Endocardial Endothelium in the Avascular Heart of the Frog: Morphology and Role of Nitric Oxide

The endocardial endothelium (EE), strategically located between the superfusing luminal blood and the underlying cardiac musculature, plays an obligatory role in controlling myocardial performance in various mammalian species in a manner analogous to the autocrine–paracrine autoregulation of smooth muscle by vascular endothelium (Brutsaert and Andries, 1992; Brutsaert et al. 1988; Smith et al. 1991). The endothelium of the coronary (micro)vasculature has also been found to regulate myocardial performance (Brutsaert et al. 1996). The secretory function of the EE involves the synthesis and release of NO and of other autacoids including prostacyclin and endothelin.

Although both microvascular coronary and endocardial endothelial cells produce common autacoids and share similar roles in the signal transduction initiated by circulating hormones, neurotransmitters or mechanical stimuli, they clearly represent two cell populations with distinct embryological origin and morphological and functional characteristics. To what extent the interactions between myocardium, microvascular coronary endothelium and EE are integrated in cardiac homeostasis and the mechanisms involved are the subject of intensive investigation. In mammalian cardiac muscle preparations, evaluation of the specific role of the EE is hampered by the coexistence of the microvascular coronary endothelium. In contrast, amphibians such as the frog possess an avascular heart, i.e. the myocardium is totally trabeculated and is supplied by luminal blood solely through the intertrabecular spaces (i.e. lacunae) (Grant and Regnier, 1926; Staley and Benson, 1968). Given the fully trabeculated wall of the ventricle and the correspondingly very high ratio of cavity surface to ventricular volume, the EE in the frog has a relatively greater mass than the EE in the compact type of ventricular myoarchitecture in mammals. The frog heart can, therefore, be used to study direct interactions between the EE and myocardium in the whole working heart, without interference from a microvascular coronary endothelium.

The aim of this paper was to explore the potential of using an avascular heart to study the role of the EE. We investigated the morphological features of the EE in the heart of Rana esculenta with particular reference to the presence of the constitutive form of NO synthase (cNOS). Because the EE of the intact working heart is continuously exposed to physical forces associated with blood flow, we used an in vitro isolated and perfused frog heart preparation working at physiological haemodynamic loads (Acierno et al. 1994) to analyse the functional role of the EE in relation to cNOS activity.

Frog hearts (N=104) were isolated from specimens of both sexes of Rana esculenta (weighing 20.8±0.7 g; mean value ± S.E.M.) and connected to a perfusion apparatus as previously described (Acierno et al. 1994). Experiments were carried out at room temperature (18–21 °C) from autumn to spring. In electrically paced heart preparations, a Grass S44 stimulator was used (single pulses of 20 V, 0.1 s) with a stimulation rate identical to the control (unpaced) rate. The saline composition was (in mmol l⁻¹): NaCl, 115; KCl, 2.5; CaCl₂, 1.0; Na₂HPO₄, 2.15; NaH₂PO₄, 0.85; anhydrous glucose, 5.6; pH was adjusted to 7.2 by adding Na₂HPO₄. The final molarity was 2.46 mmol l⁻¹ (Singh and Flitney, 1981). The saline was equilibrated with air. The mean input pressure (preload) was regulated by varying the height of the input reservoir in relation to the level of the atrium, and the minimal output pressure (afterload) was similarly controlled by adjusting the reservoir height with respect to the aortic trunk.

Pressure measurements were made through T-tubes placed immediately before the input cannula and after the output cannula, and connected to two MP-20D pressure transducers (Micron Instruments, Simi Valley, CA, USA) in conjunction with a Unirecord 7050 (Ugo Basile, Comerio, Italy). Pressure measurements were expressed in kPa and corrected for cannula resistance. Heart rate was calculated from pressure recording curves. Cardiac output was collected over 1 min and weighed; values were corrected for temperature and fluid density and expressed as volumes. Cardiac output (Q̇) and stroke volume (VS= Q̇/heart rate) were normalized per kilogram of wet body mass. Stroke volume at constant pre- and afterload in paced hearts was used as a measure of ventricular performance; changes in stroke volume under these conditions were considered inotropic effects. Ventricular stroke work (WS) was calculated in mJ g⁻¹ as (afterload − preload × stroke volume)/ventricle mass. The duration of the systolic phase (see Fig. 3A) was calculated from recording traces.

