Myocyte Apoptosis Occurs Early During the Development of Pressure-Overload Hypertrophy in Infant Myocardium

Abstract

Background

Abnormal hemodynamic loading often accompanies congenital heart disease, both before and after surgical repair. Adaptive and maladaptive myocardial responses to increased load are numerous. This study examined the hypothesis that myocyte loss occurs during compensatory hypertrophic growth in the developing infant myocardium subjected to progressive pressure overload.

Methods and Results

Pressure-overload left ventricular (LV) hypertrophy was induced in 7 to 10 day-old rabbits by banding of the thoracic aorta. LV function and mechanics were quantified by serial echocardiography and non-invasive LV wall stress analysis. LV tissue sections were examined for myocyte apoptosis, using a myocyte-specific DNA fragmentation assay, caspase-3 activation (specific fluorescent substrate), and for fibrosis (Masson’s trichrome). Significant myocyte apoptosis (198±37/10⁶ myocytes, P<0.01 vs. control) and caspase-3 activation was present in early hypertrophy when LV contractility was preserved and compensatory hypertrophy had normalized wall stress. By 6 weeks, multiple indices of LV contractility were reduced, LV wall stress was increased. Myocyte apoptosis was accelerated (361±56/10⁶ myocytes), caspase-3 activity further increased, and the estimated total number of LV myocytes was significantly reduced by 18±4%.

Conclusions

In experimental infant LV hypertrophy, myocyte apoptosis is initiated in the face of normalized wall stress and preserved contractility. The ongoing rate of apoptosis causes a measurable decrease in myocyte number that is coincident with the onset of ventricular dysfunction. It thus appears that pressure overload even at its earliest stages is not well-tolerated by the developing ventricle.

Introduction

Congenital heart disease is commonly associated with abnormal hemodynamic loading (pressure and/or volume). Because complete structural and hemodynamic correction is frequently impossible, abnormal loading is a persistent problem in many patients. In the setting of sustained pressure overload, ventricular adaptation typically progresses through stages that include compensatory hypertrophy due to the parallel addition of sarcomeres, which results in increased myocyte width and ventricular wall thickness. According to LaPlace’s Law, myocardial wall stress can be calculated from the formula: (pressure × radius)/(2 × wall thickness), and thus an increase in wall thickness can ameliorate increased wall stress caused by an increase in pressure. Because ventricular ejection is directly affected by afterload, it is believed that normalization of systolic stress (afterload) via increasing wall thickness is an important mechanism to maintain systolic performance in the face of increased systolic pressure. Failure to do so results in afterload mismatch where the degree of hypertrophy can no longer compensate for increased afterload^{1, 2}.

Numerous changes in gene expression and protein synthesis accompany this response. They include alterations in sarcomeric protein isoforms and activity, calcium handling and regulatory proteins (e.g. the sarcoplasmic reticular Ca-ATPase, SERCA-2), substrate and energy metabolism, adrenergic, muscarinic, and angiotensin receptor species and activity, growth factor and cytokine production, and stimulation of other pleiotrophic signaling pathways such as calcineurin^2–8. Abnormalities in parameters such as extracellular matrix signaling, matrix turnover, coronary blood supply, and angiogenesis have also been described and implicated in both the compensatory and pathologic hypertrophic response.

The specific molecular mechanisms underlying failure of continued hypertrophic compensation and the progression of contractile dysfunction are not well understood, especially in the immature, developing ventricle. Recently, myocyte apoptosis has been implicated in numerous states associated with reduced myocardial performance including acute and chronic ischemia, diabetes, myocarditis, adriamycin toxicity, hypertrophy, and heart failure^8–19. While the data that apoptosis occurs in these settings is significant, its functional importance remains unclear, particularly with regard to the adaptation to chronic pressure load. Several studies have quantified cardiomyocyte apoptosis in late-stage chronic hypertrophy of adult myocardium^{20, 21}, but there is no evidence about this process in the developing heart. Based upon other information that cell growth, survival, and death signaling pathways are different from adult both in the normal developing myocardium and that exposed to pressure overload, we examined the hypothesis that hypertrophic growth would stimulate myocyte apoptosis in the developing ventricle independent from the onset of heart failure.

Methods

Animal model

The institutional animal care and use committee approved these experiments. New Zealand white rabbits (aged seven to 10 days) were used to create a model of progressive left ventricular hypertrophy. After anesthesia was induced with intramuscular ketamine (20 mg/kg) and xylazine (0.5 mg/kg), a left thoracotomy was performed under sterile conditions. A 2-0 silk suture was placed around the descending aorta just distal to the ligamentum arteriosum, with care taken to make the suture as snug as possible without acutely causing stenosis of the descending aorta. After chest closure, air was evacuated from the left thoracic cavity by catheter aspiration. Following recovery from anesthesia, animals were returned to their mothers and allowed to feed in the normal manner. With growth, an aortic coarctation gradually develops. In preparation for later evaluations, weekly shaving of the right upper limb and thorax as well as sham echocardiograms were performed to accustom the animals to the procedure and reduce the stress response to actual echocardiography, thus avoiding the need for anesthesia or sedation during serial echocardiographic examinations.

