Left-right asymmetry of a nodal-related gene is regulated by dorsoanterior midline structures during Xenopus development

The development of left-right asymmetry is highly conserved throughout vertebrate species. One of the earliest morphologic indicators of left-right asymmetry is cardiac looping. Cardiac looping is the process by which the straight heart tube develops a rightward and dorsal bend, forming the anatomic relationships necessary for normal cardiac chamber and outflow tract development. Disruptions of normal left-right asymmetry are associated with cardiac defects in many species, and are a cause of significant morbidity and mortality in the human neonate. Despite the late morphologic appearance of left-right asymmetry, establishment of the left-right axis has been linked to the establishment of dorsal-ventral and anterior-posterior axes in the early embryo (Danos and Yost, 1995). The early establishment of the left-right axis and the later morphologic expression of this asymmetry suggests that there is a signaling pathway that communicates left-right axis information to the developing organ systems.

A molecular pathway potentially involved in the regulation of left-right asymmetry has recently been described in the chick (Levin et al., 1995). This pathway involves the sequential expression of several signaling molecules, including activin, sonic hedgehog (Shh) and the chick nodal-related molecule cNR-1, in an asymmetric pattern along the left-right axis. Disruption of this signaling pathway in chick by alterations in activin or Shh expression prior to cNR-1 expression is associated with an approximately 50% incidence of reversed cardiac orientation, or cardiac randomization. However, a number of the early molecules in this proposed pathway, which include activin, activin receptors and sonic hedgehog, are not expressed asymmetrically along the left-right axis in other vertebrate species (Dohrmann et al., 1993; Hemmati-Brivanlou et al., 1992; Kondo et al., 1996; Ekker et al., 1995; Collignon et al., 1996; personal observation). In contrast, homologues of the chick nodal-related gene cNR-1, which is expressed asymmetrically in the left lateral plate mesoderm later in development, are also expressed asymmetrically in other vertebrate species.

Asymmetric, left-sided expression of nodal-related genes has been described in mice and Xenopus (Lowe et al., 1996; Collignon et al., 1996; Lustig et al., 1996). In the mouse, asymmetric left-sided expression of nodal is apparent in the left lateral mesoderm by early somite stages, which precede cardiac looping. Altered expression patterns of nodal have been found in mice with mutations that result in abnormalities of cardiac situs. Mice homozygous for inv, a recessive mutation that alters embryonic turning and organ situs, exhibit abnormal patterns of nodal staining, ranging from right-sided staining associated with situs inversus (Lowe et al., 1996; Collignon et al., 1996), to a rostral truncation of staining associated with random or ambiguous cardiac looping (Collignon et al., 1996). Mouse embryos homozygous for the iv mutation have randomized left-right orientation (Seo et al., 1992; McGrath et al., 1992) and a variable pattern of nodal expression, with normal, right, bilateral and absent expression of nodal being represented in this population (Lowe et al., 1996). The normal expression pattern of the Xenopus homologue of nodal, Xnr-1, is similar to that described in chick and mouse, with asymmetric expression in the left lateral mesoderm during tailbud stages (Lowe et al., 1996; Lustig et al., 1996). These studies suggest conserved roles for the nodal gene and its homologues in vertebrate left-right patterning and in the determination of cardiac orientation.

The conserved left-right asymmetric expression of nodal-related genes and the association of altered expression of these genes with abnormalities in cardiac orientation in vertebrates implicates nodal-related genes in the establishment of normal left-right asymmetry. As described earlier, some candidate molecules for early regulation of nodal expression are asymmetrically expressed in the chick model; however, asymmetric left-right expression of these candidates has not been seen in other vertebrates. Thus, the mechanisms that regulate asymmetric nodal-related gene expression are unknown. In Xenopus, the establishment of normal left-right cardiac asymmetry is linked to normal dorsoanterior development (Danos and Yost, 1995). The work presented here examines the role of dorsoanterior structures in the regulation of the Xenopus nodal-related gene, Xnr-1. Specifically, we describe the expression pattern and variability of Xnr-1 expression in a control population ofXenopus embryos, show that the left-right asymmetry of Xnr-1 expression is altered by disruption of dorsoanterior development, identify embryonic structures required for the normal pattern of asymmetric Xnr-1 expression and determine the developmental period during which asymmetric expression of Xnr-1 is specified by repression in the right lateral plate mesoderm, both in embryos and in explants. The results indicate that asymmetric Xnr-1 expression is correlated with cardiac orientation and suggest that expression in the lateral plate mesoderm is an intermediary between a negative regulatory signal from the midline anterior structures and the cardiac primordia.

