Smooth muscle protein-22-mediated deletion of Tsc1 results in cardiac hypertrophy that is mTORC1-mediated and reversed by rapamycin

Abstract

Constitutive activation of mammalian target of rapamycin complex 1 (mTORC1), a key kinase complex that regulates cell size and growth, is observed with inactivating mutations of either of the tuberous sclerosis complex (TSC) genes, Tsc1 and Tsc2. Tsc1 and Tsc2 are highly expressed in cardiovascular tissue but their functional role there is unknown. We generated a tissue-specific knock-out of Tsc1, using a conditional allele of Tsc1 and a cre recombinase allele regulated by the smooth muscle protein-22 (SM22) promoter (Tsc1c/cSM22cre^+/−) to constitutively activate mTOR in cardiovascular tissue. Significant gene recombination (∼80%) occurred in the heart by embryonic day (E) 15, and reduction in Tsc1 expression with increased levels of phosphorylated S6 kinase (S6K) and S6 was observed, consistent with constitutive activation of mTORC1. Cardiac hypertrophy was evident by E15 with post-natal progression to heart weights of 142 ± 24 mg in Tsc1c/cSM22cre^+/− mice versus 65 ± 14 mg in controls (P < 0.01). Median survival of Tsc1c/cSM22cre^+/− mice was 24 days, with none surviving beyond 6 weeks. Pathologic and echocardiographic analysis revealed severe biventricular hypertrophy without evidence of fibrosis or myocyte disarray, and significant reduction in the left ventricular end-diastolic diameter (P < 0.001) and fractional index (P < 0.001). Inhibition of mTORC1 by rapamycin resulted in prolonged survival of Tsc1c/cSM22cre^+/− mice, with regression of ventricular hypertrophy. These data support a critical role for the Tsc1/Tsc2-mTORC1-S6K axis in the normal development of cardiovascular tissue and also suggest possible therapeutic potential of rapamycin in cardiac disorders where pathologic mTORC1 activation occurs.

INTRODUCTION

Mutations in either of two genes, TSC1 and TSC2, are causally linked to the development of tuberous sclerosis complex (TSC), a tumor-suppressor gene syndrome characterized by multiple tumors of the brain, kidney, heart and skin (1,2). Cardiac rhabdomyomas are seen in the majority of late-term TSC fetuses as well as infants (3). Aortic aneurysms also occur in TSC infants and young children (4). Hamartin and tuberin, the protein products of TSC1 and TSC2, act as a complex to suppress the activation of the small GTPase Rheb (ras homolog enriched in brain), through a GTPase-activating domain near the C-terminus of tuberin (5,6). Rheb is a major regulator of mammalian target of rapamycin complex 1 (mTORC1) such that GTP-rheb activates mTORC1, resulting in the downstream phosphorylation of p70 S6 kinase (S6K) and eukaryotic translation-initiation factor 4E-binding protein 1 (4E-BP1), to increase protein translation and cell growth (7,8). Thus, when the hamartin–tuberin complex is disrupted by loss of functional hamartin or tuberin, elevated levels of active Rheb (GTP-Rheb) lead to constitutive activation of mTORC1, which in turn leads to increased cell growth, a pathway that is evolutionarily conserved (9).

Increased protein synthesis is a cardinal feature of cardiac hypertrophy, and as in other cells, is regulated by eukaryotic translation-initiation factor 4E (eIF4E) and the ribosomal protein, S6, which are downstream of mTORC1 (9,10). Although mTORC1 receives major regulatory input from tuberin–hamartin via rheb, it is also sensitive to input from growth factors, AKT, amino acids and other nutrient levels and AMP-activated kinase (AMPK) (11). Interestingly, hypertrophic cardiomyopathy (HCM) due to glycogen storage diseases, PRKAG2 [γ-regulatory subunit of AMP-activated protein kinase (AMPK)] mutations, sarcomeric protein gene mutations and cardiac hypertrophy due to external load all appear to affect mTOR activity as a key mechanism for cardiac hypertrophy (11–14) . However, in murine models when individual components of the mTOR-p70S6K pathway have been genetically manipulated in cardiac tissues, the results have been somewhat inconsistent (15–18). Over-expression of active and dominant-negative variants of mTOR has not resulted in significant measureable differences in heart size at baseline. In addition, added stress such as exercise training or pressure overload did not induce significant ventricular hypertrophy in such models (15,16). Dominant-negative expression of phosphatidyl inostitol-3 kinase (PI3Kinase), which lies much further upstream of mTOR and directly upstream of AKT, results in a mild reduction in cardiac size only (17). More recently, cardiac-specific deletion of mTOR resulted in a severely dilated cardiomyopathy with absent hypertrophic response to an increase in afterload (19). Collectively, these studies have led to uncertainty over the exact role of mTOR in regulating heart size during normal cardiac development and in response to stress in the adult heart.

Genetic deletion of either mTOR or either of Tsc1 or Tsc2 in mice leads to early embryonic lethality but limits their value for studying adult disease. Using a tissue-specific knock out of Tsc1, we observed that constitutively activated mTORC1 in cardiovascular tissue resulted in significant cardiac enlargement that was biventricular and hypertrophic in nature, beginning as early as embryonic day (E) 15, and leading to a severe reduction in survival. Inhibition of mTORC1 by rapamycin resulted in regression of the cardiac phenotype and improved long-term survival. These data support a critical role for the Tsc1/Tsc2-mTORC1-S6K axis in the normal development of cardiovascular tissue and probable role in the pathogenesis of cardiac hypertrophy.

RESULTS

Tsc1 deletion in Tsc1c/cSM22cre^+/− embryos and adult mice

Timing of Tsc1 gene recombination

The murine primary heart tube begins development at E8.5, at which time smooth muscle protein-22 (SM22)-α is transiently expressed (20). At E9.5, SM22-α is expressed in vascular smooth muscle and remains expressed throughout adulthood (20). Multiplex ligation-dependent probe amplification (MLPA) analysis demonstrated consistent 75–80% recombination of the Tsc1 conditional to the null allele in heart DNA preparations at E15 through the day of birth (DOB) (Fig. 1). Gross cardiac enlargement was readily apparent in Tsc1c/cSM22cre^+/− embryos at E15 in comparison with littermate controls. Post-natally, a major reduction in Tsc1 expression (>90%) was observed by immunoblot analysis of tissue extracts from the right and left ventricles (RV, LV), and aorta (Ao) from Tsc1c/cSM22cre^+/− mice with no significant alteration in the expression of tuberin (Tsc2) (Fig. 1). Furthermore, elevated levels of phosphorylated S6Kinase (Thr 421/424), S6 (ser 240/244) and mTOR (ser 2448) with a trend toward reduced AKT phosphorylation (Ser 473) and increased 4EBP-1 phosphorylation (results not shown) were observed, consistent with mTORC1 activation.

