Results

Persistent truncus arteriosus and abnormal large vessels in mice lacking Alk5 in cardiac NCCs
To inactivate Alk5 in cardiac NCCs, mice homozygous for the floxed Alk5 allele (Alk5Flox/Flox) [21] were crossed with transgenic Wnt1-Cre mice [22], which were also heterozygous for the Alk5 knockout allele (Alk5KO) allele. The resulting mice heterozygous for the Alk5Flox and Alk5KO alleles, which also carried the Wnt1-Cre transgene, had the Alk5 gene specifically inactivated in NCCs (herein termed Alk5/Wnt1-Cre), while the littermates with remaining allelic combinations were phenotypically normal and served as controls (Alk5Flox/+, Alk5KO/+; Wnt1-Cre). When embryos were harvested during the last day of gestation, an expected number (25%) of Alk5/Wnt1-Cre mutants were recovered. However, all mutant offspring died either during the birth or during the first post-natal hours.
To determine, if ALK5-mediated TGF-β-signaling had a role in development of the OFT and large vessels of the aortic arch, we performed casting dye experiments on E17 embryos (Fig. 1A–D). In wild-type embryos (Fig. 1A), the aorta was clearly separated from the pulmonary trunk, and the right brachiocephalic, left carotid and left subclavian arteries branched directly off the aortic arch. In contrast, Alk5/Wnt1Cre mutant embryos consistently displayed a single prominent arterial trunk (Fig. 1C–D), while corresponding Tgfbr2 mutant embryos (Fig. 1B) displayed interrupted aortic arch, as reported earlier [8]. Approximately 40% of the Alk5 mutants had a right-sided outflow tract, with the retroesophageal arch connecting to the descending aorta and to the left subclavian artery. The carotid arteries originated either from a common bud located in the ventral side of the ascending arch, or from separate adjacent sites, as verified by serial sectioning (Fig. 1M–P). The remaining mutants displayed a left-sided aortic arch, where the right carotid arteries originated from the right lateral aspect of the ascending trunk, while the left carotid arteries budded from the ventral or right ventral aspects of the trunk (Fig. 1I–L). Both right and left subclavian arteries consistently originated from the descending part of the aortic arch. Similarly, in all mutants both left and right pulmonary arteries always branched out from the common arterial trunk. To conclude, Alk5/Wnt1-Cre mutants consistently displayed PTA, which differed significantly from the characteristic interrupted aortic arch phenotype seen in Tgfbr2/Wnt1-Cre mutants [8,9].
Figure 1  Abrogation of Alk5 in neural crest cells leads to persistent truncus arteriosus type A2. A-D, Casting-dye analysis of OFT morphogenesis at E17.0. Control (A), Tgfbr2/Wnt1-Cre mutant [8] (B) demonstrating the PTA type A4 (= truncus arteriosus with interrupted aortic arch [30]) and Alk5/Wnt1-Cre mutants demonstrating the right-sided (C) and left-sided (D) arches of the truncus. E-P, Histological cross-sections on four different levels (rostral to caudal) at E17.0. In a control (E-H), the ascending aorta (Ao) and pulmonary trunk (PT) are separated by the conotruncal septum. In Alk5/Wnt1-Cre mutants (I-P) the conotruncal septum fails to form, and either left-sided (I-L) or right-sided (M-P) aortic arch can be seen. Aberrant branching of carotid arteries from the truncus has been illustrated by black arrows (J and M). Ao, aorta; PT, pulmonary trunk; RSA, right subclavian artery; RCA, right carotid artery; LCA; left carotid artery; LSA, left subclavian artery; IAA, interrupted aortic arch; PTA, persistent truncus arteriosus.

Abnormal patterning of the pharyngeal arch arteries and aortic sac in Alk5/Wnt1Cre mutants
During cardiovascular development, the PAAs undergo a complex set of sequential asymmetric remodeling steps resulting in the left-sided aortic arch. To determine, whether ALK5-mediated signaling was involved in remodeling of PAAs, we performed intracardiac India ink injections at different developmental stages. While at E10, Alk5/Wnt1-Cre mutants did not show obvious differences in the PAAs, abnormal remodeling became obvious in mutants a day later at E11 (Fig. 2). The controls displayed the well-formed 3rd, 4th and 6th PAAs. Moreover, the carotid duct (the dorsal aorta between the 3rd and 4th PAAs) was already regressing as demonstrated by the reduced size (Fig. 2A). In Alk5/Wnt1-Cre mutants, the 3rd and 4th pairs of PAAs were bilaterally hypoplastic, whereas the 6th pair of PAAs was notably hyperplastic (Fig. 2B). Furthermore, the carotid duct was remarkably large, when compared to controls. While the controls displayed an interruption of the carotid duct at E12 and E13 as expected (Fig. 2C), the mutants demonstrated an uncharacteristic break of the dorsal aorta between the 4th and 6th pairs of PAAs (Fig. 2D).
