miRNA-processing enzyme Dicer is necessary for cardiac outflow tract alignment and chamber septation

Abstract

MicroRNAs (miRNAs) have previously been implicated in a number of developmental processes, including development of the ventricular myocardium of the heart. To determine what, if any, additional roles miRNAs play in cardiogenesis, we deleted the miRNA-processing enzyme Dicer specifically in the developing murine heart. Embryos lacking cardiac Dicer lived longer than reported in previous studies using different alleles to remove cardiac Dicer activity and displayed a highly penetrant phenotype of double outlet right ventricle with a concurrent ventricular septal defect. Before the defect’s onset, Pitx2c and Sema3c, both required for outflow tract morphogenesis, were up-regulated in Dicer-deficient hearts. Interestingly, mesenchymal apoptosis in the outflow tract normally required for outflow tract alignment was greatly decreased in the mutants, likely contributing directly to the observed phenotype. In sum, we demonstrate here a specific developmental process, that of outflow tract morphogenesis, being hindered by the deletion of miRNAs during cardiogenesis.

Congenital heart malformations, resulting from mistakes in the complex process of cardiac development, represent the most common types of defects in newborns with at least 1% of all abnormalities at birth classified as congenital heart defects (CHDs) (1, 2). Over the last few years much progress has been made in elucidating genetic pathways underlying cardiogenesis, including the identification of a large number of transcription factors necessary for orchestrating the process. At the same time, there has been an increasing awareness of the importance of posttranscriptional mechanisms regulating cellular processes during embryogenesis, including those mediated by the approximately 20–25 nt noncoding regulatory RNAs known as microRNAs (miRNAs) (3, 4). Although the functions of several specific miRNAs have been explored in the context of cardiac development, the full extent to which this class of regulatory molecule influences heart formation remains to be determined.

An important strategy for establishing the range of developmental events regulated by miRNAs is to broadly interfere with all miRNA processing, either throughout the embryo or in various specific tissues and stages of embryogenesis. This can be accomplished, for example, through inactivation of the miRNA-processing enzyme Dicer, required for the production of mature miRNAs from pre-miRNAs (5). Previously reported targeted deletion of Dicer in mice demonstrated lethality in homozygous-null embryos by approximately embryonic day 7.5 postfertilization (e7.5) (6), an early time point that precluded investigation of Dicer’s role in cardiogenesis. A more recent removal of Dicer function specifically within Nkx2.5-expressing cardiac progenitor cells resulted in developmental anomalies in the heart including an underdevelopment of the ventricular myocardium and pericardial edema, ultimately leading to cardiac failure and embryonic lethality (7). Cardiac-specific Dicer deficiency was also engineered at a later stage under the control of a promoter for the cardiomyocyte structural protein alpha myosin heavy chain (α-MHC) (8). This later-stage loss of miRNA processing resulted in misexpression of cardiac contractile proteins, disruption of the sarcomeric structure, and a consequent impairment of cardiac function. These mice rapidly developed dilated cardiomyopathy, heart failure and postnatal lethality.

These studies highlighted the information to be gained by removing Dicer function at different stages of development and maturation of an organ. Accordingly, we decided to further explore miRNA function during heart development by removing Dicer activity at a stage in between those reported in the prior two studies. To accomplish this, we also made use of a Cre transgene driven by Nkx2.5 regulatory sequences; however, we made use of a different allele (9) than the one previously used by Zhao et al. (7, 10). This transgene is also activated from the cardiac crescent stage onwards but is known to be expressed with slightly different spatiotemporal kinetics. Crossing these mice with a floxed Dicer allele (11) produced Nkx2.5-Cre^Cre/+;Dicer^flox/flox embryos lacking miRNAs in the developing heart. As hoped, due to differences in timing and breadth of expression in the two Nkx2.5-Cre constructs and perhaps also the particular strains of mice used, the survival period of these cardiac Dicer-deficient mice was longer, although they still died in utero. The longer time window allowed us to observe the effect of loss of miRNA activity on outflow tract rotation and septation, events that occur too late to be visualized in the earlier study using an alternate Nkx2.5-Cre allele and too early to be affected in the study using an α-MHC-Cre.

