High-frequency ultrasonographic imaging of avian cardiovascular development

Authors

  • Tim C. McQuinn,

    Corresponding author
    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    2. Pediatric Cardiology, Medical University of South Carolina, Charleston, South Carolina
    • Department of Cell Biology and Anatomy, Medical University of South Carolina, 173 Ashley Avenue, BSB 601, Charleston, SC 29425
    Search for more papers by this author
  • Momka Bratoeva,

    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Angela DeAlmeida,

    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Mathieu Remond,

    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Robert P. Thompson,

    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • David Sedmera

    1. Department of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, South Carolina
    2. Institute of Anatomy, First Faculty of Medicine, Charles University, Prague, Czech Republic
    3. Institute of Animal Physiology and Genetics, Academy of Sciences of the Czech Republic, Prague, Czech Republic
    Search for more papers by this author

Abstract

The chick embryo has long been a favorite model system for morphologic and physiologic studies of the developing heart, largely because of its easy visualization and amenability to experimental manipulations. However, this advantage is diminished after 5 days of incubation, when rapidly growing chorioallantoic membranes reduce visibility of the embryo. Using high-frequency ultrasound, we show that chick embryonic cardiovascular structures can be readily visualized throughout the period of Stages 9–39. At most stages of development, a simple ex ovo culture technique provided the best imaging opportunities. We have measured cardiac and vascular structures, blood flow velocities, and calculated ventricular volumes as early as Stage 11 with values comparable to those previously obtained using video microscopy. The endocardial and myocardial layers of the pre-septated heart are readily seen as well as the acellular layer of the cardiac jelly. Ventricular inflow in the pre-septated heart is biphasic, just as in the mature heart, and is converted to a monophasic (outflow) wave by ventricular contraction. Although blood has soft-tissue density at the ultrasound resolutions and developmental stages examined, its movement allowed easy discrimination of perfused vascular structures throughout the embryo. The utility of such imaging was demonstrated by documenting changes in blood flow patterns after experimental conotruncal banding. Developmental Dynamics 236:3503–3513, 2007. © 2007 Wiley-Liss, Inc.

INTRODUCTION

The chick embryo is a favorite model system for morphologic and physiologic study of the developing heart (Antin et al., 2004; Stern, 2005), with advantages that include low cost, easy visualization, and amenability to experimental manipulations (Bartman and Hove, 2005). Clark et al. (1986), using videomicroscopy and Doppler dorsal aortic velocimetry, were the first to document key parameters of the chick embryo circulation between Stages 18–29. However, videomicroscopic analysis of the chick embryo circulation is partially lost after 5 days of incubation, when the visualization of the heart becomes compromised due to the rapid growth of the chorioallantoic membrane.

Conventional optical observation provides few clues about internal cardiac structures, although transparency of the early embryo allows visualization of the outflow tract cushions and India ink injection can show intracardiac blood stream patterns (Hogers et al., 1995, 1997). Optical coherence tomography can reportedly overcome this obstacle (Yelbuz et al., 2002), but this technique is currently not widely available and presently requires gated imaging, which is very difficult to achieve in embryos except through pacing. MicroCT is of limited use in embryos in vivo because of lack of tissue contrast (Johnson et al., 2006; Butcher et al., 2007). Magnetic resonance imaging of avian embryos has been described (Hogers et al., 2001; Yelbuz et al., 2003; Zhang et al., 2003) but the equipment is highly specialized and the temporal resolution is poor for in vivo real-time imaging.

Ultrasound is a very popular non-invasive functional imaging modality in cardiology. Recent advances in ultrasound equipment design for small animal imaging overcome many of the spatial and temporal resolution barriers that prevented use for imaging embryonic and postnatal heart development (Phoon et al., 2004). High-frequency ultrasound is an investigational tool with spatial resolution of structures as small as 30 μm at sweep speeds of up to 60 Hz for 2D imaging or 1 kHz for M-mode imaging. In combination with pulsed Doppler characterization of blood flow velocity, this technology provides a highly informative platform for studying embryonic cardiovascular physiology and morphology in vivo. Multiple studies demonstrate the strengths and limitations of this technology in investigation of developmental and postnatal cardiovascular morphology and physiology in mice (Zhou et al., 2002). These studies have demonstrated informative and reproducible applicability of Doppler blood flow measurement, in particular for characterization of normal and abnormal mouse embryo cardiovascular physiology, with increasing success in characterizing in utero cardiac morphology (reviewed in Phoon, 2006).

Our goal was to determine the applicability of ultrasound biomicroscopy to the study of avian cardiac development. We have focused on the period of heart development preceding ventricular septation and compared the results with direct videomicroscopy and histological examination. The spatial (30 μm) and temporal (up to 60 frames per second in B-mode, 1 kHz in M-mode) resolution was sufficient to clearly resolve crucial morphological features of the rapidly beating, developing heart in ex ovo explant culture. Properly used, ultrasonography of avian embryos expands the array of in vivo imaging modalities during development.

