaggerated hypertrophic response to TAC (p = 0.0001), whereas, the LV hypertrophy response in Ttn^ΔIAjxn mice expressing N2BAsc titin was attenuated compared with Ttn^ΔIAjxn mice (Fig. 5 A). The degree of remodeling these mice experienced in the left ventricle was determined by measuring the left ventricular end diastolic diameter (LVDD) and the wall thickness (WTH) in diastole. These parameters were obtained using M-mode echocardiography of the left ventricle in the short axis view. By calculating WTH/LVDD we can assess the degree to which the left ventricle is concentrically remodeled in response to TAC. Robust concentric remodeling after 4 weeks of pressure overload was observed in WT and Ttn^ΔIAjxn mice with the largest values in Ttn^ΔIAjxn mice, however, remodeling was lessened significantly in Ttn^ΔIAjxn mice expressing N2BAsc titin (Fig. 5 B, see also Online Figure III with cross-sectional area analysis). In addition to reduced pathologic remodeling, Ttn^ΔIAjxn mice with upregulated compliant titin also exhibited better diastolic function as revealed by Doppler imaging (Fig. 5 C–D). This data suggests that varying the length of titin isoforms alters the responsiveness to pressure overload hypertrophy in the heart and that increasing titin compliance is advantageous for preventing diastolic dysfunction associated with afterload induced concentric remodeling of the heart.

Figure 5.

Ttn^ΔIAjxn mice have an exaggerated response to TAC that is normalized in ΔIAjxn/ΔRRM mice. A) Hypertrophy; B) Concentricity: WTH: wall thickness in diastole; LVDD: left ventricular diameter in diastole. C) E/A ratio. D) MVDT: mitral valve deceleration time. Data shown as mean ± SEM. Statistical significance calculated by one or two-way ANOVA with Bonferroni correction: * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001; (versus WT ****, versus Ttn^{ΔIAjxn ††††}). (αMHC-Cre cRbm20^ΔRRM+/− mice on aΔIAjxn background were used.)

DISCUSSION

We report for the first time a rescue of HFpEF-like symptoms in the mouse through inhibition of the titin splicing factor RBM20. The Ttn^ΔIAjxn murine model acts as a mechanical analogue of the titin-based increase in passive myocardial stiffness found in HFpEF patients⁴. Increased passive stiffness is present in the adult Ttn^ΔIAjxn mouse, in the absence of pharmacological or surgical intervention. By using the Ttn^ΔIAjxn mouse as a model of HFpEF we demonstrate that inhibiting RBM20 restores diastolic function, improves exercise tolerance and attenuates afterload induced pathologic remodeling of the left ventricle. Due to our still limited understanding of HFpEF pathophysiology, there are no specific therapies to treat this widespread disease, therefore our findings on RBM20 inhibition are clinically significant.

Cardiac titin’s I-band region is a ~1.0 MDa spring composed of three distinct elements: the tandemly arranged immunoglobulin-like (Ig) element, the cardiac specific N2B element, and the proline, glutamate, valine, and lysine rich (PEVK) element²⁴. During diastole, stretch occurs in titin’s spring region and this provides elasticity to the sarcomere²⁵. Alternative splicing of titin occurs predominantly in the spring region and is responsible for the variances found in the two main adult cardiac isoforms of titin, N2B and N2BA²⁶. The more compliant N2BA isoform has a longer extensible spring region owing to the longer PEVK and Ig segments. The N2B isoform is the dominant isoform in both mice and men, with human LV expressing N2B and N2BA isoforms at a ratio of approximately 60:40^27,4. Both isoforms extend the same distance for a given sarcomere length (near Z-disk to beginning of A-band), however, the force to stretch titin is greater in the N2B isoform attributable to its shorter contour length⁸. By reducing the level of functional RBM20 protein, additional exon incorporation into titin’s spring elements minimizes passive stiffness in a manner inversely related to the size of titin¹¹. The Ttn^ΔIAjxn murine model has 18 exons deleted from the mouse Ttn gene that encode 3 Ig domains and 11 FnIII domains that constitute a 153 kDa polypeptide fragment in titin known as the I-band – A-band junction. This model is unique from previously published titin I-band truncation models (N2B²⁸, PEVK²⁹ and Ig¹⁶) in that the entire spring region of titin remains intact, including all spring element phospho-sites. Removal of titin’s IA junction, results in anchoring of the C-terminal attachment of titin’s distal Ig spring element further away (~ 70 nm) from the Z-disk, resulting in enhanced strain in titin’s spring region and consequently an increase in force per titin molecule¹⁴. This increased force mimics the pathologic conditions of oxidative stress and deranged phosphorylation evident in HFpEF patients^{4, 30}. Interestingly, the Ttn^ΔIAjxn mouse recapitulates many characterisitcs of HFpEF, e.g., diastolic dysfunction, exercise intolerance and hypertrophy. Herein, we studied whether reducing functional RBM20 expression ameliorates these symptoms.