Proper formation of the outflow tract (OFT), which gives rise to the aorta and pulmonary artery, is critical for the division of oxygen-rich and -poor blood that optimizes cardiac efficiency. How the OFT septates, rotates, and aligns itself with the heart’s ventricles has been extensively studied in chicken and mouse embryos, and a large number of targeted deletion mouse models display OFT defects (2, 12–14). Understanding how these arise is particularly relevant as in humans over 20% of CHDs include problems in formation or positioning of the OFT and connecting major arteries. These include, among other malformations, double-outlet right ventricle (DORV) wherein both the aorta and pulmonary artery exit the right ventricle and persistent truncus arteriosus (PTA) in which the OFT fails to septate.

Although OFT defects can arise in a variety of ways, one important mechanism involves failure in the process of programmed cell death. The occurrence of apoptosis during OFT maturation is thought to be crucial for OFT rotation and shortening to properly occur (15–17). Correlation between decreased programmed cell death and OFT defects has been demonstrated in mouse targeted-deletion models such as that of Foxp1^−/− (18). However, potential miRNA-mediated regulation of these processes has not been previously explored.

Here, we use a Dicer loss-of-function approach to investigate the roles of miRNAs in OFT development and demonstrate the disruption of critical programmed cell death during OFT morphogenesis. Two genes, Pitx2c and Sema3c, known to play crucial roles in this process are aberrantly expressed in the OFT-forming region, consistent with their contributing to the observed phenotype.

Results

Cardiac Deletion of Dicer Results in DORV and Concurrent VSD.

Dicer^flox/flox embryos were harvested at multiple time points to ascertain potential cardiac defects and embryonic lethality. Table 1 illustrates the number of embryos collected and genotyped at each time point. Nkx2.5-Cre^Cre/+;Dicer^flox/flox mice proved unable to survive beyond e13.75, a later time point of lethality than had been previously reported (7) using an alternatively constructed Nkx2.5-Cre (10). Thin, improperly compacted ventricular myocardium was observed in mutant embryos (arrow, Fig. 1H, compare with G) similar to that previously reported (7) and may contribute to the embryonic lethality. We observed no significant increase in cell death in the incorrectly forming myocardium of Dicer-deficient embryos. Meanwhile, previous analysis has shown that this myocardial phenotype correlates with and may be explained by changes in expression of several structural proteins (7).

Table 1.

Nkx2.5-Cre^Cre/+;Dicer^flox/+ × Nkx2.5-Cre^+/+;Dicer^flox/flox crosses

Stages	e10.25–11.0	e11.5–12.0	e12.5–13.0	e13.5	e13.7–14.0
Nkx2.5-CreCre/+;Dicerflox/flox/total	11/47 (23%)	5/17 (29%)	32/125 (26%)	8/33 (24%)	3/29 (10%)*

^*Embryos were observed dying from e13.7 onward; those found alive were close to expiration.

Fig. 1.

Cardiac deletion of Dicer results in DORV and concurrent VSD. Hearts at e13.0 and e13.5 days postfertilization are from Nkx2.5-Cre^+/+;Dicer^flox/flox (wild-type) (A and C) and Nkx2.5-Cre^Cre/+;Dicer^flox/flox (mutant) (B and D) littermates. Note the lack of a LV-OFT connection in the mutant hearts (asterisk). Representative sections are shown at e13.0 of RV-OFT connections (E and F) and LV-OFT connections (G and H). Arrow in (H) indicates abnormally thin ventricular myocardium and asterisk indicates VSD. LV, left ventricle; RV, right ventricle; OFT, outflow tract.