RESULTS

Optimization of Imaging Conditions

We used a Vevo 660 ultrasound biomicroscope and RM708 scanhead (VisualSonics, Inc., Toronto, Canada) to study chick embryonic development. The scanning wavelength (55 MHz center frequency) and scanning frequency (up to 60 Hz) result in theoretical axial resolution of 30 μm and lateral resolution of 75 μm (Zhou et al., 2002). In ovo imaging was somewhat limited by the fairly large probe footprint (Fig. 1) and short fixed focal distance (4.5 mm) with potential compression of the chorioallantoic vasculature at later stages of development. In addition, there was a severely limited range of interrogation angles for Doppler imaging due to the egg shell. These limitations were largely overcome by using an ex ovo culture setup (Dunn et al., 1981; Tuan, 1983; Ono and Tuan, 1986; Harris et al., 2006), which allowed much freer positioning of the scanhead (Fig. 1) and additionally resulted in position of the embryo closer to the scanhead, allowing imaging with minimal mechanical interference during later stages of development (deAlmeida et al., 2007). However, as embryos increase in mass and sink into the yolk, the short working distance of the scanhead used in our studies became a limitation. The following descriptions are intended to briefly characterize the echocardiographic features at particular stages of development.

Figure 1.

Setup for avian ultrasonographic imaging. A: In ovo imaging is possible but limited primarily by probe size relative to the egg. B,D: Using an ex ovo culture setup, unhindered access is obtained, and multiple scanning interrogation angles are possible. The generous footprint of the scanhead (Echo probe) with respect to the Stage 24 embryo can be seen. The sector viewable in the image is created by side-to-side motion of the wand, which can be seen in the center of the scanhead. The tip of the wand bears an ultrasound crystal, which emits and transduces ultrasonic energy for interrogation. The in-focus image field is 4.5 mm from the crystal tip at the midpoint of the membrane. Temperature control is provided by a water-jacketed incubator connected to circulating water bath in combination with heating lamp. Desiccation of the embryo and membranes is prevented by application of pre-warmed Hepes-buffered chick Tyrode's solution. C: Appearance of embryos cultured ex ovo at 4 days (Stage 24) and 6 days (Stage 29) of incubation. Note the chorioallantoic vasculature obscuring the view of the embryo at the later stage. Scale bar = 1 mm. D shows a schematic diagram of the ex ovo imaging setup. 1: the scanhead with the moving wand, 2: coupling with Tyrode's solution, 3: the embryo with its yolk, 4: the hexagonal weight boat, 5: warm water filling the Radnoti glass incubator (6), 7: tubing coupled to circulating water bath, 8: heat lamp.

Looping Tubular Stages (Stage 9–16)

Ultrasonographic 2D imaging showed the thin muscular layer of the primitive ventricle as an echogenic rim surrounding echo-free cardiac jelly. The cardiac jelly was bounded internally by a highly echogenic endothelial layer that surrounded the lumen of the heart. Distinct points of relative proximity between the endothelium and the myocardium were present, with much reduced cardiac jelly in those locations. An invariant region of close endothelial-myocardial approximation was found at the dorsal mesocardium, while the second was roughly opposite at the original attachment of the ventral mesocardium (Fig. 2). These two points determined the plane of endocardial apposition during ventricular contraction. Additional apparent clefts in the endocardial cushion otherwise uniformly lining the ventricular myocardium were sometimes observed in the distal ventricle and proximal truncus, probably corresponding to “flutes” in the cardiac jelly as described by Fitzharris and colleagues (Fitzharris et al., 1980; Fitzharris, 1981). Histological and whole-mount immunohistochemical examination showed clear delineation of endocardial and myocardial layers, with interposed cardiac jelly containing radially arranged strands (Nakamura and Manasek, 1978) of fibrillar protein (Fig. 2).

Figure 2.

Imaging of Stage-12 chick embryonic heart. A: Embryo in New culture (C-loop). Note prominent size of the heart in comparison to the rest of the embryo, and more advanced differentiation of the cranial structures (top) compared with the caudal end where somites are still being formed. Echocardiographic cross-sectional views (B, C) show the cardiovascular structures as well as the lumen of the neural tube. For more detail, see the Supplemental Movies 1 and 2. D: Higher magnification cross-sectional views from Stage 14 ventricle, illustrating the asymmetric contraction pattern. E: Immunohistochemical staining of ventricular cross-section of a Stage-12 tubular heart shows the myocardium in red (phalloidin staining). A cleft in the cardiac jelly (area of endocardial-myocardial apposition) is indicated by an arrowhead. The cardiac jelly is filled with radially oriented fibrillar protein (JB3 anti-fibrillin staining, green), with more abundant fibers near the outer myocardial mantle (anti-tenascin staining, blue). F: M-mode imaging resolves the wall movements at the ventricular level, demonstrating the time of complete occlusion of the lumen. Time scale, 1 sec (50-msec increments); dimension scale 0.5 mm (0.1-mm increments); CJ, cardiac jelly; DA, dorsal aorta (paired); DM, dorsal myocardium; endo, endocardium; myo, myocardium; NT, neural tube; V, ventricle. Scale bar = 50 μm (E), 100 μm smallest division on echocardiographic views.

