Disruption of hyaluronan synthase-2 abrogates normal cardiac morphogenesis and hyaluronan-mediated transformation of epithelium to mesenchyme

We identified hyaluronan synthase-2 (Has2) as a likely source of hyaluronan (HA) during embryonic development, and we used gene targeting to study its function in vivo. Has2^–/– embryos lack HA, exhibit severe cardiac and vascular abnormalities, and die during midgestation (E9.5–10). Heart explants from Has2^–/– embryos lack the characteristic transformation of cardiac endothelial cells into mesenchyme, an essential developmental event that depends on receptor-mediated intracellular signaling. This defect is reproduced by expression of a dominant-negative Ras in wild-type heart explants, and is reversed in Has2^–/– explants by gene rescue, by administering exogenous HA, or by expressing activated Ras. Conversely, transformation in Has2^–/– explants mediated by exogenous HA is inhibited by dominant-negative Ras. Collectively, our results demonstrate the importance of HA in mammalian embryogenesis and the pivotal role of Has2 during mammalian development. They also reveal a previously unrecognized pathway for cell migration and invasion that is HA-dependent and involves Ras activation.

Hyaluronan, or hyaluronic acid (HA) is a linear, high-molecular-weight (mega-Dalton) polymer comprised of repeating disaccharide units of (β1→3)D-glucuronate-(β1→4)N-acetyl-D-glucosamine. HA is synthesized by integral plasma membrane glycosyltransferases and is exported directly into the extracellular space (1, 2). Although HA is chemically homogeneous, there are three distinct mammalian HA synthases (designated Has1, Has2, and Has3), encoded by related but nonlinked genes (3–9). Each synthase has distinct catalytic properties, and the distribution and abundance of each varies during development of the mouse (6, 10). These observations suggest that the different Has enzymes play distinct roles.

HA binds salt and water, expanding the extracellular space (11–14). HA is especially prominent at sites where cell migration occurs, such as pathways of neural crest cell migration and in the developing cardiovascular system. In vivo, HA interacts with other extracellular matrix molecules, typically via an HA-binding domain called the link module (15). These interactions create a supramolecular architecture of the extracellular matrix, i.e., the composite matrix network of HA, link protein, and aggrecan that plays a critical role in load-bearing articular cartilage (16–18).

In addition to its important physical properties, the overexpression of Has genes results in increased anchorage-independent growth and metastasis of transformed cells (19, 20), suggesting a link between HA and transformation. HA is also implicated in receptor-mediated cell adhesion and intracellular signaling (21, 22). Taken together, such observations suggest that HA plays a vital role in diverse cellular events, including cell migration, tissue remodeling, and metastasis. However, the near-ubiquitous distribution of HA in vivo, the biological activity of HA fragments released by degradative enzymes (23), and the inability to inhibit HA synthesis in vivo have hindered definitive analysis of the physiological roles of HA. Accordingly, we used a genetic approach to investigate the roles of HA in vivo and to identify the HA synthase that is critical during embryogenesis.

Expression of Has2 appeared to correlate with expansion of cardiac cushion tissue and subsequent transformation of endocardial cells into mesenchyme. The targeted deletion of the Has2 gene results in embryonic death at approximately E9.5, with extensive abnormalities of the developing cardiovascular system. HA is virtually absent in Has2^–/– E9.5 embryos, demonstrating that Has2 is the principal source of HA during initial organogenesis in the mouse. Interestingly, the Has2^–/– embryo closely resembles that of heart defect (hdf) mice lacking the HA-binding proteoglycan versican, showing that a composite matrix requiring HA and versican is essential for forming cardiac jelly to support cushion morphogenesis and subsequent cardiac development. Our studies also reveal a striking role for HA in cell migration and transformation. The use of a well-established explant system revealed absent cell migration and epithelium-to-mesenchyme transformation in the atrioventricular (AV) canal from Has2^–/– embryos. Normal transformation is rescued by Has2 cDNA or the addition of purified HA. The requirement for HA in endothelial migration and transformation is abrogated by the expression of constitutively active H-Ras, and rescue by HA is inhibited by dominant-negative H-Ras. Thus, HA plays two complementary roles in development: expansion of the extracellular space, providing a hydrated matrix for cell migration; and stimulation of an invasive phenotype by cardiac endothelial cells.

Has2 expression during mouse development. Has2 mRNA is expressed at least from E7.5 through birth in the mouse (6). At day E8.5, Has2 mRNA is localized predominantly in the epithelium of the foregut diverticulum, cephalic mesenchyme, the allantois, and in the myocardium and endocardium of the heart (data not shown). At E9.5, the distribution of Has2 and versican mRNAs are very similar, with prominent expression in cephalic, foregut, and periaortic mesenchyme; the septum transversum; and the cardiovascular system (Figure 1, a–d). Has2 and versican mRNAs were present in atrial and ventricular endothelium. Versican mRNA was expressed in myocardium as well, whereas Has2 mRNA was less abundant, except in the myocardium of the AV canal region. In the heart at E9.5, cardiac jelly is rich in HA (Figure 1f) and versican (Figure 1g). By E10.5, endothelial cells in the AV canal and outflow tract transform into mesenchymal cells and invade the underlying matrix. The migrating cells express high levels of Has2 mRNA (Figure 1h) and stain strongly for extracellular HA (Figure 1f, inset). Later, Has2 mRNA is expressed by mesenchymal cells during elevation of the secondary palate and by hypertrophic chondrocytes within epiphysial growth plates (data not shown); this is consistent with a role for Has2-dependent HA synthesis in the expansion of soft tissues and in chondrogenesis (24). The relatively high level of expression of Has2 mRNA and its presence at sites of HA accumulation suggest that Has2 is a major source of HA during organogenesis.