For controls, age- and litter-matched New Zealand white rabbits were sham-operated and otherwise handled identically. Echocardiography and non-invasive blood pressure determination were performed at weekly intervals in the banded and age-matched control animals (see Appendix). A separate group of normal animals was used to determine the normal relationship in the rabbit of shortening fraction (SF) to afterload (end-systolic stress, ESS) (stress shortening index, SSI), rate of circumferential fiber shortening to end-systolic stress (afterload) (VCFc to ESS; stress velocity index, SVI), and their midwall equivalents (SF_mw and VCFc_mw) compared to fiberstress (FS) to determine SSI_mw and SVI_mw. Blood pressure was determined serially and non-invasively using optical plethysmography (Finapress, Ohmeda). The Finapress was applied to the shaved right forearm after application of topical alcohol over the region of the palpable brachial artery. The accuracy of the Finapress versus direct central arterial cannulation (carotid artery) to measure arterial blood pressure and to calculate load-adjusted echocardiographic indices was validated in a separate set of experiments (see Appendix for details).

Histochemical methods

Animals were sacrificed at weekly intervals and hearts immediately perfusion-fixed in 4% paraformaldehyde in PBS, pH 7.4, then paraffin-embedded and sectioned. Mid-LV myocardial sections were de-paraffinized, hydrated in a descending alcohol series, and boiled thrice for five minutes in 1 mM EDTA, pH 8.0 (antigen retrieval) prior to staining. Slides were imaged on a Zeiss Axiovert microscope equipped with visible/UV/fluorescent objectives (4–100x), xenon light source/Sutter filter wheel, and appropriate excitation/emission filter sets. Images were recorded using a Princeton Instruments cooled CCD camera or Leica digital color camera, depending on the probe type; images were stored and analyzed using MetaMorph™ and MetaFluor™ software.

Quantification of Apoptosis

TUNEL staining of DNA stand breaks was done with enhanced fluorescent detection by labeling cleaved double stranded DNA in tissue sections following the Fluorescein FragEL™ detection kit (Oncogene). Positive and negative controls of this assay were previously been performed in in vitro cell culture models and using rabbit thoracic lymph nodes (data not shown). For DNA staining of nuclei, 10 mg/ml Hoechst 33258 in PBS/50% glycerol was used. Myocytes were identified by co-staining with mouse anti-desmin monoclonal antibody (Sigma) at a dilution of 1:100 in PBS containing 5% horse serum and detected with Alexa™ 568 goat anti-mouse secondary antibody (Molecular Probes). Using the MetaMorph software, the following parameters were quantified: 1.total nuclei (Hoechst-positive); 2. total myocyte nuclei (Hoechst-positive in desmin-positive cells); 3. number of TUNEL-positive myocyte nuclei (TUNEL+/Hoechst+/desmin+). For maximum specificity, a “positive” result (i.e. apoptotic cardiomyocyte) was counted only when a TUNEL-positive signal (green) was co-localized to both nuclear DNA (blue) that was located within a cardiomyocyte (red).

Activation of caspase-3 was used as another index of pro-apoptotic signaling. Total LV protein was extracted by homogenizing 20–30 mg of LV myocardium in 4° C lysis buffer A (consisting of 25 mM hydroxyethylpiperazine ethanesulfonic acid (HEPES; pH 7.5) 10 mM KCl, 1.5 mM MgCl₂, 1 mM ethylenediamine tetraacetic acid, 1 mM ethylene glycol bis aminoethyl ether tetra-acetic acid, 1 mM benzamidine, 1 mM dithiothreitol [DTT], and 1 mM phenylmethylsulfonyl fluoride) containing 1% (v/v) Triton X-100). After centrifugation at 100,000 × g for 60 min at 4°C, supernatant protein concentratio ns were determined using the Bradford method according to the manufacturer’s instructions (BioRad). Protein extract (25 µg) and 25 µl reaction buffer (0.1 M HEPES pH 7.0, 10% polyethylene glycol, 0.1% cholamidopropyl dimethylammonio propanesulfonate (CHAPS), 10 mM DTT were mixed with 1 µl of 5 mM fluorogenic caspase-3 specific substrate (DEVD-AFC; BioMol, Plymouth Meeting, PA). After incubation at 37°C for 2 h in 96-well fluorescence-specific plates, fluorescence was quantified in a microplate reader (Perkin Elmer, Boston, MA) at excitation and emission wavelengths of 400 and 505 nm, respectively²². Relative caspase-3 activity was expressed as fold change relative to that found in LV from 8 week-old sham-operated animals.

Assessment of Fibrosis

Tissue sections were prepared as above, stained with Masson’s trichrome, and imaged using light microscopy: images were collected with a Leica digital color CCD-camera. Total image area and the relative percentage of collagen (blue) and tissue (red) were quantified using MetaMorph™ software.