The muscle-lined conotruncal segment between the right-sided heart tube and the noncontractile midline aortic sac was short. The cardiac jelly in this segment was indistinct. The axis of the flow channel was parallel to the atrioventricular inflow and the plane of the dorsal mesocardium. The aortic sac was readily visible near the heart but its boundaries are indistinct in the portion giving rise to the first branchial arch vessels. The dorsal aortae were readily visible within the relatively echogenic head mesenchyme, but became less distinct in the body mesenchyme posterior to the heart (see Supplemental Movie 1, which can be viewed at www.interscience.wiley. com/jpages1058-8388/suppmat).

Contraction was seen as a wave of complete apposition of the endocardial surfaces with sequential transient obliteration of the atrioventricular, ventricular, and outlet chamber flow channels from posterior to anterior. Ventricular contraction was asymmetrical; the cross-section of the fully contracted ventricular segment is oval, with the major axis corresponding to the plane from the inner curvature to the outer curvature of the tube (Fig. 2, Supplemental Movie 2). The non-circular nature of the cross-section of the ventricular lumen in the early embryonic heart has been presented previously (see Fig. 6; Taber et al., 1994).

M-mode recording confirmed that there was no period of ventricular wall stasis; the walls of the ventricle were always either moving apart or rapidly contracting (Fig. 2). There was a brief acceleration of the outward movement of the distal ventricular myocardium just prior to its contraction, which marked the transiently increased volume of that segment associated with the wave of blood moving through the heart. The accelerated increase in diastolic diameter at end diastole could represent the active atrial contraction and its contribution to ventricular filling (Campbell et al., 1992). However, recent observations in zebrafish heart suggest that linear heart tubes of this type may not be peristaltic pumps (blood propelled by a myocardial wave of apposition), but impedance pumps (Forouhar et al., 2006). An impedance pump results in net forward flow of blood by generation of unidirectionally moving boundaries of low pressure in the ventricular chamber. One feature of the zebrafish heart impedance pump supporting these conditions is the presence of a bidirectional myocardial wave in the heart tube. The increase in ventricular diameter observed at end diastole may, therefore, also be consistent with the impedance pump mechanics previously defined in zebrafish (see Supplemental Movie 2 and compare to Figure 1 in Forouhar et al., 2006).

Another key feature of an impedance pump is that the velocity of blood flow is greater than the myocardial contraction wave velocity. We found Doppler signals from Stage 9–12 hearts to be too weak to address this issue due in part to the paucity of reflective elements in the blood. Within the lumen of the heart and large vessels, only scattered circulating particles were visible (Supplemental Movie 2); these particles often displayed the echo appearance of being considerably larger than the red blood cells that will fill the vascular space subsequently and may represent yolk globules or “clumps” of circulating erythroblasts (Rychter et al., 1955). Tracking of these objects frame-by-frame provided the opportunity to estimate particle velocity in the dorsal aortae. The average particle movement per heart beat was 0.16 ± 0.05 mm (mean ± SD, n = 8), and the average velocity 0.64 ± 0.15 mm/s (n = 6), which is close to values reported for vessels of similar dimensions in the yolk sac of older mouse embryos (McGrath et al., 2003). From these values, we determined that each stroke moved a volume approximately 2.5 × 10-3 mm3, within the range of previously reported filling volumes of 4.2 × 10-3 mm3 for a stage-12 chick embryonic ventricle (Hu et al., 1991) and stroke volumes of 0.01 mm3 (Hu and Clark, 1989). The blood flow appeared mostly laminar, with some eddies behind closing cushions.

Period of Chamber Differentiation (Stages 17–24)

The cardiac jelly was easily identified and remained largely echo-free in the atrioventricular canal and outflow tract, but was much reduced or absent in major portions of the ventricular segments. Blood was highly echogenic, and the net result was poor discrimination between the blood, endocardium, and myocardium in static images of some portions the ventricles by Stage 17. Cardiac jelly remained visible in much of the inner curvature following its regression elsewhere in the ventricles, consistent with continuity between inflow and outflow cushions at that location (Fig. 3, Supplemental Movies 3 and 4).

Figure 3.