Figure 1

Has2 and versican are expressed in similar domains in the E9.5 mouse. (a) Has2 expression. cm, cranial mesenchyme; fb, forebrain; mb, midbrain; fg, foregut diverticulum; 1, first branchial arch; ot, outflow tract; v, ventricle; a, atrium; st, region of developing septum transversum, including liver primordium and proepicardial organ; fm, foregut mesenchyme; am, periaortic mesenchyme; da, dorsal aorta. (b) Versican expression. (c and d) Higher-power views of the heart region boxed in a and b, respectively. Has2 and versican mRNAs are expressed in the endothelium (indicated by arrowheads) of the outflow tract (O) and myocardium (M) of the AV canal region. A, atrium; V, ventricle. (e and f) AV canal region and cardiac cushions (indicated by asterisks) of an E9.5 mouse embryo stained with hematoxylin and eosin (e) and for HA using a biotinylated HA-binding protein (f). E, endocardium. The boxed region contains endothelial cells that have transformed and are invading the underlying cushion tissue. These have abundant cell-surface HA (magnified in the inset). (g) Distribution of versican in the AV canal region, superimposing a Nomarski DIC image on a pseudocolored image of versican immunofluorescence. Note that the distribution of versican in g is similar to the distribution of HA in f. (h) Digital composite image of a Nomarski DIC image and a dark-field image of a ³⁵S-labeled in situ hybridization of Has2 mRNA in the AV canal and outflow tract region of an E10.5 heart. Mesenchymal cells within the AV canal and outflow tract cushions express abundant Has2 mRNA. The signal within the atrium results from light scattered by red blood cells, not from silver grains. Bars in a and b = 500 μm; bars in c–h = 100 μm.

Targeted inactivation of the Has2 gene. The replacement vector disrupts the coding sequence of Has2 by inserting a PGK-Neo cassette into the locus (Figure 2). This deleted a portion of intron 3 (including the splice acceptor) and the first 60 codons of exon 4, which encodes part of the cytoplasmic catalytic domain of Has2. The resultant allele encodes a truncated, enzymatically inactive protein. The replacement vector was introduced into the GK129 ES cells (25) and mice by standard methods (26). Mice heterozygous for the targeted allele were fertile and exhibited no obvious abnormalities. Heterozygous intercrosses yielded no viable offspring that were homozygous for the targeted allele (Has2^–/–), consistent with a lethal embryonic phenotype (Table 1). The first Has2^–/– embryos were identified at E11.5, having been partially reabsorbed. Occasional viable embryos were found at E10.5 with beating hearts and significant growth retardation. At day E9.5, the distribution of Has2^–/– embryos approaches mendelian frequency (Table 1). Analysis of embryonic mRNA by RT-PCR revealed only truncated transcripts of the expected length, with no full-length mRNA expressed in Has2^–/– embryos (data not shown). Although we have developed anti-peptide antibodies recognizing recombinant Has2 (6), we cannot detect Has2 protein by immunoblotting extracts of embryonic wild-type mice (data not shown). However, the truncated transcript interrupts the sequence encoding the intracytoplasmic catalytic domain. Thus, any protein product would lack HA synthase activity.

Table 1

Genotypes of mice resulting from matings of heterozygous Has2^+/– mice

The Has2^–/– embryos exhibited growth retardation and scant numbers of red blood cells, and lacked vitelline vessels in the yolk sac (Figure 3, a and b). The heart was thin-walled and relatively bloodless, and often exhibited marked pericardial swelling (Figure 3f). The visceral endoderm and mesoderm forming the yolk sac was not fused except at discrete foci (see insets in Figure 3a and 3b, showing wild-type and Has2^–/– sacs, respectively), giving the yolk sac of Has2^–/– embryos a characteristic punctate appearance (Figure 3b). Red blood cells were present within the space formed by the unfused visceral endoderm and mesoderm (Figure 3b). The hearts of viable Has2^–/– embryos beat with a distinctive to-and-fro motion of red blood cells. This is consistent with absent AV cushions as well as partial or complete obstruction of the outflow tract. Somites were present, albeit distorted, and other structures including the first pharyngeal pouch and otic placodes were also present (Figure 3, c and d). Some of the E9.5 Has2^–/– embryos had failed to turn, and exhibited posterior defects as well as cephalic mesenchyme abnormalities (data not shown). Whole-mount immunostaining for PECAM, a marker for endothelial cells, revealed a marked reduction in vessels in Has2^–/– embryos (compare Figure 3f with Figure 3e).

Figure 3

Abnormalities exhibited by Has2^–/– embryos. (a) Yolk sac of an Has2^+/– embryo. (b) Yolk sac from an Has2^–/– littermate. Cross-sections of the yolk sac stained with hematoxylin and eosin are shown in the inset. Note the presence of vitelline vessels (VV) containing nucleated red blood cells in the yolk sac of the Has2^+/– embryo. The endoderm and mesoderm are not fused in the Has2^–/– embryo, and the red blood cells are free within this space. (c and d) Representative wild-type and Has2^–/– embryos at E9.5. Note the diminished size, the bloodless heart, and distorted somites of the Has2^–/– embryo. (e and f) E9.5 wild-type and Has2^–/– embryos stained for the endothelial marker PECAM. Note the absence of an organized vascular network expressing PECAM in the Has2^–/– embryo. P, pericardium; E, endoderm; M, mesoderm; OpP, optic placode; OtP, otic placode; first and second pharyngeal pouches are numbered. Bars in c–f = 500 μm.

Has2 is required for HA production at E9.5 and for cardiac morphogenesis. Has2^–/– embryos have strikingly compacted extracellular space, and do not stain with alcian blue for acidic glycosaminoglycans (Figure 4, a and b). A specific biotinylated probe verified the absence of HA (Figure 4d). Quantitative analysis of pooled embryos using FACE analysis demonstrated a 96% reduction in HA in Has2^–/– embryos compared with wild-type littermates (Table 2). An image of a representative FACE analysis is shown in Figure 5. Wild-type littermates (Figure 5, lanes 2 and 3) clearly yield much higher levels of the fluorescent disaccharide ΔDiHA (which is specific to HA) than do the Has2^–/– embryos (Figure 5, lanes 4 and 5). Other extracellular-matrix components, including collagen I, fibronectin, laminin, and versican were all present in Has2^–/– embryos. However, their organization was altered in the compacted extracellular spaces (data not shown). The heart lacked cardiac jelly and cushions, although there was a characteristic constriction in the region of the AV canal (Figure 4b, arrows). The myocardium lacked trabeculae even though the ventricular wall was several cell layers thick. These observations confirmed a lack of HA production, compaction of the extracellular space, and lack of endocardial cushions and trabeculae in the hearts of Has2^–/– embryos.