Figure 2  Abnormal patterning of the PAAs in Alk5/Wnt1-Cre mutants. Left lateral view after intracardiac ink injections to visualize the developing PAAs at E11.0 (A,B), E12.0 (C, D) and E13.0 (E, F) in controls (A, C, E) and Alk5/Wnt1-Cre mutants (B, D, F). Arrow in A points to the regressing carotid duct. Asterisk in B depicts the corresponding structure in the mutant with no signs of regression. Asterisk in D illustrates the aberrant regression of the dorsal aorta between the 4th and 6th PAAs. PT, pulmonary trunk; Ao, Aorta; TA, truncus arteriosus. Around E11.5, the aortic sac normally forms a distinctive T-shaped structure, as seen in frontal sections of the control sample in Fig. 3(A,C). Subsequently, the right horn of this structure transforms into the prospective brachiocephalic artery, while the left horn together with the left 4th PAA gives rise to the definitive aortic arch [23]. In Alk5/Wnt1-Cre mutants, the T-shaped aortic sac failed to form (Fig. 3B,D). Instead, the truncus bifurcated to a left and right arm, which further branched to the PAAs, particularly to the predominant pair of 6th PAAs (Fig. 3B,D). The observed phenotype is consistent with the absence or severe hypoplasia of structures derived from the aortic sac in late stage embryos (E17), e.g., the missing brachiocephalic artery and severe shortening of the ascending truncus as shown in the Figure 1.
Figure 3  Abnormal Aortic Sac in Alk5/Wnt1-Cre mutants. Alk5/Wnt1-Cre mutants (B, D) fail to form the typical T-shaped structure of the aortic sac seen in controls at E11.5. (A, C). A-B, frontal image of ink-injected embryos; C-D, frontal sections (H&E staining). Arrows in A and B point to the level of section shown in C and D (red arrows in C and D point to the aortic sac of the control and mutant, respectively).

Cardiac NCCs deficient in Alk5 can populate the outflow tract
Next we used the R26R lineage-tracing assay to determine whether CNCCs could appropriately populate the outflow tract region. Briefly, Alk5Flox/Flox mice were crossed with the ROSA26 Cre reporter mice, and subsequently Alk5Flox/Flox;R26R(+/+) females were crossed with Alk5KO/WT;Wnt1-Cre males. The resulting embryos had the NC-lineage permanently labeled with β-galactosidase expression, and displayed identical phenotypes to those obtained without the R26R reporter. Staining of embryos for β-galactosidase at E8-E11 did not reveal detectable differences in NCC migration between mutants and controls (data not shown). Similarly, serial transverse sectioning of whole mount embryos (E10-E12) and subsequent analysis of positively stained cells in the OFT region demonstrated that CNCCs deficient in Alk5 were capable of populating the PAAs, aortic sac and conotruncal ridges at a level comparable to that of controls (Fig 4). To conclude, the observed phenotypes in Alk5/Wnt1-Cre mutants were certainly not due to defective migration of CNCCs to the pharyngeal and outflow tract regions.
Figure 4  Normal cardiac NCC migration in Alk5/Wnt1-Cre mutants. The OFT of controls (A, C, E) and Alk5/Wnt1-Cre mutants (B, D, F) display similar staining patterns when analyzed using the R26R lineage tracing assay at E11.0. A-B, whole mount staining (left lateral image); C-F, transverse sections on the level of the 4th (C, D) and 6th (E, F) PAAs. Arrows (A-F) point to the most proximal location staining positive for the β-galactosidase activity.