Imaging of Stage-17 chick embryonic heart. A: Projection image of confocal stack of whole mount specimen stained with anti-myosin (MF20, green) and propidium iodide (PI, nuclear stain, red). Pharyngeal arches are indicated with arrowheads. B: A single confocal section from the stack. C: A three-dimensional reconstruction. Ventricular trabeculae are indicated with arrows, outflow tract cushions with asterisks. D: A frontal view of the heart; note that due to echogenicity of the blood; delineation of the endocardium is impossible on a single frame. E, E': High-power views in the same orientation showing different phases in cardiac cycle, with myocardium and endocardium indicated by green and red arrows, respectively. Note that there is some residual ventricular volume even at maximum contraction. For more detail, see the Supplemental Movies 3 and 4. F: A cross-section through the pharyngeal region; the arteries of the pharyngeal arches are indicated by arrowheads (compare with A). G: M-mode imaging of atrioventricular cushion function; the time of complete occlusion of the lumen is indicated by arrows. There is a thin layer of acellular jelly just underneath the endocardium. Note that in this mode, the blood in the lumen is distinguishable by its speckled appearance. H: The flow across the atrioventricular canal; passive component of the ventricular filling is predominant. Cooling the embryo by 3° caused some systolic regurgitation (red arrow) that was reversible by warming it back to physiological temperature. a, active phase of ventricular filling; Br, brain; AV, atrioventricular; iAVC, inferior atrioventricular cushion; OT, outflow tract; p, passive phase of ventricular filling; sAVC, superior atrioventricular cushion; V, ventricle. Scale bars = 100 μm.(Fig. 2).

The segments of the tube heart could easily be distinguished at these stages with 2D imaging. The right and left atrial chambers showed significant volume expansion, had thicker walls, and were more contractile than in previous stages. The right atrial appendage in particular could be seen protruding into the inner curvature in intimate association with ventricular and outflow segments of the heart. The atrioventricular segment was shorter and marked by the persistence of the cardiac jelly (still predominantly acellular and echo-free). The right and left ventricles could be best identified by their position relative to the interventricular furrow along the inner curvature and by their relative position in the curved ventricular mass along the outer curvature.

The outlet segments of the heart were particularly well seen, including the “dogleg” bend between the proximal and distal outflow segments, the transition from the contractile to noncontractile tissue in the distal outlet, the aortic sac, and the branchial arch vessels (Fig. 3D, Supplemental Movie 3).

Functionally, the regression of the ventricular cardiac jelly coincided with the loss of complete apposition of the endocardial surfaces in the ventricles and the first evidence of a significant residual ventricular volume after contraction. Complete endocardial apposition continued in the atrioventricular junctions and outlet segment of the heart (Fig. 3G, Supplemental Movie 4).

As noted, the echogenicity of the blood was indistinguishable from other embryonic soft tissues and resulted in difficulty visually locating the endocardial surface with precision in still frames. However, observation of 2D echo video recordings of blood flow readily revealed directionality of blood flow, including regurgitation if present. In addition, low velocity laminar flows and vortical elements in slowly moving blood in the cardiac chambers could easily be seen on 2D imaging when the flow velocities were so low as to make Doppler insensitive. It should be noted, however, that these phenomena were only seen in abnormal embryos or as a response to conditions such as hypothermia (Fig. 3H).

Pulsed Doppler sampling showed biphasic inflow profiles with dominance of the early filling wave (Fig. 3H). Outflow signals were monophasic. The conversion from biphasic to monophasic flow was accompanied by significant flow acceleration.

Period of Ventricular Septation (Stages 25–34)

Early stages of heart development (up to Stage 25) in the avian embryo were readily visualized by conventional light microscopy (Figs. 1, 2); however, even at these stages echocardiography provided useful cross-sectional images and spatially accurate measurement of blood flow velocities in the developing heart and vessels. B-mode images showed vigorous beating of the heart and allowed distinction of the cardiac cushions and forming septa. The muscular part of the interventricular septum was distinguishable by stage 24 (Fig. 4) and the interventricular foramen therefore also readily identified.

Figure 4.

Cardiac morphology at Stage 24 revealed by ultrasound and histology. A,B: The four-chamber view. C,D: Transverse sections at the level of the outflow tract. E: The pressure gradient across the band (placed 24 hr ago at Stage 21) measured during a Doppler sweep from the ventricle to the outflow tract. F: A different Stage-21 heart imaged confocally with position of the band indicated by red arrow. G,H: Monophasic flow across the outflow tract cushions in Stage-27 heart and diastolic regurgitation (red arrow) due to placement of a constricting loop of 10-0 nylon 48 hr earlier. LA, left atrium; LV, left ventricle; OT, outflow tract; RA, right atrium; RV, right ventricle. Scale bar for histological sections = 1 mm; smallest division on echo pictures = 100 μm.

When the vascularized chorioallantoic membrane starts to obscure the embryo by incubation day 5 (Stage 27; see also Fig. 1), the advantages of the ultrasound images come to the fore. At this stage, specific details of developing valve motion can be analyzed, as previously demonstrated for cushion mechanics (Hu et al., 1991; Butcher et al., 2007). Since the embryos were now mostly covered with the chorioallantoic membrane, the optical visibility of the heart was minimal (Fig. 1). The heart rate could still be deduced from video recordings, but the ventricular contour was barely traceable even with green light imaging in the monochrome mode. Also, the position of the embryos started to change, with their backs turning to the surface, especially in ovo. This contributed to complete impracticality of direct optical imaging past Stage 29.

