Temporally Distinct Six2 Positive Second Heart Field Progenitors Regulate Mammalian Heart Development and Disease

Abstract

The embryonic process of forming a complex structure such as the heart remains poorly understood. Here, we show that Six2 marks a dynamic subset of second heart field progenitors. Six2-positive (Six2⁺) progenitors are rapidly recruited and assigned, and their descendants are allocated successively to regions of the heart from the right ventricle (RV) to the pulmonary trunk. Global ablation of Six2⁺ progenitors resulted in RV hypoplasia and pulmonary atresia. An early stage-specific ablation of a small subset of Six2⁺ progenitors did not cause any apparent structural defect at birth but rather resulted in adult-onset cardiac hypertrophy and dysfunction. Furthermore, Six2 expression depends in part on Shh signaling, and Shh deletion resulted in severe deficiency of Six2⁺ progenitors. Collectively, these findings unveil the chronological features of cardiogenesis of which mammalian heart is built sequentially by temporally distinct populations of cardiac progenitors, and gain insights into the late-onset congenital heart disease.

In Brief

The process of building multiple components of the mammalian heart remains poorly understood. Zhou et al. identify and characterize a population of temporally distinct Six2-positive cardiac progenitors and suggest that the heart is made progressively by adding new components to the preexisting structures.

Graphical Abstract

INTRODUCTION

Understanding how an organ arises from a cell is a central theme of developmental biology. In principle, an organ might arise from a homogeneous pool of progenitors (Figures S1A and B), or a heterogeneous pool of progenitors (Figures S1C and D). The progenitor pool could be permanent (progenitors are self-renewing) or changing (progenitors are continuously recruited and assigned to specific fates). Descendants of an individual progenitor might disperse across space through time (Figure S1A) or allocate in a spatiotemporally restricted manner (Figures S1B–C). These principal modes of organ formation have radically different features with regard to the genealogical history of cell types within the mature organ (Figure S1a–d).

The mammalian heart originates from two primary sources of progenitors known as the first heart field (FHF) and second heart field (SHF)(Evans et al., 2010, Meilhac et al., 2014). Mesoderm posterior 1 (Mesp1) is expressed in all cardiac progenitors and is a key regulator of cardiac specification (Bondue and Blanpain, 2010). Prospective clonal analysis of Mesp1-positive (Mesp1⁺) progenitors reveals that the FHF and SHF progenitors are specified at the onset of gastrulation, and are assigned to specific anatomic structures of the heart (Lescroart et al., 2014, Devine et al., 2014). The undifferentiated highly proliferative SHF progenitors express LIM homeodomain transcription factor Islet1 (Cai et al., 2003). Mouse embryos without Islet1 fail to form the SHF derivatives including the right ventricle (RV), much of the atria and the outflow tract (OFT). The subpopulation of SHF progenitors that contribute to the definitive OFT, which comprises the sub-arterial myocardium, the non-myocardial arterial roots with the semilunar valves, and the intrapericardial arterial trunks (Figure S2), are located at the ventral pharyngeal arch apparatus behind embryonic OFT (Waldo et al., 2005, Waldo et al., 2001, Mjaatvedt et al., 2001, Verzi et al., 2005). In addition to Islet1, this subpopulation of SHF progenitors expresses transcription factors including Tbx1 (Zhang et al., 2006, Nowotschin et al., 2006), Six1 and Eya1 (Guo et al., 2011). Mutations of these genes are closely linked to human DiGeorge syndrome, a common congenital heart disease (CHD) with a spectrum of OFT anomalies including persistent truncus arteriosus (PTA) and tetralogy of Fallot (TOF) (Lindsay et al., 1999, Jerome and Papaioannou, 2001, Merscher et al., 2001, Xu et al., 2004, Arnold et al., 2006). Other subpopulations of SHF progenitors have also been identified that contribute to structures including the atrial myocardial wall, atrial septum, sinus venosus and pulmonary veins (Meilhac et al., 2014). Collectively, there is an apparent heterogeneity of cardiac progenitors, particularly the SHF, which contribute to different anatomic components of the heart.