Figure 4

E9.5 Has2^–/– embryos lack alcian blue–staining glycosaminoglycans and HA in cardiac jelly. The cardiac jelly of wild-type embryos is rich in acidic glycosaminoglycans (blue stain in a) and HA (brown stain in c). In contrast, Has2^–/– embryos totally lack alcian blue–stained material (b) and HA (d). The heart of the Has2^–/– embryo has a characteristic constriction at the AV canal region (indicated by the arrows), but no endocardial cushions, which are indicated by the asterisks in a and c. Bars in a and b = 100 μm; bars in c and d = 250 μm.

Figure 5

FACE analysis of E9.5 embryo extracts for HA. Lane 1 contains disaccharide standards. Lanes 2–5 represent 5% of a single embryo (lanes 2 and 3 are wild-type; lanes 4 and 5 are Has2^–/–). The arrow indicates the ΔDiHA disaccharide derived from HA. Note the marked reduction of the ΔDiHA band in the individual Has2^–/– samples compared with wild-type controls. Additional analyses of material pooled from wild-type or Has2^–/– embryos and run at higher concentrations (equivalent to 40–60% of a single embryo) gave similar results (data not shown; see Table 1).

Table 2

FACE analysis of ΔDiHA content of wild-type and Has2^–/– embryos_______Genotype ΔDiHA^A

Scanning electron microscopy revealed surface and internal detail of the hearts of wild-type and Has2^–/– embryos (Figure 6). In the wild-type heart at E9.5, the common atrium has moved dorsal and anterior to the ventricle. The left and right ventricles are becoming distinct, and the outflow tract is dividing into the aortic and pulmonary vessels by formation of the aortopulmonary septum (Figure 6b). The heart from an Has2^–/– embryo has a small atrium, a swollen pericardial space (Figure 6e), and marked reduction of the right ventricle and outflow tract (Figure 6, d–f). The wild-type heart has AV cushions and trabeculation of the ventricle myocardium, whereas the Has2^–/– embryo lacks endocardial cushions and has a compacted ventricle wall devoid of trabeculations (Figure 6, i and j).

Figure 6

Ultrastructure of wild-type and Has2^–/– E9.5 mouse hearts. Scanning electron micrograph of the external structure of the heart from a wild-type (a–c) and an Has2^–/– embryo (d–f). Specimens were viewed from the left side (a and d), the front (b and e), and the right side (c and f). Note the apparent absence of the presumptive right ventricle and outflow tract in the Has2^–/– embryo compared with the wild type. Scanning electron microscopy of cross-sections of hearts from wild-type (g and h) and Has2^–/– embryos (i and j) reveal a lack of AV cushions and a compacted ventricle wall lacking trabeculations (arrowheads). There is a constriction at the site of the AV canal in the Has2^–/– embryo. LV, left ventricle; RV, right ventricle; AoP septum, aortic pulmonary septum. Asterisks indicate left posterior atrial wall. h and j are higher magnifications of g and i, respectively.

The Has2^–/– phenotype closely resembles that of the hdf mouse (27), which lacks versican, an HA-binding proteoglycan (28). Examination of Has2 and versican mRNAs by in situ hybridization reveals that both are expressed in the heart (Figure 1, a–d). Moreover, the cardiac jelly is rich in HA and versican (Figure 1, f and g). Thus, the common phenotype resulting from disruption of the Has2 and versican genes demonstrates that both matrix molecules are essential for formation of cardiac jelly and endocardial cushions.

Transformation of cardiac endothelium to mesenchyme does not occur in the absence of Has2 and is restored by adding exogenous HA or by expressing Has2 cDNA. At 4–5 weeks gestation in the human, and by E9–10 in the mouse, the cardiac jelly expands rapidly in the AV canal and outflow tract, forming the cardiac cushions. Initially, the AV cushions function as valves ensuring unidirectional blood flow. Then, a subset of endothelial cells lining the cushions detaches from adjacent cells, migrates, and transforms into mesenchymal cells that invade and remodel the cushions into the tricuspid and mitral valves and the membranous portion of the interventricular septum. This process requires soluble signals released from the underlying myocardium interacting with receptor tyrosine kinases in the endothelium (29, 30). We could not assess AV canal formation in situ because of the absence of cushions in Has2^–/– embryos. However, in vitro collagen gels support endothelium-to-mesenchyme transformation of AV canal explants, duplicating the in situ events in AV canal morphogenesis. This system has been extensively validated and used to elucidate the signals regulating AV canal development (30–33). In vitro, this process is characterized by migration of endothelial (endocardial) cells as an epithelial sheet, followed by retraction and separation of the endothelium (“activation”). Finally, the activated endothelial cells undergo transformation, invading the underlying collagen gel and expressing a repertoire of mesenchymal genes de novo, such as α–smooth muscle actin. For clarity, we will refer to the entire process as AV canal morphogenesis, and will refer to each step with the terms “endothelial migration,” “activation,” and “transformation.” It is important to recognize that most published studies of in vitro AV canal morphogenesis use the chick. In the mouse, we find that formation of a distinct endothelial sheet occurs only in AV canal explants from embryos at E10–10.5. We studied E9.5 embryos to avoid artifacts associated with the morbidity of the Has2^–/– embryo. At this age, a definitive endothelial sheet is rarely seen (Camenisch et al., unpublished results).

