Main

LV pressure overload is a frequent physiopathologic component of several congenital heart defects. Depending on the degree and level of the anatomic cause for such LV pressure overload, myocardial consequences will be more or less severe during the antenatal and neonatal lives. With the advent in high-resolution echocardiography, prenatal diagnosis of such a lesion can be suspected with a high degree of sensibility in most cases (1), attention being first drawn to the ventricular asymmetry with a particularly enlarged RV. Indeed, in the human fetus with LV obstructive lesion, there is an RV hypertrophy followed after birth by LV hypertrophy and/or hypoplasia. Experimental studies trying antenatally to simulate this condition failed to reproduce the human natural history of such cardiovascular defects and invariably obtained an LV hypertrophy without right heart enlargement (2, 3). These studies, therefore, focused on the mechanisms of LV hypertrophy and demonstrated that it was mainly related to hyperplasia of cardiomyocytes without significant proliferation of interstitial cells. Although it is well known that cardiac growth in humans occurs by hyperplasia in the first few months after birth (4), more recent studies of human fetal hearts suggest that the switch from hyperplasia to hypertrophy may occur before birth (5). In an attempt to better understand morphometric variations of LV cardiomyocytes, namely hypertrophia versus hyperplasia, in prenatal LV pressure overload, we used the fetal lamb model of isthmic coarctation described by Burrington (2) and Fishman et al. (3). Also, because the normal LV is known to have very limited inotropic and lusitropic reserve, we aimed to investigate Ca2+ homeostasis in those prenatally hypertrophied LV by assessing the SERCA 2a that controls calcium reuptake by the longitudinal sarcoplasmic reticulum from the cytosol during diastole.

METHODS

Induction of CoA.

Sixteen pre-Alpes pregnant ewes (mean, 92 ± 0.7 d of gestation) were fasted for 12 h, then sedated with intramuscular ketamine and atropine (0.5 mg), and anesthetized with sodium pentobarbital (15 mg/kg i.v.).They were then intubated and ventilated with intermittent positive ventilation pressure (MMS 107 ventilator, Paris, France) at 15 breaths/min with a tidal volume of 12 mL/kg, fraction of inspired oxygen (Fio2) 100%; general anesthesia was maintained during surgery with halothane (0.5–2%). All animal procedures were in keeping with the principles of laboratory animal care formulated by the National Society for Medical Research and the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (National Institutes of Health Publication No. 85–23, revised 1985). The uterus was exteriorized through a midline laparotomy. The chest of one of the fetuses (nine fetuses total) was exposed via a short hysterotomy. The aortic arch of the fetus was approached via a left thoracotomy. Cotton string was then placed around the transverse aortic arch distal to the ovine trunk and proximal to the ductus arteriosus. The twins (eight fetuses) underwent a sham operation that consisted of the same surgical procedure without band placement around the aorta. After fetal chest closure, the fetus was returned to the uterus. All maternal incisions were closed. After surgery, the ewes were allowed to recover, and prophylactic antibiotics were administered for 3 d (3 M penicillin G procaine and gentamicin sulfate 5 mg/kg i.m.). Diclofenac (300 mg/d for 3 d) was routinely administered for tocolysis.

Fetal studies.

Between 27 to 37 d after the first operation (nine lambs in the first group before 30 d and eight lambs in the second group after 30 d) (Table 1), cesarean section and a new fetal left thoracotomy were performed for catheterization. Millar catheters were introduced into the left and right ventricles, left and right atria, ascending and descending aorta, and pulmonary trunk, and pressures and peak positive and negative dP/dt were measured (Uniflow-Baxter transducers, Gould Brush 2800 recorder). The pressure gradient through the aortic arch was assessed after ductus arteriosus cross-clamping. Blood gases in cardiac cavities and great vessels were measured with a radiometer blood gas analyzer (ABL 30, Jarre-Jacquemin). Left and right ventricular outputs were recorded by using a Doppler probe placed around the ascending aorta and the pulmonary artery trunk before the pulmonary artery bifurcation (model T106 M, Transonic Systems). The combined ventricular output (CVO) index was calculated as follows: CVO index = QLV + QRV/body weight (kg), where CVO index is the combined ventricular index in L·min−1·kg−1, QLV is LV output, and QRV is RV output. The fetuses were then killed with intracardiac potassium chloride and umbilical cord ligation and rapidly autopsied. The heart, great vessels, lungs, and liver were weighed, measured, and photographed. Right and left ventricular free walls were rinsed in cold saline (150 mM NaCl), immediately frozen in liquid nitrogen, and kept at −80°C until use. Interventricular septum was fixed in 10% formol overnight at 4°C and then embedded in paraffin.