Ultrasonography of Stage 29 revealed dual cardiac outflows and further growth of the interventricular septum. The interventricular foramen is smaller than Stage 24 but still readily seen. In short-axis views, the two ventricles were clearly separated. The echogenicity of the left ventricular endocardial surface was greater and allowed more precise detection of the luminal boundaries than for the right ventricle, presumably due to the many coarse trabeculae associated with the right ventricular endocardium (Fig. 5, Supplemental Movie 5). Details of ventricular trabeculae (with the exception of major sheets in the left ventricular apex; Fig. 5C and Supplemental Movie 6) were beyond spatial resolution limits.

Figure 5.

Cardiac morphology at Stage 29. Four-chamber views at different levels are shown (A–D). Note that the major left ventricular trabecular sheets (arrows in C, D) can be resolved on the echocardiographic pictures (and better yet at Supplemental Movies 5 and 6). The cranial part of the interventricular foramen is marked with arrows. Ao, aorta; LA, left atrium; LV, left ventricle; Pu, pulmonary artery; RA, right atrium; RV, right ventricle. Scale bars = 1 mm in B, 100 μm in D, G, and 100 μm smallest division in A, C, E, and F.

All veins entering the heart (left and right superior caval, inferior caval, single pulmonary) were also clearly seen based on echogenic blood streams moving in them. Both in vivo and in vitro imaging was feasible even in post-septation stages until hatching.

Echocardiographic Imaging of Abnormal Hemodynamics

To obtain baseline values of ventricular functional parameters, we performed serial investigations of 8 normal embryos at Stages 24 and 27. At Stage 24, the heart rate was 155 ± 33 beats per minute, end-diastolic volume 3.3 ± 0.8 mm3, end-systolic volume 2.0 ± 0.6 mm3, and maximal outflow velocity 14.3 ± 1.5 mm/s. Between Stages 24 and 27, the end-diastolic volume increased by 95%, end systolic volume by 120%, and maximal outflow velocity by 36%. There was also an increase of 50% in the stroke volume, and 31% increase in the cardiac output. At Stages 24, 27, and 29, we performed imaging of embryos subjected to the experimental hemodynamic interventions of conotruncal banding (Clark et al., 1984, 1989; Sedmera et al., 1999). In embryos banded 24 h before (at Stage 21), an outflow tract Doppler velocity gradient across the band (Fig. 5E) could subsequently be measured in most cases. In a significant proportion (6 of 8 vs. 0 of 8 controls) imaged at Stage 27, regurgitation across the band or through the atrioventricular valves (1/8) was noted. While the ventricular volumes were within normal range and ejection fraction was preserved (38 ± 10 vs. 38 ± 12 at Stage 24 and 34 ± 11 vs. 32 ± 8 at Stage 27, p = NS), the parameter that showed the clearest difference between control and banded hearts was the maximal outflow tract velocity, which was significantly increased at both Stage 24 and 27 (+120%). The hearts imaged at stage 29 showed persisting gradients and impaired ventricular septation (larger interventricular foramen, absence of alignment of the aorta with the left ventricle). Further applicability of echocardiographic imaging was recently demonstrated on post-septation Stage 34 hearts in experimentally induced left ventricular hypoplasia (deAlmeida et al., 2007). Detailed description of later fetal stages, valve maturation, and ventricular functional performance, is thus omitted.

DISCUSSION

General Considerations

The technical requirements for successful ultrasound evaluation under physiologically desirable conditions are achievable in early avian embryos. Temperature control that does not desiccate the exposed yolk surface is critical and several strategies are potentially compatible with ultrasonic imaging. Meticulous temperature control together with care taken during the imaging process enables repeated imaging of the same embryos. Access to the embryo from a variety of imaging planes is made possible by use of a shell-less culture system. Access in ovo is also achievable but the viewing angles are much more limited due to the size of the scanhead and its fixed focal length (4.5 mm for the 55-MHz scanhead used in these studies). As magnification increases, the apparent zone of focus decreases and at the highest practical magnification focus was generally limited to the object of study being within 1 to 1.5 mm of the focal plane.

Acoustical coupling is a critical issue in ultrasonography of avian embryos. Commercial ultrasound gels were lethal, creating serious heart rate abnormalities much more quickly than hypoxia alone could explain. An ideal acoustical coupling medium for embryonic use would have high surface tension (to support meniscus formation [“bridging”] between the scanhead and the surface of the egg membranes, and thus minimize the volume of solution required), be physiologically compatible, and have a lower specific gravity than the yolk of the egg. Albumin is a potential choice, but it suffered from having a higher specific gravity than the yolk, quickly flowing from the surface of the yolk and elevating the yolk and embryo, thereby resulting in failure to maintain the correct working distance between the scanhead and the embryo.

Additionally, the viscous nature of the albumin made bubble entrapment a continuous threat to image quality. Buffered saline solutions, in particular Tyrode's-HEPES solution, were found to provide highly acceptable nontoxic acoustical coupling. Although yolk has a lower specific gravity than buffered saline solutions, their difference in specific gravity is low enough to significantly slow the rate of redistribution of volume and they were less prone to bubble entrapment. As embryos mature, however, and their oxygen consumption requirements increase, the impact of an aqueous buffer bathing the chorioallantoic membranes becomes a limitation to the scanning time possible before functional deterioration in cardiac status becomes apparent.