In AV canal explants from E9.5 wild-type embryos, numerous endothelial cells migrated over the surface of the collagen gel, transformed, and invaded the gel matrix (Figure 7, a and b). The mesenchymal cells expressed characteristic mesenchymal markers, including fibulin-1 and fibulin-2 (data not shown) (34, 35). In contrast, endothelial-cell migration was absent in AV canal explants from Has2^–/– embryos (Figure 7d). The production of diffusible signaling molecules by the myocardium (33, 36, 37) and the responsiveness of the endothelial cells to signals are temporally and spatially regulated (29). Either (or both) could be deficient in Has2^–/– embryos. However, if migration or transformation were lacking solely due to the absence of HA, then providing exogenous HA or restoring Has2 function should correct the defect. To test this, we transfected Has2^–/– explants with Has2 cDNA. An empty control vector had no effect on migration or transformation in Has2^–/– explants (Figure 7d), whereas transfection with Has2 cDNA restored both migration and transformation (Figure 7e). Moreover, supplementing the culture medium (Figure 7f) or the collagen gel with HA (Figure 7g) also restored the normal phenotypic response in Has2^–/– AV explants. Interestingly, exogenous HA was fully effective at ng/mL concentrations, implicating a high-avidity interaction mediating its biological effects (data not shown). (Boiling the HA had no effect on its ability to restore AV canal morphogenesis in vitro.) Finally, conditioned medium from AV explant cultures from wild-type embryos also restored migration, epithelium-to-mesenchyme transformation, and invasion in Has2^–/– AV explants (data not shown). The transforming activity was abrogated by treatment with hyaluronidase (data not shown). The rescue of endothelial migration and transformation in AV explant cultures from Has2^–/– hearts demonstrates that neither the production of defined transforming activity by the myocardium (i.e., TGF-β) nor endothelial-cell competence is deficient in Has2^–/– embryos. The simplest interpretation is that HA is required for migration and possibly for subsequent transformation of the endothelial-cell population into mesenchyme. Notably, AV canal endothelial migration, transformation, and invasion is not affected in versican-deficient hdf mice, despite their lack of cardiac jelly (27).

Figure 7

AV canal morphogenesis is deficient in Has2^–/– AV canal explants and is restored by exogenous HA or activated Ras. Top panel: AV canal explant morphogenesis in vitro. AV canal explants from E9.5 wild-type (a–c) or Has2^–/– embryos (d–h) were cultured on collagen gels. Explants from wild-type embryos exhibit abundant endothelial cell migration and invasion (image a is focused on the surface of the gel; b is focused below the surface). In contrast, there is no endothelial-cell migration in AV canal explants from E9.5 Has2^–/– embryos (d). Transfection with dominant-negative (DN) Ras cDNA significantly (P < 0.001) reduces endothelial migration and invasion in AV explants from wild-type embryos (c). Because migration and invasion begins during the 16-hour incubation before transfection, the degree of inhibition is probably underestimated (see Methods). AV canal explants from Has2^–/– embryos exhibit comparable morphogenesis after transfection with Has2 cDNA (e), in the presence of HA in the media (f), or in the collagen gel (g). Transformation in Has2^–/– embryos is also rescued by transfecting with constitutively active Ras (h). M, myocardium. Nomarski DIC optics. Bar in h = 200 μm. Bottom panel: Quantification of AV canal transformation in the presence or absence of either dominant-negative Ras (S17N) or constitutively active Ras (Q61L). The dominant-negative Ras significantly inhibited cell migration and invasion in viable wild-type explants, whereas constitutively active Ras restored cell migration and invasion in AV explants from Has2^–/– embryos to the same degree as wild-type explants. The scoring method is outlined in Methods. ^AP < 0.001.

The requirement for HA in AV canal morphogenesis is removed by expressing constitutively active Ras, and HA-mediated AV canal morphogenesis in vitro requires Ras. Unregulated Ras activity leads to hyperproliferation of mesenchymal cells and AV cushion defects (30). We assessed the effect of expressing dominant-negative Ras in AV canal explants. Dominant-negative Ras significantly reduced the migration and transformation of endothelial cells in wild-type AV canal explant cultures, mimicking the Has2^–/– phenotype (Figure 7c). Thus, a pathway acting through Ras is important for transformation and invasion by cardiac endothelium. Next, we tested whether expressing constitutively active Ras could negate the requirement for HA in AV canal morphogenesis in explants from Has2^–/– embryos. Active Ras restored the normal phenotypic response in Has2^–/– AV explants to wild-type levels (Figure 7h). Thus, expression of activated Ras circumvents the requirement for HA in migration and epithelium-to-mesenchyme transformation. Finally, to determine if exogenous HA is acting via pathways involving Ras activation, we assessed the effect of inhibiting Ras on migration and transformation in Has2^–/– explants treated with HA. As shown in Figure 8, dominant-negative Ras significantly inhibited transformation mediated by HA in Has2^–/– explants. The endothelial cells adopted an epithelial phenotype and migrated over the collagen gel, but did not transform into mesenchyme or invade the collagen gel. This result suggests that a pathway requiring HA and leading to Ras activation is essential for endothelial transformation during AV canal morphogenesis. Endothelial migration, although dependent upon HA, occurs irrespective of Ras activity.

Figure 8

Dominant-negative Ras inhibits HA-mediated endothelial-cell invasion in Has2^–/– AV canal explants. These images were obtained by laser scanning confocal microscopy after staining for α−smooth muscle actin. Exogenous HA (0.75 mg/mL of medium) was added to the culture medium in both Has2^–/– explants. (a) A collapsed Z series of 100 μm showing the characteristic transformation to mesenchyme and invasion of the collagen gel in the presence of exogenous HA. The dotted line indicates the previous location of the myocardium, which was removed. (b) Characteristic effect of transfection with dominant-negative Ras. In contrast to the rescued AV explant, an epithelial sheet has migrated over the surface of the collagen gel, but there are no invading mesenchymal cells. Similar results were obtained in three independent experiments.

Has2 is the major source of HA during embryogenesis, and HA is essential in cardiac organogenesis. The lethality in Has2^–/– embryos underscores the critical role of HA in cardiovascular morphogenesis and demonstrates that Has2 is the major source of HA during this period of embryonic development. A functional cardiovascular system is required to sustain life beyond E9.5–10 (38). HA appears at E5.5 in the yolk cavity, and by E6.5 is present in all embryonic cavities, the embryonic and extraembryonic mesoderm, and the allantois (16, 17). Either another Has gene (i.e., Has1 or Has3) must function earlier in development to replace Has2 function, or HA is dispensable before cardiac morphogenesis; this is currently being evaluated. Phenotypically, we cannot distinguish Has2^–/– embryos earlier than approximately E8.5. Targeting of the other Has genes has been accomplished, and Has1^–/– and Has3^–/– animals are viable and fertile (unpublished data). Furthermore, compound Has1, Has3 homozygous-null mice are viable and fertile, underscoring the pivotal role of Has2 during embryogenesis. We are currently creating Has1^–/–, Has2^+/–, Has3^–/– animals to assess whether haploinsufficiency occurs with only one functional Has2 allele.