Aortic sac and aortico-pulmonary septal defects in Alk5/Wnt1Cre mutant embryos
Septation of the outflow tract lumen begins in a cranial-to-caudal direction, starting distally in the aortic sac and proceeding toward the heart [24]. Initially, the condensed mesenchyme derived from the NC forms in the base of the aortic sac between the origins of 4th and 6th PAAs. Subsequently, two prongs of the developing aortico-pulmonary (AP) septum extend into the truncal cushions and the aortico-pulmonary septation complex crosses the aortic sac cranially. In ink-injected control embryos at E11.5, a characteristic conotruncal transition separating the truncus and conus could be seen as a twisted configuration, resulting from a change in orientation of the truncal and conal cushions (Fig. 5). In contrast, in Alk5/Wnt1-Cre mutants the outflow tract appeared unusually straight, failing to demonstrate the distinct conotruncal transition (Fig. 5B,D). This assay also clearly showed a dramatic reduction in the size of the aortic sac. Histological analysis of control samples displayed the characteristic rotation of the aortic sac and truncal OFT at the level where the AP septation takes place and verified the presence of the distinctive condensed AP-septal mesenchyme, which gradually divided the OFT to the aorta and the pulmonary trunk (asterisks in Fig. 6A). R26R lineage tracing showed that this tissue is derived from the NC, while immunostaining for α-SMA showed differentiation into smooth muscle (Fig 6B). In Alk5/Wnt1-Cre mutants the characteristic rotation of the aortic sac and truncal OFT fails to take place (Fig. 6G–L), and a properly formed AP-septum was not detectable (Fig. 6G,H). R26R lineage tracing demonstrated that the defects were not due to failure of NCCs to reach the OFT region. NC-derived cells around the abnomally bifurcated aortic sac, the abnormally large sixth PAAs and the truncus demonstrated strong αSMA staining (Fig. 6H,J,L). Recently, we showed that the NC-specific mutants of the related type I receptor, Alk2, display PTA as well [12]. In Alk2/Wnt1-Cre mutants, the rotation of the aortic sac and truncal OFT failed to occur (Fig. 6M–R) as seen in Alk5/Wnt1-Cre mutants. However, in Alk2 mutants the 6th pair of the PAAs was grossly hypoplastic, and while the Alk2/Wnt1-Cre mutants displayed a noticeable amount of septal tissue between the 4th and 6th PAAs (Fig. 6M,N), the condensed septal mesenchyme lacking Alk2 failed to extend the prongs into the truncal cushions and to form the AP septum. Concurrently, the 6th PAAs were losing their patency, which may have further contributed to the failed AP septation (Fig. 6M,O,Q). While CNCCs managed to migrate to the aortic sac and the truncal cushion level (Fig. 6N,P,R), immunostaining for αSMA appeared much weaker when compared to controls and Alk5 mutants, implying that ALK2-mediated signaling is involved in smooth muscle cell differentiation as previously suggested [12]. To conclude, while both Alk2 and Alk5 mutants demonstrate a failure in both the rotation of the aortic sac and the truncal OFT, and in the formation of the AP septum, the pathogenetic mechanisms behind these defects appear remarkably different.
Figure 5  The truncal OFT fails to rotate in Alk5/Wnt1-Cre mutants. Left (A, C) and right (B, D) lateral images of ink-injected control (A-B) and mutant (C-D) embryos at E11.5 demonstrate the abnormally straight OFT in mutants lacking the typical conotruncal transition (black arrow in A vs. black arrowhead in C) seen in control. Red arrowhead (C) points to the abnormally shaped aortic sac. Red "s", aortic sac; t, truncus; c, conus.
Figure 6  Signaling via ALK5 and ALK2 controls different aspects of aortico-pulmonary septation. Frontal sections from distal (top row) to proximal (bottom row) of the control (A-F), Alk5/Wnt1-Cre mutant (G-L) and Alk2/Wnt1-Cre mutant (M-R) samples (E11.5). A, C, E, G, I, K, M, O, Q, H&E staining; B, D, F, H, J, L, N, P, R, double staining for αSMA (brown) and β-galactosidase (green; R26R reporter assay). 6, the 6th PAA; AS, aortic sac; TA, truncus arteriosus; Ao, Aorta; PT, pulmonary trunk; Asterisks in A, B, M and N depict the AP septal mesenchyme.