Albino X. laevis females were induced to ovulate by injection of 50 units of pregnant mare’s serum gonadotropin (Sigma) into the dorsal lymph sac followed by injection of 800 units of human chorionic gonadotropin (Sigma) 24 hours later. Eggs were expressed from the ovulating females and fertilized with minced testis from pigmented males suspended in a modified Ringer’s solution at one-third strength (R/3). The fertilized eggs were dejellied in 2% cysteine, pH 8.0. and incubated in R/3 at either 15°C or 21°C. Embryos were staged for use in experiments according to Nieuwkoop and Faber (1967). Heart orientation was scored in embryos anesthetized with 0.05% benzocaine, all other interventions were performed in R/3. Embryos were fixed for in situ hybridization in MEMFA (Danos and Yost, 1995; Harland, 1991) for 1-2 hours and stored in methanol at −20°C.

Antisense RNA probes for Xnr-1 were synthesized from a cDNA insert of Xnr-1 in pBS, (Jones et al., 1995; provided by C. Wright), linearized with PstI and transcribed with T7 RNA polymerase. Probes were labeled with digoxigenin-rUTP (Boehringer Mannheim Biochemical). RNA probe synthesis and whole-mount in situ hybridization were performed using a standard protocol for Xenopus embryos (Sive et al., 1995). Embryos were scored for Xnr-1 expression and photographed. Selected embryos were embedded in paraffin and sectioned at 20-25 μ m in the transverse plane, mounted on glass slides and photographed using light microscopy.

Fertilized eggs were dejellied, placed on quartz slides in R/3 and irradiated for 14-16 seconds with an ultraviolet light at 254 nm within 30 minutes of fertilization (Scharf and Gerhart, 1980). Embryos were either fixed at stage 24-26 for in situ hybridization or scored at stage 42-46 for dorsoanterior development and direction of heart looping. Dorsoanterior development was scored by a standard Dorsoanterior Index (DAI) as follows: DAI 5 - normal appearing embryos; DAI 4 - reduced forehead, close set eyes; DAI 3 - shortened embryo, cyclopic; DAI 2 - severely microcephalic, no retinal pigment, DAI 0-1 - acephalic (Kao and Elinson, 1988; Danos and Yost, 1995).

Two dorsal tissue extirpation methods were used. In broad dorsal tissue extirpations, anterior notochord and some anterolateral tissues were removed from embryos at stage 15 (early neural fold) and stage 20 (neural tube stage), as previously described (Danos and Yost, 1996). Narrow extirpations were performed at stage 15 by removing only anterior notochord, hypochord and directly overlying prospective neural floorplate. Because of neural tube closure, comparable narrow extirpations could not be done at stage 20. Embryos were either fixed at stage 24-26 for in situ hybridization or scored for cardiac orientation at stage 42-46.

Right and left lateral plate mesoderm, overlying ectoderm and underlying endoderm were explanted from embryos at stage 15-16 and 20-21, using a tungsten wire or eyebrow knife. In a subset of embryos, endoderm was removed from the explants and discarded. Explants were cultured in R/3 at 15° C, fixed when sibling embryos were at stage 24 and assayed for Xnr-1 expression by whole-mount in situ hybridization.

Whole-mount in situ hybridzations were scored for pattern of Xnr-1 expression and categorized as left (normal expression pattern), right (mirror-image expression pattern), bilateral (expression in lateral plate mesoderm on both right and left sides), or none (no staining in the lateral plate mesoderm). Explants were scored for presence of Xnr-1 expression. Appropriate sibling controls were processed in parallel to ascertain normal assay function. Statistical significance between groups was tested using a standard z-test of independent proportions, accepting P<0.05 or smaller as significant difference.