There are profound defects in vasculogenesis in the Has2^–/– embryos involving the yolk sac and systemic vasculature. Similar abnormalities are seen in the yolk sacs of mice lacking molecules that mediate cell-matrix and cell-cell interactions, including α5-integrin (39) and N-cadherin (40). Interestingly, yolk sac pathology is seen in mice that lack N-Ras function and are heterozygous for a null mutation in K-ras (41). These abnormalities can also be associated with pericardial edema (38), but we believe that the vascular abnormalities result from the absence of HA, because despite near-identical cardiac phenotypes, hdf mice do not have yolk sac defects (Corey Mjaatvedt, personal communication). The early lethality of Has2^–/– embryos makes it difficult to directly assess the role of Has2 in the development of other embryonic structures. This will require more sophisticated in vivo genetic approaches, such as generation of chimeric animals or tetraploid rescue.

A composite matrix is required for expansion of the extracellular space during cardiac cushion formation. Abnormal formation of the endocardial cushions results in AV septal defects, the second most common cardiac defect in infants (42). The similar phenotype of HA-deficient Has2^–/– and versican-deficient hdf mice reveals for the first time the critical importance of a composite matrix containing HA in AV canal morphogenesis. Clearly, neither molecule alone is sufficient for this process. HA interacts with other matrix molecules, including inter-α-trypsin inhibitor (43), link protein (44), and fibulins (45). Fibulin is present in cardiac cushions (34), and binds both HA and versican (45). These interactions, in conjunction with our observations, underscore the diversity and potential importance of composite matrices in the development of the AV cushions. The expression of endocardial cushion matrix components in the AV canal is probably controlled by activity of upstream transcription factors implicated in cardiac morphogenesis. Potential candidates include the basic HLH transcription factor Hand2, which is required for proper morphogenesis of the right ventricle and outflow tract/conus (46, 47), and epicardin, another basic HLH transcription factor (48). We are currently analyzing the Has2 gene locus for regulatory elements that are responsible for Has2 expression in the developing heart.

HA is required for endothelial-cell migration and transformation in AV canal morphogenesis. A striking and unexpected outcome of our studies is the requirement for HA in cell migration, transformation, and subsequent invasion. Whereas previous studies have implicated HA in cardiac development (31, 49, 50), our investigation demonstrates an absolute requirement for HA during AV canal morphogenesis in vivo and in vitro. This is the first report of a developmental system in which cell migration and transformation have an absolute dependence on HA. The fact that dominant-negative Ras inhibits endothelial cell migration and transformation in normal AV explants confirms the requirement for Ras activity during AV morphogenesis (30). Moreover, constitutively active Ras circumvents the requirement for HA in endothelial-cell transformation in Has2^–/– embryo explants. Finally, we show that the ability of HA to rescue normal endothelial transformation to mesenchyme in Has2^–/– AV canals is inhibited by dominant-negative Ras. This indicates that the HA induction converges on Ras activation during transformation of these cardiac endothelial cells. Interestingly, treatment of Has2^–/– AV explants with HA in the presence of dominant-negative Ras supports endothelial cell migration, but not transformation (Figure 8). Therefore, endothelial cell migration is independent of active Ras signaling, but is still dependent on HA. Therefore, two sequential steps require HA: migration of endothelial cells, which may not require Ras, followed by their transformation into mesenchyme, which appears to require Ras signaling.

What is the mechanism through which HA acts? One possibility is that HA binding to a cognate receptor, e.g., CD44, activates downstream pathways that converge on Ras activation. Although CD44-deficient mice are not reported to exhibit any cardiovascular phenotype (51), we have observed retarded endothelial-cell transformation in vitro with AV explants from CD44^–/– embryos (data not shown), as well as partial inhibition of transformation using a monoclonal antibody blocking CD44-HA binding (52). Additionally, HA may serve as a costimulatory molecule acting in concert with known receptor-mediated signaling pathways. This is a well accepted paradigm for integrin matrix receptors, which act synergistically with growth factor receptors (53–56). Given the ubiquitous distribution of HA in cardiac jelly and the sharply restricted domain of endothelial-cell transformation, we favor the latter hypothesis. Provocatively, mice with null mutations in either an ErbB ligand, heregulin-1 (neuregulin-1), or in the cognate receptors ErbB3 and ErbB4 exhibit cardiac phenotypes resembling those of the Has2^–/– and hdf embryos (57–59). Heregulin-1 is expressed in AV canal endothelial cells; ErbB3 is expressed in the subjacent AV canal mesenchyme (57). Binding of HA to CD44 activates Ras cooperatively with members of the receptor tyrosine kinase ErbB family (21, 60), and HA activates mitogen-activated protein kinase via Ras (61). HA binding to CD44 also activates Rac1 and promotes formation of lamellipodia in mouse epithelial cells (62). Collectively, these observations suggest that an HA-dependent, receptor-mediated signaling event involving HA, CD44, and ErbB receptors/ligands may be essential for AV canal morphogenesis. Collectively, our data indicate that HA acts by stimulating intracellular signaling pathways, including Ras, and that additional or alternative receptors for HA may compensate for CD44.

Our results demonstrate that HA plays two distinct and complementary roles during AV canal morphogenesis. First, HA is a necessary component of the extracellular supramolecular array expanding the extracellular matrix, a requisite for formation of cardiac jelly and the endocardial cushions. Second, although the cognate receptor (or receptors) must be defined, HA acting in concert with additional factors stimulates the Ras-dependent intracellular signaling that is necessary for endothelial cell transformation to form mesenchymalized cushions. These observations reveal a novel mechanism in which an extracellular matrix component, HA, acts through a receptor complex to induce Ras-dependent signaling and cell transformation. It remains to be seen whether HA plays similar, dual roles in the development of other embryonic structures and in tumor progression by expanding the extracellular space and inducing cell migration and transformation.