Although the cardiac chambers underwent basic morphogenesis, by e13.0–e13.5 hearts homozygous for the deletion of Dicer demonstrated 79% penetrance (15 of 19 examined embryos) of an apparent cleft-like division where a connection between the left ventricle and aorta would normally have existed (Fig. 1 B and D; asterisks). Histological analysis of serial sections taken through the heart revealed that the right ventricle was connected to both the pulmonary artery and aorta (Fig. 1 F and H, respectively) in a demonstration of DORV, with a concurrent ventricular septal defect (VSD) (Fig. 1H, asterisk). DORV leaves no direct outlet for oxygen-rich blood in the left ventricle, and hence it is forced into the right ventricle through the VSD. Neither Nkx2.5-Cre^+/+;Dicer^flox/Dicer^flox littermates nor Nkx2.5-Cre^Cre/+;Dicer^flox/+ littermates displayed any aberrant phenotypes. Thus, complete removal of Dicer activity in the heart is required for the observed cardiac defects to manifest.

Altered Gene Expression During Outflow Tract Development Following Dicer Deletion.

Recent predictions suggest that as much as 20–30% of vertebrate genomes may be influenced by miRNA regulation (19). To attempt to pinpoint specific genetic causes of the observed DORV/VSD defect, we undertook a candidate gene approach using RNA in situ hybridization for several genes known to be important for heart development. Most of the genes tested (see Methods for list) showed no change in expression levels in Dicer-deficient hearts as compared to wild-type littermates. The homeobox transcription factor Pitx2c, however, demonstrated a significant, site-specific increase in RNA levels in the absence of Dicer at e10.0–e10.5 in the OFT and adjacent ventricular wall (Fig. 2 A and B, arrowheads) that continued as late as e12.5–e13.0 (Fig. 2 C–F). Pitx2c is known to play an important role in the establishment of OFT positioning relative to the ventricles, and its absence leads to defects in the OFT including a significant percentage of DORV (20, 21). Intriguingly, ectopic myocardial Pitx2 expression in iv/iv mutant mice—demonstrating cardiac looping randomization and an assortment of subsequent defects—has been shown to correlate specifically with DORV (22), reminiscent of the Pitx2c up-regulation and DORV phenotype shown here.

Fig. 2.

Modified gene expression post-Dicer deletion. Pitx2c mRNA levels are up-regulated in mutant hearts in comparison with wild-type littermates from as early as e10.25 (A and B) through e12.7 (C and D), and e13.0 (E and F). Arrowheads in (B) and (D) indicate up-regulation in outflow tract and adjacent ventricular myocardium. By comparison, Sema3c expression is first seen up-regulated at e12.5 in mutant hearts (H) in comparison with wild-type littermates (G). Arrowheads in G and H indicate outflow tract expression; arrows in H indicate ventricular expression. LV, left ventricle; Ao, aorta; PA, pulmonary artery; OT, outflow tract.

In trying to understand the molecular consequences of altered Pitx2c expression, we noted previous work that has demonstrated its regulation of Semaphorin3c (Sema3c) in the OFT myocardium. Sema3c is a signaling molecule known to play a variety of roles including regulating axonal guidance, cell survival, and migration of neural crest cells (23, 24), and mice lacking Sema3c manifest a PTA phenotype in the OFT (25).

It has been shown that although the targeted deletion of murine Pitx2c does not affect Sema3c expression levels at e10.5 or e11.5, by e12.5 Sema3c mRNA is nearly completely absent in Pitx2c^−/− outflow tracts (20). These data suggest a role for Pitx2c in maintaining appropriate levels of Sema3c in the developing OFT myocardium. Consistent with this possibility, we found that Sema3c RNA expression was markedly up-regulated at e12.5 in the OFT (Fig. 2 G and H, arrowheads) and adjacent ventricular wall (Fig. 2H, arrows) in Dicer-deficient embryos as compared to wild-type littermates. These were locations where we had observed the earlier up-regulation of Pitx2c, suggesting that Pitx2c activation may partially contribute to the observed DORV/VSD via up-regulation of Sema3c.

Programmed Cell Death During Outflow Tract Alignment Is Disrupted.