The process whereby diverse cardiac progenitors assemble multiple anatomic components of the heart remains to be defined. Conceptually, cardiogenesis may follow one or several of the above-mentioned principal models (Figure S1). Here, we provide evidence that the RV and OFT originate primarily from a Six2-positive (Six2⁺) subpopulation of the SHF. The behavior and distribution of descendants of Six2⁺ progenitors indicate that new progenitors are rapidly recruited and assigned to specific components of the heart, and further suggest that the presumptive early cardiogenic field comprises the changing or temporally distinct pools of progenitors. We provide additional evidence demonstrating that this mode of formation has major implications in OFT development and late-onset heart disease.

RESULTS

Six2 is dynamically expressed in a subpopulation of the SHF

We have shown previously that Six1 is a downstream effector of Tbx1 during SHF development (Guo et al., 2011). Six2 is closely related to Six1 among all family members (Li et al., 2002, Li et al., 2003). We were therefore interested in examining whether Six2 is also expressed in the SHF (Figure 1A–F). Six2 transcripts were detected initially at e7.0-e7.5 cardiac crescent stage embryos by whole mount RNA in situ hybridization approach, and the expression pattern was restricted to two bilateral discrete domains (Figure 1A and B). At e8.0 or somatic stage (ss) 2 (Figure 1C), strong Six2 signal was observed in the cephalic mesoderm flanking but not in the developing heart tube. A similar overall expression pattern was maintained at e8.25/ss5 and e8.75/ss9 (Figure 1D and E). At e9.25/ss14, the Six2 expression domain extended to the developing pharyngeal arches (PAs) (Figure 1F). A series of histological sections of e8.75/ss9 and e9.25/ss14 embryos revealed strong Six2 signal in the splanchnic mesoderm of the dorsal pericardial wall (Figure 1G and H), which belongs to the SHF. At both of these stages, Six2 expression was restricted to isolated regions. Specifically, the Six2 expression domain was localized bilaterally along the mediolateral axis and clearly absent from midline tissues including the dorsal mesocardium (DM, Figure 1G). Six2 expression appeared to be stronger on the left versus the right side of SHF at e8.75. Along the rostrocaudal axis, Six2 was detected in the discrete subdomain of SHF (Figure 1H, brackets). Its expression was clearly absent from the rostral half and the most caudal one quarter of SHF at e9.25. Six2-expressing cells were further characterized using Six2^GC knock-in mice, which express an enhanced GFP (eGFP) and Cre fusion protein (GC) from the endogenous Six2 locus (Wang et al., 2013). The overall staining pattern of GFP-specific antibody was similar to that detected by Six2 RNA in situ hybridization, confirming that GC is a dependable surrogate for endogenous Six2. We found that some Six2⁺ cells were also positively stained by Islet1-specific antibody (Figure 1I). In addition, Six2 and Islet1 were co-expressed in the pharyngeal groove ectoderm and the pharyngeal pouch endoderm (Figures 1G and I). High levels of Six2 transcripts were also detected at the tracheoesophageal complex (TEC) and the frontonasal prominence (FNP) at e9.25/ss14 (Figures 1F and H). Together, these findings suggest that Six2 is transiently expressed in the SHF.

Figure 1. Six2 is expressed in the subdomains of SHF.

(A-F) Six2 gene-specific whole mount RNA in situ hybridization of mouse embryos from e6.0 to e9.25/ss14.

(G) Serial cross sections from rostral to caudal direction of whole mount stained embryos shown in E. All sections are orientated with the left side on right and the dorsal on top.

(H) Serial sagittal sections from middle to the left side of the stained embryos shown in F. All sections are orientated with the rostral on top and the dorsal on left. Dash line, SHF; bracket, the Six2-expressing subdomain.

(I) Immunofluorescence staining of Six2^GC embryos at e9.0 using eGFP (surrogate of Six2, Green) and Islet1 (Red) specific antibodies. Blue, DAPI counter staining; white, co-staining of all three colors (white arrowhead).

(J-M) Whole mount X-gal staining of Six2^GC;R26R^lacZ embryos (J, e10.5) and serial histological frontal sections of the stained embryos (K-M). Pink, eosin counter staining; blue, X-gal staining. White arrow points to the pharyngeal arch mesoderm core.

(N) Microdissected heart of Six2^GC;R26R^mTmG embryos at e10.5. Green, positive signal of the Six2⁺ genetic lineages; Red, negative signal; yellow, green cells located on the top or bottom of red cells.

(O) Three dimensional (3D) imaging of X-gal staining (volren red) of Six2^GC;R26R^lacZ double heterozygous embryos at e10.5. White arrow points to the pharyngeal arch mesoderm core.