Marv Ruona and Julie Jensen provided graphics support. Judy Bradley, Suresh Savarirayan, and the transgenic core staff assisted with transgenics, and M. Anita Jennings provided histology services. We would also like to acknowledge helpful discussions with Ray Runyan, Joyce A. Schroeder, and Maria Valencik. Vincent Hascall and Maria Benavides at the Cleveland Clinic assisted with the FACE analysis. This work was supported by NIH grants 1RO1 AR-44689 and T32 HL-07897, the Desert Mountain Affiliate of the American Heart Association, and the Mayo Foundation for Medical Education and Research.

Andrew P. Spicer’s present address is: Rowe Program in Genetics, Department of Biological Chemistry, University of California, Davis, California, USA.

Todd D. Camenisch and Andrew P. Spicer contributed equally to this work.

1. Philipson, LH, Westley, J, Schwartz, NB. Effect of hyaluronidase treatment of intact cells on hyaluronate synthetase activity. Biochemistry 1985. 24:7899-7906.
   View this article via: PubMed CrossRef Google Scholar
2. Ng, KF, Schwartz, NB. Solubilization and partial purification of hyaluronate synthetase from oligodendroglioma cells. J Biol Chem 1989. 264:11776-11783.
   View this article via: PubMed Google Scholar
3. Watanabe, K, Yamaguchi, Y. Molecular identification of a putative human hyaluronan synthase. J Biol Chem 1996. 271:22945-22948.
   View this article via: PubMed CrossRef Google Scholar
4. Itano, N, Kimata, K. Expression cloning and molecular characterization of HAS protein, a eukaryotic hyaluronan synthase. J Biol Chem 1996. 271:9875-9878.
   View this article via: PubMed CrossRef Google Scholar
5. Shyjan, AM, Heldin, P, Butcher, EC, Yoshino, T, Briskin, MJ. Functional cloning of the cDNA for a human hyaluronan synthase. J Biol Chem 1996. 271:23395-23399.
   View this article via: PubMed CrossRef Google Scholar
6. Spicer, AP, McDonald, JA. Characterization and molecular evolution of a vertebrate hyaluronan synthase gene family. J Biol Chem 1998. 273:1923-1932.
   View this article via: PubMed CrossRef Google Scholar
7. Spicer, AP, et al. Chromosomal localization of the human and mouse hyaluronan synthase genes. Genomics 1997. 41:493-497.
   View this article via: PubMed CrossRef Google Scholar
8. Spicer, AP, Olson, J, McDonald, JA. Molecular cloning of a gene encoding a third mammalian hyaluronan synthase, Has3. J Biol Chem 1997. 272:8957-8961.
   View this article via: PubMed CrossRef Google Scholar
9. Spicer, AP, Augustine, ML, McDonald, JA. Molecular cloning and characterization of a putative mouse hyaluronan synthase. J Biol Chem 1996. 271:23400-23406.
   View this article via: PubMed CrossRef Google Scholar
10. Itano, N, et al. Three isoforms of mammalian hyaluronan synthases have distinct enzymatic properties. J Biol Chem 1999. 274:25085-25092.
    View this article via: PubMed CrossRef Google Scholar
11. Scott, JE, Heatley, F. Hyaluronan forms specific stable tertiary structures in aqueous solution: a 13C NMR study. Proc Natl Acad Sci USA 1999. 96:4850-4855.
    View this article via: PubMed CrossRef Google Scholar
12. Haddon, CM, Lewis, JH. Hyaluronan as a propellant for epithelial movement: the development of semicircular canals in the inner ear of Xenopus. Development 1991. 112:541-550.
    View this article via: PubMed Google Scholar
13. Camaioni, A, Salustri, A, Yanagishita, M, Hascall, VC. Proteoglycans and proteins in the extracellular matrix of mouse cumulus cell-oocyte complexes. Arch Biochem Biophys 1996. 325:190-198.
    View this article via: PubMed CrossRef Google Scholar
14. Tirone, E, D’Alessandris, C, Hascall, VC, Siracusa, G, Salustri, A. Hyaluronan synthesis by mouse cumulus cells is regulated by interactions between follicle-stimulating hormone (or epidermal growth factor) and a soluble oocyte factor (or transforming growth factor beta1). J Biol Chem 1997. 272:4787-4794.
    View this article via: PubMed CrossRef Google Scholar
15. Kohda, D, et al. Solution structure of the link module: a hyaluronan-binding domain involved in extracellular matrix stability and cell migration. Cell 1996. 86:767-775.
    View this article via: PubMed CrossRef Google Scholar
16. Brown, JJ, Papaioannou, VE. Ontogeny of hyaluronan secretion during early mouse development. Development 1993. 117:483-492.
    View this article via: PubMed Google Scholar
17. Fenderson, BA, Stamenkovic, I, Aruffo, A. Localization of hyaluronan in mouse embryos during implantation, gastrulation and organogenesis. Differentiation 1993. 54:85-98.
    View this article via: PubMed Google Scholar
18. Watanabe, H, et al. Mouse cartilage matrix deficiency (cmd) caused by a 7 bp deletion in the aggrecan gene. Nat Genet 1994. 7:154-157.
    View this article via: PubMed CrossRef Google Scholar
19. Kosaki, R, Watanabe, K, Yamaguchi, Y. Overproduction of hyaluronan by expression of the hyaluronan synthase Has2 enhances anchorage-independent growth and tumorigenicity. Cancer Res 1999. 59:1141-1145.
    View this article via: PubMed Google Scholar
20. Itano, N, Sawai, T, Miyaishi, O, Kimata, K. Relationship between hyaluronan production and metastatic potential of mouse mammary carcinoma cells. Cancer Res 1999. 59:2499-2504.
    View this article via: PubMed Google Scholar
21. Bourguignon, LY, et al. Interaction between the adhesion receptor, CD44, and the oncogene product, p185HER2, promotes human ovarian tumor cell activation. J Biol Chem 1997. 272:27913-27918.
    View this article via: PubMed CrossRef Google Scholar
22. Banerji, S, et al. LYVE-1, a new homologue of the CD44 glycoprotein, is a lymph-specific receptor for hyaluronan. J Cell Biol 1999. 144:789-801.