Evaluation of Ttn^ΔIAjxn mice with and without N2BAsc expression was performed at the cell and organ levels. The increased stiffness of cardiac myocytes from Ttn^ΔIAjxn mice (Fig. 1 E) are in agreement with previous reports showing that altering the size of titin directly affects passive stiffness^{11, 14}. That this causes increased LV chamber stiffness is supported by multiple chamber-level measurements. The greater ratio of the early mitral inflow (E) to late atrial kick (A) revealed by pulse-wave Doppler imaging is indicative of increased diastolic stiffness (Fig. 2 C). Moreover, LV chamber stiffening abbreviates the time of deceleration of the early mitral inflow³¹ and the reduced deceleration time in Ttn^ΔIAjxn mice (Fig. 2 D) also support restrictive filling. In accordance with echocardiography, cathererization of the left ventricle revealed a steeper slope of the EDPVR demonstrating diastolic stiffness in the LV of Ttn^ΔIAjxn mice. Remarkably, inhibiting RBM20 both chronically and acutely in Ttn^ΔIAjxn mice restored these parameters back to WT levels (Figs. 2 and 3). Assessing inducible expression of compliant titin is imperative since HFpEF patients typically develop symptoms later in adult life, therefore, an acute therapy to restore diastolic function in the elderly HFpEF patient is necessary. Upon inducing compliant titin isoform expression in Ttn^ΔIAjxn mice, diastolic dysfunction was corrected and these mice preformed better in exercise tolerance studies (as further discussed below). Thus, our hemodynamic studies in mice with inhibited RBM20 suggest that this intervention is beneficial for restoring compliance to the restrictive left ventricle.

The advantage of RBM20 inhibition was further evidenced by exercise testing. Exercise is considered a beneficial therapy in heart failure patients³², but how titin compliance relates to exercise capacity is not well understood. Since our models are cardiac specific for RBM20 reduction, we could assess exercise performance as it directly relates to increasing the strain in cardiac titin. A limitation to note is that the Ttn^ΔIAjxn mouse deletes the IA junction in both skeletal and cardiac muscles (it is a global KO) and that future control studies are necessary to test exercise tolerance in a cardiac-specific Ttn^ΔIAjxn mouse. When Ttn^ΔIAjxn mice were challenged with treadmill running, mice with upregulated compliant titin (raloxifine group) performed significantly better those that did not (vehicle control). Exercise intolerance leading to fatigue and shortness of breath has a major impact on the quality of life in HFpEF patients, therefore our finding of improved exercise performance through upregulating titin compliance is encouraging.

Titin has been hypothesized to function as biomechancial sensor and our present studies support that titin mechanosensation contributes to increased afterload induced LV hypertrophy. We constricted the transverse aorta of mice to induce acute pressure overload stress to the heart. Compared to WT controls, Ttn^ΔIAjxn mice exhibited a 50% greater increase in LV mass characterized by robust concentric remodeling (Fig. 5). In contrast, RBM20 inhibited Ttn^ΔIAjxn mice (with reduced titin strain) had the least amount of LV mass incease in respone to TAC (Fig. 5 A). Hypertrophy signaling in the myocyte is extremely complex³³ and mechanotransduction remains incompletely understood. We explored a previously proposed titin-based signal transduction mechanisms that involved the titin binding proteins FHL1 and FHL2. Four-and-a half LIM domain (FHL) proteins are expressed in striated muscle where they facilitate diverse protein-protein interactions involved in signal transduction³⁴. Previous studies have shown localization of FHL1 and FHL2 at the N2B element in titin^{35, 36}. Futhermore, regulation of extracellular signal-regulated kinase 2 (ERK2) has been reported by both FHL1 and FHL2 in cardiomyocytes^{37, 38}. It has also been reported that in response to pressure oveload, FHL1 null mice display an attenuated hypertrophy response³⁵ whereas the TAC response of FHL2 null mice does not significantly differ from WT³⁴. Our data are in agreement with these findings as FHL1 is significanlty upregulated and FHL2 protein levels remain unaltered in response to TAC (see Online Figure IV). Thus, FHL1 might be involved in the increased hypertrophy response in the Ttn^ΔIAjxn model and its normalization when N2BAsc is expressed. Further studies are necessary to investigate titin as a sensor for mechanotransduction including the roles of FHL1 and FHL2.

