Mesodermal cells acquire the potential to become myocardial cells soon after gastrulation. The primary cardiomyocytes are arranged in a horseshoe-shaped crescent that fuses in the midline area of the body axis, giving rise to a tubular heart (for a review see Fishman and Chien, 1997). With further development, five different morphological and functional compartments can be distinguished in the embryonic heart: inflow tract, atrium, atrioventricular canal, ventricle, and outflow tract (Moorman and Lamers, 1994). As the heart begins to loop, myosin heavy chain (MHC) and myosin light chain (MLC) isoforms become confined to distinct cardiac segments (Franco et al., 1998; Lyons, 1994; Moorman and Lamers, 1994). Regional specification, however, is likely to occur at an earlier stage since expression of different atrial-specific and ventricular-specific myosin isoforms becomes regionalised in avian species prior to the morphological identification of these cardiac compartments (Yutzey and Bader, 1995; Yutzey et al., 1994). Furthermore, an iroquois-related homeobox gene, cIrx-4, which modulates MHC expression, is confined to the prospective ventricular myocardium at the linear heart tube stage (Bao et al., 1999; Bruneau et al., 2000).
Within the fetal heart, left and right atrial and ventricular chambers are morphologically different (Brown and Anderson, 1999; Webb et al., 1996). Some genes such as MCK, connexin40, connexin43, and a number of MHC and MLC isoforms display transient left and right differences in ventricular expression patterns (Kelly et al., 1998a; Lyons, 1994; Van Kempen et al., 1991, 1996; Zammit et al., 2000). Transgenic mice carrying regulatory sequences of sarcomeric genes have also revealed transcriptional differences between the left and right atrial and ventricular compartments (Biben et al., 1996; Kelly et al., 1995; Kelly et al., 1998b; Kuisk et al., 1996; Ross et al., 1996;), providing potential molecular markers to examine the acquisition of left and right identity during cardiogenesis (Franco et al., 1997; Kelly et al., 1999). More recently, Pitx-2 has been reported to show left/right differences in the atrial myocardium (Franco et al., 2000; Campione et al., 2001; Schweickert et al., 2000).
The heart is the first embryonic structure to display morphological left-right asymmetry. Under normal conditions (situs solitus), the cardiac tube loops towards the right side of the embryo (D-loop). Several genes (nodal, lefty-2, Pitx-2) are expressed asymmetrically in the lateral plate mesoderm (LPM) prior to cardiac looping in mice (Campione et al., 1999; Collignon et al., 1996; Logan et al., 1998; Meno et al., 1996; Piedra et al., 1998; Yoshioka et al., 1998; for a review see Burdine and Schier, 2000). Interestingly, Pitx-2 is preferentially expressed in the left side of the cardiac tube just before cardiac looping (Campione et al., 1999, 2001; Logan et al., 1998; Ryan et al., 1998). It has been proposed that Pitx-2 plays a role in processing embryonic left-right signalling, contributing to the determination of the asymmetries of the heart and gut (Campione et al., 1999; Logan et al., 1998; Piedra et al., 1998; Ryan et al., 1998).
Abnormal specification of cardiac laterality has been described in several mutant mice strains such as iv (Hummel and Chapman, 1959), inv (Yokoyama et al., 1993), legless (Singh et al., 1991), and brachyury (King et al., 1998). Hummel and Chapman (1959) showed that embryos homozygous for the iv gene present inverted cardiac looping in approximately 50% of cases. More recently, it has been shown that several cardiac malformations such as double-outlet right ventricle (DORV), persistent truncus arteriosus (PTA), and atrial isomerism are observed with a relatively high incidence in the iv/iv background (Icardo, 1990; Icardo and Sanchez de Vega,1991; Seo et al., 1992). Evidence from a number of studies of mice carrying null alleles for components of the left/right cascade suggests that left-right situs and cardiac looping are independent events (King et al., 1998; see for a review Burdine and Schier, 2000). However, the precise timing of the events that underlie the specification of left and right atrial situs remain unknown due to the lack of left/right molecular markers.