Figure 1.

Timing and extent of recombination in the Tsc1 gene in Tsc1c/cSM22cre^+/− mice. (Left panel) MLPA of DNA extracted from the hearts of Tsc1c/cSM22cre^+/− embryos (cc+) (n = 14) and Tsc1c/wSM22cre^+/− controls (c/w+) (n = 8) showing the percentage recombination of the conditional allele (c) is 75–80% at E15 through the DOB. Note the percent recombination reflects the amount of conditional allele that has been successfully recombined to the knockout allele and does not measure the wild-type allele (w), and is therefore equal in c/w+ and c/c+ embryos. Heart weight, embryo weight and percentage of heart weight relative to embryo weights are also shown. (Right panel) Immunoblot and densitometric analysis of lysates derived from the RV, LV and aorta (Ao) in 3-week-old mice. Tsc1c/cSM22cre^+/− mice (cc+) (n = 6) and littermate controls (ctrl; Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (n = 6) are compared. The proteins assessed are Tsc1 (hamartin), Tsc2 (tuberin), pS6K (phospho-S6 kinase), Thr 421/424, pS6 (phospho-S6), Ser 240/244, pERK (phospho-p42/44 ERK), Thr202/Tyr204, pAkt (phospho-Akt), Ser 473, pmTOR (phospho-mTOR), Ser 2448 and actin (loading control).

Together these data suggest that significant knockdown of Tsc1 in the heart begins during embryogenesis and persists throughout adulthood. Moreover, post-natal assessment of Tsc1 deletion in all cardiovascular tissues was also associated with robust downstream activation of mTORC1.

Distribution of gene recombination

To assess the distribution of SM22-regulated cre recombinase expression, we used a lacZ reporter allele in the ROSA26 locus (21). Beta galactosidase staining (blue staining) indicates recombination and expression of the LacZ gene. Figure 2 shows strong blue staining in all four chambers of the heart, including the majority of cardiac myocytes, in the smooth muscle layers of large vessels, including the aorta, coronary arteries, pulmonary arteries, renal and cerebral arteries. There was no staining in the vascular endothelium. Staining was also seen in the bronchial smooth muscle cells and in occasional cells in the kidney and brain.

Figure 2.

Distribution of recombination due to the SM22cre allele. β-Galactosidase staining of cryosections of the aorta, heart, brain, kidney and lung from 2-week-old (n = 8) SMA22cre/RosaR26 β-gal mice. β-Galactosidase expression (blue) is extensive throughout the heart and vascular SMCs, and rare in other cell types.

Survival and clinical signs in Tsc1c/cSM22cre^+/− mice

Tsc1c/cSM22cre^+/−mice were born in Mendelian ratios with the expected distribution of genotypes and gender. All control mice had normal development and weight gain with normal survival through 12 months of age. Tsc1c/cSM22cre^+/− mice were notably lighter but heart weights were significantly heavier than those of their littermate controls at all time points measured, 0–5 weeks (Fig. 3A). In addition, cardiac weight continued to rise in the mutant mice from 2 to 5 weeks of age, whereas it reached a plateau in controls beginning at ∼2 weeks of age. The median survival of Tsc1c/cSM22cre^+/− mice was 2.8 weeks with a trend toward earlier death in female mice but none survived beyond 6 weeks (Fig. 3B). Typically, 1–2 days prior to death, Tsc1c/cSM22cre^+/− mice developed a prodrome of respiratory distress, reduced activity and then suddenly died without apparent pre-terminal event. Gross examination done during the prodromal phase revealed significant cardiac enlargement (Fig. 3A), as described above. However, no evidence of left heart failure was seen in these mice (frothy phlegm, pulmonary congestion or hemorrhage, or body edema). These results suggested that the early mortality observed in Tsc1c/cSM22cre^+/− mice was associated with the observed cardiomyopathy.

Figure 3.

Reduced body weight and survival in Tsc1c/cSM22cre^+/− mice. (A) Representative images of body and heart size in Tsc1c/cSM22cre^+/− (cc+) and Tsc1c/wSM22cre^+/− control mice (Ctl). Body weight and heart weight from birth (0 week) through 5 weeks and body weight to tibial length ratio at 4 weeks are shown. (B) Kaplan–Meier curves showing survival of Tsc1c/cSM22cre^+/− mice (cc+) (n = 30) compared with littermate controls of mixed genotype (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (n = 30) (left panel) and segregated by gender (right panel). In younger mice, the gender was not clearly identified (N/A).

Characterization of the cardiomyopathy

Cardiac pathology

Examination (at all ages) of Tsc1c/cSM22cre^+/− mice revealed grossly enlarged hearts within the thoracic cavity with upward displacement of the right ventricle (RV) (Fig. 3A). Gross and histopathologic sections at 4 weeks confirmed increased LV wall thickness, with two patterns, asymmetric and symmetric, significant bulging of the interventricular septum compromising the RV cavity and loss of the normal striated ventricular appearance (Fig. 4A). Myocyte disarray (a non-specific feature that is common in cardiomyopathies), fibrosis (typical of HCM due to sarcomeric gene mutations) and vacuolated myocytes (suggestive of glycogen accumulation in cardiomyopathies due to PRKGA2 mutations) were not seen in Tsc1c/cSM22cre^+/− hearts by hematoxylin and eosin (H&E). The interstitial matrix did not display expansion with any obvious inflammatory infiltrate and coronary arteries appeared normal. In keeping with our gross observations, both RV and LV plus septum (LV+S) weights were significantly elevated in Tsc1c/cSM22cre^+/− mice, confirming biventricular enlargement, with a significantly reduced RV:LV+S ratio (Fig. 4A) suggesting an LV-dominant cardiomyopathy.

Figure 4.