Xnr-1 expression in a control population of embryos was determined by in situ hybridization with an antisense RNA probe (Jones et al., 1995). Embryos at varied developmental stages, ranging from stage 7 (blastula) to stage 38 (late tailbud) were analyzed for expression pattern. The earliest left-right asymmetric expression was seen at stages 17-19 (late neural fold to neural tube stages) in the left posterior region of the embryo, just lateral to midline structures. By stages 21-23, the expression of Xnr-1 spread anteriorly just lateral to the midline (Fig. 1A,B) and, by stages 23-26, was prominent in the anterior portion of the left lateral mesoderm (Fig. 1C-E). In embryos with heavier staining, Xnr-1 expression was occasionally seen on the right posterior aspect lateral to the notochord, but did not extend anteriorly or laterally into the lateral plate mesoderm on the right side. By stage 26-27, Xnr-1 expression in the posterior region of the embryo decreased (Fig. 1F). Anterior expression of Xnr-1 decreased by stage 28-30. Asymmetric staining patterns were not apparent later in development (through stage 38).

The variability in normal Xnr-1 expression in a control population was assessed by whole-mount in situ hybridization in 324 embryos at stages 21 through 26. Xnr-1 expression was left-sided and asymmetric, or ‘normal’ in 88% of the embryos. Variations of Xnr-1 expression in the control population at these stages included right-sided expression only in approximately 1% of embryos, bilateral expression in 8% of embryos and no Xnr-1 expression in 3% of the embryos assayed. As discussed later, this rate of abnormal Xnr-1 expression patterns correlates well with a 3-8% rate of spontaneous cardiac reversals in untreated batches of Xenopus embryos (Danos and Yost, 1995; unpublished observations).

In order to determine if left-right cardiac orientation is correlated with Xnr-1 expression patterns in Xenopus, embryos were treated with UV irradiation. UV irradiation in the first cell cycle disrupts both dorsoanterior and left-right axis formation and has been shown to randomize cardiac orientation at a DAI score of 3 (Danos and Yost, 1995). UV-treated, partially ventralized embryos at stage 24-26 were scored for DAI and assayed for Xnr-1 expression by in situ hybridization (Figs 2, 3). DAI 5, or treated embryos with normal dorsoanterior development, were found to have normal, asymmetric left-sided expression of Xnr-1 in 86% of the cases, which is not significantly different from the control population (Figs 2, 3A). DAI 4 embryos had an increased incidence of abnormal Xnr-1 expression, with only 50% of embryos exhibiting normal leftsided asymmetric expression, while 31% of embryos exhibited bilateral Xnr-1 expression (Figs 2, 3B). At DAI 3, no normal Xnr-1 expression was seen (Figs 2, 3C). 49% of DAI 3 embryos exhibited bilateral Xnr-1 expression, and 51% had no detectable Xnr-1 expression. DAI 2 embryos also had a high percentage of abnormal Xnr-1 expression (Figs 2, 3D) and embryos at less than DAI 1 were predominately without Xnr-1 expression at stage 24 (Fig. 2).

A normal, asymmetric Xnr-1 expression pattern was characteristic of embryos with normal dorsoanterior development (DAI 5), while there was an increased incidence of abnormal and often symmetric expression patterns in embryos with lower DAI scores (Fig. 3A-D). The percentage of embryos exhibiting abnormal Xnr-1 expression at DAI 3-5 correlated well with cardiac reversal rates obtained from sibling embryos scored for DAI and cardiac orientation at stage 42-46. In this group of embryos, cardiac reversal rates were 3% in an untreated control group (n=148); 2% in treated embryos at DAI 5 (n=63); 9% in DAI 4 embryos (n=22); and 48% in DAI 3 embryos (n=33). Cardiac orientation could not be scored in the majority of DAI 2 embryos and no cardiac structures were visible in DAI 0-1 embryos. Thus, diminished dorsoanterior development was correlated with an altered Xnr-1 expression pattern, as well as altered cardiac orientation. Of note, at DAI 3, where Xnr-1 expression was either bilateral or absent, cardiac orientation was randomized.

Extirpation of dorsoanterior structures during early neurulation results in cardiac randomization, while extirpation of these structures after completion of neurulation does not alter cardiac orientation (Danos and Yost, 1996). This suggests a developmental stage-dependent regulatory function of dorsoanterior structures in the establishment of cardiac orientation prior to heart formation and looping. In order to determine whether asymmetric Xnr-1 expression is dependent on dorsal midline cells in a stage-specific manner, three series of dorsoanterior midline extirpations were performed.