Embryonic body density increases with respect to the yolk during maturation, resulting in the embryo sinking into the yolk as it grows. This is a relevant concern to the fixed focal length of the scanhead technology currently available for ultrasound biomicroscopy. At the earliest stages, the embryonic mass and density are trivial and the embryo is suspended above the superior surface of the yolk. Imaging is challenging due to the requirement for 4.5 mm of acoustical coupling material between the scanhead and the very buoyant yolk/embryo combination. Imaging required placing 200–500 μl of buffered chick Tyrode's solution over the embryo and quickly lowering the scanhead to make contact with the coupling medium. The scanhead was then elevated to the viewing distance while carefully adding further buffer, thus forming a surface tension-dependent “bridge” of acoustical coupling medium between the embryo and the scanhead membrane.

Conversely, at Stage 24 and beyond, the embryos are sunken into the egg yolk far enough that some pressure on the chorioallantoic vessels is needed to obtain the requisite 4.5 mm from the heart; this results in valvar regurgitation and a reversible bradycardia or other arrhythmias in most animals after a variable amount of time. Scanheads with longer focal distances, with somewhat lessened resolution, are commercially available and would potentially reduce these problems. If scanheads with smaller footprints become available in the future, they may also result in less tendency to generate physiological artifact.

Ventricular Contraction and Cardiac Jelly

Our data demonstrate that the distribution of the cardiac jelly, and therefore the level of approximation of the endothelium and myocardium, is not uniform around the circumference of the ventricular myocardium. Well-defined zones of proximity are present at the myocardium adjacent to the dorsal myocardium and on the opposite side of the ventricular lumen in the early tube heart. These regions function as “hinges,” defining the line of complete apposition of the endocardium during ventricular contraction. The myocardial contour at peak ventricular filling for a given cross-section can be represented as a circle. In contrast, the systolic ventricular cross-section is an oval (Fig. 2), with the major axis represented by the axis of endocardial apposition. The fact that the opposite endocardial surfaces contract to fully occlude the cardiac lumen in a wave from the atrioventricular canal to the muscularized outflow tract has implications for description of cardiac mechanics at these stages (Barry, 1948; Patten et al., 1948). Our data demonstrate that in the pre-Stage 16 tube heart there is a wave of apposition of the endocardial surface from the atrioventricular canal to the aortic sac that obliterates the ventricular lumen at each point sequentially. There is no residual non-ejected volume under normal conditions until the ventricular cardiac jelly regresses and cardiac ventricular outgrowth (“ballooning”) begins, at approximately Stage 16–17 (Moorman and Christoffels, 2003). This renders concepts of myocardial mechanics derived from global function models of the piston-pump heart dubious for hearts prior to Stage 16. Overall, our observations correlate well with the data of Forouhar et al. (2006), in suggesting that improved imaging technology will lead to revised models of the mechanics of the preseptated, pre-ballooning embryonic heart.

Observations of Embryonic Blood Streams

Our observations of only a few isolated circulating cells at the early stages are consistent with the low hematocrits documented in the earliest stages of the circulation by the studies of Rychter et al. in chicken embryos (Rychter et al., 1955) and more recently also in the mammalian circulation (McGrath et al., 2003). At later stages, echogenicity of blood increases significantly, because of nucleation and dimensions of the red blood cells. This results in deterioration of contrast in static images, but viewing of movies allows one to distinguish static from moving parts, a very useful feature when observing low-velocity venous or capillary flows (see Supplemental Movie 3).

The observed switch in relative contribution of passive and active filling in favor of the active component corresponds well with published results (Hu et al., 1991; Butcher et al., 2007), and may correlate with developing pectinate muscles that strengthen the atrial walls (Sedmera et al., 2006) as well as with changing ventricular compliance related to remodeling of trabecular architecture (Sedmera et al., 2000).

Anomalies in intracardiac blood flow patterns induced experimentally were readily visualized. Flow velocity increases across stenotic regions were readily demonstrated, indicating pressure gradients between ventricle and the conotruncus. Imaging at room temperature with non-physiological heart rates (60–70 bpm) showed regurgitation in both atrioventricular and outflow valves. Small regurgitant jets were often encountered in imaging valves at stage 29 and greater, but became less common in older embryos as the semilunar valve cusps became more pliant. The classic investigations of the Clark and Keller labs in chick embryos were the first to explore Doppler ultrasonic technology and its applicability to embryonic cardiovascular physiology (Clark et al., 1986; Hu and Clark, 1989; Keller et al., 1990). Multiple investigators have subsequently shown that mouse embryo physiology can be investigated and implied through Doppler techniques, especially with the assistance of improved 2D guidance, and Doppler-detected regurgitant flow in mammalian embryos is highly correlated with structural malformations and/or myocardial dysfunction, as first noted by Gui et al. (1996) and reviewed by Phoon (2006). In our report, we demonstrate that in chick embryos a high level of morphologic detail can be added to the analysis of cardiovascular structure and function in avian embryos.