Estimation of Myocyte number

The total number of myocytes in the LV free wall of 7 week-old banded and sham-operated animals was estimated as described by Kajustra et al.¹⁴. Briefly, this requires enzymatically dissociated for: 1. estimation of the frequency of mono-, bi-, and multi-nucleated myocytes (Hoechst staining): 2. Confocal microscopic estimation of the cell volume of these myocyte populations by staining with FITC and propidium iodide (to label nuclei); cell area was computed and thickness measured by optical sectioning in the Z plane. Next, the total ventricular volume was calculated as the weight of LV free wall myocardium /1.06 g/ml, the specific gravity of muscle tissue. This was corrected for the percentage of replacement fibrosis (Masson’s trichrome) to yield the volume of viable tissue, which was then used with the estimation of the volume fraction of myocytes in the viable tissue to yield the absolute volume of myocytes in the LV free wall. Next, the volume percentage of mono-, bi-, and multi-nucleated myocytes in the tissue was calculated, which in combination with the absolute volume of myocytes in the tissue allowed estimation of the aggregate volume of the different myocyte populations, and from this the number of myocytes in each population was estimated and summed to yield the total number of myocytes in the ventricle.

Statistics

All results are expressed as mean±one standard deviation unless otherwise stated. To adjust for age-related and growth-related changes in ventricular mechanics, echocardiographic variables were expressed as Z-scores (normal deviates) relative to the distribution of each variable in the control animals. A p-value ≤ 0.05 was considered statistically significant. Comparison of groups over time was performed using two way analysis of variance (ANOVA) (group and time) with repeated measures for time. Subsequent intergroup testing utilized Scheffe’s method for post hoc pairwise comparisons. Fold changes in caspase-3 activity between banded and age-matched sham controls were compared using the Wilcoxon rank-sum procedure. Comparison with the normal population (using Z-scores) was performed using a univariate t-test. Statistical analysis was performed on commercially available software (SAS Proc Mixed, SAS Institute Inc. Cary, NC).

Results

Control echocardiographic data

Thirteen normal rabbits were assessed serially from age three to six weeks during which time there was a nearly 4-fold increase in body mass. Heart rate, SF, VCFc, SF_mw, VCFc_mw, mass:volume ratio, systolic blood pressure, end-systolic and peak systolic stress, and end-systolic and peak fiberstress were found to have no significant change over time. The progressive increase in left ventricular end diastolic dimension, posterior wall thickness, and body weight reflects the rapid growth seen in these animals in the first seven weeks of life. Values from normal animals were used for comparison of banded animals to normal, aged-matched controls, as well as to describe the normal growth-related evolution of these variables. Using the measures of ESS obtained during afterload manipulation the normative date for the VCFc-ESS (SVI), SF_mw-FS_mw (SSI_mw) and VCFc_mw-FS_mw (SVI_mw) relationships were determined. Table 1 summarizes these data over time.

Table 1. Echocardiographic Parameters in control and banded Animals.

Echocardiographic indices in normal sham-operated rabbits measured at 3 to 7 weeks of age. Data are mean±SD, n=13 each.

Variable	Week 3		Week 4		Week 5		Week 6		Week 7
	control	banded	control	banded	control	banded	control	banded	control	banded
SSI (Z-score)	0.2±1.2	−0.1±1.1	0.1±0.9	0.1±0.9	0.4±0.8	0.1±1.7	0.8±1.1	−1.0±2.0	0.7±0.9	−1.6±1.9*
SSImw (Z-score)	0.1±0.8	0.6±1.3	0.2±0.9	−0.6±1.0	0.9±0.9	−0.8±1.3*	0.8±1.1	−1.3±1.6*	1.1±1.2	−1.4±1.0*
SVI (Z-score)	0.2±1.1	−0.4±0.2	0.4±0.8	−0.6±0.6	0.4±0.6	−1.2±0.2*	0.6±1.2	−1.6±0.4*	0.6±0.9	−1.7±0.5*
SVImw (Z-score)	0.2±0.7	0.2±1.7	−0.2±1.6	−0.6±1.3	0.4±1.2	−1.1±1.3	0.8±1.4	−1.4±1.2*	0.9±1.2	−1.6±1.1*
SF (%)	36.0±6.0	34.2±6.0	34.7±4.2	37.0±5.5	32.4±4.2	32.9±13.5	36.6±8.3	24.1±14.9	37.3±5.5	15.1±11.5*
SFmw (%)	19.5±3.1	21.1±5.1	19.5±3.8	18.0±4.2	15.8±3.2	14.8±6.4	21.1±5.0	11.8±7.7*	20.9±4.3	8.6±6.8*
VCFc (circ/sec)	1.8±0.4	1.6±0.4	1.7±0.3	1.8±0.5	1.6±0.2	1.6±0.7	1.8±0.5	1.1±0.7	1.8±0.3	0.7±0.5
VCFcmw (circ/sec)	1.0±0.2	1.0±0.3	1.0±0.2	0.9±0.2	0.8±0.2	0.7±0.3	1.1±0.3	0.5±0.3	1.0±0.2	0.4±0.2
Heart rate (b/min)	284±32	292±37	309±47	299±42	310±25	305±37	281±42	306±42	286±31	293±33
SBP (mmHg)	80±19	98.3±22.5	83±30	89.8±17.5	83±13	108.5±14.5	89±13	107.7±26	87±11	107±24.7
LVEDD (cm)	0.8±0.1	0.79±0.12	0.9±0.1	0.84±0.13	1.0±0.1	1.03±0.31	1.0±0.2	1.28±0.42	1.0±0.2	1.57±0.57*
Body weight (kg)	0.35±0.28	0.34±0.05	0.54±0.46	0.61±0.08	0.86±0.20	0.81±0.12	1.33±0.24	1.15±0.22	1.59±0.48	1.34±0.38