Broad extirpations of anterior notochord and dorsoanterior structures were performed on embryos at stage 15 or stage 20 as previously described (Danos and Yost, 1996). Stage 15 extirpations resulted in predominately bilateral Xnr-1 expression (Fig. 4A,C). In contrast, stage 20 extirpations of the same tissues did not alter normal asymmetric Xnr-1 expression (Fig. 4B,C). In previous studies, broad extirpation of dorsal midline structures at stage 15 is associated with randomization of cardiac orientation, while extirpation at stage 20 does not perturb cardiac situs (Danos and Yost, 1996). To further define the essential dorsal structures required for left-right development, narrow extirpations of only notochord, overlying neural floor plate and hypochord were performed at stage 15, leaving somites and other more lateral structures intact. This technique of more limited extirpation is not possible at stage 20 because of the overlying neural tube. A 36% incidence of cardiac reversals was obtained when narrow extirpations were performed at early neural fold stages (n=59), consistent with an increased incidence of bilateral Xnr-1 expression using this technique (Fig. 4C).

These experiments demonstrate that removal of dorsoanterior structures, including notochord, has a developmental stage-dependent effect on Xnr-1 expression as well as cardiac orientation. Removal of the dorsoanterior midline structures at stage 15, whether through a broad extirpation of tissue or a narrow extirpation of just notochord and overlying neural floor plate, is associated with bilateral expression of Xnr-1 (Fig. 4C) and an increased incidence of cardiac reversals. Removal of analogous structures at stage 20 has no effect on normal asymmetric Xnr-1 expression (Fig. 4B,C) or cardiac orientation.

The results of dorsal midline extirpation experiments support the hypothesis that notochord and/or neural floorplate and hypochord are involved in the regulation of Xnr-1 expression. In order to investigate the dependence of Xnr-1 expression in both the right and left lateral mesoderm on midline dorsal structures, lateral plate mesoderm was explanted from both right and left sides of the embryo, cultured and assayed for Xnr-1 expression. Since endoderm has been shown to have a role in induction of cardiac primordia in Xenopus embryos (Nascone and Mercola, 1995), these experiments were performed with and without endoderm to investigate any additional regulatory effect of endoderm on Xnr-1 expression. In contrast to the asymmetric expression pattern in untreated embryos, Xnr-1 was expressed in both left and right lateral plate mesoderm explants obtained at stage 15 (Fig. 5A,B; Table 1), with no statistically significant difference in expression between the two groups. At stage 15, Xnr-1 expression in left lateral plate mesoderm was slightly enhanced by the presence of endoderm (P<0.10) and Xnr-1 expression in right lateral plate mesoderm was unaffected by the presence of endoderm. These results suggest that the presence of dorsoanterior midline structures inhibits the expression of Xnr-1 in the right lateral plate mesoderm at later stages in development and that endoderm is not required after stage 15 for asymmetric Xnr-1 expression.

To assess the timing of specification of Xnr-1 expression in the lateral plate mesoderm, left and right lateral plates were explanted with and without endoderm at stage 20 and assayed for Xnr-1 expression at stage 24. Left explants expressed Xnr-1 (Fig. 5C). In contrast, very few of the right lateral plate mesoderm explants expressed Xnr-1 (Fig. 5D; Table 1). The difference between left and right lateral plate mesoderm was statistically significant (P<0.001, using the z-test of independent proportions). Endoderm had no effect on Xnr-1 expression in either group at stage 20 (Table 1). These data are consistent with the hypothesis that the inhibitory effect of dorsal midline structures on Xnr-1 expression in the right lateral plate mesoderm is stage-dependent, and occurs after stage 15 (early neural fold) and before stage 20 (neural tube stage). Thus, the specification of left-right asymmetric Xnr-1 expression in lateral plate mesoderm occurs by stage 20 and coincides with the developmental period during which specification of cardiac left-right orientation is dependent on midline structures (Danos and Yost, 1996).

Left-right asymmetric expression of nodal-related genes has been described in embryos of three vertebrate species, mouse, chick and frog, and appears to be highly conserved (Levin et al., 1995; Collignon et al., 1996; Lowe et al., 1996; Lustig et al., 1996; this study). Here, alterations of the normal left-right asymmetric expression patterns of a nodal-related gene in Xenopus, Xnr-1 (Lowe et al., 1996; Lustig et al., 1996), are shown to be correlated with alterations in cardiac left-right development. Furthermore, results of extirpation and explant studies in Xenopus embryos indicate that the regulation of asymmetric Xnr-1 expression is dependent on signals from midline cells and that the right lateral plate mesoderm becomes specified during neurulation to repress Xnr-1 expression.