Cardiac enlargement and reduced cardiac function in Tsc1c/cSM22cre^+/− mice. (A) Upper panel: Gross anatomy and H&E-stained axial sections from Tsc1c/cSM22cre^+/− (cc+) mice hearts and littermate controls (Control) (n = 8, each group). LV and RV are labeled; note both asymmetric and symmetric hypertrophic cardiac phenotypes and bulging of the interventricular septum into the RV cavity. In the asymmetric phenotype, the hypertrophic anterolateral wall (arrow) and thinner posterior wall (arrowhead) of the LV are indicated. Higher magnification (×4, ×20) views of the LV show loss of striations but no obvious fibrosis or infiltrate in Tsc1c/cSM22cre^+/− (cc+) mice. Lower panel: RV and LV+S weights in Tsc1c/cSM22cre^+/− (cc+) mice and their littermate controls (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctrl) over time (0–5 weeks) (n = 6 per group, each time point) are compared. Values are expressed as actual weight, with body weight as a denominator (RV/BW, LV + S/BW) and as a ratio relative to one another (RV/LV + S ratio). (B) Echocardiographic images (axial and M-mode) are shown with measurements of the anterior (A) and posterior (P) wall thickness as well as LVEDD, LVESD and fractional index. Tsc1c/cSM22cre^+/− mice (cc+) (n = 6) and their littermate controls of mixed genotype (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctrl) (n = 6) are compared.

Cardiac function

Echocardiographic analysis was performed in 4-week-old mice (Fig. 4B and Table 1). The usual echocardiographic appearance of the murine heart was easily observed in control animals. In contrast, massive hypertrophy of the anterior LV wall prohibited easy localization of the LV cavity in Tsc1c/cSM22cre^+/− mice and the RV was not clearly imaged due to its unusual displacement. Hypertrophy of the anterior wall was consistently observed compromising LV cavity size particularly during diastole. LV end-systolic diameter (LVESD) was reduced but not significantly (Tsc1c/cSM22cre^+/− 0.76 mm ± 0.31 versus controls 0.92 mm ± 0.36; P = 0.059); however, LV end-diastolic diameter (LVEDD; Tsc1c/cSM22cre^+/− 1.69 mm ± 0.51 versus controls 2.87 mm ± 0.22; P < 0.001) and calculated fractional index (Tsc1c/cSM22cre^+/− 53% ± 0.19 versus controls 68% ± 0.13; P < 0.001) were significantly lower in Tsc1c/cSM22cre^+/− mice. These results suggest that loss of Tsc1 in cardiac myocytes resulted in significant cardiac dysfunction, dominated by LV hypertrophy and diastolic dysfunction.

Table 1.

Echocardiographic measurements in Tsc1c/cSM22cre^+/− mice compared to controls

Group	n	m/f	Age (weeks)	Anterior wall (mm), mean ± SD	Posterior wall (mm), mean ± SD	LVEDD (mm), mean ± SD	LVESD (mm), mean ± SD	Fractional index (%), mean ± SD
Ctrl	6	3/3	4.2	1.23 ± 0.24	1.39 ± 0.38	2.87 ± 0.22	0.92 ± 0.36	68 ± 0.13
cc+	6	3/3	4.2	2.25 ± 0.41	1.26 ± 0.38	1.69 ± 0.51	0.76 ± 0.31	53 ± 0.19
P-value				<0.001	0.057	<0.001	0.059	<0.001

Ctrl, control (Tsc1c/wSM22cre^+/−; Tsc1c/cSM22cre^−/−; Tsc1w/wSM22cre^+/− ); cc+, Tsc1c/cSM22cre^+/− ; m/f, male/female; anterior wall, thickness of the anterior wall of the LV measured in millimeters; posterior wall, thickness of the posterior wall of the LV measured in millimeters. Values are expressed as mean ± SD and statistical significance is shown.

Cardiac histology

In keeping with that observed on H&E, neither collagen deposition or glycogen accumulation was detected by trichrome and periodic acid schiff (PAS) stain, respectively (Fig. 5A). To determine whether the increased cardiac mass in the hearts of Tsc1c/cSM22cre^+/− mice was due to an increase in cell number or increased cell size, we assessed LV myocytes for markers of cellular proliferation (Ki67) and apoptosis (TUNEL) and performed nuclear counts on H&E-stained LVs (Fig. 5A). Ventricular myocytes from Tsc1c/cSM22cre^+/− mice did not show any appreciable alteration in Ki67 or TUNEL stains. Quantification of the surface area of ventricular myocytes revealed a significant increase in area in Tsc1c/cSM22cre^+/− mice (Fig. 5B) (Tsc1c/cSM22cre^+/− 442.89 ± 29.64 versus controls 290.27 ± 9.16; P < 0.001). Cell counts performed on dissociated ventricular myocardium revealed a significant increase in cell number in Tsc1c/cSM22cre^+/− mice compared with controls (Fig. 5B) (Tsc1c/cSM22cre^+/− 369/mg versus controls 219/mg ± 7; P < 0.05). These results are in keeping with the expected effects of mTORC1 activation on cell size and number and suggest that both increased ventricular myocyte size and cell number contribute to the increased cardiac muscle mass observed in Tsc1c/cSM22cre^+/− mice.

Figure 5.

Cell growth and apoptosis, cell size and cell number in Tsc1c/cSM22cre^+/− mice. (A) Paraffin-embedded sections of the LV were stained with trichrome (collagen), PAS (glycogen), Ki67 (proliferation) and TUNEL (apoptosis). Positive (pos) control samples were used to confirm marker expression. Tsc1c/cSM22cre^+/− (cc+) mice (n = 8) and their littermate controls (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctrl) (n = 8) are compared. (B) Surface area of LV myocytes measured on ImageproPlus software is shown; littermate control (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctl) and Tsc1c/cSM22cre^+/− mice (cc+) were compared (n = 8, each group). Number of ventricular myocytes counted from hearts of Tsc1c/cSM22cre^+/− mice (cc+) and littermate controls (Tsc1c/wSM22cre^+/−) (ctl) is shown (n = 4, each group). Results are expressed as total number of myocytes counted in each sample and number of myocytes normalized to the original sample weight; statistical significance is indicated.

Vascular smooth muscle effects of loss of Tsc1

Immunoblot analysis revealed significant reduction (>90%) in the levels of hamartin expressed in the aortic smooth muscle similar to that observed in the myocardium (Fig. 1). However, no obvious gross pathology was observed in the aorta (Fig. 6A) apart from one incidental case demonstrating irregular smooth muscle layers within the aortic wall. However, by 4 weeks of age, there was a striking elevation in mean systolic pressure in Tsc1c/cSM22cre^+/− mice compared with control mice (Tsc1c/cSM22cre^+/− 136.9 mmHg ± 8.6 versus controls 101.0 mmHg ± 4.4; P < 0.01) as measured by the tail cuff. Given the disparity between the degree of hypertension and the lack of excessive muscularization of aortic smooth muscle, we tested the vasoconstrictive response of aortic rings to phenylephrine. The force generated from aortas lacking Tsc1 was significantly greater than that generated by control mice (Tsc1c/cSM22cre^+/− 0.310 g ± 0.05 versus controls 0.198 g ± 0.02; P < 0.05). These data suggest that increased vessel contractility, rather than increased muscle mass within the vessel wall, contributed to the elevated systemic pressures seen in Tsc1c/cSM22cre^+/− mice. Interestingly, systemic pressures measured at 2 weeks of age using carotid catheterization showed relative hypotension or normotension (Tsc1c/cSM22cre^+/− 78.9 mmHg ± 12.7 versus controls 97.8 mmHg ± 6.7) that had progressed to hypertension by 4 weeks of age (Fig. 6A).