Abbreviations: LV, left ventricular; SF, shortening fraction; SF_mw, midwall shortening fraction; VCFc, rate of circumferential fiber shortening; VCFc_mw, midwall equivalent of rate of circumferential fiber shortening; SSI, stress shortening index; SSI_mw, midwall equivalent of stress shortening index; SVI, stress velocity index; SVI_mw, midwall equivalent of stress velocity index; SBP, systolic blood pressure; LVEDD, LV end-diastolic dimension.

^*p<0.05 compared to age-matched sham operated controls by ruskal-Wallis (Z-scores and SF) or ANOVA (SF, VCFc, SBO, LVEDD)

Serial Ventricular Mechanics After Aortic Banding

No significant alteration in shortening fraction was noted at the time of aortic banding. The rate of increase in body weight was not different from controls. Detailed serial myocardial mechanics were available in 36 rabbits from week 3 to 7 and are summarized in Table 1. The earliest detected change was an increase in LV mass:volume ratio (Figure 1), which was associated with a significant fall in peak systolic stress compared to age-matched controls (Figure 2) by week 4. Thereafter, mass:volume ratio progressively fell, resulting in elevated peak systolic stress and reflecting a failure of adequate hypertrophy (Figure 1 and Figure 2). Cardiac performance, as measured by SF and VCFc, declined progressively over time, reaching significantly lower levels by week 6 and 7 (Table2). Similar patterns were seen with SF_mw and VCFc_mw reaching significantly lower levels by week 5. Myocyte contractility, as quantified by the SSI and SVI relationships, although initially sustained in the normal range fell at week 6 (Table 1). The use of midwall indices of contractility, which avoids the overestimation of cardiac performance seen with endocardial indices in hypertrophied myocardium, revealed significant changes. SVI_mw and SSI_mw fell significantly by week 5 and continued to decline at 6 and 7 weeks. These serial changes reflect an initial period of adequate hypertrophy and later onset of inadequate hypertrophy and depression in myocardial contractility with concomitant chamber dilation.

Figure 1.

Effects of aortic banding of infant rabbits on LV mass:volume ratio determined by serial echocardiography. Data are mean±SD, n=6–10 each. ^#p<0.05 vs. age-matched controls; *p<0.05 vs. 4 week-old banded. LV mass:volume ratio increases early in hypertrophy and then progressively declines.

Figure 2.

Effects of aortic banding of infant rabbits on LV peak systolic stress (PkSS). Data are mean±SD, n=6–10 each. PkSS is reduced in banded animals at 4 weeks in association with increased LV mass (see Figure 1). Failure of continued hypertrophy in the face of increasing afterload results in a progressive increase in PkSS. *p<0.05 vs. controls.

Cardiomyocyte Apoptosis

Representative myocardial tissue sections with labeled nuclei (Hoechst-blue), myocytes (desmin-red), and nuclei undergoing DNA fragmentation (TUNEL-green) are shown in Figure 3a. Quantification of TUNEL-positive cardiomyocytes in this model is shown in Figure 3b. The baseline number of TUNEL-positive nuclei was in the range 26 to 48 per million cardiomyocytes and did not change in the sham-operated animals. Most interestingly, a significant increase in myocyte apoptosis was apparent as early as 2 weeks after aortic banding (approximately 3 to 4 weeks of age); this was occurring at a time when LV mass was increasing, LV wall stress had normalized, and contractility was preserved (compare to Figure 1 and Figure 2 and Table 1). There was similarly an increase in LV tissue caspase-3 activity at this early time point (Figure 4). The rate of myocyte apoptosis as indicated by TUNEL and activation of caspase-3 in the LV continued to increase, in association with inadequate hypertrophic compensation, increasing wall stress, decreasing contractility, and chamber dilatation.

Figure 3.