In all experiments, hearts were initially allowed to maintain a spontaneous rhythm. The afterload pressure was set at 3.92 kPa (40 cmH₂O) and the input pressure was adjusted to obtain a cardiac output of approximately 110 ml min⁻¹ kg⁻¹ wet body mass. These values were within the physiological range (Shelton and Jones, 1965). Cardiac output, heart rate and aortic pressure were simultaneously measured during the experiments. Hearts that did not stabilize within 10 min from the onset of perfusion were discarded. Ten hearts were discarded (8.8 %). The parameters of cardiac performance under basal conditions were measured after 20 min of perfusion; in the treated hearts, the 20-min control period was followed by a further 20 min period during which hearts were perfused with drug-enriched saline containing a given concentration of N^G-nitro-L-arginine methyl ester (L-NAME), Methylene Blue (MB), 8-bromo-cGMP, sodium nitroprusside (SNP) or 3-morpholinosydnonimine (SIN-1) (SNP and SIN-1 were used in a darkened perfusion apparatus to prevent degradation). SNP was added cumulatively from 10⁻⁷ to 10⁻⁴ mol l⁻¹ and SIN-1 from 5×10⁻⁸ to 10⁻⁵ mol l⁻¹.

When the heart was stabilized in the basal condition after 10–15 min of perfusion, 0.1 ml of Triton X-100 (0.05 % in saline) was introduced through the aortic trunk, to avoid damage to the atrium, using the following procedure: the inflow was closed, the afterload was increased to approximately 7 kPa, and Triton X-100 was injected through a needle inserted into the output cannula. After 3–4 isovolumetric systoles, the inflow was reopened and the outflow was adjusted to the control value, the perfusion being continued with saline. This procedure induced filling of the ventricle as a result of temporary incompetence of the valve. Preliminary experiments with Evans Blue dye showed that, using this procedure, there was no backflow into the atrium. Parameters of cardiac performance were measured after 20 min of perfusion with saline following a stabilization period of 30–35 min from the onset of perfusion. Control experiments with afterload increased to 7 kPa and injection of 0.1 ml of saline instead of Triton X-100 (0.05 %) did not alter heart rate (58±3 beats min⁻¹, mean ± S.E.M., N=5), cardiac output (112±18 ml min⁻¹ kg⁻¹) and stroke volume (2.0±0.1 ml kg⁻¹) (data not shown), indicating that Triton X-100 per se and not the perfusion manipulation affected cardiac pump performance.

In the group of detergent-pretreated hearts tested for drug effects, the 20 min period of perfusion with saline was followed by a further period of perfusion with the drug-enriched saline, after which the performance parameters were measured. In this group, the total duration of perfusion was between 55 and 60 min.

The results are expressed as mean values ± one standard error of the mean (S.E.M.). Each heart received only one concentration of the drug being tested, under control conditions, after Triton X-100 exposure or in the presence of L-NAME. Since each heart thus represented its own control, the statistical significance of differences was assessed on actual parameter changes using the paired Student’s t-test. Comparisons for the cumulative response to SNP and SIN-1 were made using Bonferroni’s correction. Percentage changes were calculated as the mean ± S.E.M. of percentage changes obtained from individual experiments.

Isolated working hearts were perfusion-fixed for 10–20 min at room temperature with 4 % paraformaldehyde in a Hepes-buffered solution containing (in mmol l⁻¹): CaCl2.2H2O, 1.2; MgCl₂.6H₂O, 0.5; KCl, 2.7; KH₂PO₄, 1.5; NaCl, 136.9; NaH₂PO₄.H₂O, 8.1; Hepes, 25.0; D-glucose, 5.0; pH 7.2. Tissue samples were rinsed in buffer and then dissected from the ventricular and atrial walls. Tissue samples were stained en block for en face confocal scanning laser microscopy as previously described (Andries and Brutsaert, 1993). Filamentous actin was stained by Bodipy-phallacidin (Molecular Probes; N=4 hearts). The von Willebrand factor (vWF; N=2) and constitutive nitric oxide synthase (cNOS; N=2) in endothelial cells were labelled using commercial antibodies (polyclonal antibody against human vWF and secondary antibodies coupled to FITC, Dakopatts; polyclonal antibody against a synthetic peptide of a sequence derived from bovine cNOS, amino acids 599–613, Affinity Bioreagents). Stained tissue blocks were mounted in a small chamber on a slide filled with Slowfade (Molecular Probes) and observed with a Biorad MRC-600 confocal scanning laser microscope equipped with an Argon ion laser and using the BHS single-channel filter block.