HFpEF is a complex disease with multiple etiologies that induce restrictive filling in the left ventricle of which titin-based stiffening is likely to be only one of several underlying causes. Our present work revealed that increasing the compliance of titin’s spring region has a beneficial effect on diastolic function of the heart in a titin specific HFpEF mouse model. HFpEF patients suffer from diastolic dysfunction that leads to a myriad of symptoms including exercise intolerance and reduced quality of life³⁹. Innovative therapies are urgently needed to address diastolic stiffness and our study reveals novel insights into the pathophysiology of HFpEF and the central role played by titin. Lengthening the spring region of titin through alternative splicing reduces passive myocardial stiffness and improves a HFpEF-like phenotype in the Ttn^ΔIAjxn mouse. Diastolic function was recovered both chronically and acutely resulting in an enhanced exercise performance compared to littermate controls. Furthermore, afterload induced pathologic remodeling was attenuated in mice expressing N2BAsc titin. Future efforts should be focused on testing RBM20 inhibition in larger animal models to determine how much inhibition is required to achieve benefit in diastolic function without causing side-effects, especially when inhibition of RBM20 is long-term. In summary, our work supports that ameliorating diastolic dysfunction and pathological chamber remodeling might be achievable through therapeutically targeting titin.

Novelty and Significance.

What Is Known?

• Tuning mechanisms in the elastic protein titin that alter titin’s compliance are deranged in heart failure patients with preserved ejection fraction (HFpEF).

• The titin-based murine model Ttn^ΔIAjxn has increased diastolic stiffness and HFpEF-like symptoms (diastolic dysfunction, exercise intolerance).

• RBM20, a splicing factor of titin, regulates the compliance of titin in a dose-dependent manner.

What New Information Does This Article Contribute?

• Reduction of RBM20 by 40–45% in Ttn^ΔIAjxn mice is sufficient to greatly increase titin compliance.

• Reduction of RBM20 in Ttn^ΔIAjxn mice recovers diastolic function, and improves exercise tolerance and pressure-overload induced ventricular remodeling.

Increasing titin’s compliance improves HFpEF-like symptoms in the Ttn^ΔIAjxn mouse model. HFpEF affects greater than half of all heart failure patients, yet no effective therapies are available. This study reveals that inhibiting RBM20 increases titin compliance and is a potential therapeutic HFpEF target.

Nonstandard Abbreviations and Acronyms

E/A

LV pulse wave Doppler E-wave: A-wave ratio

Ea

arterial elastance

Eccentricity

LVDD/WTH

EDP

end‐diastolic pressure

EDPVR

end diastolic pressure volume relationship

EDV

end‐diastolic volume

ESP

end‐systolic pressure

ESPVR

end‐systolic PV relationship

ESV

end systolic volume

FHL1

four and a Half LIM domains 1

FHL2

four and a Half LIM domains 2

FS

fractional shortening

HF

heart failure

HFpEF

heart failure with preserved ejection fraction

IAjnx

I-band – A-band junction

Ig

immunoglobulin-like

LA

left atrium PV

LV

left ventricle

LVIDd

left ventricular internal diastolic diameter

LVIDs

left ventricular internal systolic diameter

LVPWd

left ventricular posterior wall thickness in diastole

LVPWs

left ventricular posterior wall thickness in systole.

MHC

myosin heavy chain

MV

DT mitral valve deceleration time

PEVK

proline, glutamate, valine, and lysine

PKG

protein kinase G

PTM

posttranslational modification

RBM20

RNA binding motif 20

RRM

RNA recognition motif

SERM

selective estrogen receptor modulator

SL

sarcomere length

TAC

transverse aortic constriction

WTH

wall thickness

αBC

alpha-B crystallin