Figure 6.

Systemic and RV pressures in Tsc1c/cSM22cre^+/− mice. (A) Representative images in 4-week-old mice of H&E-stained aortas, tail cuff tracings (left panel), tail cuff systolic pressure (mmHg) (n = 24, each group) and contractile force generated from aortic rings (g) (n = 6, each group) (middle panel) and systolic pressure measured by carotid catheterization (mmHg) (n = 6, each group) from Tsc1c/cSM22cre^+/− mice (cc+) and their littermate controls (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctrl) are shown. Note that, rarely, vessels showed irregular smooth muscle layers (insert; ×100). (B) Representative images of Geisen-stained (elastin) lung tissue (n = 8, each group), RVSP (mmHg) (n = 16, each group) and wet:dry lung weight ratios (n = 16, each group) from Tsc1c/cSM22cre^+/− mice (cc+) and their littermate controls (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctrl) are shown. Distribution of the muscularization of the distal pulmonary arteries are compared between the indicated groups; results are expressed as a percentage of the total number of vessels counted.

Examination of pulmonary pathology showed that there was no obvious or consistent muscularization of large pulmonary arteries or accompanying bronchi (data not shown). In contrast, significant muscularization of small peripheral pulmonary vessels was observed as evidenced by a significant increase in fully muscularized vessels (Tsc1c/cSM22cre^+/− 42.57% ± 6.19 versus controls 24.62% ± 3.7; P < 0.05) and a simultaneous reduction in non-muscularized and partially muscularized vessels (none: Tsc1c/cSM22cre^+/− 20.29% ± 6.3 versus controls 32.94% ± 6.2; P < 0.05; partial: Tsc1c/cSM22cre^+/− 37.14% ± 7.9 versus controls 42.93% ± 6.6; P < 0.05). RV systolic pressures (RVSPs) were significantly elevated in Tsc1c/cSM22cre^+/− mice (Tsc1c/cSM22cre^+/− 23.06 mmHg ± 1.1 versus controls 18.08 mmHg ± 0.6; P < 0.01). We also measured wet:dry lung weights in Tsc1c/cSM22cre^+/− mice at 4 weeks and during the prodrome phase that occurred immediately prior to death. Consistent with the lack of gross evidence of alveolar edema, there was no difference in wet:dry lung weights, arguing against significant pulmonary edema or decompensated LV failure at 4 weeks or during the prodrome phase.

Kidney pathology

We did not appreciate any significant difference in kidney size but cysts were appreciated in a significant number of Tsc1c/cSM22cre^+/− mice (Fig. 7). Pathologic analysis revealed cystadenomas in ∼70% of animals. The extent of cystadenomatous change was modest and no evidence of hypertensive glomerulopathy was observed. Interestingly, an identical phenotype of cystic adenomatous change within the kidney is observed in murine models where Tsc1 is expressed in a heterozygous fashion. Moreover, we noted β-galactosidase staining in the occasional renal epithelial cell when using SM22-α cre recombinase to drive knockout of the lacZ reporter allele (Fig. 2). This observation together with the epithelial derivation of adenomatous tumors suggested that the renal phenotype was due to minor leak of the SM22-α cre allele to renal epithelium. Last, a significant reduction in total creatinine levels and consequential increase in the BUN/creatinine ratio was observed (Fig. 7). These values seemed probably to be due to the lower skeletal muscle mass as reflected by a lower body weight in Tsc1c/cSM22cre^+/− mice.

Figure 7.

Renal pathology and function in Tsc1c/cSM22cre^+/− mice. Gross and histopathologic H&E-stained sections of kidneys showing cystadenomatous tumors in Tsc1c/cSM22cre^+/− (cc+) mice (n = 16) compared with littermate control (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (ctrl) mice (n= 16). BUN and creatinine levels measured in the serum of both groups (n = 8, each group) are also shown.

Rapamycin therapy

Rapamycin is an inhibitor of mTORC1, which has been highly effective in treating tumors that spontaneously arise in Tsc1^+/− mice. The activation of mTORC1 in the heart led us to test the effect of rapamycin on the survival and phenotype observed in these mice (Fig. 8). Two different dosage regimens were tested (4 mg/kg fives times per week; 4 mg/kg three times per week) beginning at 2 weeks of age. A third cohort was treated late in the phenotype, at 3 weeks of age with rapamycin 4 mg/kg three times per week; a small subset of these mice underwent rapamycin withdrawal at 6 weeks. A major survival benefit to Tsc1c/cSM22cre^+/− mice was conferred by all rapamycin treatment strategies with no deaths in any of the treatment cohorts (Fig. 8A). Analysis of Tsc1c/cSM22cre^+/− mice after treatment (approximate age 4 weeks) demonstrated significant regression of systemic and pulmonary hypertension as well as reversion of body and heart weights to normal (Fig. 8A). Pathologic examination at ∼ 6 weeks of age also confirmed striking regression of the LV wall hypertrophy and a normal appearance to LV and RV cavities (Fig. 8B). Moreover, heart weights rapidly and significantly declined with rapamycin treatment. In parallel, improvement in cardiac output as indicated by an increase in the LVEDD (Tsc1c/cSM22cre^+/− 1.90 mm ± 0.19 versus 2.12 mm ± 0.37) and fractional index (Tsc1c/cSM22cre^+/− 68% ± 0.08 versus 69% ± 0.02) was observed (Fig. 8C and Table 2) together with a reduction in the level of phosphorylated S6 in ventricular myocyte extracts (Fig. 8D). The survival benefit of rapamycin was maintained until at least 6 months (data not shown). However, once discontinued, the survival benefit persisted until 8 weeks after the stop date (∼14 weeks of age) when a short 1–2-day prodromal illness prior to death re-emerged together with cardiac enlargement at levels similar to that observed in non-rapamycin-treated Tsc1c/cSM22cre^+/− mice (Fig. 8B). These results suggest that the effect of rapamycin was truly therapeutic, resulting in the regression of an established cardiac hypertrophy, and that continued therapy is required to maintain this regressive response.