To detect damage to endothelial cells and myocytes in control frog hearts (N=2) and after experimental protocols with Triton X-100 (N=2), hearts were perfused for 10 min with a saline solution containing 8×10⁻⁶ mol l⁻¹ ethidium homodimer-1 (Molecular Probes). This viability tracer enters only dead cells and binds to DNA and remaining RNA. Unbound tracer was washed away by continuous perfusion with saline. Hearts were perfusion-fixed and processed for en face confocal microscopy as described above. Dissected tissue samples were stained for F-actin or with TOPRO, a nuclear stain that allows visualization of nuclei from all cells in cardiac tissue. Double-labelled images were obtained using the dual-channel filter blocks (A1 and A2) of the Biorad MRC-600.

Ventricular and atrial trabeculae were both covered with endothelial cells that were positively labelled for vWF (Fig. 1A–C). Sections through trabeculae revealed that no cells were stained for vWF within the trabeculae (Fig. 1D). En face optical sections of the EE showed extensive vWF staining adjacent to labelled nuclei. The shape of the stained structures ranged from small spherical particles to small tubule-like particles and included larger, more amorphous, mainly juxtanuclear structures (Fig. 1B,C). In many sections, broad regions corresponding to the peripheral parts of EE cells were almost completely devoid of staining.

F-actin labelling by Bodipy-phallacidin intensively stained myocytes in the ventricular and atrial trabeculae. Since myocytes and EE cells are very closely apposed, very thin optical sections had to be used to observe the actin filaments in EE cells. In a limited number of EE cells, it was possible to see that the actin filaments formed a thin peripheral band (Fig. 1E) outlining a hexagonal cell shape or a less-regular and more elongated cell shape. Some short, thin actin filaments were present in the endothelial cytoplasm. Nuclei of EE cells were frequently visible as dark circular or oval areas framed by the actin staining in myocytes.

Staining for cNOS was restricted to endothelial cells enveloping the trabeculae in ventricular and atrial tissue. Sections through the trabeculae demonstrated specific staining only in EE cells. En face optical sections of EE showed cytoplasmic labelling outlining the circular or oval-shaped nuclei (Fig. 1F). There was usually a densely stained area near the nucleus. Controls in which the primary antibody was omitted or in which polyclonal antibodies were used that did not react with frog tissue showed a low level of background staining throughout the EE and myocardium and a few granule-like structures.

To assess the viability of cardiac tissue in control hearts and after Triton X-100 treatment, working hearts were, prior to fixation, treated with ethidium homodimer-1, which passes through damaged cell membranes and reacts with DNA and the remaining RNA. Superficial en face sections of ventricular trabeculae from control hearts, stained with TOPRO after fixation, showed a regular pattern of circular or oval-shaped nuclei in the green channel (Fig. 2A); no staining was observed in the red channel of the confocal microscope. These nuclei represent viable EE cells. Similarly, sections deeper into trabeculae demonstrated only green highly elongated nuclei of viable cardiac myocytes (Fig. 2B). After actin labelling, myocytes had a normal striation pattern (Fig. 2E). In atria, a few EE cells were occasionally damaged, as shown by the presence of yellow endothelial nuclei in merged dual-channel images (Fig. 2C).

The morphology of frog hearts after Triton X-100 treatment was virtually identical to that of control hearts (Fig. 2D,F). Only very occasionally were damaged EE cells present in ventricular tissue (Fig. 2D). No damaged myocardial cells were observed in ventricular tissue; myocytes had a normal actin pattern (Fig. 2F).