Comparison With Other Imaging Modalities

Embryonic ultrasonography contrasts favorably with standard video techniques (Clark et al., 1986). There is potentially better time resolution in videomicroscopic image recording but the ability to measure and characterize real-time intracardiac structures and motion from multiple viewing planes is impossible in the chick embryo without ultrasonography. Additional methods of analysis recently described include MRI (Yelbuz et al., 2003) and microCT (Butcher et al., 2006, 2007; Johnson et al., 2006). However, the later two techniques are of limited use in vivo (Hogers et al., 2000, 2001). To some extent, an alternative modality could be optical coherence tomography (Yelbuz et al., 2002), but this technology is not widespread, and off-the-shelf systems are not yet available (for comparisons of embryonic imaging modalities, see Table 1). For comparison of quantitative measurements derived from different techniques, several considerations need to be taken into account. For example, the values of heart rate, end-diastolic and end-systolic ventricular areas reported by Keller and associates using video microscopy (Keller et al., 1991) show that their embryos were considerably smaller (younger) than ours. Nevertheless, the values are in the same order of magnitude. Additional considerations have to be taken for the calculated parameters, such as cardiac output. Internal structure of the embryonic ventricle resembles a sponge, with the proportion of the actual lumen being only ∼37% of the epicardial area (Sedmera et al., 1998), so the volumes derived from tracking the epicardial contour, both on video or echo recordings, over-estimate the actual values.

Table 1. Comparison of Advantages and Disadvantages of Different Imaging Modalities Applicable to the Developing Hearta
TechniqueResolutionAdvantagesShortcomingsReferences
  • a

    All modalities useable in vivo require setup for maintaining constant physiological temperature and humidity. In most techniques, in-plane resolution is superior to z-resolution.

Ultrasound biomicroscopy30 μm, 15 (1 ms in M-mode)Real time, Doppler velocimetryLow contrast, limited focal plane, 3D only with gatingdeAlmeida et al., 2007; Phoon, 2006
OCT (optical coherence tomography)1 μm, 10 msReal time, 4D dattasetsLimited depth of viewYelbuz et al., 2002 Filas et al., 2007
Micro-CT10 μm3D datasets, endovascular contrast imagingCurrently for fixed contrasted specimens onlyButcher et al., 2007
Micro-MRI25 μm3D volumes, gadolinium contrast improves signal-to-noise ratioLimited useability for live 3D imaging (immobilization, gating)Hogers et al., 2000, 2001; Yelbuz et al., 2003; Zhang et al., 2003
Video microscopy−2 μm, <15 msInexpensive, real-timeLimited information in Z, only for optically unhindered heartsKeller et al., 1990, 1991, 1996
Confocal microscopy (live)1 μm, 10 msReal time, 4D datasets, fluorescent probesNeeds small, transparent specimensMiller et al., 2005; Forouhar et al., 2006
Histology + 3D reconstruction<1 μm in plane, Z-depends on section thicknessOff-line, inexpensive, possibility to use specific stainingNot for live specimens, 3D reconstruction labor intensiveMiller et al., 2005; Mommersteeg et al., 2007

Further validation of the utility of embryonic echocardiography is provided by its ability to detect structural and functional anomalies in the model of conotruncal banding (Clark et al., 1984); in particular, valve regurgitation, expected based on abnormal valvular morphology (Sedmera et al., 1999; Reckova et al., 2003), has not been documented before. Outflow tract regurgitation across the band, observed in a significant proportion of the banded embryos (Fig. 4), could help explain the dilated phenotype sometimes observed in addition to wall thickening in this model (Sedmera et al., 1999). Abnormal cardiac structure and function was also documented recently in the model of left atrial ligation (deAlmeida et al., 2007). Flow velocity values between 10–40 mm/s correlated well with values reported previously from the dorsal aorta (Dunnigan et al., 1987). Minimally invasive, repeated echocardiographic investigation provides a unique opportunity to correlate abnormal structure and function in animal models of cardiac dysmorphogenesis, indeed a purpose for which this method is originally intended (Phoon et al., 2004). Since this landmark study in embryonic mice, the parameters of the system have improved significantly: higher sampling frequency results in improved spatial resolution (down to 30 μm) and temporal resolution (in B-mode, 60 frames/s).

Correlation with cross-sectional histology shows that resolution of the system is adequate for visualization of intracardiac structures including cardiac jelly, cushions, valvar leaflets, large arteries and veins, and major trabecular sheets in the left ventricle. Finer detail, especially in the right ventricle, was obscured because of similar echogenicity of blood and fine trabecular network. However, the ability to study a particular area of interest throughout the cardiac cycle resulted in rapid detection of malformations (ventricular septal defect, dysplastic valvular leaflets), especially those associated with streaming blood. In addition, it is possible to calculate functional parameters of the left ventricle at stages of development when optical in ovo recordings are technically impossible.