(P) A virtual sagittal section of X-gal staining (white) obtained from the 3D image (O). Red dash line, SHF; insert, whole mount view; white dash line, section plane shown in P.

The scale bar is 250μm.

See also Figures S3 and movie1

AS, aortic sac; C, conus; CM, cephalic mesoderm; DM, dorsal myocardium; Ect, ectoderm; End, endoderm; FG, foregut; FL, forelimb; FNP, frontonasal prominence; Ht, heart; IFT, inflow tract; LA, left atrium; LV, left ventricle; Md, mandible; Mx, maxillary; NC, nephrogenic cord; NT, neural tube; OFT, outflow tract; PA, pharyngeal arch; PE, pharyngeal endoderm; RA, right atrium; RV, right ventricle; S, stomach; SHF, second heart field; T, truncus; TE, tracheal endoderm; TEC, tracheoesophageal complex.

To determine unequivocally that Six2⁺ progenitors are directly involved in the SHF development, we indelibly labeled them and their descendants using the Six2^GC Cre driver (Wang et al., 2013) and Cre-dependent R26R reporters (Soriano, 1999, Muzumdar et al., 2007, Madisen et al., 2010), and then analyzed the distribution of all labeled cells, i.e., Six2⁺ genetic lineages (Figures 1J–M, S3 and movie 1). As shown in Figure 1J, X-gal stained cells were found extensively in the developing heart tube at e10.5. Serial histological sections of Six2^GC;R26R^lacZ embryos revealed that the entire OFT myocardium through the presumptive truncus and conus, the OFT endocardium, and the majority of primitive RV myocardium were derived from Six2⁺ progenitors (Figures 1K–M). Very few X-gal stained cells were detected in the developing left ventricle (LV) or atria. These observations were confirmed using a different Cre reporter (Muzumdar et al., 2007), R26R^mTmG, where the Six2^GC genetic lineage was labeled by membrane-localized green fluorescent protein (mG) and directly visualized using epifluorescence microscopy (Figures 1N). The detailed distribution pattern of Six2⁺ genetic lineages was further characterized using high-resolution episcopic microscopy (HREM) (Figures 1O, P and movie 1). In addition to the RV and OFT, the HREM 3 dimensional (3-D) images revealed that Six2⁺ genetic lineages also contributed to the pharyngeal mesoderm cores and the tracheoesophageal mesenchyme (TEC). Virtual sagittal sections obtained from the 3-D reconstructed embryos revealed that, unlike the restricted expression pattern of Six2 in the SHF (Figure 1G and H), descendants of Six2⁺ progenitors were distributed broadly in the SHF towards the OFT at e10.5 (Figure 1P, red dash line). Only few labeled cells were observed in the developing atrial septum. Collectively, these findings demonstrate that Six2 transiently expresses in the subpopulation of SHF progenitors that contribute to the RV and OFT.

Six2 marks the rapidly changing pools of progenitors

To determine whether Six2⁺ progenitors are a collection of homogeneous or heterogeneous progenitors, we next interrogated the behavior of Six2⁺ progenitors at different developmental stages (Figure 2, S4 and S5). Here, we used the Six2^GCE mouse line in which Six2 coding sequence was replaced by GCE fusion gene (eGFP and tamoxifen-activated CreERT2)(Wang et al., 2013). Six2^GCE mice were crossed with R26R^lacZ reporter mice using a two-hour mating strategy to better estimate the timing of conception. A single dose of tamoxifen was administered via oral gavage to indelibly label Six2⁺ cells at 11 different embryonic stages between e6.0 and e10.0. Because of stage variation among littermates, perdurance of tamoxifen activity and variation of plug time, the exact timing of cell labeling could vary as much as one day. For instance, if tamoxifen is given at e6.0, the exact time of cell labeling would be between e6.0 and e7.0, and so on. The labeled cells were chased and analyzed at stages between e8.5 and e17.5 (Figure 2, supplementary Figures S4, S5, Table S1 and S2).

Figure 2. Six2⁺ progenitors consist of a temporally distinct heterogeneous subset of second heart field.

(A) A experimental scheme of pulse-chase indelible labeling of Six2⁺ progenitors and their descendent using the tamoxifen (TM)-inducible lacZ genetic lineage marker. Green dots, stage of TM treatment (10–100 μg/g of body weight via oral gavage); red dots, stage of X-gal staining analysis.