Time courses for stroke volume and heart rate are shown in Fig. 3, both under control conditions (N=5) and after treatment with Triton X-100 (0.05 %) (N=5) in unpaced preparations. In control conditions, the performance of the frog heart was stable for at least 1 h, after which cardiac output decreased linearly (Acierno et al. 1994) Within 5 min of treatment with Triton X-100, both VS and Q increased and stabilized at these higher values within . 10–15 min. Stroke volume increased by 35±5 ml kg⁻¹, Q increased by 23±8 % (data not shown) and Fh decreased significantly (by 10±2.7 beats min⁻¹) for at least 1 h. In addition, there was a similar decrease of fH (11±1.6 %) in a series of unpaced experiments with Triton X-100 (N=12) in which the effects of detergent were tested only during the first 20 min of perfusion with saline after Triton X-100 treatment (data not shown).

To avoid changes in performance secondary to changes in heart rate (the ‘staircase’ effect), the experiments were repeated in paced hearts. Under these conditions, treatment with Triton X-100 caused a significant increase in cardiac output (from 112 to 121 ml min⁻¹ kg⁻¹, P<0.05); this corresponded to an increase in cardiac output and stroke volume of 14±3.5 %; WS increased by 12±3.3 % (Fig. 4).

The L-arginine analogue N^G-nitro-L-arginine methyl ester (L-NAME) was used to inhibit endogenous NO production in the isolated working frog heart. When L-NAME (10⁻⁴ mol l⁻¹) was administered (Fig. 5A), stroke volume of the paced heart increased by 5.3±2.5 % and stroke work increased by 6±2.6 %. Endogenous NO thus appears to have a mild depressant effect in the frog heart. The stimulatory effect of L-NAME was abolished by prior treatment with Triton X-100 (Fig. 5B).

Since NO activates guanylate cyclase activity, the effect of increased intracellular cGMP levels was mimicked by exogenous administration of 8-bromo-cGMP, a poorly hydrolysable analogue of cyclic GMP (Fig. 6A). Administration of 10⁻⁶ mol l⁻¹ 8-bromo-cGMP induced a significant decrease in VS and WS (7.9±2.8 % and 9±2.8 %, respectively). Administration of 10⁻⁶ mol l⁻¹ Methylene Blue, which at this concentration blocks NO-mediated activation of guanylyl cyclase and therefore decreases endogenous cGMP production, increased stroke volume by 8.5±2.9 % and stroke work by 8.8±3.5 % (Fig. 6B).

SNP, an exogenous NO donor administered in cumulative concentrations (Fig. 7A), significantly depressed the performance of unpaced hearts at 10⁻⁴ mol l⁻¹, with a 22±3.6 % decrease in stroke volume, i.e. from 2.0±0.07 to 1.7±0.10 ml kg⁻¹, and with a 21±4 % decrease in stroke work, i.e. from 4±0.37 to 3.2±0.35 mJ g⁻¹. At this concentration, SNP decreased heart rate by 13±2.8 % (P<0.025; N=5) (data not shown). In unpaced hearts, the SIN-1 dose–response curve (Fig. 7B) showed that VS and WS decreased significantly at all concentrations used without any significant variation in fH (data not shown). The percentage changes with respect to control values were from 15±1.8 % (at 5×10⁻⁸ mol l⁻¹) to 26±3 % (at 10⁻⁵ mol l⁻¹) for VS and from 16±1.2 % to 30±6 % for WS at the same concentrations.

In paced preparations, a single administration of 10⁻⁴ mol l⁻¹ SNP induced a decrease of stroke volume from 2.0±0.07 to 1.6±0.10 ml kg⁻¹ (22±3.2 %) and of stroke work from 4.7±0.27 to 3.6±0.5 mJ g⁻¹ (23±4.8 %). The depressant effect of SNP was still present, but markedly attenuated, after prior treatment with Triton X-100 or in the presence of 10⁻⁴ mol l⁻¹ L-NAME (Fig. 8). Table 1 illustrates our findings on the duration of the pressure transient or of systole (in ms) measured, as shown in Fig. 3, from the pressure traces. The negative inotropic effects of SIN-1 and 8-bromo-cGMP were accompanied by a reduction in the duration of systole, while L-NAME and MB prolonged systole. Similarly, Triton X-100 also had the effect of prolonging systole.