Transgenic mice in which regulatory sequences of the myosin light chain 3F (MLC3F) gene drive nlacZ expression in the right atrium and left ventricle provide the first marker that permits molecular distinction of left and right cardiac compartments (Franco et al., 1997; Kelly et al., 1995). We have investigated whether the specification of abnormal handedness affects the transcriptional potential of each cardiac compartment in normal and congenitally malformed hearts. We have also examined the earliest stage at which left and right differences can be detected during atrial specification. For this purpose, we have crossed 3F-nlacZ-2E transgenics with iv mice. We provide evidence that transcriptional differences between the left and right atrial appendages are first observed soon after cardiac looping, as revealed by expression patterns of the MLC3F transgenes and fate mapping experiments. Our data also demonstrate that transcriptional specification of the cardiac chambers is independent of the direction of heart looping.
We crossed 3F-nlacZ-2E transgenic mice with iv mutant mice. The presence of the transgene had no observable effect on the iv/iv mutant phenotype, nor did the presence of the mutation modify overall expression of the transgene (see Fig. 1). Within the iv/iv background, approximately 50% of the embryos display inverted cardiac looping (Hummel and Chapman, 1959). Of 212 embryos examined from iv/3F-nlacZ-2E backcrossed with iv/iv mice, 47 (22%, expected 25%) had inverted heart loops and 116 expressed the transgene (55%, expected 50%). Similarly, of 198 embryos examined from iv/3F-nlacZ-2E intercrossed with iv/3F-nlacZ-2E, 24 (12.1%, expected 12.5%) had inverted heart loops and 162 expressed the transgene (82%, expected 75%). Transgene expression was analysed under three different conditions, normal cardiac looping (D-loop), inverted cardiac looping (L-loop), and atrial isomeric hearts, including specimens with either D-loop or L-loop phenotypes.
Expression Pattern of the 3F-nlacZ-2E Transgene in Normal and Inverted Hearts
In D-loop hearts (E10.5), the expression pattern of the 3F-nlacZ-2E transgene is predominantly confined to the right atrial appendage, the atrioventricular canal, and the left ventricle (Franco et al., 1997; Fig. 1). Within the left ventricle, a patch of nonexpressing cells can be observed in the dorsal aspect of the ventricular free wall adjacent to the atrioventricular junction (see Fig. 2A). In L-loop hearts, the nlacZ expression pattern is a mirror-image, i.e., transgene expression is confined to the morphological right atrial appendage (located on the left side of the embryo), the atrioventricular canal, and the morphological left ventricle (located on the right side of the embryo) (Fig. 1).
Transgene Expression in Isomeric Hearts
The arrangement of the atrial appendages can be determined in fetal stages by morphological examination of the junction between the appendage (left and right) and the sinus venosus-derived component of each atrium (Seo et al., 1992; Webb et al., 1996). Within the iv background, a small proportion (4.8%) of mice display atrial isomerism (Seo et al., 1992). In this study, our aim was to examine whether differential left and right transgene expression in the atrial appendages correlates with left and right appendage identity. For this purpose, we analysed 220 specimens ranging from E14.5 to E18.5, of which 28 showed morphological evidence of right atrial appendage isomerism (RAI) and four showed morphological evidence of left atrial appendage isomerism (LAI). The presence of right and left atrial appendage isomerism was observed independently in both D-looped and L-looped hearts.
Transgene expression in RAI hearts was confined to the morphological left ventricle, atrioventricular canal, and both atrial appendages, as illustrated in Figure 2. No expression of the transgene was observed in the myocardium of the symmetrically branched superior caval veins (Figs. 2 and 3). Such an expression pattern was also observed at earlier stages (E10.5/E12.5), prior to detectable differences in left and right atrial appendage morphology (Fig. 3). Interestingly, an extensive β-galactosidase positive area was observed in the dorsal region of the atrioventricular canal, which bifurcated to both morphological right atrial appendages (Fig. 2F).
In LAI hearts, nlacZ expression was mostly confined to the morphological left ventricle and atrioventricular canal (Fig. 3C). Scattered β-galactosidase positive cells were observed in the most caudal aspect of the atrial appendages, but the overall atrial appendage was mainly negative for nlacZ expression (Fig. 3C). No bifurcation of transgene expression was observed in the dorsal aspect of the atrioventricular junction (data not shown).
Within the 220 specimens analysed, in no case was abnormal expression of the transgene documented in the ventricular chambers, independently of whether the specimens were D-looped or L-looped or whether isomerism of the atrial appendages was observed.