(B and C) Representative images (B) and summary (C) of the X-gal staining patterns of Six2^GCE; R26R^lacZ mouse hearts at e17.0 that have received a single dose of TM at the indicated stage.

(D and E) Representative images (D) and summary (E) of X-gal staining patterns of Six2^GCE; R26R^lacZ embryos and dissected hearts at e11.0. These embryos have received a single dose of TM at the indicated stages from e6.3 to e9.0. Left panels in D, whole embryos; middle two panels in D, ventral and right side views of the microdissected hearts; right panel in D, the schematic representation of the ventral view of hearts, Blue, positive signal of X-gal staining. Dash lines demarcate the outflow tract region. The scale bar is 1mm.

See also Tables S1–S2

AT, aortic trunk; C, conus; DA, ductus arteriosus; TEC, tracheoesophageal complex; If, infundibulum; LV, left ventricle; PT, pulmonary trunk; RV, right ventricle; T, truncus; TM, tamoxifen.

Descendants of Six2⁺ progenitors were readily detectable at e14.0 and e17.0 when tamoxifen was given at e6.0 or cell labeling between e6.0 and e7.0 (Figures 2B, C, and Table S1, n=4). Labeled cells were predominantly located at the RV and occasionally the LV. Notably, no labeled cells was located at the intrapericardial arterial trunks. When tamoxifen was administered at e8.3, labeled cells were found in the pulmonary trunk and the ductus arteriosus but not the RV (n=7). Overall, descendants of Six2⁺ progenitors between e6.0 and e10.0 were observed progressively from the RV to infundibulum to pulmonary trunk and lastly, the ductus arteriosus (Figure 2C and supplementary Table S1).

To understand the process through which Six2⁺ genetic lineages were allocated to the heart, we next analyzed the distribution of labeled cells at e11.0 (Figures 2D, E and supplementary Table S2). At this stage, the OFT consists of a single unseptated myocardial channel that is subdivided into the conus proximally and the truncus distally. When TM was given at e6.3, the majority of labeled cells were found at the developing RV and conus (n=5). No labeled cell was detected within the truncus. When TM was given at e7.0, the distribution shifted to the conus and truncus (n=8). Few labeled cells were detected in the developing RV. While labeled cells were observed occasionally in the pharyngeal arch and the frontonasal prominence, none was found in TEC or the aortic sac region at these stages. However, when TM was given at e8.0, most labeled cells distributed at the TEC, pharyngeal arches, and frontonasal prominences (n=5). A few of them were found in the conus and truncus. When TM was given at e9.0, no labeled cell was observed in the RV, conus or the truncus (n=5). Instead, all of the labeled cells were found in the TEC and frontonasal prominences. Collectively, these findings suggest that Six2 marks a pool of heterogeneous and temporally distinct subpopulation of SHF progenitors, which are rapidly recruited and assigned to form different components of the heart from the RV to the pulmonary trunk.

The intrapericardial aortic and pulmonary trunks have distinct embryonic origins

Embryonic cardiac OFT is a single tube with a myocardial wall. The definitive OFT, however, has two separate channels, each of which has three distinct segments (Figure S2). The entire embryonic OFT comes from Six2⁺ genetic lineages at e10.5 (Figure 1J–P). Unexpectedly, only a fraction of the intrapericardial aortic trunk was populated by Six2⁺ genetic lineages at e17.0 despite that the entire intrapericardial pulmonary trunk was labeled (Figure 3A–D). Six2⁺ genetic lineages also contributed extensively to the ductus arteriosus, suggesting that Six2 is also expressed in a subset of neural crest cells. To determine whether the Six2-negative neural crest cells might contribute to the aortic trunk smooth muscle, we analyzed the distribution of Wnt1-Cre⁺ neural crest lineages (Epstein et al., 2000, Jiang et al., 2000, Olaopa et al., 2011). Indeed, neural crest lineages contributed much more extensively to the intrapericardial aortic trunk than to the pulmonary trunk (Figures 3E–H). As expected, Wnt1-Cre⁺ neural crest lineages also contributed to the ductus arteriosus smooth muscle wall formation (Figure 3E). To further confirm these observations, both Six2⁺ and Wnt1-Cre⁺ genetic lineages were examined using a different reporter, R26R^tdT, to label the lineages with a fluorescent protein variant tdTomato (tdT) (Madisen et al., 2010) (Figures 3I–K). Tissue sections were stained with smooth muscle and endothelial cell-specific markers SMA and CD31, respectively. The pulmonary trunk smooth muscle originated from Six2⁺ genetic lineages while the aortic trunk smooth muscle cells largely come from Wnt1-Cre⁺ neural crest lineages (Figure 3K). A thin layer of smooth muscle cells of both arterial trunks was co-labeled by Wnt1-Cre⁺ and Six2⁺ genetic lineage markers. The adventitial and endothelial cells of both arterial trunks were derived from Six2⁺ but not Wnt1-Cre⁺ genetic lineages.