Apoptosis in LV myocardium of control sham-operated and banded infant rabbits. a. representative fluorescent photomicrographs of 5 µm sections of the left ventricular wall from banded rabbits. Sections were labeled with Hoechst to stain nuclear DNA (blue), desmin to identify cardiomyocytes (red), and DNA strand breaks (TUNEL-green). Following image acquisition, images were subjected to 2-D “no neighbors” deconvolution using a kernel that estimates the 3-D point spread function based upon the optics and imaging system. A and B, banded animal at 3 weeks; C and D, banded animal at 7 weeks. Arrows indicate apoptotic nuclei localized to myocytes; asterisks demonstrate TUNEL-positive staining in non-myocytes. b. assessment of myocyte apoptosis in sham vs. banded animals using this approach. Data are mean±SD, n=6–8 each. The difference between sham-operated and banded animals was significant at all ages.

Figure 4.

LV myocardial caspase-3 activity measured by cleavage of the specific fluorescent substrate DEVD-AFC, in age-matched sham operated control (open bars) and banded hypertrophying (solid bars) hearts. Data are mean ± SD, n=5–7 per group, and expressed as arbitrary fluorescent units relative to that measured in 7-week controls. (*p=0.04, ^#p=0.03, ⁺ p=0.008).

LV Myocyte Number and Fibrosis

There were no measurable differences in LV myocyte numbers within the sham-operated group; myocyte volume at 7 weeks of age was 22±5% greater than at 3 weeks (p<0.05). Compared to age-matched sham-operated animals at 7 weeks of age, total LV myocyte number decreased by 18±4% (p<0.05) and myocyte volume increased an additional 43±8% (p<0.05) in banded animals.

The relative proportion of collagen and muscle tissue n LV myocardium was assessed by Masson’s trichrome staining. Collagen represented approximately 6±2% of LV myocardial area. This was not changed with increasing age. Interestingly, there was also no significant difference between late hypertrophy (7 weeks) and age-matched controls in the LV area taken up by collagen (p>0.1).

Discussion

Improved surgical techniques and younger age of repair have improved outcome in congenital heart disease. Despite advances in operative techniques and myocardial preservation methods, progressive ventricular dysfunction remains one of the most important problems that hamper long-term outcome in the treatment of congenital heart disease. A growing body of evidence indicates that apoptosis is responsible for some degree of myocyte loss and myocardial dysfunction resulting from hypoxia, ischemia, reperfusion, hypertrophy, heart failure, and inflammation^{9–14, 16, 18}. The major findings in the present study in the developing, pressure-loaded left ventricle are that significant myocyte apoptosis: 1) begins with the onset of abnormal loading, 2) is not attenuated by normalization of wall stress during the period of compensatory hypertrophy, and 3) is sufficient to result in a measurable decrease in cardiomyocyte number as the ventricular dysfunction progresses to failure.

Chronic pressure overload induced failure of the adult heart has long been suspected to be related to exacerbated cardiomyocyte apoptosis²³. This link has been demonstrated in several experimental models^{21, 22}, and very recently mitochondrial proteins such as apoptosis inducing factor (AI)F as well as myogenic transcriptional regulators such as MyoD have been implicated in the molecular mechanism^{24, 25}. However, neither the time course nor the interrelation of cardiomyocyte apoptosis with the intricate changes in in vivo myocardial mechanics in response to pressure overload have been systematically analyzed. Moreover, the developing infant heart is likely to respond to pressure overload in a manner that is fundamentally different from the one observed in adult myocardium. The amount of myocyte apoptosis found in the present study is quantitatively similar to that measured in adult studies of pacing-induced heart failure, diabetes, and ischemic cardiomyopathy^{14, 17}. However, the degree to which apoptotic myocyte loss contributes to myocardial dysfunction and pathologic remodeling in any of these states remains uncertain. The effects of myocyte loss upon ventricular function in the growing ventricle are likely to be at least as great as in the adult due to the significant increases in cardiac size and function that must occur to permit body growth. The physiologic remodeling and hypertrophy that accompany normal growth require maintaining an adequate number of myocytes, which are generally believed to be terminally differentiated and hence incapable of significant replacement. However, given the recent appreciation of the potential for cardiomyocyte regeneration by cell cycle re-entry or via resident myocardial precursor cells, studies of myocyte proliferation as a possible mechanism to compensate for the hypertrophy-induced cell loss are clearly warranted.

The sensing and transduction mechanisms by which abnormal load is perceived and leads to the hypertrophic response remain uncertain, as are specific mechanisms responsible for myocyte apoptosis during hypertrophy. Current work favors connections between the myocyte cytoskeleton and the extracellular matrix, most likely focal adhesion kinases and integrins; signaling protein kinases within the extracellular matrix, altered sarcomeric function, altered substrate metabolism, ionic fluxes, and local neurohumoral mechanisms may also be involved in initiating the hypertrophic response². Downstream from the initiating signals, local production of cytokines, growth factors, catecholamines, and other compounds (e.g. endothelin, angiotensin) stimulate myriad protein kinases and phosphatases, thereby promoting numerous quantitative and qualitative changes in gene expression. In many ways these responses recapitulate the fetal cardiomyocyte gene program, and result in substantial alterations in cardiomyocyte signaling and functional pathways^2–6. The molecular events occurring during the hypertrophic response that could alter cell survival are numerous, and include imbalances in cell survival and death pathways such as those related to G-protein signaling, tumor necrosis factor-α, phophoinositide 3-kinase (PI3-K), calcineurin, and gp130, as well as altered calcium homeostasis, mitochondrial and substrate metabolism, and ischemia due to insufficient capillary development. We found that apoptotic cell death is initiated during early compensatory hypertrophy and ultimately leads to a net loss of approximately 18% of the myocytes in our rabbit model.