Transgene Expression During Early Cardiac Looping
The expression of the 3F-nlacZ-2E transgene in hearts with atrial isomerism at fetal stages is high in both atrial appendages in RAI and low in both atrial appendages in LAI. Moreover, the fact that such a regionalised expression pattern is seen at earlier stages suggests that the expression of 3F-nlacZ-2E provides a molecular marker for investigating the earliest signs of left and right atrial appendage identity. To look further at the early onset of left and right atrial appendage specification, we investigated nlacZ expression in wild type 3F-nlacZ-2E and in iv/3F-nlacZ-2E embryos. These embryos were compared, at similar stages, to 3F-nlacZ-9 embryos, which show expression in both left and right atria and ventricles but not in the embryonic outflow tract or inflow tract myocardium (Franco et al., 1997, 2000).
In the early symmetrical cardiac tube of 3F-nlacZ-9 embryos, β-galactosidase positive cells are distributed throughout the visible tube, including the left and right sinus horns (Fig. 4E). At this stage, 3F-nlacZ-2E expression is broadly similar, also extending bilaterally within the sinus horns, but with a small nonexpressing region medially, particularly in the most caudal area, overlying the foregut (Fig. 4A). As looping begins, 3F-nlacZ-9 expression is observed throughout most of the heart, with the outflow tract becoming visible as a nonexpressing region (Fig. 4G,H). The outflow tract is similarly β-galactosidase negative in early 3F-nlacZ-2E looping hearts: expression of this transgene, however, is weaker in the left than in the right sinus horn (Fig. 4C,D). In these hearts, a β-galactosidase negative area stretches from the most caudal portion of the left sinus horn to a more medial region of the prospective atrioventricular canal. At this stage, the presumptive ventricular area is present whereas no morphological indications of the primary atrium can be detected.
We examined 3F-nlacZ-2E transgene expression in iv/iv embryos during the looping phase, in comparison with wild-type 3F-nlacZ-2E and 3F-nlacZ-9 embryos. We found a subset of iv/3F-nlacZ-2E embryos (<10%) in which β-galactosidase staining extended bilaterally and evenly within the sinus horns, without any areas of non-expression (Fig. 4I). This staining pattern is remarkably similar to that seen in wild-type 3F-nlacZ-9 embryos at the same stage (Fig. 4H). This β-galactosidase pattern in iv/iv embryos corresponds to what would be expected of RAI hearts, suggesting that the earliest distinction between left and right atrial appendages takes place at the early looping stage. The rest of iv/3F-nlacZ-2E embryos display a β-galactosidase expression pattern comparable, or mirror-image, to wild type 3F-nlacZ-2E.
Fate Mapping of the Presumptive Left Atrial Precursor Region
The observation of nonstaining regions in early 3F-nlacZ-2E hearts poses two questions: 1.) are the non-staining medial cells displaced leftwards as looping starts, to become the left-sided unstained atrial region? and 2.) what is the fate of the left-side area?
To answer these questions, we labelled five different early hearts by DiI injection on each experimental setting, then cultured embryos to postlooping stages. In all cases in which dye was injected medially into a symmetrical heart tube, the labelled cells were located in the left ventricle after looping (Fig. 5A,A'). In contrast, dye injected into the left horn region that does not express the 3F-nlacZ-2E transgene was found in the left side of the atrium, postlooping (Fig. 5C,C'). When hearts were injected at intermediate positions between these two regions, labelled cells were found in the area of the atrioventricular canal (Fig. 5B,B'). From these experiments, we conclude that the early medial nonexpressing cells are fated to be left ventricle and are not related to the nonstaining left sinus horn myocardium, which will contribute to the left atrium. In line with these findings, the most medial part of the linear heart tube, constituted of a cluster of β-galactosidase negative cells gives rise to the patch of negative cells in the most dorsal aspect of the left ventricular free wall in wild-type 3F-nlacZ-2E transgenics.