Our description of Xnr-1 left-right asymmetric expression patterns in Xenopus embryos, with expression in the left lateral plate mesoderm during tailbud stages, is consistent in general with others (Lowe et al., 1996; Lustig et al., 1996). However, examination of a large population of embryos revealed variability. Less than 1% of the population exhibited a reversal of the normal expression pattern. A larger percentage (11%) exhibited loss of normal asymmetric Xnr-1 expression: either bilateral Xnr-1 expression in both the left and right lateral plate mesoderm or bilateral absence of Xnr-1 expression in the lateral plate mesoderm. In chick, bilateral expression or absence of cNR-1 expression has been correlated with randomization of cardiac orientation (Levin et al., 1995). Similarly, mouse iv/iv homozygotes have randomized cardiac orientation (Seo et al., 1992; McGrath et al., 1992) and a high incidence of either symmetric or absent expression patterns of nodal (Lowe et al., 1996).

If the determination of cardiac left-right orientation is dependent on normal Xnr-1 asymmetric expression, then early experimental interventions that result in randomization of cardiac orientation in Xenopus should also alter the expression patterns of Xnr-1. This is the case in four separate interventions. First, Xnr-1 expression is altered in embryos with diminished dorsoanterior development (Figs 2, 3). As the DAI score is decreased to 3, there is an increased incidence of bilateral Xnr-1 expression patterns. More severely ventralized embryos lacked Xnr-1 expression, which probably reflects the severe disruption of axis formation and the loss of lateral plate mesoderm. Second, Xnr-1 expression is bilateral and cardiac orientation is randomized in embryos from which midline structures, including notochord, hypochord and floorplate, were extirpated at stage 15 (Fig. 4). Third, injection of RGD peptides into early gastrula randomizes left-right orientation (Yost, 1992) and results in bilateral Xnr-1 expression (data not shown). Fourth, early perturbations in the maternal Vg1 signaling pathway result in randomization of left-right orientation and bilateral expression of Xnr-1 (Hyatt et al., 1996). Thus, there is a strong correlation between altered Xnr-1 expression and randomization of cardiac left-right orientation in Xenopus. This suggests two possibilities: one, that asymmetric Xnr-1 expression in the lateral plate regulates specification of cardiac left-right orientation; or two, that asymmetric Xnr-1 expression and cardiac left-right orientation are parts of separate, parallel pathways that are both regulated by a common factor. Recently, bilateral expression of the nodal-related gene cNR-1 in the chick has been shown to alter cardiac orientation (M. Levin and C. Tabin, personal communication) implicating the nodal-related genes in the regulatory pathway of cardiac left-right determination. Overexpression of Xnr-1 in 4-cell Xenopus embryos (Jones et al., 1995) or in single cells on the left or right side of 8-to 16-cell embryos (data not shown) results in hyperdorsalization. These results reflect an earlier developmental role in dorsoventral axis formation (Jones et al., 1995) but complicate the assessment of a direct role for Xnr-1 in cardiac left-right orientation, since cardiac orientation is dependent on dorsoventral development (Danos and Yost, 1995) and is perturbed in embryos with secondary axes (Hyatt et al., 1996).

It is not known what signals in the early embryo are necessary for the establishment of asymmetric nodal-related gene expression in the lateral plate mesoderm, such that Xnr-1 expression is on in the left and off in the right. Here, extirpation and explant culture experiments address two issues: what cell interactions are necessary for specification of Xnr-1 asymmetric expression and when does the lateral plate become specified, as a result of these interactions, for asymmetric expression of Xnr-1. Extirpation of dorsal midline cells at stage 15 results in altered expression of Xnr-1 (Fig. 4). Somewhat surprisingly, when midline cells are removed, Xnr-1 is expressed ectopically in the right lateral plate mesoderm, so that Xnr-1 expression becomes bilateral. The timing of the midline extirpations is crucial; extirpation at stage 15 results in bilateral Xnr-1 expression and randomized cardiac asymmetry, whereas extirpation of the same midline cells at stage 20 does not (Fig. 4). Broad extirpations at stage 15 gave no increase in incidence of altered Xnr-1 expression patterns relative to narrow extirpations at the same stage, implicating the most medial structures, the neural floor plate, notochord and hypochord, in the regulation of asymmetric Xnr-1 expression.