In the present work, we have carried out a morphological characterization of the avascular heart of the frog, with particular reference to cNOS distribution. We have also made a functional evaluation of basal NO production and of its myocardial effects in relation to the specific contribution of the EE. In frog heart, the ventricle covered by the EE is highly trabeculated, resulting in a much higher endothelial surface to myocardial volume ratio than in the compact ventricular wall of higher vertebrates. Moreover, only a thin layer of extracellular material separates the EE cells from the underlying myocytes (Staley and Benson, 1968). As in mammals, the EE cells are characterized by a largely peripheral distribution of F-actin, with vWF localized around the nucleus (Brutsaert and Andries, 1992).

Previously, cNOS has been detected in the endocardial and coronary vascular endothelium of rat hearts using the NADPH-diaphorase histochemical technique (Ursell and Mayes, 1993) and by immunohistochemistry (Andries et al. 1996). Nitric oxide production by the EE has been demonstrated in cultured EE cells and in isolated cardiac valves of mammalian species (Mohan et al. 1996). We believe the current work to be the first demonstration of the presence of cNOS in the EE of the frog heart. Given the very close proximity of the EE and underlying myocytes, cNOS activity in the EE could exert a direct effect on myocardial contractility.

Studies with isolated and perfused mammalian heart preparations have shown that cNOS is associated with the continuous generation of NO at a relatively modest basal level (nanomolar concentrations). Although this basal NO production is considered critical for the preservation of cardiac and vascular homeostasis (Moncada et al. 1991), there is no consensus as to the effect of physiological concentrations of NO on heart rate and myocardial contractility in the whole heart. While earlier studies suggested that the inotropic action of NO donors (e.g. SNP) or NOS inhibitors (e.g. L-NMMA) in isolated hearts was secondary to the effects of coronary perfusion (Amrani et al. 1992; Fort and Lewis, 1991), the possibility that these contractile effects could in part be due to direct NO diffusion into the myocardium was not excluded. The inotropic action of increased NO levels has been variously reported to be negative (Balligand et al .1993; Meulemans et al. 1988; Grocott-Mason et al. 1994; Paulus et al. 1994), positive (Mohan et al. 1996; Kojda et al. 1996) or absent (Weyrich et al. 1994). These conflicting results may be at least partly explained by the different levels of performance of mammalian hearts, i.e. isolated unloaded myocyte, papillary muscle or intact heart (in vitro or in vivo), in association with various levels of sympathetic or parasympathetic tone, myocardial cGMP content (Mohan et al. 1996) and interactions between several endothelium-derived mediators such as prostaglandins (Mohan et al. 1995).

Here, we demonstrate that the NO donor SNP exerts a negative inotropic effect in a poikilotherm working heart which is paralleled by a positive inotropic effect obtained when NOS activity is inhibited by L-NAME. These inotropic effects are coupled to changes in the duration of the pressure transient, or systolic duration, that correspond to the finding in mammals that cardiac endothelium and the NO–cGMP pathway affect myocardial performance mainly by influencing twitch duration, by shifting the time of onset of relaxation (Brutsaert et al. 1988; Smith et al. 1991). In frog ventricular myocytes, NO donors, in particular SIN-1, exert a concentration-dependent regulation of cardiac Ca²⁺ current, I_Ca, controlled by cGMP, and thus linked to the activation of guanylyl cyclase (Mèry et al. 1993). The negative inotropic effect of micromolar concentrations of SNP and of the more specific NOS inhibitor SIN-1 may indeed result from inhibition of I_Ca by activation of the cGMP-stimulated cyclic AMP phosphodiesterase. Our finding of a negative inotropic effect of the cGMP analogue 8-bromo-cGMP confirms previous data obtained in isolated frog ventricular trabeculae (Singh and Flitney, 1981).