We have analysed mice showing compartment-specific transgene expression in the left ventricle, atrioventricular canal and right atrium (3F-nlacZ-2E transgenic mice; Kelly et al., 1995) in the iv background (Hummel and Chapman, 1959). This cross permits us to examine the emergence of transcriptional differences between left and right components of the developing heart. Transgene expression in inverted hearts (L-loop) demonstrates that the transcriptional identity of left and right atrial and ventricular cardiac segments correlates with the morphological identity of each segment.
We show that 3F-nlacZ-2E embryos are differentially expressed in the right and left atrial as well as ventricular components regardless of the direction of cardiac looping; transgene expression remains confined to the morphological right atrial appendage myocardium independent of its position relative to the embryonic left-right axis. Similarly, embryos that display isomeric atrial appendages show transgene expression according to the morphological identity of the atrium. This is the first study in which a marker that molecularly distinguishes left and right atrial appendages has been used to investigate early chamber identity. Interestingly, transgene expression in isomeric hearts is always confined to the trabeculated atrial appendages but is not observed in the venous myocardial component (caval veins or pulmonary veins myocardium). These observations support the notion that the atrial appendages and the venous components are distinct morphological and transcriptional entities (Franco et al., 2000).
Regionalized expression of the 3F-nlacZ-2E transgene is observed from an early stage of cardiac looping (E8.5) onwards. Analysis of transgene expression in a subset of iv/3F-nlacZ-2E embryos demonstrates bilateral expression in the left and right sinus horns. We suggest that these embryos would have developed right atrial isomeric hearts. In support of this conclusion, injection of DiI into the most posterior (caudal) region of the left sinus horn labels the left atrial appendage myocardium. Left and right atrial appendages, therefore, appear to be established by the early looping stage.
During embryogenesis, the first morphological evidence of asymmetry is observed as a rightward looping of the cardiac tube. Mutations in the left-right dynein gene generate randomization of embryonic laterality in iv mutant mice (Supp et al., 1997), in which asymmetric expression of genes such as nodal, lefty-2, and Pitx-2 is randomized (Campione et al., 1999; Lowe et al., 1996; Piedra et al., 1998). Our observations suggest that the acquisition of left and right atrial myocardial identity precedes chamber morphogenesis. The nodal/lefty-2 signalling pathway may, therefore, be involved in the specification of the left and right atrial appendages. Recently, it has been reported that null mutants of the lefty-1 gene, which reveal bilateral expression of lefty-2, nodal, and Pitx-2 in the LPM, show left thoracic isomerism (Meno et al., 1998). Gene targeting experiments have demonstrated the role of the activin receptor IIB and Pitx-2 on the determination of right pulmonary isomerism (Oh and Li, 1997; Gage and Camper, 1999; Lu et al., 1999; Kitamura et al., 1999). These data, in conjunction with our findings on the early acquisition of left and right atrial appendage identity, support the hypothesis that bilateral expression of the nodal/lefty-2/Pitx-2 signalling pathway can give rise to left atrial appendage isomerism whereas absence of nodal/lefty-2/Pitx-2 signalling results in right atrial appendage isomerism.
In contrast to our observations of molecular isomerism of the atrial compartments, we detected no examples of ventricular isomerism (out of 220 iv/iv transgene positive embryos). Since isomeric ventricles have not been described in the literature (Becker and Anderson, 1983), these findings suggest that the acquisition of right (pulmonary) and left (systemic) ventricular identity is not dependent predominantly on left-right patterning. Indeed, it has been previously postulated, based on fate mapping experiments (De la Cruz et al., 1989; Brown and Anderson, 1999), that anteroposterior patterning plays a role on the acquisition of right and left ventricular chamber identity. Furthermore, expression of transcription factors dHAND and eHAND, which are initially regionalised along the anteroposterior axis of the tubular heart and later in the left and right embryonic ventricles, has been shown to define systemic and pulmonary ventricular identity independent of situs in inv mutant mice (Thomas et al., 1998). The myocardium composing the early symmetrical tubular heart appears, therefore, to be predisposed to give rise to the systemic and pulmonary ventricular compartments. Consistent with data from cell-labelling experiments in the chick (de la Cruz et al., 1898), our DiI labelling data in the mouse suggest that prospective atrial myocardial cells have yet to become incorporated into the cardiac tube. This difference in timing of differentiation suggests that the atrial precursors may still be susceptible to right-left signalling pathways. This hypothesis is underscored by our fate mapping experiments, which show that only the systemic and pulmonary ventricles are present at early stages of cardiac tube formation, whereas only after cardiac looping has started, do the most caudal myocardial cells map to the atrial components.