Figure 8.

Reversal of the phenotype with rapamycin. (A) Kaplan–Meier curves showing survival of Tsc1c/cSM22cre^+/− mice (cc+ without Rapamycin) and their littermate controls (Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−) (Controls) given 4 mg/kg of subcutaneous rapamycin using three different treatment schedules (cc+ Rapamycin 5 times weekly starting at 2 weeks (n = 15); cc+ Rapamycin 3 times weekly starting at 2 weeks (n= 15); cc+ Rapamycin 3 times weekly starting at 3 weeks (n = 8) [late rescue]). The following measurements are compared in rapamycin (Rapa) and vehicle-treated Tsc1c/cSM22cre^+/− mice and littermate controls (ctrl) at ∼4 weeks: body weight (BW), RV/BW, LV+S/BW, RV/LV+S ratio, systolic pressure (mmHg) and RVSP (n = 15, each group). Statistical significance is indicated (*). (B) Images of gross hearts and heart weight: body weight ratio (HW:BW) in: 2-week- and 4-week-old untreated Tsc1c/cSM22cre^+/− mice (cc+) and Tsc1c/wSM22cre^+/− control mice (control); ∼6-week-old rapamycin-treated mice depicting the regressive effect of rapamycin with 4 weeks of therapy; 14-week-old mice following rapamycin withdrawal (at 6 weeks of age) depicting re-emergence of cardiac enlargement off therapy in Tsc1c/cSM22cre^+/− (cc+) mice. The lower graphs show heart weights in rapamycin-treated (Rap) and vehicle-treated (Veh) Tsc1c/cSM22cre^+/− (cc+) and Tsc1c/wSM22cre^+/− (ctrl) mice (0–6 weeks) where treatment was started in one cohort at 2 weeks and in another cohort at 3 weeks; and heart weight following rapamycin withdrawal. (C) Echocardiographic images (axial and M-mode) and measurements of the anterior and posterior wall thickness as well as LVEDD, LVESD and fractional index in rapamycin-treated (Rapa) Tsc1c/cSM22cre^+/− mice (cc+) and controls (ctrl) (n = 6, each group) showing no significant differences between the groups. (D) Western immunoblot of lysates derived from the LV in vehicle- and rapamycin-treated Tsc1c/wSM22cre^+/− control mice (c/w+) and Tsc1c/cSM22cre^+/− mice (cc+) (n = 6, each group). The proteins assessed are pS6 (phospho-S6), pAkt (phospho-Akt) and actin (loading control).

Table 2.

Echocardiographic measurements in rapamycin-treated Tsc1c/cSM22cre^+/− mice compared to controls

Group	N	m/f	Age (weeks)	Anterior wall (mm), mean ± SD	Posterior wall (mm), mean ± SD	LVEDD (mm), mean ± SD	LVESD (mm), mean ± SD	Fractional index (%), mean ± SD
Ctrl	6	3/3	4.5	1.21 ± 0.39	1.35 ± 0.35	2.12 ± 0.37	0.59 ± 0.16	69 ± 0.02
cc+	6	3/3	4.5	1.20 ± 0.23	1.41 ± 0.69	1.90 ± 0.19	0.64 ± 0.09	68 ± 0.08
P-value				>0.05	>0.05	>0.05	>0.05	>0.05

Ctrl, control (Tsc1c/wSM22cre^+/−; Tsc1c/cSM22cre^−/−; Tsc1w/wSM22cre^+/−); cc+, Tsc1c/cSM22cre^+/−; m/f, male/female; anterior wall, thickness of the anterior wall of the LV measured in millimeters; posterior wall, thickness of the posterior wall of the LV measured in millimeters. Values are expressed as mean ± SD and statistical significance is shown.

DISCUSSION

Constitutive activation of mTORC1 is consistently found when the Tsc1/2 complex is disrupted by loss of either of Tsc1 or Tsc2. Consistent with an important role for this protein complex in growth control within vascular tissues is the occurrence of cardiac rhabdomyomas in the majority of, and abdominal aortic aneurysms less commonly in, TSC individuals (3,4,22,23). In this context, the present study was designed to examine the role of Tsc1/2 in cardiovascular tissue, using an SM22-driven Tsc1 conditional knockout approach. The resulting clinical phenotype was that of severe cardiac hypertrophy which began at least as early as E15 that progressed post-natally, and was associated with systemic hypertension and early death, all of which regressed significantly with the administration of the mTORC1 inhibitor, rapamycin.

Although differences in experimental approach can lead to variable outcomes, genetic manipulation of components of the PI3K-AKT-mTOR-p70S6K signaling pathway within cardiac tissues has not yielded a consistent phenotype of either cardiac hypoplasia or hypertrophy (15–19). Interestingly, transgenic mice over-expressing constitutively active mTOR do not develop cardiac hypertrophy despite the evidence of downstream constitutive activation of S6 in cardiac myocytes (16). Similarly, over-expression of kinase-dead mTOR or p70S6K did not result in cardiac hypoplasia, nor did their expression prevent the development of cardiac hypertrophy in response to exercise (16). Although the expression of endogenous mTOR could have interfered with the phenotype, these data were in direct contrast to our observations and the observations by Zhang et al. (19), who eliminated the expression of endogenous mTOR in adult murine hearts and found that cardiac tissue promptly underwent significant apoptosis and autophagy followed by four-chamber dilatation. In keeping with our findings that myocyte number and size were increased in mTOR-activated adult hearts, mTOR ablation resulted in reduced ventricular myocyte size and number (19). Also in keeping with our observations was the evolution of a dead heart and consequent rapid course to early death within 4 weeks following mTOR ablation. Together, these results are complementary with our data and suggest that, whether constitutively active (our study) or ablated (19), mTOR is critical for the regulation of myocyte size, cardiac mass and survival. Thus, when mTOR is inactive, the myocardium becomes dysfunctional by undergoing apoptosis, leading to a dilated and dead heart (19); in contrast, when overactive such as that observed in this study, cardiac muscle mass increases, leading to an HCM; and in both models, rapidly leading to death.

