Morphogenetic adaptation of the looping embryonic heart to altered mechanical loads
Abstract
The biophysical mechanisms that drive and regulate cardiac looping are not well understood, but mechanical forces likely play a central role. Previous studies have shown that cardiac torsion, which defines left–right directionality, is caused largely by forces exerted on the heart tube by a membrane called the splanchnopleure (SPL). Here we show that, when the SPL is removed from the embryonic chick heart, torsion is initially suppressed. Several hours later, however, normal torsion is restored. This delayed torsion coincides with increased myocardial stiffness, especially on the right side of the heart. Exposure to the myosin inhibitor Y-27632 suppressed both responses, suggesting that the delayed torsion is caused by an abnormal cytoskeletal contraction. This hypothesis is supported further by computational modeling. These results suggest that the looping embryonic heart has the ability to adapt to changes in the mechanical environment, which may play a regulatory role during morphogenesis. Developmental Dynamics 235:1822–1829, 2006. © 2006 Wiley-Liss, Inc.
INTRODUCTION
Cardiac looping is one of the first visible morphologic breaks in left–right symmetry in the vertebrate embryo. During looping, the initially straight heart tube deforms rightward into a curved tube, creating the basic pattern of the mature heart. Looping abnormalities often lead to congenital heart defects (Srivastava and Olson, 1997; Harvey, 1998), and considerable progress has been made recently in identifying the genetic and molecular pathways that regulate looping (Brand, 2003; Linask, 2003). However, despite nearly a century of dedicated research (Patten, 1922), the biophysical mechanisms that translate these signals into physical form still are poorly understood (Männer, 2000; Taber, 2005).
It long has been known that mechanical forces play an important role in morphogenesis of various tissues and organs, including the heart (Nakamura et al., 1980; Taber, 1995, 2001). Assimilating a wide range of experimental evidence, the developmental biologist L.V. Beloussov has postulated that embryonic tissues respond actively to mechanical stimuli (Beloussov, 1998). According to his hyperrestoration hypothesis, perturbations in mechanical stress induce an active response that changes tissue shape and causes a new stress perturbation, which elicits a new response, and so on, until the proper form is created. He has shown that this idea can explain several observations for embryos undergoing various morphogenetic events, including cleavage, gastrulation, and neurulation (Beloussov, 1998). Whether mechanosensitive processes are involved in cardiac looping has not yet been investigated.
In this study, we illustrate the ability of the embryonic heart to sense and respond to mechanical stimuli and explore the role of this behavior during the first phase of looping, called c-looping. During c-looping, the heart tube undergoes ventral bending and dextral torsion to create a c-shaped tube, with the outer (convex) curvature normally directed toward the right side of the embryo (Männer, 2000). Previous work has shown that bending is driven primarily by morphogenetic forces that arise within the heart tube (Butler, 1952; Manning and McLachlan, 1990; Latacha et al., 2005), but torsion is caused by loads exerted on the heart by neighboring tissues (Voronov et al., 2004). The present study illustrates the inadequacy of this dichotomous perspective, revealing a causal relationship between forces acting on the heart and those developed within the heart.
In particular, we investigated the response of the embryonic chick heart when compressive loads due to the splanchnopleure (SPL, see Fig. 1) are removed near the onset of looping. Our results suggest that, after several hours without twisting, the heart generates active contractile forces, stronger on the right side, that restored normal rightward looping. This behavior indicates that the embryonic heart has the ability to respond and adapt to perturbations in normal morphogenetic mechanisms. Such built-in redundancy is crucial for minimizing congenital heart defects.
RESULTS AND DISCUSSION
This study examined the response of the embryonic heart to altered mechanical loads during the morphogenetic process of c-looping. In the chick, c-looping occurs between stages 10 and 12 of Hamburger and Hamilton (1951), or approximately 35–48 hr of a 21-day incubation period. During this time, the primitive heart tube consists of three layers: an outer, two-cell-thick layer of myocardium containing developing cardiomyocytes; a relatively thick layer of extracellular matrix called cardiac jelly; and a one-cell-thick, inner layer of endocardium (Fig. 1). The heart is constrained by the omphalomesenteric veins (OVs) at its caudal end, the conotruncus at its cranial end, the SPL on its ventral surface, and the dorsal mesocardium (DM) along its dorsal extent (Figs. 1, 2A). All of these constraints apply mechanical loads that likely influence looping morphology (Manasek, 1983; Voronov and Taber, 2002; Voronov et al., 2004; Ramasubramanian et al., 2005).