Figure 3. The intrapericardial arterial trunks have distinct embryonic origins.

(A-H) X-gal staining of the dissected hearts and four chamber view sections (note, atria were removed) of Six2^GC;R26R^lacZ (A-D) and Wnt1^Cre;R26R^lacZ (E-H) embryos at e17.0. Blue, positive signal of X-gal staining; pink, eosin counter staining; white dash line, pericardium. The scale bars correspond to 1mm (A,E), 250μm (B,F) and 25μm (C,D,G,H), respectively.

(I and J) Immunofluorescence staining of the arterial trunk vessels of Six2^GC;R26R^tdT (1 and 2) and Wnt1^Cre;R26R^tdT (3 and 4) embryos at e17.5 with anti-SMA antibody (green, I) or anti-CD31 antibody (Green, J). Red, the genetic marker of Six2⁺ lineages (1 and 2) or Wnt1-Cre⁺ lineages (3 and 4). Yellow, co-localization of red and green; Blue, DAPI counter staining. The scale bar is 25μm.

(K) Schematic representations of embryonic origins of the aortic and pulmonary trunks. Red, the Six2⁺ genetic lineages; green, the Wnt1-Cre⁺ genetic lineages; yellow, originated from cells positive for both Six2 and Wnt1-Cre.

(L-N’) Serial cross sections (L-N) and accompanying diagrams (L’-N’) of X-gal staining (Blue) of Six2^GC;R26R^lacZ embryos at e9.5 that have received a single dose of tamoxifen (TM) at e8.6. Pink, eosin counter staining; blue dot, X-gal positive cells; arrowhead, left caudal SHF. The scale bar is 200μm.

(O-Q”) Whole mount X-gal staining of Six2^GC;R26R^lacZ double heterozygous embryos (O-Q) and microdissected hearts (O’-Q’) with accompanying diagrams (O”-Q”). All embryos have received a single dose of tamoxifen at e8.6 but analyzed at different stages as indicated. Blue dot/circle, the size and number of which correlate to the abundance of X-gal stained cells. The scale bar is 1mm.

See also Figures S5

AT, aortic trunk; Ad, adventitia; AS, aortic sac; DA, ductus arteriosus; DM, dorsal myocardium; En, endothelium; FG, foregut; LV, left ventricle; MD, mandible; OFT, outflow tract; PT, pulmonary trunk; RV, right ventricle; SHF, second heart field; SM, smooth muscle cells;

Six2⁺ progenitors are committed to the pulmonary trunk fate as early as e7.0 and peak around e8.5 (Figure 2C). Since Six2 is not expressed in the embryonic OFT at any time (Figures 1E and F), these findings suggest that the Six2⁺ pulmonary trunk progenitors likely reside outside the developing heart tube. To examine this possibility, we pulse-labeled them at e8.3 or e8.6 and chased the descendants from e9.5 to e12.5, the critical period of pulmonary trunk morphogenesis (Figures 3L–Q and S5). At e9.5, the labeled cells were scattered around the pharyngeal arch apparatus (Figure S5). Few of them were located at the developing frontonasal prominences. Analysis of serial histological sections revealed that labeled cells were concentrated at the caudal left region of the dorsal pericardial wall (Figure 3L–N, n=11). At later stages between e10.5 and e12.5, whole-mount images demonstrated strong labeling of the pharyngeal arches and frontonasal prominences (Figure 3O–Q). Microdissected hearts from these embryos indicated that few labeled cells were localized within the OFT at e10.5. Instead, a group of labeled cells resided outside and caudal to the OFT (Figure 3O–O”, n=4). A day later at e11.5, these labeled cells became closer to the OFT (Figure 3P–P”, n=5). By e12.5, labeled cells were found at the developing pulmonary trunk (Figure 3Q–Q”, n=4). Together, results from these prospective analyses indicate that the Six2⁺ pulmonary trunk lineages are recruited progressively to the heart from the caudolateral region of the dorsal pericardial wall.