Alk5/Wnt1-Cre mutants display increased apoptosis of post-migratory neural crest cells
As described above, Alk5/Wnt1-Cre mutants displayed an inadequate amount of AP-septal tissue in the base of the aortic sac between the origins of 4th and 6th PAAs. To analyze whether this phenotype resulted either from defective CNCC proliferation or inappropriate apoptosis, we used BrdU and TUNEL staining, respectively. While CNCC proliferation was not affected in Alk5 mutants (data not shown), we could detect a dramatic increase in the number of TUNEL positive cells in tissues surrounding the aortic sac including the site where the AP-septum forms (Fig. 7A–C). Dual staining for lacZ and TUNEL positive cells demonstrated that these cells were postmigratory CNCCs; this phenotype was already clearly detectable at E10.5. These results were confirmed by using immunostaining for cleaved caspase-3, another marker for apoptosis (Fig. 7I,J). In the chick, apoptotic neural crest-derived cells have also been found at the sites, where the prongs of the AP septum penetrate into the OFT cushion mesenchyme [25,26]. Thus, we compared apoptosis patterns also on the more proximal level, but found no detectable differences at E11.0 between Alk5/Wnt1-Cre mutants and controls (Fig. 7D,E). Unlike in Alk5/Wnt1-Cre mutant embryos, increased apoptosis of NC-derived cells is not responsible for the observed defects in the OFT septation in corresponding Alk2 mutants (Fig. 7C,E).
Figure 7  Aberrant apoptosis of NCCs in Alk5/Wnt1-Cre mutants. TUNEL (A-H) and Cleaved Caspase-3 (I, J) staining at E11.0 demonstrates a notable increase in apoptosis in Alk5/Wnt1-Cre mutants (B, H, J) on the aortic sac level when compared to controls (A, G, I) or Alk2/Wnt1-Cre mutants (C) (frontal sections), while sections on the OFT level do not demonstrate differences between controls (D) and Alk5 (E) or Alk2 (F) mutants. G,H, TUNEL staining of lacZ-stained embryos demonstrates that apoptotic cells are of neural crest origin. G, control; H, mutant. AS, aortic sac, arrows point to clusters of apoptotic cells surrounding the aortic sac. To conclude, our results suggest that in Alk5/Wnt1-Cre mutants a noticeable increase in apoptosis coincides with the abnormal patterning of the PAAs and the aortic sac, and with the failed AP-septation. These data support a specific role for ALK5 signaling, either directly or indirectly, in CNCC survival, since a similar apoptosis of NC-derived cells is not seen in Tgfbr2/Wnt1-Cre mutants [8,9].

Pharyngeal organs fail to migrate in Alk5/Wnt1-Cre mutants
In addition to the cardiac OFT, development of pharyngeal organs, i.e., the parathyroid glands and the thymus was also abnormal in Alk5/Wnt1-Cre mutants (see Figs. 1 and 8). Normally the thymus develops from the third pharyngeal pouch endoderm and migrates caudally to its final location in the superior mediastinum as seen in controls at E14 (Fig. 8A,B). In contrast, the thymic primordia of the Alk5 mutant littermates failed to descend caudally, and were located bilaterally in the neck region, where they were surrounded by neural crest-derived mesenchyme (Fig. 8D,E). The fate determination assay demonstrated that the thymic primordia were equally populated by NCCs both in controls and Alk5 mutants (Fig. 8B,E). Likewise the parathyroid glands failed to migrate normally in Alk5/Wnt1-Cre mutants. During normal development, the parathyroids first migrate in association with the thymic primordia, until they reach the thyroids in the neck region as seen in Fig 8C. In Alk5/Wnt1-Cre mutants, the parathyroids remained associated with the thymic primordia, and despite this abnormal rostral location, expression of parathyroid hormone was indistinguishable between Alk5 mutants and controls at E14 (Fig. 8C,F). To conclude, the observed pharyngeal organ phenotypes were also in striking contrast to those seen in Tgfbr2/Wnt1-Cre mutants [8,9].
Figure 8  Pharyngeal organs fail to migrate in Alk5/Wnt1-Cre mutants. At E14.0, the thymus was not detectable in the superior mediastinum (asterisks) in Alk5 mutants (D), when compared to controls (A). Serial sectioning revealed that the thymic primordia had failed to descend and were seen bilaterally in the upper pharyngeal region (E, F) surrounded by neural crest derived mesenchyme (blue staining cells in E). In controls, the parathyroid glands were properly associated with the thyroid glands (arrows in C), while in Alk5 mutants the parathyroid glands were associated with the thymic primordia (arrows in F). However, both controls and mutants expressed parathyroid hormone (PTH) at comparable levels (blue staining in C and F). A and D, hematoxyllin and eosin staining; B and E, R26 R lineage tracing assay – counterstaining with eosin; C and F, section in situ hybridization for PTH – counterstaining with eosin. T, thymus; Th, thyroid; asterisks in D depict the absence of the thymic primordia; asterisks in E and F depict the tongue.