    View this article via: PubMed CrossRef Google Scholar
23. McKee, CM, et al. Hyaluronan fragments induce nitric-oxide synthase in murine macrophages through a nuclear factor kappaB-dependent mechanism. J Biol Chem 1997. 272:8013-8018.
    View this article via: PubMed CrossRef Google Scholar
24. Nishida, Y, Knudson, CB, Nietfeld, JJ, Margulis, A, Knudson, W. Antisense inhibition of hyaluronan synthase-2 in human articular chondrocytes inhibits proteoglycan retention and matrix assembly. J Biol Chem 1999. 274:21893-21899.
    View this article via: PubMed CrossRef Google Scholar
25. Philpott, KL, et al. Lymphoid development in mice congenitally lacking T cell receptor alpha beta-expressing cells. Science 1992. 256:1448-1452.
    View this article via: PubMed CrossRef Google Scholar
26. George, EL, Hynes, RO. Gene targeting and generation of mutant mice for studies of cell-extracellular matrix interactions. Methods Enzymol 1994. 245:386-420.
    View this article via: PubMed Google Scholar
27. Yamamura, H, Zhang, M, Markwald, RR, Mjaatvedt, CH. A heart segmental defect in the anterior-posterior axis of a transgenic mutant mouse. Dev Biol 1997. 186:58-72.
    View this article via: PubMed Google Scholar
28. Mjaatvedt, CH, Yamamura, H, Capehart, AA, Turner, D, Markwald, RR. The Cspg2 gene, disrupted in the hdf mutant, is required for right cardiac chamber and endocardial cushion formation. Dev Biol 1998. 202:56-66.
    View this article via: PubMed CrossRef Google Scholar
29. Eisenberg, LM, Markwald, RR. Molecular regulation of atrioventricular valvuloseptal morphogenesis. Circ Res 1995. 77:1-6.
    View this article via: PubMed Google Scholar
30. Lakkis, MM, Epstein, JA. Neurofibromin modulation of ras activity is required for normal endocardial-mesenchymal transformation in the developing heart. Development 1998. 125:4359-4367.
    View this article via: PubMed Google Scholar
31. Bernanke, DH, Markwald, RR. Effects of hyaluronic acid on cardiac cushion tissue cells in collagen matrix cultures. Tex Rep Biol Med 1979. 39:271-285.
    View this article via: PubMed Google Scholar
32. Runyan, RB, Markwald, RR. Invasion of mesenchyme into three-dimensional collagen gels: a regional and temporal analysis of interaction in embryonic heart tissue. Dev Biol 1983. 95:108-114.
    View this article via: PubMed CrossRef Google Scholar
33. Brown, CB, Boyer, AS, Runyan, RB, Barnett, JV. Requirement of type III TGF-beta receptor for endocardial cell transformation in the heart. Science 1999. 283:2080-2082.
    View this article via: PubMed CrossRef Google Scholar
34. Bouchey, D, Argraves, WS, Little, CD. Fibulin-1, vitronectin, and fibronectin expression during avian cardiac valve and septa development. Anat Rec 1996. 244:540-551.
    View this article via: PubMed CrossRef Google Scholar
35. Miosge, N, Sasaki, T, Chu, ML, Herken, R, Timpl, R. Ultrastructural localization of microfibrillar fibulin-1 and fibulin-2 during heart development indicates a switch in molecular associations. Cell Mol Life Sci 1998. 54:606-613.
    View this article via: PubMed CrossRef Google Scholar
36. Boyer, AS, et al. TGFbeta2 and TGFbeta3 have separate and sequential activities during epithelial-mesenchymal cell transformation in the embryonic heart. Dev Biol 1999. 208:530-545.
    View this article via: PubMed CrossRef Google Scholar
37. Boyer, AS, Erickson, CP, Runyan, RB. Epithelial-mesenchymal transformation in the embryonic heart is mediated through distinct pertussis toxin-sensitive and TGFbeta signal transduction mechanisms. Dev Dyn 1999. 214:81-91.
    View this article via: PubMed CrossRef Google Scholar
38. Copp, AJ. Death before birth: clues from gene knockouts and mutations. Trends Genet 1995. 11:87-93.
    View this article via: PubMed CrossRef Google Scholar
39. Yang, JT, Rayburn, H, Hynes, RO. Embryonic mesodermal defects in alpha 5 integrin-deficient mice. Development 1993. 119:1093-1105.
    View this article via: PubMed Google Scholar
40. Radice, GL, et al. Developmental defects in mouse embryos lacking N-cadherin. Dev Biol 1997. 181:64-78.
    View this article via: PubMed CrossRef Google Scholar
41. Johnson, L, et al. K-ras is an essential gene in the mouse with partial functional overlap with N-ras [erratum 1997, 11:3277]. Genes Dev 1997. 11:2468-2481.
    View this article via: PubMed Google Scholar
42. Rubin, EM, Ishida, BY, Clift, SM, Krauss, RM. Expression of human apolipoprotein A-I in transgenic mice results in reduced plasma levels of murine apolipoprotein A-I and the appearance of two new high density lipoprotein size subclasses. Proc Natl Acad Sci USA 1991. 88:434-438.
    View this article via: PubMed CrossRef Google Scholar
43. Bost, F, Diarra-Mehrpour, M, Martin, JP. Inter-alpha-trypsin inhibitor proteoglycan family: a group of proteins binding and stabilizing the extracellular matrix. Eur J Biochem 1998. 252:339-346.
    View this article via: PubMed CrossRef Google Scholar
44. Binette, F, Cravens, J, Kahoussi, B, Haudenschild, DR, Goetinck, PF. Link protein is ubiquitously expressed in non-cartilaginous tissues where it enhances and stabilizes the interaction of proteoglycans with hyaluronic acid. J Biol Chem 1994. 269:19116-19122.
    View this article via: PubMed Google Scholar
45. Aspberg, A, Adam, S, Kostka, G, Timpl, R, Heinegard, D. Fibulin-1 is a ligand for the C-type lectin domains of aggrecan and versican. J Biol Chem 1999. 274:20444-20449.
    View this article via: PubMed CrossRef Google Scholar
46. Thomas, T, Yamagishi, H, Overbeek, PA, Olson, EN, Srivastava, D. The bHLH factors, dHAND and eHAND, specify pulmonary and systemic cardiac ventricles independent of left-right sidedness. Dev Biol 1998. 196:228-236.