The Heart Stiffens in Response to Reduced Loading
To alter the mechanical environment of the looping heart, the SPL was removed from stage 10 hearts by means of dissection. This procedure removed significant compressive radial and longitudinal forces on the heart tube (Voronov and Taber, 2002), immediately resulting in elongation and a decrease in diameter (Fig. 2A,B). When cultured through c-looping (10–12 hr), SPL-lacking hearts appeared similar in morphology to normal stage 12 embryos (see Fig. 4C,F), with regular heartbeat and vascular bed formation in the surrounding tissue. Although the SPL healed during culture, the hole remained sufficiently large that the heart remained ventrally exposed. Culturing beyond 12 hr, however, resulted in full closure of the wound.
Regional mechanical properties of the looping heart were measured using microindentation (Zamir et al., 2003). Characteristic stiffness k was defined as the slope of the force–displacement curve at an indentation depth of 30 μm (see the Experimental Procedures section). As shown by Zamir et al. (2003), myocardial stiffness is nearly an order of magnitude greater than that of the underlying cardiac jelly. Moreover, the deformation is confined to a relatively small region near the indenter. These results indicate that regional heart stiffness does not depend on gross geometry of the heart and is dominated by the material properties of the myocardium.
Hearts cultured from stage 10 to 12 without the SPL were nearly four times stiffer (k = 0.96 ± 0.38 mdyn/μm) than embryos cultured to stage 12 with SPL intact (k = 0.25 ± 0.12 mdyn/μm; P < 0.001, Fig. 3A). As shown below, increased heart stiffness was detected as early as 4 hr after SPL dissection; however, the 10- to 12-hr incubation period was chosen to couple this response with normal c-looping.
Notably, the stiffness of hearts cultured without the SPL was similar to that of hearts cultured in isolation with all external loads removed (Rémond et al., 2006), whereas hearts from embryos cultured with SPL intact were similar in stiffness to those taken from embryos incubated to stage 12 in ovo (Zamir et al., 2003; Fig. 3A). These results suggest the following: (1) removing the SPL induces a significant increase in heart stiffness, and (2) the culture method with SPL intact mimics the in ovo environment sufficiently, producing no measurable biomechanical artifacts. We conclude, therefore, that the increase in heart stiffness was caused by SPL removal and not culturing.
To confirm that the response to SPL removal was due to mechanical factors and not other changes, such as a loss of molecular signals, we replaced the SPL forces with forces due to surface tension. In these experiments, the SPL was removed and the embryos were cultured using a method that exposes the ventral surface of the embryo directly to 5% CO2 atmosphere (see the Experimental Procedures section). Surface tension acutely increased heart diameter by an amount similar to that normally caused by compressive loads exerted by the SPL (Fig. 2A–C), suggesting a comparable magnitude of compression.
When stage 11 embryos were cultured for 4 hr under surface tension, SPL removal induced a small but not statistically significant stiffening response (k = 0.39 ± 0.09 mdyn/μm; Fig. 3B). On the other hand, when a thin layer of liquid medium was placed over similarly prepared stage 11 embryos to eliminate surface tension, 4 hr of culturing increased stiffness significantly in SPL-lacking hearts (k = 1.03 ± 0.38 mdyn/μm; P = 0.006; Fig. 3B). Hence, removing the SPL did not result in measurable stiffening under conditions where the heart remained ventrally compressed. These results suggest that the stiffening response is induced by changes in external loading and not simply by loss of the SPL.
Finally, we note that it is unlikely that any possible loss of fluid pressure in the celoemic cavity surrounding the heart was responsible for the observed response, as surface tension does not restore hydrostatic pressure lost with SPL removal. Also, the slightly increased stiffness of SPL-lacking hearts exposed to surface tension may be because surface tension replaces radial forces, but not longitudinal forces, that normally are exerted by the SPL.
Increased Heart Stiffness Is Caused by Cytoskeletal Contraction
Accepting that the heart senses and responds to mechanical stimuli, we next sought to determine the source of the observed fourfold increase in heart stiffness. Because contractility is associated with increased mechanical stiffness (Wakatsuki et al., 2000), we speculated that the stiffening was caused by an active contractile response in the myocardium.