Study Limitations

Some of the limits of the system are caused by laws of physics (resolution, focal depth, low contrast between blood and heart walls), but some could be overcome by technical improvements. The Vevo system uses a sector scanning technology in its current scanhead system, which, while state-of-the-art, produces undesired vibrations, has a fixed working distance, and rather large footprint (Fig. 1). Enhancement by conversion to a solid-state crystal system would potentially result in reduced vibration and footprint, greater latitude in depth of focus, and perhaps color flow mapping. Coloring flow mapping is a useful feature especially for presentation of static images of embryonic cardiovascular structures. As the grey-scale signal intensity of the blood, myocardium, and other soft tissues are very similar, it can be difficult to measure myocardial or chamber dimensions on still frame images of some cardiac structures. Endocardial cushion tissue and valves, however, are generally distinctive in echo-texture at all developmental stages examined and readily distinguished from the blood or myocardium.

In conclusion, we found ultrasound biomicroscopy useful in the imaging of the developing heart, and uniquely valuable at later stages of development. Echocardiographic study of pre-septated heart physiology in avian embryos provides a valuable foundation for investigation of abnormal cardiac development in transgenic mouse models. The imaging platform we report here, while without the spatial resolution of optical coherence tomography, does not require gating for image acquisition and is widely available. Anticipated improvements of instrumentation should make this imaging modality even more practical as an imaging tool for the developmental biologist.

EXPERIMENTAL PROCEDURES

Egg incubation and Ex Ovo Culture

Fertile white Leghorn chicken eggs were purchased from Sunkist hatchery in Sumter, SC, and delivered via courier in a temperature-controlled manner in order to ensure egg viability and quality. The eggs were incubated blunt end up in a forced-draft constant-humidity incubator at 37.5°C with continuous rocking and studied at Hamburger-Hamilton (Hamburger and Hamilton, 1951) stages 24 (4 days) to 34 (8 days) of a 46-stage (21-day) incubation period. Embryos that were dysmorphic were excluded from further study.

Shell-less culture of chick and quail embryos was performed according to described protocols (Auerbach et al., 1974; Tuan, 1983) with minor modifications. Eggs pre-incubated for 2 days were aseptically explanted into 64/51 mm (top/base) hexagonal weigh boats. The weigh boats were gently placed inside sterile 150×15-mm Petri dishes containing sterile, distilled water. The Petri dishes were then incubated inside a still-draft cell culture humidified incubator at 37.5°C until subsequent ultrasound imaging.

Conotruncal banding, an established experimental procedure causing hemodynamic perturbations (pressure overload) in chick embryos, was performed at Stage 21 as described in detail previously (Clark et al., 1984, 1989; Sedmera et al., 1999).

Photography and Videomicroscopy

Photographs of the equipment were taken using a Canon G5 digital camera. Images and movies were taken of all embryos prior to ultrasound imaging. The embryos were photographed under a Leica MZ125 dissecting microscope with 0.5× Plan Apo objective using the same camera mounted via Scopetronix MaxView Plus universal attachment kit. Standard video camera (Automaticam, Microimage Video Systems) connected to high-resolution video monitor (Sony) and SVHS video tape recorder was used for general purposes.

Setup for Ultrasound Imaging

Access to the embryo from a variety of imaging planes was made possible by use of ex ovo culture system. For imaging, we used a RM708 scanhead (center frequency 55 MHz) on the Vevo660 ultrasound biomicroscopy system (VisualSonics, Toronto, Canada; Fig. 1). Temperature of the embryos during examination was maintained at 37.5 ± 0.5°C using a combination of Radnoti circulating water jacketed incubator and a heat lamp. Setup for in ovo imaging was essentially the same; the egg was held stable in a clay mold, opened on its side, and almost fully submerged in the warm water. Acoustic coupling between the probe and yolk surface was maintained by sterile, warm Tyrode's-HEPES solution.

Histological Examination

After the last ultrasound recording, the embryos were dissected out of the membranes and fixed by immersion in Dent's fluid (80% methanol, 20% dimethylsulfoxide) for at least 24 h. The specimens were then processed for paraffin histology, and sectioned in planes matching the echocardiographic views. Photographs were taken after hematoxylin-eosin staining on an Olympus BH2 compound microscope fitted with Olympus DP50 digital CCD camera. Early tubular hearts up to Stage 18 were processed for whole mount double immunohistochemistry as described previously (Germroth et al., 1995; Miller et al., 2005), and either sectioned on vibratome at 200 μm after embedding in polyacrylamide, or cleared in benzyl alcohol-benzyl benzoate prior to confocal imaging. Three-dimensional reconstructions were done in Amira software.

Acknowledgements

We thank Ms. Aimee Phelps for histological assistance, Mr. Wade Reardon for help with echocardiographic data quantification, and Prof. Oldrich Eliska for a critical reading of the manuscript. This work was conducted (in part) in a facility constructed with support from the National Institutes of Health, Grant C06 RR018823 from the Extramural Research Facilities Program of the National Center for Research Resources. D.S. was the recipient of a Purkinje Fellowship.

Ancillary