In adult models, a low but significant frequency of cardiomyocyte apoptosis has been found in experimental hypertension and during the transition to failure (but not during adaptive compensation) in experimental aortic stenosis, pacing-induced heart failure, as well as in humans with dilated cardiomyopathy and diabetes^{10, 14, 16–18, 22}. In the present report, the presence of both cardiomyocyte-specific TUNEL staining and LV tissue caspase-3 activity represent relatively strong evidence that one or more pro-apoptotic programs are activated early in infant LV as it is responding to increased pressure loading. It is important to note that both results, although significant, are relatively modest in absolute quantitative terms compared to what has been reported in a variety of ischemia-reperfusion and other models. For example, 8 week old rabbit LV subjected to 20 min of normothermic ischemia followed by 2 hours of reperfusion demonstrates approximately a 7 to 12 fold increase in caspase-3 activity (data not shown). In addition, these indices represent but a snapshot in time of a process that most likely is continuing and evolving over many days to weeks. Clearly, it could be useful to try to establish the quantitative relationship between the rate of apoptosis and the magnitude of ultimate cell loss. This computation, however, requires knowledge of the duration of TUNEL or other apoptotic signals in a cell that undergoes programmed cell death, which is not known and currently difficult or impossible to measure in vivo models. However, there is increasing evidence that a relatively modest degree of apparent cardiomyocyte apoptosis similar to that observed in the present study can have progressive and deleterious contractile consequences over time^{22, 26}. One recent study in sheep subjected to LV pressure overload also demonstrated onset of caspase activation during the apparent “compensated” phase of LV hypertrophy which markedly increased during the development of LV failure²². In terms of evolving contractile dysfunction, it is also important to note that caspase activation in cardiomyocytes can potentially result in the cleavage of contractile proteins (and hence contribute to the development of contractile dysfunction) independent of and/or in conjunction with effects leading to cell death²⁷.

The present study is the first to identify an increase in apoptotic cardiomyocytes during the early phase of adaptive hypertrophy. This is an important finding because it suggests that normalizing wall stress in the absence of other interventions is not sufficient to attenuate pro-apoptotic signals active in pressure-loaded myocardium. This result is consistent with recent evidence that normalizing wall stress is not essential to preserve contractile function and prevent ventricular dilatation in chronic pressure-overload^{6, 28}.

There are several important limitations to the present study: TUNEL staining has well-known difficulties, particularly in terms of specificity. However, the microscopic method used did identify apoptotic myocytes with known sensitivity and specificity that have been used and validated in under similar circumstances by others^{14, 16, 17}. TUNEL and other methods that detect DNA damage can show false-positive results, particularly in settings of necrosis (e.g. ischemia); however, the occurrence of ischemia and necrosis would seem to be quite unlikely in early and compensated stages of hypertrophy. Moreover, we were able to support the likelihood of increased apoptotic events using a sensitive and specific fluorescent substrate for caspase-3; unfortunately, when this assay is performed on whole LV tissue it cannot reveal the cell type(s) responsible for any measured increase in caspase activity. However, these results, when taken together with the subsequent measured decrease in myocyte number, strongly support the conclusion that substantial cardiomyocyte apoptosis was occurring in this model.

As noted above, the pathways involved in hypertrophy-associated programmed cell death are likely to be multiple and interwoven. A related issue that must be addressed in the future is whether the observed frequency of myocyte apoptosis and reduced number of cardiomyocytes are important causes of the progression to failure that should be targeted for inhibition, or instead is facilitating the controlled and beneficial (as opposed to necrosis, cell lysis, and inflammation) removal of damaged and dysfunctional myocytes and is therefore protective. It should also be emphasized that our findings can not be directly translated to the setting of pressure-overload hypertrophy in the adult heart. In our model, the animal slowly grows into in aortic band, developing progressive constriction. These kinetics cannot be mimicked in adult rabbits. Similarly, the situation in manifest end-stage heart failure is probably very different from that during compensated hypertrophy, but this was not a focus of the present study.

In summary, this study indicated that myocyte apoptosis is induced by pressure-overload in the infant myocardium, and is not attenuated during the period of compensated hypertrophy and normalized wall stress. From a clinical standpoint, these data suggest the pressure-load, even at its earliest stages, in not tolerated by the developing left ventricle. They therefore support early and complete hemodynamic correction in the infant whenever possible.