To examine this possibility, cytoskeletal contraction due to nonmuscle myosin II was inhibited using the Rho kinase inhibitor Y-27632 (Narumiya et al., 2000). In a recent study, we examined the effects on looping of several myosin inhibitors (BDM, ML-7, Y-27632, and blebbistatin). Of these, Y-27632 proved most effective in inhibiting cytoskeletal contraction (Rémond et al., 2006). Embryos were cultured from stage 10 to stage 12 without the SPL, and microindentation was performed to confirm an increase in stiffness (Fig. 3C). Next, 12.5 μM Y-27632 was added to the testing medium, and microindentation was repeated after 45-min exposure. The same was done for embryos cultured with the SPL left intact. In SPL-lacking hearts exposed to Y-27632, stiffness decreased to the level of in ovo hearts (Fig. 3C). However, no measurable change in stiffness was found with treatment in hearts cultured with SPL intact. The ratio of stiffnesses before and after exposure to Y-27632 was 0.99 ± 0.48 for embryos cultured with the SPL intact and3.14 ± 0.31 for embryos cultured without SPL (Fig. 3C). This difference was statistically significant (P < 0.001).
These results suggest that (1) the heart tube undergoes little or no cytoskeletal contraction during normal c-looping, and (2) the myocardium actively contracts in response to decreased mechanical loading. The first observation is consistent with recent evidence showing that c-looping does not require contraction (Rémond et al., 2006). Recently, Latacha et al. (2005) suggested that actin polymerization drives the bending component of c-looping.
The second observation has more complex implications. In the normal embryo, the SPL compresses the heart in the ventral–dorsal direction, causing the essentially incompressible cardiac jelly to expand laterally. This deformation stretches the myocardium, adding to the tension normally caused by hydration pressure exerted outward by the glycosaminoglycan-rich cardiac jelly. This additional tension would be greatly diminished when the SPL is removed. If the heart tube functions according to Beloussov's hyperrestoration hypothesis (Beloussov, 1998), then the observed contraction may be an adaptive response that restores (and possibly overshoots) normal myocardial tension.
Our experiments with Y-27632 suggest that the observed response is mediated by Rho kinase, which has been implicated recently in multiple cellular responses to mechanical perturbation, including cytoskeletal reorganization (Kaunas et al., 2005), stem cell differentiation (McBeath et al., 2004), and growth and proliferation (Nelson et al., 2005). It is believed that transduction of mechanical stimuli to the cytoskeleton occurs through integrin signaling and that this process is responsible for the reported Rho kinase activation. Our results confirm an important role for Rho kinase as a downstream target in mechanotransduction. However, further studies are necessary to identify the upstream agents in the observed contractile response of the embryonic heart and ultimately to determine the mechanosensitive receptor from which this response originates.
Splanchnopleure Removal Leads to Delayed Cardiac Torsion
During the course of the above studies, we observed a puzzling phenomenon. When the SPL over the heart was removed, the heart did not rotate for up to 6 hr, as indicated by myocardial labels that remained along the ventral midline of the heart tube (Fig. 4D,E). This result is consistent with the study of Voronov and Taber (2002). On continued culturing, however, the heart began to twist, eventually reaching its normal configuration with the labels located along the outer curvature (Fig. 4F). Moreover, when the SPL was removed from partially looped stage 11 hearts, the hearts untwisted within minutes and resumed normal torsion up to 12 hr later. This behavior was unexpected, as Voronov and colleagues have shown that the SPL plays an important role in the torsional component of c-looping (Voronov and Taber, 2002; Voronov et al., 2004). These authors, however, followed looping for only 6 hr after SPL removal at stage 10.
Asymmetric Cytoskeletal Contraction Drives Delayed Cardiac Torsion
The delayed cardiac torsion occurred in roughly the same time frame as the contractile response after SPL removal. These results led us to wonder whether these two findings were related. In other words, does the contractile response to mechanical cues serve a useful morphogenetic function, rather than just restoring normal tissue stress?
To address this question, cytoskeletal contraction was blocked by culturing stage 10 embryos in medium containing 40 μM Y-27632. (It is notable that 40 μM Y-27632 was needed to inhibit contractility in whole embryos, whereas only 12.5 μM was required to reduce stiffness during microindentation of isolated hearts. As shown by Latacha et al. [ 2005], the presence of extracardiac tissue requires a higher drug concentration to obtain the effects seen in isolated hearts. We have also confirmed that cytoskeletal architecture remains unaffected at this concentration [Rémond et al., 2006]). Torsion was visualized through the motions of fluorescent labels in the myocardium. In embryos cultured with the SPL intact, looping was unaffected by exposure to the drug (Fig. 5A–C), corroborating recent evidence from our laboratory that nonmuscle myosin II is not necessary for normal c-looping (Rémond et al., 2006). In hearts where the SPL was removed, however, exposure to Y-27632 resulted in ventral (and some rightward) bending, but little or no rotation (Fig. 5D–F). Hence, contraction is required for the torsional component of c-looping only when the normal driving forces supplied by the SPL are perturbed. These observations suggest the existence of an adaptive contraction-based mechanism for cardiac torsion in the absence of an SPL. This built-in redundancy ensures that errors in cardiac looping are minimized, so that heart malformation is unlikely.