To determine the mechanistic causes of the DORV/VSD, fundamental processes that play important roles in OFT maturation were examined. We failed to find any difference in cell proliferation, as assayed by H3B staining, between mutant and wild-type littermate embryos at e12.0, e12.5, or e13.0. Similarly, using TUNEL staining for apoptosis, we observed no significant difference between wild-type and mutant OFT during early time points. In mice, apoptosis in the OFT has been previously characterized as beginning around e12.5 and peaking a day later (17). We observed a similar pattern with no apparent apoptosis before e12.5. At e12.5, small numbers of stained cells were observed with no significant difference between wild-type and mutant embryos (Fig. 3 A and B). By e13.0 and e13.5, however, widespread apoptosis was evident in wild-type OFT tissue (Fig. 3 C and E), but the number of stained cells was greatly decreased in Dicer-deficient OFTs (Fig. 3 D and F). Cellular morphology, as well as costaining for muscle actin, suggested that the cells normally undergoing apoptosis at these time points in wild-type embryos were predominantly mesenchymal cushion tissue. Staining of serial sections demonstrated increased Sema3c RNA expression in mutant embryos to be directly adjacent to the area of decreased cell death at e13.0 (Fig. 3 G and H), suggesting that increased levels of Sema3c may help inhibit mesenchymal apoptosis. Quantitative analysis at e12.75–e13.0 revealed a nearly fivefold decrease in mesenchymal cell death in comparison with wild-type OFT tissue (Fig. 3I).

Fig. 3.

Programmed cell death during outflow tract alignment is disrupted. TUNEL-positive cells (green) and muscle actin costain marking cardiomyocytes (red) in outflow tract mesenchyme at e12.5 (A and B), e13.0 (C and D), and e13.5 (E and F). TUNEL-positive cells were few in number in both wild-type (A) and mutant (B) OFT tissue at e12.5, prevalent in wild-types at e13.0 and e13.5 (C and E), and markedly reduced in mutants at e13.0 and e13.5 (D and F). Side by side comparison of Sema3c mRNA and outflow tract apoptosis (G and H) in mutant embryos shows Sema3c expression directly adjacent to the area of decreased cell death. (I) Quantitation of mesenchymal apoptosis at e12.75–e13.0 in wild-type and mutant OFT. Bars indicate 95% C.I. *P < 0.001.

Discussion

The relatively recent discovery of miRNAs in the early 1990s and subsequent studies since then have dramatically altered our understanding of gene regulation. However, the small size of miRNAs, unclear binding specificities, and other factors have made it difficult to pinpoint specific functions for individual miRNAs. While work in this field is now progressing rapidly, here we chose to directly determine the relevance of miRNAs to cardiogenesis by deleting one of the necessary components of miRNA processing, the enzyme Dicer.

One of the initial conditional deletions of Dicer, done in the developing limb mesoderm, demonstrated no aberrations in basic patterning of the limb, analogous to the intact basic patterning observed here in the heart. However, although the limb deletion exhibited greatly increased cell death (11), the Dicer deletion detailed herein led to decreased cell death, essentially the reverse effect. The difference may, in part, be explained by the fact that in the limb mesenchyme, the increase in cell death can be attributed to a late, nonspecific mitotic defect (11). In contrast, the decrease in cell death we observe is confined to a small subset of cardiac tissue and exhibits spatiotemporal specificity, suggesting a specific effect mediated by changes in gene regulation during OFT development.

Interestingly, the expression of many molecular markers was not perturbed (at least at the level observable by in situ hybridization) in the absence of Dicer activity and hence miRNA processing. This stands in contrast to the extremely large number of genes predicted to be miRNA targets, up to 1/3 of those in the genome (26–28), but is consistent with the previous observation that many miRNAs are expressed in reciprocal cell populations from those expressing their target mRNAs (29, 30). Such a dichotomy of RNA and miRNA expression may act to provide robustness to developing systems rather than major regulation of gene activity within domains of expression. The same may largely be true for the molecular pathways orchestrating cardiogenesis as indicated by the fact that many aspects of the heart formed normally in this and other studies in which Dicer activity was removed. Nonetheless, as our analysis affirms, some miRNA targets are critical for normal cardiac formation.