MLC3F transgenic lines carry the nlacZ reporter gene under the transcriptional control of regulatory elements of the MLC3F gene. 3F-nlacZ-2E mice (Kelly et al., 1995) harbour a construct with 2 kilobases (kb) of upstream sequence in front of the MLC3F promoter, driving expression of an nlacZ reporter gene and the 3′ MLC1F/3F enhancer element placed 3′ to the polyadenylation site. 3F-nlacZ-9 contains a 9-kb fragment of DNA upstream of the MLC3F transcription initiation site (Franco et al., 1997). A second muscle-specific enhancer (approximately at −4 kb from the MLC3F transcription initiation site) is present within this 9-kb sequence (Kelly et al., 1997). Details of transgene construction and the characterisation of skeletal and cardiac muscle expression during development and in the adult have been reported elsewhere (Franco et al., 1997; Kelly et al., 1995, 1997).
Generation of 3F-nlacZ-2E Mice in the iv Background
Heterozygous 3F-nlacZ-2E mice were crossed with homozygous iv/iv mice. The first generation of offspring (F0) was selected for transgene expression by β-galactosidase coloration of tail tip preparations. F1 3F-nlacZ-2E positive mice (iv/3F) were backcrossed with homozygous iv/iv mice, or intercrossed with each other, and embryos ranging from embryonic day (E) 7 to E18 were collected. The day of vaginal plug was taken as E0.5. Embryos were excised from the uterus and the thoracic wall was removed (E12 to E16) exposing the heart to allow maximal penetrance of fixatives and reagents. Specimens were briefly fixed in 4% freshly prepared formaldehyde (30 min to 1 hr) and rinsed twice in phosphate-buffered saline (PBS).
In Toto X-Gal Histochemical Staining
Incubation in X-gal solution at 37°C was performed for periods of 30 min to overnight (Sanes et al., 1986; Franco et al., 2001). Subsequently, embryos were postfixed in 4% freshly made formaldehyde for 4 hr to overnight and conserved in 70% glycerol in PBT (PBS with 0.1% Tween-20).
X-Gal Histochemical Staining on Tissue Sections
Embryos were rinsed in increasing concentrations of sucrose (10, 20, and 30% in PBS) for 2 hr at each step, embedded in OCT (Miles Inc.) and frozen. Freeze cryo- stat serial sections of 7–10 μm were cut, mounted onto gelatin-coated slides and stored at −20°C until use. Incubation in X-gal solution at 37°C was performed for periods of 30 min to overnight as detailed elsewhere (Sanes et al., 1986; Franco et al., 2001). Sections were counterstained with azofloxine for 5 min, dehydrated in a graded ethanol series, and mounted in Entellan (Merck).
X-Gal Histochemical Staining and Immunohistochemistry
Cryostat sections were allowed to warm to room temperature for 30 min. Incubation in X-gal solution at 37°C was performed for periods of 30 min to overnight as described elsewhere (Franco et al., 1997). Several sections were processed for SERCA2 immunohistochemical detection as described in Franco et al. (2001). Briefly, sections were hydrated in decreasing ethanol concentrations, rinsed in PBS, and incubated overnight with a primary antibody against SERCA2. Subsequently, sections were thoroughly washed in PBS, incubated with an alkaline phosphatase-coupled secondary antibody overnight. Coloration of alkaline phosphatase activity was revealed with incubation in NBT/CIBP solution for approximately 20 min. Sections were dehydrated and mounted in Entellan.
Embryos ranging from 3 to 15 somites were collected, transferred to Hank′s solution, and injected with DiI using a needle drawn from glass capillary tubing. For each experimental timepoint, 5 embryos were used. Labelling experiments were video-taped during injection. Embryos were cultured for 24 hr in vitro in 75% rat serum, 25% T6 medium in 5% CO2, 20% O2/75% N2 in rolling bottles. DiI labelled tissues were photographed in unfixed embryos.
We are indebted to Hata Zavrelova and Corrie de Gier-de Vries for technical support, and to Dr. Frank Wuytack for his kind supply of SERCA2 antibody.