Finally, we investigated the specific mechanism by which cytoskeletal contraction produces the delayed rightward torsion. One possibility, as suggested by Itasaki et al. (1991) for normal looping (with intact SPL), is that contraction (in response to loss of wall stress) is stronger along circumferential fibers on the right side of the heart tube than on the left. The net force pulls the heart rightward, and it is forced to rotate by the DM, which acts as a pivot point. If indeed the contraction in the heart tube is asymmetric, microindentation should reveal regional variations in stiffness, with the right side of heart tube stiffer than the left side.
To explore this possibility, microindentation was performed 4 hr after SPL removal and culturing from stage 10, near the onset of the delayed torsion. In these hearts, the right lateral side was found to be stiffer than the left side (kright/kleft = 2.44 ± 1.59; Fig. 3D). In contrast, consistent with the stage 12 data of Zamir et al. (2003), no significant difference was seen between the two sides when the SPL was left uncompromised during the 4 hr of culture before indentation (kright/kleft = 1.09 ± 0.53). Supporting our hypothesis, these data suggest that cytoskeletal contraction was stronger on the right side than on the left side only in SPL-lacking hearts (P = 0.039).
To assess the physical feasibility of asymmetric contraction as a cause of cardiac torsion/rotation, we created a plane strain computational model for a cross-section of the heart tube. The model consists of layers representing the myocardium and cardiac jelly, and the DM was fixed to a rigid constraint representing the foregut (Fig. 6A). The finite element method was used to analyze the model, which consisted of 1,420 second-order quadrilateral elements. When circumferential contraction was simulated in the right half of the myocardium and DM, considerable rightward rotation occurred (Fig. 6B). It is important to note that Itasaki et al. (1991) originally suggested this mechanism for torsion during early c-looping, but later experiments showed the OVs and SPL are primarily responsible for early torsion (Voronov et al., 2004). Here, we suggest that such a mechanism may in fact occur, but only if normal SPL loading is perturbed.
It is unclear how a relatively symmetric change in loading can induce an asymmetric mechanical response. This mechanical asymmetry is likely indicative of a more subtle biological asymmetry. Left–right asymmetries in gene expression have been widely documented in the embryonic heart (Srivastava and Olson, 1997; Harvey, 1998; Mercola and Levin, 2001; Brand, 2003; Linask, 2003), but the ability of mechanical cues to stimulate asymmetry has not previously been shown.
Crucial Contraction Occurs Near the Ends of the Heart Tube
Finally, we note that, by the time late-onset torsion occurred in our experiments, the DM had ruptured everywhere except near the ends of the heart tube as is normal during development (Romanoff, 1960). This observation suggests that the important torsion-generating contraction originates near the ends of the tube, i.e., in or near the OVs or conotruncus.
To examine this idea, we removed the SPL and one or both OVs, and cultured the embryos from stage 10 to 12. Dissection of the left OV resulted in a reversal of looping direction (leftward looping) in 71% of the embryos (Fig. 7A,A′). However, when only the right OV was severed, or when both veins were severed, normal (rightward) looping was observed in all embryos (Fig. 7B,B′,C,C′). Statistical significance was confirmed by χ2 test (P < 0.025). These results support previous notions that the OVs play a key role in establishing left–right directionality, with both OVs tending to push the heart to the opposite side (Voronov et al., 2004). In addition, we speculate that asymmetric contraction of the conotruncus ensures rightward looping when both veins are absent. This latter hypothesis is supported by microindentation data for the conotruncus, which revealed a similar left–right asymmetry in stiffness for SPL-lacking hearts (results not shown). Hence, the contractile response discovered in the heart tube appears to extend cranially into the conotruncus. Interestingly, Männer (2004) has demonstrated recently that rotating the ends of an elastic tube produces morphology remarkably similar to that of the looping heart.
In summary, the results of this study suggest that the early embryonic heart recognizes and responds to changes in its mechanical environment. The ability to respond to mechanical stimuli appears critical for proper c-looping under abnormal loading conditions and highlights the importance of mechanical feedback in early cardiac development. Moreover, previous studies of c-looping had suggested that the forces that drive bending are generated within the heart, whereas external forces are responsible for torsion. Our new findings suggest a coupling between these two types of force, i.e., removing external loads induces internal contractile forces. With recent discoveries in “mechanotranscription” (mechanical regulation of gene expression) during early development (Farge, 2003; Brouzes and Farge, 2004), it is clear that a truly comprehensive theory of morphogenesis will require supplementing traditional approaches of experimental biology with biomechanical testing and modeling.