The observed aberrant expression of genes known to be required for proper OFT development, although other tested genes remained unaffected, points to a precise effect. Pitx2c and Sema3c will certainly not be the only affected genes in a broad deletion of Dicer, but it is intriguing to speculate on their particular roles in regulating levels of apoptosis during OFT alignment. The removal of Smad4 signaling in cardiac neural crest cells (CNCCs) has been suggested to non-cell autonomously down-regulate Sema3c and other molecules in non-CNCC OFT in concert with defects such as PTA and tract elongation (31), illustrating the crosstalk among multiple cell types necessary for OFT formation. In addition, Sema3c has a known capacity to promote cell survival such as that demonstrated with non-cell autonomous effects on cerebellar granule neurons and endothelial cells in vitro (32, 33). These and other data lend a clue as to how up-regulation of genes in OFT myocardium could lead to aberrant non-cell autonomous signaling and thus decreased cell death in adjacent mesenchyme.

Several groups have now reported the presence and potential functions of miRNAs in the heart during development and adult hypertrophic responses. Results from our work and that of others suggest that, despite the presence of numerous miRNAs, many fundamental building blocks of cardiac morphogenesis such as the number of chambers and initial formation of component layers are not severely affected. However, some of the complex morphogenetic events that allow for correct formation of the mammalian heart’s detailed architecture are indeed flawed in the absence of miRNAs. At least in the heart, then, there is a compelling analogy between the fine-tuning of gene expression that miRNAs are often suggested to regulate and the resultant fine detailing of cardiac morphogenesis that is necessary for 4-chambered, dual-outflow cardiac functionality in mammals. As tools to study miRNAs continue evolving and become more powerful, new details will surely emerge regarding miRNA-based regulation of cardiac development and of embryogenesis as a whole.

Methods

Generation of Dicer Conditional Mice.

Nkx2.5-Cre mice (9) and Dicer^flox/flox mice (11) have been previously described and were intercrossed to produce Nkx2.5-Cre; Dicer^flox/flox mice that were genotyped as described. Nkx2.5-Cre^Cre/+;Dicer^flox/flox (mutant) embryos were compared in all analyses to Nkx2.5-Cre^+/+;Dicer^flox/Dicer^flox (wild-type) littermates.

RNA In Situ Hybridization.

Section in situ hybridization was done with antisense riboprobes as previously described (34, 35). RNA expression compared between wild-type and mutant littermates included that of the following genes known to be involved in cardiogenesis: Fgf8, GATA4, Hand2, Mef2c, Mlc2v, Pitx2c, PlexinA2, Sema3c, Tbx1.

Immunohistochemistry and TUNEL.

Some embryos were embedded in paraffin after fixation in 4% paraformaldehyde whereas others were fixed with 4% paraformaldehyde followed by 10% and 30% sucrose gradients for cryosectioning. Serial paraffin-embedded sections were used for H&E staining, RNA in situ hybridization, HHF35 staining, and TUNEL staining; cryosections were used for H3B staining and TUNEL staining. Apoptosis was assayed using ProMega’s DeadEnd Fluorometric TUNEL System (#G3250), cell proliferation with α-phosphohistone H3B (Upstate Biotechnology, # 06–570) and muscle actin with HHF35 (Dako # M0635). Quantitation for apoptosis was done by counting the number of nonmyocardial apoptotic cells divided by the total number of DAPI-stained nonmyocardial cells per high magnification field of view with four levels assayed for each embryo. Seven wild-type and seven mutant embryos were assayed. IMARIS software (Bitplane) was used to assist in identification and counting of nonmyocardial cells.

Statistical Analysis.

Statistical calculations were performed using t test of variables (two-sample t test assuming unequal variances) with 95% confidence intervals.