The positive inotropic effect resulting from inhibition of guanylate cyclase by MB may be due to a reduction of cGMP concentration in the preparation. Taken together, these effects demonstrate that, in the isolated frog heart, there is both basal release of NO and basal guanylate cyclase activity. No cNOS staining was observed in the myocardium, but we cannot exclude the presence of cNOS in the epicardium which, however, has a much smaller volume ratio than the EE. Neural NOS, found in nerve cells in the frog heart, occurs mainly in the atria; it is not found in the ventricle (Clark et al. 1994). Thus, morphological evidence suggests that the EE cell is the major cNOS-expressing cell type in the frog heart. Consequently, in basal conditions, the EE of the frog heart may produce NO in amounts sufficient to exert a negative inotropic effect on the underlying myocardium. We used a lower concentration of Triton X-100 than those used in mammalian preparations (Andries et al. 1991), because treatment with 0.5 % Triton resulted in ventricular asystole. Treatment with 0.05 % Triton, as in Li et al. (1993), affected only the ventricular luminal surface and did not influence the mechanical performance of the atrial myocardium or its endothelial secretory function. However, since substances added to the normal Ringer’s solution superfused the entire heart, they might have modulated ventricular performance indirectly through their effect on the mechanical or secretory activity of the atrium or atrial endothelium.

Exposure of the ventricular luminal surface to Triton X-100 did not result in any morphological change in either endothelium or myocardium, but caused a positive inotropic effect and resulted in the loss of response to L-NAME. Thus, exposure to Triton X-100 depressed basal release of NO from the EE. In line with the abolition of vascular endothelium-dependent relaxation induced by Triton X-100 in isolated rat hearts despite the continuing presence of NOS enzyme activity (Panda et al. 1996), this depressant effect may be attributable to an interruption of the signal transduction pathway that normally activates cNOS in the EE. The loss of cNOS enzyme activity in the EE after exposure to Triton X-100 could be related to changes in intracellular [Ca²⁺]. An alternative hypothesis is that Triton X-100 could impair the sensing surface of the EE, with a subsequent decrease of stimulation as a consequence of shear strain. Interesting in this context is the report of the selective disappearance of single proteins (e.g. the cholinergic receptors) in isolated preparations and in cell cultures (Marchenko and Saga, 1994). In addition, muscarinic receptors are known to be highly sensitive to their microenvironment, as shown by the loss of antagonist selectivity after detergent solubilization (Hulme et al. 1990). Indeed, in the same frog heart preparation, Triton X-100 treatment abolished the inotropic responses to exogenous acetylcholine (Gattuso et al. 1995). The frog heart is particularly sensitive to mechanical stresses that induce the release of several factors that have a marked effect on cardiac performance (Singh and Flitney, 1981; Tung and Zou, 1995).

In isolated rat right atria, N-nitro-L-arginine (10⁻⁴ mol l⁻¹) and its methyl ester derivative (10⁻³ mol l⁻¹) did not exert any chronotropic effect, while the NO donor SIN-1 had a bradycardic effect only at concentrations as high as 10⁻³ mol l⁻¹ (Kennedy et al. 1994). In cultured rat cardiomyocytes, the bradycardic effects of carbachol (Balligand et al. 1993) and of interleukin-1 beta (Roberts et al. 1992) were inhibited by MB and L-NMMA (N^G-monomethyl L-arginine). We show that, while neither MB nor 8-bromo-cGMP significantly influenced isolated frog heart rate, the NO donor SNP, at relatively high concentrations (10⁻⁴ mol l⁻¹), exerted a negative action on cardiac automatism. Intriguingly, Triton X-100 and L-NAME reduced heart rate. A search of the literature failed to provide data that could explain these effects, which appear to contrast with the positive inotropic actions exerted by both the detergent and the NOS inhibitor.

In conclusion, our results suggest that the EE in working frog heart in vitro produces NO in amounts sufficient to modulate ventricular performance. Furthermore, the presence of vWF and F-actin in the frog EE suggests that they could also be involved in haemostasis and in transendothelial permeability. The role of the EE as a source of NO and as a cardiac mechanosensor of blood-flow-related forces is still obscure. The different types of ventricular myoarchitecture (compact versus trabeculated) and the blood supply (luminal versus vascular endothelium) in vertebrates complicates the study of EE control of myocardial performance. The avascular heart of the frog, isolated and working at physiological loads, is a promising tool with which to explore the specific role of the EE, free from the effects of the vascular endothelium.

The authors acknowledge crucial methodological input from the two anonymous reviewers to this paper. This work was supported by a grant from Programma Nazionale Ricerche in Antartide (PNRA) 1993–1995, awarded to B.T.