    View this article via: PubMed CrossRef Google Scholar
47. Srivastava, D, et al. Regulation of cardiac mesodermal and neural crest development by the bHLH transcription factor, dHAND [erratum 1997, 16:410]. Nat Genet 1997. 16:154-160.
    View this article via: PubMed CrossRef Google Scholar
48. Robb, L, et al. Epicardin: a novel basic helix-loop-helix transcription factor gene expressed in epicardium, branchial arch myoblasts, and mesenchyme of developing lung, gut, kidney, and gonads. Dev Dyn 1998. 213:105-113.
    View this article via: PubMed CrossRef Google Scholar
49. Krug, EL, Runyan, RB, Markwald, RR. Protein extracts from early embryonic hearts initiate cardiac endothelial cytodifferentiation. Dev Biol 1985. 112:414-426.
    View this article via: PubMed CrossRef Google Scholar
50. Baldwin, HS, Lloyd, TR, Solursh, M. Hyaluronate degradation affects ventricular function of the early postlooped embryonic rat heart in situ. Circ Res 1994. 74:244-252.
    View this article via: PubMed Google Scholar
51. Schmits, R, et al. CD44 regulates hematopoietic progenitor distribution, granuloma formation, and tumorigenicity. Blood 1997. 90:2217-2233.
    View this article via: PubMed Google Scholar
52. Miyake, K, Underhill, CB, Lesley, J, Kincade, PW. Hyaluronate can function as a cell adhesion molecule and CD44 participates in hyaluronate recognition. J Exp Med 1990. 172:69-75.
    View this article via: PubMed CrossRef Google Scholar
53. Rosales, C, O’Brien, V, Kornberg, L, Juliano, R. Signal transduction by cell adhesion receptors. Biochim Biophys Acta 1995. 1242:77-98.
    View this article via: PubMed Google Scholar
54. Lin, TH, Chen, Q, Howe, A, Juliano, RL. Cell anchorage permits efficient signal transduction between ras and its downstream kinases. J Biol Chem 1997. 272:8849-8852.
    View this article via: PubMed CrossRef Google Scholar
55. Howe, AK, Juliano, RL. Distinct mechanisms mediate the initial and sustained phases of integrin-mediated activation of the Raf/MEK/mitogen-activated protein kinase cascade. J Biol Chem 1998. 273:27268-27274.
    View this article via: PubMed CrossRef Google Scholar
56. Schwartz, MA, Baron, V. Interactions between mitogenic stimuli, or, a thousand and one connections. Curr Opin Cell Biol 1999. 11:197-202.
    View this article via: PubMed CrossRef Google Scholar
57. Meyer, D, Birchmeier, C. Multiple essential functions of neuregulin in development [erratum 1995, 378:753]. Nature 1995. 378:386-390.
    View this article via: PubMed CrossRef Google Scholar
58. Erickson, SL, et al. ErbB3 is required for normal cerebellar and cardiac development: a comparison with ErbB2-and heregulin-deficient mice. Development 1997. 124:4999-5011.
    View this article via: PubMed Google Scholar
59. Gassmann, M, et al. Aberrant neural and cardiac development in mice lacking the ErbB4 neuregulin receptor. Nature 1995. 378:390-394.
    View this article via: PubMed CrossRef Google Scholar
60. Ilangumaran, S, Briol, A, Hoessli, DC. CD44 selectively associates with active Src family protein tyrosine kinases Lck and Fyn in glycosphingolipid-rich plasma membrane domains of human peripheral blood lymphocytes. Blood 1998. 91:3901-3908.
    View this article via: PubMed Google Scholar
61. Serbulea, M, et al. Hyaluronan activates mitogen-activated protein kinase via Ras-signaling pathway. Int J Oncol 1999. 14:733-738.
    View this article via: PubMed Google Scholar
62. Oliferenko, S, Kaverina, I, Small, JV, Huber, LA. Hyaluronic acid (HA) binding to CD44 activates Rac1 and induces lamellipodia outgrowth. J Cell Biol 2000. 148:1159-1164.
    View this article via: PubMed CrossRef Google Scholar
63. Broekelmann, TJ, Limper, AH, Colby, TV, McDonald, JA. Transforming growth factor beta 1 is present at sites of extracellular matrix gene expression in human pulmonary fibrosis. Proc Natl Acad Sci USA 1991. 88:6642-6646.
    View this article via: PubMed CrossRef Google Scholar
64. Baraniuk, JN, Shizari, T, Sabol, M, Ali, M, Underhill, CB. Hyaluronan is exocytosed from serous, but not mucous cells, of human nasal and tracheobronchial submucosal glands. J Investig Med 1996. 44:47-52.
    View this article via: PubMed Google Scholar
65. Seo, JW, Kim, EK, Brown, NA, Wessels, A. Section directed cryosectioning of specimens for scanning electron microscopy: a new method to study cardiac development. Microsc Res Tech 1995. 30:491-495.
    View this article via: PubMed CrossRef Google Scholar
66. Spicer, AP, Rowse, GJ, Lidner, TK, Gendler, SJ. Delayed mammary tumor progression in Muc-1 null mice. J Biol Chem 1995. 270:30093-30101.
    View this article via: PubMed CrossRef Google Scholar
67. van den Hoff, MJ, et al. Myocardialization of the cardiac outflow tract. Dev Biol 1999. 212:477-490.
    View this article via: PubMed CrossRef Google Scholar
68. Allgood, VE, Zhang, Y, O’Malley, BW, Weigel, NL. Analysis of chicken progesterone receptor function and phosphorylation using an adenovirus-mediated procedure for high-efficiency DNA transfer. Biochemistry 1997. 36:224-232.
    View this article via: PubMed CrossRef Google Scholar
69. Calabro, A, Hascall, VC, Midura, RJ. Adaptation of FACE methodology for microanalysis of total hyaluronan and chondroitin sulfate composition from cartilage. Glycobiology 2000. 10:283-293.
    View this article via: PubMed CrossRef Google Scholar
70. Calabro, A, Benavides, M, Tammi, M, Hascall, VC, Midura, RJ. Microanalysis of enzyme digests of hyaluronan and chondroitin/dermatan sulfate by fluorophore-assisted carbohydrate electrophoresis (FACE). Glycobiology 2000. 10:273-281.
    View this article via: PubMed CrossRef Google Scholar