Explantation of the lateral plate mesoderm and culturing in the absence of the rest of the embryo gave results that concur with the stage-dependent influence of midline structures on the expression of Xnr-1 in the right lateral plate mesoderm. Both left and right lateral plate mesoderm autonomously expressed Xnr-1 when explanted at stage 15 and grown in culture. In contrast, when explanted at stage 20, right lateral plate mesoderm did not express Xnr-1. Although the endoderm slightly improved the rate of expression of Xnr-1 in the left lateral plate mesoderm at stage 15, endoderm appears to have no major role in the regulation of Xnr-1 in this assay system (Table 1). The positive effect of endoderm in the early left lateral plate explants may be due to an increase in the overall health of the explant as opposed to a specific regulatory effect.

The extirpation and explantation results suggest that a negative regulatory signal is emitted from the midline between stages 15 and 20 to suppress expression of Xnr-1 in the right, but not left, lateral plate mesoderm. Since lateral plate mesoderm on both sides of the embryo expresses Xnr-1 in the absence of midline cells, it is likely that they receive an earlier inductive signal to initiate the capability of Xnr-1 expression, which is then countered on the right side by an antagonistic signal from the midline. Explantation of lateral plate mesoderm at stage 15 or extirpation of midline structures at stage 15 separates the right lateral plate mesoderm from the proposed suppressing signal emitted from the midline and allows subsequent ectopic expression of Xnr-1.

Previous embryological and genetic experiments in Xenopus and zebrafish indicated that dorsal midline cells are necessary for normal cardiac left-right orientation (Danos and Yost, 1995, 1996). Given the geometry of the embryo, it is not clear how signals from the dorsal midline are transmitted to cardiac precursor cells at the ventral midline. The results from the extirpation and explantation studies here indicate that the stage-dependent effect of midline structures on Xnr-1 expression is concurrent with or slightly precedes the stagedependent effects of these structures on the specification of cardiac orientation (Danos and Yost, 1996). Midline cells regulate both asymmetric Xnr-1 expression and cardiac orientation between stages 15 and 20. The lateral mesoderm is positioned between the dorsal midline and the ventral midline, and might serve as an intermediary for passing signals between cells on opposite sides of the embryo. Thus, we propose that signals pass from the midline, possibly via somitic tissue, to the lateral plate to specify Xnr-1 expression. Once Xnr-1 asymmetry is specified, it can then either directly or indirectly signal the heart primordia to specify cardiac leftright orientation. In this model, cardiac primordia that are presented with either bilaterally symmetric expression of Xnr-1 or with no Xnr-1 expression would not be specified for left-right orientation and, as a population, would be randomized for left-right orientation.

Although left-right body axis formation is initiated early in development (Danos and Yost, 1995; Yost, 1991, 1995; Hyatt et al., 1996), the left-right signaling pathway is probably dependent on subsequent checkpoint regulation by the midline cells at later stages of development and thus can be altered by experimental manipulations until the point of specification of left-right orientation within an organ primordium. This checkpoint regulation would allow for consistent alignment of the left-right axis in each of many organ primordia relative to the dorsal-ventral and anterior-posterior axes. Little is known about the mechanisms by which developing organ systems, including the heart, respond to these left-right signals to generate morphologic asymmetry.

In summary, the variability in left-right expression of Xnr-1 in a untreated populations is consistent with observed rates of spontaneous cardiac reversals. Four experimental interventions in early embryos, known to cause cardiac randomization in X. laevis, also alter Xnr-1 expression, with a resulting increase in bilateral expression patterns. The molecular nature of the signals from the notochord and neighboring midline structures that are necessary for the specification, during neurula stages, of subsequent Xnr-1 asymmetric expression is currently being investigated.

The authors thank M. L. Condic for suggestions on data analysis and comments on the manuscript, Brian Hyatt for general discussions regarding left-right asymmetry, Judy Liddell for technical assistance, Bruce Lingren for advice on statistical analysis, and K. Sampath and C. Wright for Xnr-1 DNA. We would also like to thank M. Levin, K. Sampath, C. Tabin and C. Wright for sharing results prior to publication. J. L. L. was supported by fellowships from Variety Club Children’s Association and the American Academy of Pediatrics. M. C. D. was supported by a NSF-RTG predoctoral traineeship and a U. of M. Dissertation Fellowship. This work was supported in part by a Grant-In-Aid from the American Heart Association and was done in part during the tenure of an Established Investigatorship from the American Heart Association to H. J. Y.