EXPERIMENTAL PROCEDURES
Embryo Culture
Fertilized White Leghorn chicken eggs (Sunrise Farms, Catskill, NY) were incubated in a humidified atmosphere at 38°C for 35 to 40 hr, yielding embryos of stage 10 and 11 according to the system of Hamburger and Hamilton (1951).
The technique used to harvest and culture embryos has been described previously (Voronov and Taber, 2002). Unless stated otherwise, embryos were placed in a liquid medium (89% Dulbecco's Modified Eagle's Media [Sigma], 10% chick serum [Sigma] and 1% penicillin/streptomycin/neomycin [PSN, Invitrogen]) and incubated in an atmosphere containing 95% O2 and 5% CO2. Embryos were submerged entirely to eliminate the effects of surface tension, which can influence looping (Voronov and Taber, 2002).
A second culture method was used to exert surface tension forces on the embryo, as described by Voronov and Taber (2002). To do so, embryos were planted onto 5 ml of semisolid agar medium (warm egg albumin, 0.3% agar [Sigma] in Ringer's Solution, and 1% PSN [Sigma]) after extraction from the egg. These embryos were incubated at 37.5°C with the surface exposed to an atmosphere containing 5% CO2.
Microindentation
Mechanical stiffness was measured using a custom-built microindentation device consisting of a piezoelectric motor and a flexible glass beam. The motor moves the beam horizontally, and a tip at end of the beam contacts the heart, deforming the tissue and deflecting the beam. Beam deflection and indentation depth were determined from the motion of the tip, which was measured from captured video. Then, with indentation force computed from beam deflection and the measured beam stiffness, force–displacement curves were constructed. Heart stiffness was given by the slope of a nonlinear regression fit to the force–displacement curve at an indentation depth of 30 μm (unless stated otherwise). See Zamir et al. (2003) for more details.
Hearts were indented in phosphate buffered saline at 38°C, with 4 × 10−4 mM verapamil (Sigma) added to arrest the hearts in diastole. Agreement of control data with measurements of Zamir et al. (2003) revealed that neither temperature (between 25°C and 38°C) nor verapamil affected the passive heart stiffness. Unless stated otherwise, indentation tests were performed at the outer curvature of hearts at stage 12, dissected free from the embryo by severing the omphalomesenteric veins and conotruncus. For experiments where left and right sides were indented, hearts at stage 10+/11− were tested in the intact embryo with the SPL removed to gain access.
Labeling and Imaging
Tissue labels were used to visualize rotation of the heart tube. As the heart rotates, myocardial labels displace from the ventral midline at stage 10 (see Fig. 4A) toward the outer curvature (Fig. 4B,C). To create labels, the fluorescent dye 1,1′, di-octa- decyl-3,3,3′,3′,-tetramethylindo-car- bocyanine perchlorate (DiI; Molecular Probes, D-282) was injected into small groups of cells in the myocardium, as described in Voronov et al. (2004). Dye was propelled through a fine pulled-glass micropipette by a pneumatic pump (PicoPump PV830, World Precision Instruments). Fluorescent and brightfield images were taken with a video camera (Retiga 1300) installed on a fluorescent microscope (Leica DMLB). Additional brightfield images were obtained by CCD camera (COHU, Model 4915) mounted on a dissecting microscope (Leica MZ8). Images were processed using Adobe Photoshop and ImageJ programs. For injection without removing the SPL, small holes were made in the SPL to access the heart. These holes healed rapidly, before any additional interventions.
Statistics
All data are reported as mean ± SD. Statistical analysis was performed using SigmaStat software (SPSS Science). Stiffness values were compared by t-test.
Computational Modeling
Nonlinear finite element modeling was done using the commercial software ABAQUS. Myocardium and cardiac jelly were assumed to be isotropic, pseudoelastic, and incompressible, with material properties defined by an exponential strain–energy density function taken from Zamir and Taber (2004). Cytoskeletal contraction was included by (1) increasing the active modulus by threefold (based on stiffness changes measured by microindentation) and (2) by altering the zero-stress configuration (50% shortening) of each material element (by means of the user subroutine UMAT), as described in Voronov et al. (2004).
Acknowledgements
The authors thank Evan Zamir and Kim Latacha for their help with the microindentation device and histology (Fig. 1), respectively, as well as Mathieu Rémond for providing force–displacement data for an isolated heart (Fig. 3A). L.A.T. and A.R. were funded by NIH grants.