Appendix–Echocardiographic Methods

Blood Pressure

Determination of the accuracy of optical plethysmography was performed on eight occasions in rabbits between 4 and 6 weeks of age. After anesthesia was induced with intramuscular ketamine (20 mg/kg) and xylazine (0.5 mg/kg), the right upper limb was shaved and topical alcohol applied. Dissection of the neck vessels and direct cannulation of the carotid artery was achieved using a 24 gauge polyethylene catheter. A two Fr. High-fidelity (Millar, Houston, TX) catheter was inserted in line with the fluid filled catheter. The central arterial waveform was captured digitally using a standard analog-to-digital converter at a sampling rate of 200 Hz. Simultaneous right upper limb brachial arterial blood pressure was determined using optical plethysmography (Finapress, Ohmeda, Englewood, CO) The Finapress was applied to the shaved right forearm after application of topical alcohol over the region of the palpable brachial artery.

Comparison of the non-invasive arterial waveform with the high-fidelity direct central measurement was performed using customized software. Beat-to-beat waveform segmentation was followed by peak and trough detection to calculate systolic and diastolic values. Determination of the variation (bias) of optical plethysmography compared to Millar measurement was expressed as the mean and the standard deviation of the difference between non-invasive and invasive values. To estimate the magnitude of error (accuracy) the mean±standard deviation of the absolute differences was calculated. To quantify the magnitude of error in relationship to derived variables of stress, a comparison of Finapress-derived versus Millar-derived end-systolic stress and peak systolic stress was also performed. The bias, accuracy, and mean percentage absolute error were then calculated.

Spectral analysis of power and frequency components was performed on both the non-invasive and invasive arterial waveforms. Spectral analysis was performed using a non-parametric spectral estimation technique based on a discrete Fourier transform. Specifically a modified averaged periodogram method was used to decrease the high variance incurred with direct application of the discrete Fourier transform. A 30-second data segment was used for power spectral analysis. Transfer function (magnitude and phase) and power spectrum was assessed.

A comparison of the direct high-fidelity central arterial systolic and diastolic blood pressure and those recorded by Finapress was determined. The non-invasive waveform revealed a narrower pulse pressure and generally higher diastolic blood pressure. Bias as measured by mean and one standard deviation for the error was −1.1±4.0 and 0.85±3.2 mmHg for systolic and diastolic blood pressure, respectively. Accuracy, the absolute mean and standard deviation of the error, was 5.9±4.7 and 9.2±6.9 mmHg for systolic and diastolic blood pressure, respectively.

Fourier analysis revealed similar frequency components for both Millar and Finapress waveforms. A similar time course of the pressure waveforms was documented, supported by the transfer magnitude approximating one. Non-invasive waveform contains less power at the beat frequency and its harmonics suggesting that it does underestimate the pulse pressure. A phase lag consistent with the more peripheral measurement of the Finapress was noted.

To better appreciate the error incurred using the Finapress blood pressure in the determination of derived stress indices, calculation of derived myocardial indices at the time of invasive versus non-invasive blood pressure comparison was performed. The non-invasively derived indices were compared to invasively derived indices using the same measures of chamber size and wall thickness. Overall there was a small under representation of both stress measures (end-systolic and peak-systolic). Bias was −4.6±8.8 and −10.5±16.3 g/m² and accuracy was 6.95±7.1 and 14.1±13.2 g/m² for end-systolic and peak systolic stress, respectively. The mean percentage absolute error was 11.0±7.4% for end-systolic stress and 8.15±4.9% for peak systolic stress.

Echocardiography

Weekly assessment of cardiac function was performed from the time of aortic banding. Non-invasive blood pressure assessment was limited by animal size and was first measured at 3 weeks of age. All earlier studies were limited to two-dimensional and M-mode echocardiographic assessment. Thereafter, weekly evaluations were performed using echocardiographic and optical plethysmography. Transthoracic echocardiography was performed using an Accuson 128 or Hewlett-Packard Sonos 1500 Cardiac Imager equipped with a 7–7.5 MHz transducer. Adequate ultrasound surface-interface coupling was achieved by initial shaving of the anterior surface of the thorax and application of ultrasound gel. All imaging was performed under light restraint without sedation. Pulsed Doppler interrogation of the left ventricular outflow tract from an apical view with simultaneous electrocardiogram was recorded at high speed (100 mm/s) hardcopy printout from the strip chart recorder, for determination of ejection time. Two-dimensional cross-sectional imaging and M-mode interrogation of the left ventricular short axis at the level of the papillary muscles were performed. High-speed (100 mm/s) hard copy two-dimensionally directed echocardiographic M-mode recordings of the left ventricular minor axis were recorded simultaneously with electrocardiogram and non-invasive peripheral arterial pulse waveform using optical plethysmography (Finapress)

Determination of the normal endocardial SSI and SVI relationships and their midwall equivalents (SSI_mw SVI_mw) was undertaken in 13 control animals between weeks 3 and 7. To determine the normal relationship of the function indices to afterload, assessment at resting afterload was followed by determination at increased afterload during atropine and methoxamine administration. The animals were premedicated with atropine, 0.01 mg/kg (minimum dose of 0.1 mg) intravenously, to maintain stable heart rate during subsequent studies with repeat dosing as needed. After several stable recordings to determine baseline blood pressure and afterload, an intravenous infusion of methoxamine (25 µg/kg/min) was begun. A gradual increase in systolic blood pressure to 36–60 mmHg above baseline was obtained. A total of 56 measurements were obtained.