Over the last decade, the role of mTOR in the control of cell and organ size has been well characterized, acting to regulate cellular protein translational machinery. mTORC1 consists of mTOR, raptor and mLST8/GβL, and when activated phosphorylates p70S6K and 4EBP1, both of which, through downstream effects, augment general and cap-dependent translation (24,25). Genetic interference of the TSC proteins, Tsc1 and Tsc2, has been used by many investigators, including us, to manipulate and constitutively activate the mTORC1 complex in vivo (reviewed in 26,27). However, loss of the Tsc1/Tsc2 complex is also associated with feedback inhibition of the activation of AKT, through the degradation of IRS proteins and reduction in platelet-derived growth factor receptor expression (28). Consistent with this pathway in the dilated cardiomyopathy observed in mTOR-ablated myocardium, there were elevated levels of phosphorylated AKT (19). In contrast, cardiac enlargement due to LKB1 knockout was not associated with significant changes in phosphorylated AKT (29). In our model, the reduction in AKT activation was only mild and appeared unlikely to be contributing to the pathology seen. Although it could be argued that our phenotype may be independent of mTORC1 activation, the combined observations by our group and those by Zhang et al. (19), together with the dramatic rapamycin response and the re-emergence of a cardiomyopathy upon withdrawal of rapamycin in our model, all suggest that mTORC1 activation was the major, if not sole, pathway mediating the cardiac hypertrophy.

HCM is characterized by myocyte disarray and interstitial fibrosis (14). It is inherited as a Mendelian autosomal dominant trait and caused by mutations in 1 of at least 10 genes encoding a number of components of the cardiac sarcomere (14). Recently, mutations in the γ-2 regulatory subunit of the adenosine monophosphate-activated protein kinase (PRKAG2) have been reported to cause cardiac hypertrophy that mimics HCM. PRKAG2 mutations inactivate AMPK, which in turn inhibits glycogen synthase as well as Tsc1/2 complex function (11,30), resulting in reduced vacuolated myocytes from glycogen accumulation and Tsc1/2-mediated mTORC1 activation (reviewed in 27). Similarly, cardiac-specific deletion of LKB1, a serine/threonine kinase that lies upstream of AMPK, resulted in a dilated cardiomyoathy that was associated with mTOR and S6K activation but no vacuoles to suggest glycogen accumulation were observed (29). It is notable that previously one of us (D.J.K.) described a model of cardiac-specific Tsc1 deletion using the myosin light chain 2v promoter, in which a dilated cardiomyopathy and early death was seen at 6 months of age (31). In the present study, we did not find the histologic features of HCM, PRKAG2 mutations or LKB1 knockout. The histologic findings were more consistent with a general increase in myocyte mass, due to mTORC-1-mediated increase in cell size, cell number and/or pressure overload. The increase in heart weight in E15 embryos strongly argues against pressure overload from systemic hypertension as a major cause of the initial development of cardiac hypertrophy. Furthermore, the discrepancy between the absence of ki67 staining and the increased numbers of cardiac myocytes counted in 4-week-old Tsc1c/cSM22cre^+/− mice also suggest that the increased cardiac mass from a proliferative process occurred early (approximately before 2 weeks). However, significant hypertension that developed post-natally between 2 and 4 weeks of age suggests that pressure overload on an already hypertrophied heart may compound and accelerate the process, resulting in premature death.

The treatment of cardiac hypertrophy with rapamycin in murine models and in humans, post-cardiac transplant, has been described (32–36). Rapamycin can decrease heart size by up to 68% and is less effective in mice with decompensated hypertrophy (35). In the present study, wet:dry lung weights were unchanged, suggesting that the clinical phenotype was one of compensated cardiac hypertrophy in keeping with the higher rapamycin response rates observed by other groups. However, in our model, cardiac hypertrophy was present for a considerable period of time (approximately at least 3 weeks) prior to therapy, and the observed regression was 100% complete, significantly greater than that observed in other models. We also observed normalization of systemic hypertension with rapamycin, suggesting that concomitant treatment of hypertension may have potentially compounded the reversibility of the observed established cardiomyopathy induced by loss of Tsc1. However, data from human patients after cardiac transplant (34) and data from spontaneously hypertensive rats (36) and uremic mice (37) suggest that the ability of rapamycin to significantly reduce LV mass is independent of its blood-pressure-lowering effects. Moreover, Siedlecki et al. (37) have shown an identical effect of rapamycin in uremic cardiomyopathy that was independent of renal function. In this context, our data showing the significant reduction in levels of phosphorylated S6 in rapamycin-treated hearts together with the rapid and considerably significant regression of cardiac hypertrophy all suggest that rapamycin, independent of its effect on hypertension or kidney function, had a direct, independent beneficial effect on cardiac myocytes that ultimately translated into improved cardiac function.

The premature mortality described in this study is striking and parallels temporally that observed in cardiac-specific mTOR (19) and LKB1 deletion models (29). Death was sudden in our model, suggesting an acute event but our analyses do not support pulmonary edema or decompensated left heart failure as the cause of death. Furthermore, routine histology of the brain and kidney did not suggest that hypertensive cerebrovascular accident or hypertensive nephropathy as major contributors to the observed premature mortality. Echocardiographic analysis revealed an obliterative LV cavity, suggesting that death was more likely to be due to a sudden decrease in LV output. As an alternative, we noted significant septal bulging that compromised RV cavity size, which, in the context of pulmonary hypertension, could also result in sudden death from obstructive shock due to a failing RV. Interestingly, rhythm disturbances have been described in patients with TSC (38). In mice overexpressing kinase-dead mutant of mTOR, prolongation of the PR interval has been shown (16) and death in mice with αMHC-specific PRKAG2 mutations and LKB1 mutations is associated with ventricular pre-excitation phenomena (11,29). Although electrophysiologic studies were not performed in our model, it is conceivable that death was due to a sudden arrhythmia in the setting of mTOR activation, or sudden hemodynamic compromise initiated by the RV or LV, or a combination of these mechanisms.

In conclusion, loss of functioning Tsc1 in cardiac myocytes and vascular smooth muscle results in a severe phenotype that reflects the pathologic consequences of constitutively activated mTORC1 during development. Furthermore, our data support a possible role for Tsc1/2 complex and mTOR in the development of cardiac hypertrophy and systemic hypertension more broadly. Rapamycin therapy for this condition was found to be remarkably effective, similar to its benefit in other TSC mouse models affecting other organs. We suggest that rapamycin may be beneficial in the cardiac tumors and arrhythmias that are seen in TSC patients, and may have benefit more broadly for cardiac disorders in which mTORC1 activation is an important disease mechanism.