Table 1 Morphometric findings before and after 30 d of fetal ventricular pressure overload NS = not significant.
Full size table

Histology.

Microtome sections of paraffin-embedded heart samples were stained with hematoxylin and eosin. The mean transverse diameter of 100 cardiomyocytes was measured in 10 randomly selected regions of the interventricular septum of each fetus by light microscopy at a magnification of ×200 (DML-8 Leica microscope, DXCI-938 CCD Sony camera, Visiolab software).

Morphometric analysis was performed with a computer (NS 15000, Nachet-vision, Paris, France). Total nuclear surface area and length were determined using the WIN 15000 program WINSEQ software (Microvision Instruments) on slices stained with Harris hematoxylin after treatment with 1% periodic acid at a magnification of ×40. Percentage binucleation was assessed with SAISAM software (Microvision Instruments) at a magnification of ×100 on slices stained with hematoxylin and eosin. Fibrosis was assessed on slices stained with Sirius red.

PCNA and Ki67 are two markers of proliferation expressed during proliferating phases of the cell cycle and, therefore, are assessed to investigate proliferative activity of the tissues. Staining using these antigens was performed on paraffin-embedded sections. After retrieval by microwave-oven processing, the proliferating cell nuclear PC10 MAb (Dako A/S, Copenhagen, Denmark) was used at a dilution of 1/20, and the Ki67 antigen MIB-1 MAb (BioGenex, San Ramon, CA, U.S.A.) was used at a dilution of 1/15 or MM1 mouse MAb (Novocastra, Newcastle, UK) was used at a dilution of 1/50. Detection was accomplished using the streptavidin biotinylated horseradish peroxidase method (Dako). Diaminobenzidine tetrahydrochloride 0.06% in PBS containing 0.03% hydrogen peroxide was used as a chromogen. In control experiments, the incubation steps with anti-PCNA were omitted.

Probes.

We used a rat SERCA 2a-specific cDNA probe named RH39-A corresponding to a 628-bp Eco RI-Pst I fragment of the RH39 probe (a gift from A.M Lompré). A 20-mer oligonucleotide complementary to part of the sequence of rat ribosomal 18S RNA was used to normalize the Northern blots. The latter and an oligo dT complementary to polyA+ mRNA were used for dot blots.

Northern blot and dot-blot analysis.

Northern blots were performed with 10 μg of total RNA from fetal LV for the 17 fetuses. Total RNA was purified as described by Chomczinsky and Sacchi (6). The blots were successively hybridized with the cDNA probe RH39-A and the 18S rRNA-specific oligonucleotide. The cDNA probes were labeled with [α32P]-dCTP by random priming (Megaprime labeling kit, Amersham), and the oligonucleotide probes were labeled with [γ32P]-ATP by using T4 kinase (Boehringer Mannheim GmBH, Germany). For the dot blots, serial dilutions (3, 1.5, 0.75, and 0.375 μg of total RNA) were spotted onto a nylon membrane (Hybond N+, Amersham), and dots were successively hybridized with oligonucleotide 18S RNA and an oligo dT. Autoradiograms were scanned and analyzed with the Starwise imaging system (Imstar); the expression level of SERCA 2a-specific mRNA in each fetus was normalized to the 18S rRNA level and is expressed as the SERCA/18S ratio.

Western blot analysis.