Analysis of Echocardiography

The non-invasive arterial pulse tracing, the left ventricular endocardial border of the septum, and the endocardial and epicardial borders of the posterior wall were hand digitized using a bit pad interfaced to a microcomputer-based digitizing station with custom software. This system is programmed to adjust the sampling rate of the tablet to 200 Hz, which is adequate to obtain 20 to 25 non-aliased harmonics of heart rates of 240 to 300 beats/min. After data input, the pulse transmission delay is corrected by electronically aligning the pulse upstroke to the time of the onset of ejection as determined from the left ventricular outflow Doppler sample. End systole was defined as the time of cessation of antegrade flow as determined from the left ventricular outflow track Doppler signal. The time of end systole was used to ascertain end-systolic pressure from the blood pressure recording.

From the digitized data, the following measurements were obtained by averaging three cardiac cycles: 1. pressure during left ventricular ejection, calculated by assignment of diastolic pressure to the minimum and systolic pressure to the maximum of the pulse tracing; 2. left ventricular internal diameter; 3. left ventricular posterior wall thickness; 4. left ventricular midwall dimensions by the method of Shimizu²⁹; 5. left ventricular meridional wall stress in g/cm², calculated throughout ejection according to Grossmann¹:

$$\mathit{\text{systolic}}\mathit{\text{stress}}=\frac{P\times D\times 1.35}{h\times (1+\frac{h}{D})\times 4}$$

where P is the left ventricular pressure in mmHg, D is left ventricular endocardial cross-sectional diameter in cm, h is posterior wall thickness in cm, and 1.35 is the conversion factor from mmHg to g/cm², and ventricular fiberstress (FS) according to the equation of Regen³⁰:

$$\mathit{\text{FS}}=\frac{1.35\times P(\frac{h}{\mathit{\text{ln}}(D+h)}-\mathit{\text{ln}}D)}{2\times h}$$

From the continuous data, end-diastolic values for short-axis dimension and wall thickness were taken at the time of maximal left ventricular dimension, and end-systolic values for short-axis dimension, wall thickness, blood pressure, meridional wall stress and FS were determined at the time of end-systole on the Doppler tracing. Meridional peak systolic stress and FS were obtained from the continuous systolic stress calcultation¹, ³⁰–32. Similar calculations were performed for peak and end-systolic FS. Left ventricular ejection time was measured from the left ventricular outflow Doppler tracing and adjusted to the heart rate of 60 beats/min by dividing by the square root of the R-R interval on the ECG. The left ventricular circumferential fiber shortening fraction and midwall shortening fraction were as calculated with use of short-axis dimensions as (end-diastolic dimension – end-systolic dimension)/end-diastolic dimension. VCFc was calculated as the SF divided by the rate-adjusted ejection time. Midwall VCFc was similarly calculated using midwall equivalents. The LV mass:volume ratio was estimated from determination of left ventricular epicardial and endocardial cross-sectional area as determined from M-mode measurements.

Stress-velocity and stress-shortening indices

The relation between rate-adjusted mean velocity of shortening and end-systolic meridional stress (SVI) has been shown to be a an afterload-adjusted, preload-independent index of contractility whether obtained as a global or regional index³³, ³⁴. The relation is inversely linear and in normal humans the slope of the relationship for any one person is parallel to the slope of the relation for the total study group. The value of the SVI for each banded rabbit was determined relative to the calculated distribution of this index in normal subjects. The SVI is reported as a normal deviate (Z-score) relative to the distribution in normals. Thus, the SVI=(X₁−µ)/σ where X₁=measured rate-adjusted mean velocity of circumferential fiber shortening, µ=the group mean rate-adjusted mean velocity of circumferential fiber shortening for the measured end-systolic meridional stress, and σ=the standard error of the regression for the group stress-velocity index at that level of end-systolic meridional stress³⁵. A SVI of less than negative two (i.e. two standard deviations below the normal mean value) was considered to represent myocardial dysfunction. Similarly the SSI is the relation of SF and ESS. This index of contractility is preload sensitive³⁰, ³³. The SSI is correspondingly quantified as the normal deviate of SF for the given ESS, obtained in a manner analogous to that described for the SVI. Fiberstress is stress independent of radial forces³⁰ and was used relative to midwall cardiac performance measures (SF_mw and VCF_mw) to determine load independent measures of contractility (SSI_mw and SVI_mw).