MATERIALS AND METHODS

Animal experimentation

All animal experimentation and procedures were performed in accordance with the Guide for the Humane Use and Care of Laboratory Animals and were approved by the Tufts and/or the Children's Hospital Boston Animal Care and Use Committee. We generated mice with loss of Tsc1 in cardiovascular tissue, using a conditional allele of Tsc1 (Tsc1^c) (31) and an SM22 allele expressing cre recombinase (SM22cre) (39). Using mixed strain Tsc1c/cSM22cre^−/− and c57BL/6 Tsc1w/wSM22cre^+/+ (obtained from Jackson Laboratories, Bar Harbor, ME, USA), we applied three different breeding strategies to generate mice with the desired genotype, Tsc1c/cSM22cre^+/−. Littermates of target animals with the following genotypes were used as controls: Tsc1w/wSM22cre^+/+, Tsc1c/wSM22cre^+/−, Tsc1c/cSM22cre^−/−. Polymerase chain reaction genotyping of the conditional, wild-type and null alleles of Tsc1 was performed as described previously (31).

Analysis of recombination rates by MLPA

The timing and extent of recombination at the Tsc1 locus was determined by MLPA of DNA obtained from heart tissues of Tsc1c/cSM22cre^+/− and Tsc1c/wSM22cre^+/− embryos removed at E15 and E17 and at the DOB. Timed matings were performed to obtain embryos at these ages, which were confirmed by Theiler stage analysis. At sacrifice, embryonic hearts were dissected and DNA was extracted using a Qiagen Gentra Puregene Mouse Tail Kit (Valencia, CA, USA) and subjected to MLPA (40). Relative levels of the three alleles of Tsc1 (wildtype, w; conditional, c; knockout, k) were determined as described (23). MLPA analysis was performed by J.G. and D.J.K., who were blinded to the results of the MLPA analysis.

LacZ staining in post-natal tissues

To assess distribution of recombination, Tsc1w/wSM22cre^+/+ mice were bred with mice bearing the conditional RosaR26 β-galactosidase allele (21). Mice were sacrificed and the following tissues removed in 2-week-old mice: heart, lungs, brain, kidneys and aorta. Tissues were fixed overnight with 1% paraformaldehyde, followed by immersion in 30% sucrose for at least 24 h before freezing. Ten-micrometer sections were stained with 1 mg/ml of 5-bromo-4-chloro-3-indolyl β-d-galactopyranoside and counterstained with nuclear fast red. Sections were imaged using a Nikon Eclipse E400 microscope, and images were captured using Spot software v4.0.5.

Standard histology

At 4 weeks of age, mice of the Tsc1c/cSM22cre^+/− genotype and corresponding littermate controls were sacrificed, weights determined, tibial length measured and lungs, kidneys, the heart, brain and aorta were harvested. The hearts were weighed, the LV plus septum (LV+S) was dissected from the RV and each ventricle re-weighed. RV/BW, LV/BW LV+S/BW, RV/tibial length, LV+S/tibial length ratios were calculated. One subset of samples was snap-frozen in dry ice and stored in liquid nitrogen for immunoblot analysis. An additional subset was inflation-fixed with 10% formalin (23 cmH₂O), hearts removed and sectioned axially at the level of the mitral valve, and corresponding sections were subsequently obtained for H&E staining, Masson trichrome stain (Ventana, Tuscon, AZ, USA) and PAS stain (Poly Scientific, Bayshore, NY, USA). Myocyte analysis was performed at the mid-papillary level and was restricted to myocytes located in the middle third of the LV wall. Myocytes located near the epicardium or the papillary muscles were not analyzed. Morphometric analysis of the lungs and aorta are outlined below in the hemodynamic measurements section. Kidneys were sectioned longitudinally through the calyx as described previously (41). Both gross and microscopic pathology were read by an observer blinded to the genotype.

Immunoblot analysis

Tissue extracts of the LV plus septum, RV and aorta were prepared by homogenization as described previously (41). To minimize contamination for aortic extracts, the aorta was dissected along its length, endothelium and adventitial layers were scraped and removed prior to freezing and homogenization. Equal amounts of protein were loaded and analyzed by immunoblotting as described previously (42). Primary antibodies used were against: hamartin, tuberin, actin (Santa Cruz, CA, USA), phospho-S6 (Ser 240/244), phospho-S6K (Thr 421/424), phospho-AKT (Ser 473) and phospho-mTOR (Ser 2448) (Cell Signaling Technology, MA, USA). Semi-quantitative densitometric analysis was performed using ImageQuant software (Molecular Dynamics, Sunnyvale, CA, USA). Densitometric analysis for each protein is expressed as fold amounts of the indicated protein relative to control and is representative of at least three immunoblots. Data are expressed as fold-activation relative to control. Images of immunoblots were obtained and cropped using Adobe Photoshop 6.0 (Adobe Systems, Inc., San Jose, CA, USA).

Immunohistochemistry

Immunohistochemistry was performed on paraffin-embedded sections of the LV and aorta as described previously (43). In brief, slides were deparaffinized, blocked with 1% bovine serum albumin, incubated with the following: Ki67 (Ventana), TUNEL (Roche, Indianapolis, IN, USA) or appropriate control and then counterstained with hematoxylin (Dako, Carpinteria, CA, USA). After staining, slides were viewed on a Nikon Eclipse E400 microscope, and images captured using Spot software v4.0.5. Images were interpreted by an observer blinded to the genotype.

Hemodynamic measurements

Systemic pressures were measured using a non-invasive tail cuff pressure system (ADInstruments, PowerLab NIBP Controller, Colorado Springs, CO, USA). Mice were awake and restrained. The cuff was placed at the proximal end of the tail and was inflated to 280 mmHg to occlude measured pulse pressure. The cuff was deflated slowly until a consistent pulse pressure returned and this pressure was recorded as systolic blood pressure (SBP). Mice underwent a brief period of training prior to undergoing this procedure and an average of three recordings per animal was measured. RVSP was measured in anesthetized mice with a transthoracic needle stick, attached to a Statham pressure transducer (44). Mice were then euthanized by cutting the abdominal aorta; formalin (10%) was infused and the lungs and heart were inflation-fixed at 23 cmH₂O. Lungs were cut longitudinally in the longest axis from the apex to the base and parallel with the mediastinum, ensuring that the main pulmonary artery and main stem bronchus were included as a central reference point in each sample. Samples were paraffin-embedded and Geisen's stain was used to identify elastin fibers in vessel walls. The percentage of muscularized peripheral pulmonary vessels was calculated in vessels <100 mm in diameter which are not expected to be muscularized. Vessels were characterized as non-muscular (no smooth muscle identifiable), partially muscularized (smooth muscle identifiable in less than three-fourths of the vessel) or fully muscularized (smooth muscle identifiable in more than three-fourths of the vessel) (45). Vessels were interpreted by two observers both blinded to the genotype.