Western blot analysis was performed for four pairs of twins (four banded lambs and four sham). Fifty milligrams of LV was homogenized at 4°C with a P21 glass homogenizer in 1 mL of medium containing 100 mM KCl, 30 mM Tris-HCl (pH 7.4), and 5 mM sodium azide. Each homogenate was centrifuged at 1250 ×g for 2 min. The supernatant was collected, and the pellet was rehomogenized in 0.5 mL of isolation medium and centrifuged as above. The supernatants from the two spins were pooled and centrifuged at 10,000 ×g for 5 min. Proteins (20 mg) were separated by SDS-PAGE in reducing conditions on a 7.5% gel. The blot was preincubated with monoclonal anti-SERCA-2 antibody diluted 1:5000 (Affinity BioReagents MA3–910) and with peroxidase-labeled anti-mouse Ig diluted 1:3000 (Amersham). The reaction was revealed by using the Renaissance Western blotting reagent from NEN.

Ca2+ uptake by sarcoplasmic reticulum.

For eight pairs of twins (eight banded lambs and eight sham), calcium uptake was measured as previously described (7). Fifteen to eighty milligrams of homogenate protein was added to 0.5 mL of medium containing 0.1 M KCl, 20 mM HEPES (pH 7.25), 5 mM sodium azide, 5 mM ATP, 6 mM MgCl2, 10 mM potassium oxalate, 50 mM EGTA, and 50 mM 45CaCl2. Calcium uptake was linear up to 15 min. No time-dependent calcium uptake was registered in the absence of MgATP. The reaction was quenched in 7 min by filtration through a 0.45-mm Millipore filter. The filter was washed twice with 3 mL of solution containing 0.1 M KCl, 20 mM HEPES (pH 6.2), and 1 mM EGTA, and absorbed radioactivity was measured by scintillation counting.

Statistical analysis.

Values are expressed as mean ± SD. Data were analyzed using 2-way ANOVA, which allowed us to distinguish among an effect of the coarctation (sham versus CoA), an effect of the duration of the coarctation (before and after 30 d), and an interaction between the two factors. p values < 0.05 denote significant differences.

RESULTS

Induction of CoA and hemodynamic analysis.

None of the ewes or fetuses died. Surgery resulted in severe obstruction of the isthmic aorta in all banded fetuses and in none of the sham-operated twins, as shown by the decrease in the diameter of the preductal aorta (mean 0.8 ± 0.4 mm versus 4.2 ± 0.6 mm, p < 0.0001). As well, isthmic obstruction was confirmed by the considerable reduction of the isthmic aorta versus the ascending aorta (19% in sham versus 80% in CoA, p < 0,0001). The heart rate was similar in the two groups. LV systolic pressure and ascending aorta systolic pressures were higher in the banded group than in the sham group (LV pressures, 57.9 ± 4.6 and 43.6 ± 2.2 mm Hg, p < 0.05; ascending aorta pressures, 52.3 ± 4.7 and 37 ± 2.5 mm Hg, p < 0.05). There was a trend toward increased systolic pressures in the RV and pulmonary artery, but it did not reach statistical significance. Diastolic ventricular pressures were similar in both groups, as were positive or negative LV dP/dT. CVO was not affected by isthmic obstruction, and neither was the percentage of LV output to CVO (33.8 ± 3.7 versus 39.7 ± 2.5%, p = 0.08). The intracardiac shunting portion of CVO, assessed as the ratio of RV output to LV output, was higher in the banded group versus sham (2.38 ± 0.33 versus 1.76 ± 0.16, p < 0.05). Oxygen saturation in the left heart cavities was increased in the banded group compared with the sham group, without reaching statistical significance.

Cardiac hypertrophy.

LV wall thickness and ventricular septum thickness were significantly higher (Table 1) in the banded group than in the sham controls (LV thickness, 7.67 ± 0.29 versus 6.13 ± 0.40 mm, p < 0.005; septum, 7.67 ± 0.41 versus 6.50 ± 0.33 mm, p < 0.05). The effect of duration of the coarctation was not significant using 2-way ANOVA.

Histologic findings.

The histologic aspect of myocardial tissue was similar in the two groups (Fig. 1, A–D). In particular, there was no fibrosis or interstitial cell deposition on the endocardial layer. Statistical analysis on the mean transverse diameter of 100 cardiomyocytes obtained for each lamb showed that LV pressure overload had a significant effect (p < 0.05), as well as the duration of LV pressure overload (p < 0.001), and that there was a significant interaction between the two factors (p < 0.001). When analyzing each experiment case by case, we were able to find that at 30 d after fetal surgery, there was an inversion in the ratio of mean transverse diameter of cardiomyocytes obtained from banded or sham lambs. Thirty days after fetal surgery, the mean transverse diameter of cardiomyocytes was 4.8 ± 0.1 μm in the banded group (Fig. 1D) compared with 6.6 ± 0.1 μm in the sham-operated twins (Fig. 1C), whereas, before 30 d after fetal surgery, the mean transverse diameter of cardiomyocytes was 7.2 ± 0.1 μm in the banded group (Fig. 1B) compared with 5.7 ± 0.1 μm in the sham-operated twins (Fig. 1A). A diagram presents the diameter of myocytes as a function of time of CoA (Fig. 2). Statistical analysis of these data demonstrated that the curves of diameter of myocytes from CoA and from sham intersect between 29 and 30 d of postoperative periods. In banded lambs, therefore, we considered two subgroups of experiments: before and after 30 d after surgery. Statistical analysis of these data are summarized in Table 1. Morphometric studies failed to demonstrate a significant change in the total number of nuclei or nuclear surface area. However, the mean percentage of binucleated cells in the banded group versus sham appears to decrease 30 d after aortic band placement (3.25 ± 1.71 in CoA versus 7.50 ± 1.29 in sham, p < 0.05).

Figure 1
figure 1

Photomicrographs of LV cardiomyocytes stained with PCNA. (A) and (B), Less than 30 d after surgical procedure; (C) and (D), more than 30 d after surgical procedure. (A) and (C), Sham-operated fetuses; (B) and (D), coarcted fetuses. Bar = 30 μm. Mean transversal diameter is 5.7 ± 0.06 μm in (A), 7.2 ± 0.07 μm in (B), 6.6 ± 0.09 μm in (C), and 4.8 ± 0.05 μm in (D).

Full size image
Figure 2
figure 2

Diagram presenting myocyte diameter evolution according to the length of postoperative periods in nine CoA and eight sham. □ = CoA, • = sham.

Full size image

PCNA labeling was detected in all heart preparations, but the expression of the antigen was similar in all groups and subgroups. We failed to detect the expression of Ki67 antigen with the MAb used.

SERCA 2a analysis.

A single hybridization signal was obtained with each sample using the RH39-A probe. The RH39-A probe detected a band at 4.4 kb, corresponding to SERCA 2a mRNA. The LV SERCA/18S ratio was markedly lower in the coarcted group than in the sham-operated twins [0.53 ± 0.20 versus 1.86 ± 0.43;p < 0.05 (Fig. 3)].

Figure 3
figure 3

Northern blot hybridization with a SERCA 2a cDNA probe. Ten milligrams of total RNA from LV (VG) and RV (VD) was loaded on each lane: sham ventricle (S), CoA ventricle (CoA). The blot was then hybridized with an 18S oligonucleotide to check the quantity of RNA in each lane.

Full size image

The RV SERCA/18S ratio was lower in the coarcted group than in the sham-operated twins (0.74 ± 0.14 versus 1.17 ± 0.25), but the difference was not significant. In four pairs of twins studied by Western blot, no significant difference in SERCA 2a protein levels was observed. Ca2+ uptake studies showed no significant difference between the coarcted group and the sham controls (data not shown). Dot-blot analysis with an oligo-dT probe and an 18S probe showed that the polyA+ mRNA to 18S ratio was increased in the coarcted group compared with the sham controls (without reaching statistical significance) but suggested an increase in expression of mRNA in the coarcted group.