Tail cuffs are not commercially available for small mice. Thus, for the accurate measurement of systemic pressures in 2-week-old mice, carotid artery pressures were performed. In brief, following anesthetization, the carotid artery was isolated and clamped distally. A Millar catheter (1/1.4 F) (Millar Instuments, Houston, TX, USA) was inserted into the carotid artery, the clamp was removed and the catheter was advanced 2–3 mm. The systolic pressure was recorded for 4–5min using PowerLab (ADInstruments). Acceptable and reproducible waveform analysis confirmed good catheter placement and validated the accuracy of systolic pressure measurements in 2-week-old mice. Additional validation was performed by measuring systolic pressures in 4-week-old mice using the same catheterization technique which was compared and correlated closely with tail cuff pressures.

Measurement of vasoconstriction response

We used a modification of the technique described by Hill et al. (44) for rats. The aorta was dissected in its full length and five 2–3 mm rings were excised and suspended on tungsten wires in a vessel bath containing Earle's balanced salt solution at 37°C. Vessel rings were hung at a tension of 0.5 g, and after 1 h the vasoconstrictive response to phenylephrine (10⁻¹⁰–10⁻⁴m) was measured. Multiple measurements were made per ring and averaged per animal.

Assessment of myocyte surface area

Measuring the surface area of ventricular myocytes was conducted on H&E sections of the LV obtained from Tsc1c/cSM22cre^+/− mice and littermate controls, using Image-Pro Plus software (Ver 7.0; MediaCybernetics, MD, USA). Hearts were cut axially below the mitral valve, and corresponding sections were stained with H&E. Analysis was restricted to myocytes located in the middle third of the LV wall. Myocytes located near the epicardium or the papillary muscles were not analyzed. Myocytes that were sectioned axially and contained a nucleus were selected, the cell membrane of individual myocytes was outlined and the number of pixels per myocyte determined. At a minimum, 50 myocytes from five separate regions were measured per sample. Using a conversion factor, the surface area was calculated from the total pixel number per myocyte and expressed in square micrometers. Surface area measurements were performed by an individual blinded to the genotype (H.H.).

Transthoracic echocardiography

Mice were anesthetized with 2.5% isoflurane gas, placed supine on a heating pad and transthoracic echocardiography was performed using a Philips Sonos 5500 Cardiac Vascular Echo Ultrasound machine. Measurements were made according to the leading-edge method of the American Society of Echocardiography (46). Images and measurements were interpreted by an experienced murine echocardiographer (B.R.) who was blinded to the genotype. Measurements included: LVEDD, LVESD and fractional index (LVEDD − LVESD/LVEDD).

Murine ventricular myocyte dissociation

Hearts were extracted from 4-week-old Tsc1c/cSM22cre^+/− and Tsc1c/wSM22cre^+/− mice anesthetized with pentobarbital (Nembutal, 120 mg/kg) (47,48). Ventricles were cut axially into 0.3 mm thick sections and digested in 0.0625 or 0.125% trypsin in Hank's balanced salt solution (HBSS; Atlanta Biologicals, Lawrenceville, GA, USA) overnight at 4°C. The sections were treated with Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (10% FBS in DMEM; Mediatech, Inc., Manassas, VA, USA) to inhibit trypsinization. Ventricular cells were dissociated through repeated digestions with warm HBSS containing collagenase type II (1 mg/ml; Worthington Biochemical Corp., Freehold, NJ, USA) at 37°C. The digestions were consolidated on ice for each ventricle, then centrifuged at 1000 r.p.m. for 4 min, decanted and resuspended in 10% FBS in DMEM twice. The final cell suspension was plated twice on 0.1% gelatin-coated plates for 60 min each to enrich for ventricular myocytes, and then processed through a Beckman Coulter Z1 Coulter Particle Counter at 1:50 dilution (Beckman Coulter, Inc., Brea, CA, USA). Cell counts were expressed as total cell counts and also normalized to ventricle weight.

Rapamycin treatment

Rapamycin (LC Laboratories, Woburn, MA, USA) was dissolved at 20 mg/ml in ethanol and stored at −20°C for up to 1 month. Each dose was prepared by being diluted in 5% Tween-80, 5% polyethyleneglycol 400 to 1.2 mg/ml solution. Tsc1c/cSM22cre^+/− mice and their littermate controls were injected with rapamycin at a dosage of 4 mg/kg subcutaneously. One group of mice was treated 5 days per week and another 3 days per week when treatment began at post-natal day 14; treatment continued for at least 6 weeks (and in a small cohort for 6 months) for survival analyses. In another group, treatment began at post-natal day 21 and continued for at least 6 weeks for survival analyses. A subset of mice was treated at 4 mg/kg for 5 days/week, and at ∼4–6 weeks of age, the subset was anesthetized, SBP and RSVP were measured, the heart and aorta were then harvested and analyzed for heart weight and gross and microscopic pathology. A small cohort was treated, until ∼4 weeks of age, echocardiography was performed, the cohort was then euthanized, and fresh tissues from the heart collected for homogenization and immunoblot analysis. A final cohort was treated for 3 weeks (3–6 weeks of age) with rapamycin, which was then stopped and the mice followed for survival analysis, hearts were removed and weighed peri-mortem in this group.

BUN/creatinine measurements

Direct measurement of urea (BUN) and creatinine in serum samples drawn from 4-week-old mice just prior to sacrifice was performed in a select number of Tsc1c/cSM22cre^+/− mice and compared with samples drawn from littermate controls. Levels of BUN and creatinine were measured using ELISA assay kits and performed according to the manufacturer's instructions (Biochain, Hayward, CA, USA).

Statistical analysis

Results are expressed as mean ± SD. Statistical analysis using a non-parametric Kruskal–Wallis ANOVA with Dunn's multiple comparison test for follow-up analysis was performed using SigmaStat ver 3.11.0 software.

Conflict of Interest statement. G.A.F. acts as chair for the Data Safety Board of Novartis, Inc.

FUNDING

This work was supported by the National Institutes Health/National Heart Lung and Blood Institute (1R01 HL089671-01 to G.A.F.); the American Heart Association (Grant in Aid 0855935D to G.A.F., Scientist Development Grant 10SDG2630161 to B.R.) and the National Institutes Health/National Institute of Neurological Disorders and Stroke (2R37NS031535 to D.J.K.).