DISCUSSION

LV pressure overload is a frequent component of several congenital heart defects. The anatomic cause for such hemodynamic perturbation may be at each level of the LV outflow tract including the ascending aorta and aortic isthmus. One of the most frequent causes is the isthmic aortic coarctation where the coronary and cerebral blood flows are exclusively provided by LV output and descending aortic flow is provided partially or totally by a patent ductus arteriosus. In this study as others (2, 3), we tried to reproduce a model for fetal preductal aortic coarctation by placement of an initially loose string around the transverse aorta, which became constrictive as the fetus grew in utero. The initial operation was performed at midgestation, and morphologic and hemodynamic control was done 1 mo later. However, as others, we were not able to reproduce the natural history of human fetal coarctation because this model did not lead to RV enlargement, which is an important finding in human fetal coarctation. In the present era, antenatal diagnosis of human aortic coarctation, although difficult, can be suspected as soon as wk 18 of gestation and is probably present but not detectable even before. Therefore, the hemodynamic perturbations caused by such lesion, namely an increase in RV output secondary to an increase in the intracardiac shunting portion of the combined ventricular output, may have long-lasting effects resulting in right heart enlargement. The hemodynamic data presented in this study partially support this theory. We found an increase in the intracardiac shunting portion of the CVO, and the RV output showed a trend toward an increased output. We suspect the reason for being unable to reproduce the human fetal coarctation is time related. Indeed, the fetuses were operated on at midgestation, and a loose band was placed around the transverse aorta. We believe that this band became restrictive much later, and then morphologic control was performed, giving no chance to a time-related effect. On the other hand, in a series of experiments not reported in the present study, we found that acute occlusion of the transverse aortic arch at midgestation resulted in 100% fetal mortality. In 1978, Burrington (2) applied the same approach of loose band around the transverse aortic arch to create a model of fetal aortic coarctation at 78 d of gestation. LV hypertrophy was present in all experiments; unfortunately, there was no information regarding RV hypertrophy.

Our data support at least that in the very early stages of LV pressure overload induced during fetal life, there are mild hemodynamic perturbations that are associated with LV hypertrophy. LV weight to body weight ratio was increased by 65%, and the LV wall was thickened by 25% compared with twins controls. In an attempt to better define the morphologic patterns or LV hypertrophy, we found a biphasic response of cardiomyocytes to LV pressure overload, and this response was time dependent. In short time experiments (<30 d after fetal surgery), the mean transverse diameter of cardiomyocytes was significantly increased versus sham; on the other hand, in long-lasting experiments (>30 d after fetal surgery), it was significantly decreased versus sham-operated twins. There was, however, no significant difference in LV wall thickness whatever the duration of the experimental model. These results suggest that in the former group, myocyte hypertrophy was responsible for LV enlargement, whereas, in the latter, myocyte hyperplasia was the mechanism for cardiac enlargement. This was further documented by an apparent decrease of mean percentage of binucleated cells in the banded group versus sham, although not significant, without any significant change in the total number of nuclei or modification in the total surface area between the two groups (see below). This ability to divide is a characteristic feature of fetal cardiomyocytes (3, 8, 9) but is usually believed to be lost soon after birth (10), with a switch from hyperplasia to hypertrophy. More recent studies of human fetal hearts suggest that this switch occurs even before birth (5). During the neonatal conversion from hyperplastic cell growth to hypertrophic growth in the rat heart, myocytes divide and then undergo a last nuclear division without cytokinesis, leading to myocyte binucleation (11). The latter observation, therefore, corresponds to the arrest in cell division. Interestingly, in our model after an initial hypertrophic phase, the fetal cardiomyocytes appear to undergo hyperplasia. This is in agreement with recent studies demonstrating an increase of myocardial cell proliferation by aortic banding in guinea pig hearts (12). To further document the mechanism of cardiomyocyte hyperplasia, we performed morphometric measurements as proliferating cells are expected to show nuclear modifications. The ratio of either mononucleated or multinucleated cells to the total number of nuclei can provide an estimation of the total number of myocytes (13). We, therefore, found a decreased number of binucleated myocytes in the long-lasting banded group without, however, reaching statistical significance, probably due to the small number of experiments. Combined with an identical number of nuclei, a similar total nuclear surface area, this suggests an increased number of cardiomyocytes in the long-lasting banded group. Furthermore, as binucleated cells are thought to reflect the end of cytokinesis and to delineate the switch from myocyte hyperplasia to hypertrophy, the elevated percentage observed in the sham group may suggest an arrest of mitosis in this group. In an attempt to better document cell proliferation, cell proliferation markers Ki67 antigen and PCNA were used. PCNA was detected in all groups without the expected significant increase in pressure overloaded lambs. Indeed, although PCNA expression is widely accepted as a marker for DNA synthesis, it could also be detected in cells undergoing DNA repair or apoptosis (14). Also, PCNA is expressed only at the G1-S boundary of the cell cycle (15, 16), selecting, therefore, only those cells in this phase. As the cell divisions in the myocardium are not synchronous, the use of PCNA as a marker for cell proliferation may underestimate the true state of proliferation. Therefore, to increase the sensibility for detection of cell proliferation, we used the cell proliferation marker Ki67 antigen, which is expressed during more cell cycles (late G1, S, G2, and M phase). Unfortunately, this marker did not cross-react with the fetal lamb myocardium.

The probe RH39-A (for SERCA 2a) was constructed with the rat coding sequence to increase the likelihood of cross-hybridization with the uncloned sheep sequence. Indeed, the coding sequence of the SERCA 2a gene is 90% identical among rat, rabbit, and human (17). The specificity of the RH39-A probe was confirmed by the detection of a single band of 4.4 kb, which corresponds to the size of SERCA mRNA in mammals. The lack of fibrosis and cellular infiltration supported the use of the SERCA/18S ratio to assess SERCA 2a gene expression in cardiomyocytes. As SERCA 2a is the only isoform expressed in the cardiomyocytes of all mammals so far studied (18, 19), we did not try to identify other isoforms.

Our results demonstrate a significant decrease of SERCA 2a mRNA level in CoA compared with sham. In comparison with models of postnatal ventricular hypertrophy, the reduction in the steady state SERCA 2a mRNA level in the fetal ventricle is reminiscent of that observed in models of ventricular hypertrophy induced after birth (20, 21). However, the fetal ventricle is characterized by immaturity of the sarcoplasmic reticulum. Calcium exchanges are mainly controlled by the Na+-Ca2+ exchanger of the plasma membrane (22, 23), and SERCA 2a plays an important role only at the end of gestation and after birth (18). Therefore, the regulation of SERCA 2a expression may be different in fetal and adult heart, in normal and pathologic conditions. It has been recently demonstrated that during the physiologic neonatal period, the increase of SERCA 2a gene products is regulated through posttranscriptional mechanisms (24).

Unlike models of postnatally induced myocardial hypertrophy, no correlation was observed between the reduction in the SERCA/18S ratio and the extent of myocardial hypertrophy and hemodynamic parameters reflecting ventricular inotropy (dP/dt max) and lusitropy (dP/dt min). A correlation between LV lusitropy (dP/dt min) and calcium uptake from cardiac sarcoplasmic reticulum was demonstrated in fetal rabbits (25). In our model, the lusitropy is preserved despite the decrease of SERCA 2a mRNA. These differences could be explained by the fact that the reduction in SERCA 2a mRNA does not necessarily imply a decrease in the density of sarcoplasmic reticulum calcium pumps or less efficient function. Other regulatory mechanisms, for example at the translational or posttranslational levels, may be involved in the hypertrophied ventricles (26). This is in agreement with our findings that in the sets of twins studied for protein expression (n = 4) and Ca2+ uptake (n = 8), no difference was observed between the banded and sham-operated groups. Discrepancy between SERCA 2a protein levels and mRNA levels has also been observed in adult heart. The failing human heart was found to contain normal SERCA 2a protein levels and 50% reduced mRNA levels (27), suggesting that altered sarcoplasmic reticulum function in human heart failure is not caused by altered expression of this protein. The reduced SERCA 2a mRNA levels in human heart failure are well documented (28–30), but the corresponding protein level remains controversial (31, 32).

In conclusion, preductal coarctation of the aorta in utero induces LV hypertrophy. Hemodynamic and histologic results showed that this hypertrophy resulted from cardiomyocyte hypertrophy when the duration of banding was <30 d and from hyperplasia when the duration of banding was >30 d, suggesting a shift of the response of cardiomyocytes to ventricular overload. Our results also indicate that LV overload leads to molecular changes that appear to be an initial event of the evolutionary pattern. Indeed, gene expression is modified as reflected by the reduction in SERCA 2a mRNA in pressure overloaded fetal lamb LV at all stages of banding. This reduction in SERCA 2a mRNA was observed despite ontogenic maturation, which is normally associated with an increase in SERCA 2a mRNA. Two different processes are involved in the genetic program of pressure overloaded fetal lamb LV: ontogenic maturation and pressure overloaded-induced hypertrophy. The result of the two appears to override and/or slow myocardial maturation.