The Shh signaling pathway regulates behavior of Six2⁺ progenitors

The endoderm-derived Shh signal is essential for OFT development (Goddeeris et al., 2007, Dyer and Kirby, 2009), and the Shh-receiving cells contribute directly to the pulmonary trunk and infundibulum (Hoffmann et al., 2009). We therefore tested the hypothesis that Shh regulates development of Six2⁺ progenitors by examining initially whether Shh regulates Six2 gene expression. While strong Six2 gene expression was detected in Shh heterozygous control embryos at e7.0-e7.5, only weak signal was detected in Shh homozygous mutant littermate (Figure 4A and B). At e9.0, Six2 expression was detected but was visibly reduced at the rostral and caudal expression subdomains of Shh mutant embryos (Figures 4C–H). Serial histological sections of these stained embryos revealed that Six2 expression was reduced in the SHF, particularly on the caudal left side of the SHF. Therefore, Six2 gene expression depends in part on Shh signal.

Figure 4. The Shh signaling pathway regulates development of Six2⁺ progenitors.

(A-H) Six2 expression in Shh^+/− (A, C, E and F) and Shh^−/− (B, D, G and H) embryos at e7.0 - e7.5 (A and B) and e9.0 (C-H). Dash lines in C and D, cross section planes shown in E-H. white arrows point to reduced Six2 expression.

(I-M) Phospho-histone H3 (pHH3) immunofluorescence of cross sections of e9.0 Shh^+/− (I and J) and Shh^−/− (K and L) embryos. Blue, DAPI counter staining. pHH3 stained cells (green) were quantified based on their location at the rostral or caudal subdomains of the SHF and the result is shown in M. *, p<0.05, Student’s t-test; ns, not significant.

(N-S) The distribution of Six2 genetic lineages (green) in Six2^GC;Shh^+/−;R26R^mTmG control (N-P) and Six2^GC;Shh^−/−;R26R^mTmG (Q-S) mutant embryos at e10.5. Red, the non-Six2 lineages; yellow, green Six2 lineages layered with the red non-Six2 lineages. Dash lines in N and Q indicate the cross section planes shown at the level of OFT (O and R) and caudal SHF (P and S).

The scale bar is 250μm.

AS, aortic sac; At, atrium; DM, dorsal myocardium; En, endoderm; FG, foregut; FNP, frontonasal prominence; L-SHF, left-second heart field; Md, mandible; OFT, outflow tract; PA2/3, pharyngeal arch 2/3.

Reduced expression of Six2 in Shh mutants suggests that Shh may regulate development of Six2⁺ progenitors. Indeed, percentage of phospho-histone H3-positive mitotic cells was significantly reduced in Shh mutants at the caudal SHF (Figures 4I–M and S9, n=4). Consistent with previous reports (Washington Smoak et al., 2005, Goddeeris et al., 2007, Hoffmann et al., 2009, Dyer and Kirby, 2009), the proliferation rate of the rostral SHF was comparable between control and Shh knockouts. Next, we indelibly labeled Six2⁺ progenitors with the mG genetic lineage marker and analyzed their behavior in Shh mutant embryos at e10.5 (Figure 4N–S). Whole mount analysis demonstrated a dramatic reduction of mG-labeled cells in the frontonasal prominences of Shh mutants, coinciding with the severe forehead defect. The rostral SHF had comparable numbers of mG-labeled cells (Figures 4O and R). However, contribution of Six2⁺ genetic lineages to the caudal SHF was dramatically reduced (Figures 4P and S), reminiscent of the proliferation defect of the caudal SHF. Taken together, these findings suggest that Shh is a critical upstream regulator of Six2 and is required for proliferation and deployment of the caudal SHF progenitors.

Global ablation of Six2⁺ progenitors causes severe cardiovascular malformations

To examine directly whether Six2⁺ progenitors are essential for cardiovascular development, we genetically ablated them by expressing diphtheria toxin fragment A (DTA) (Voehringer et al., 2008) in Six2⁺ progenitors and their descendants (Figure 5). To validate the experimental strategy, we first stained Six2^GC;R26R^DTA compound heterozygous embryos with LysoTracker Red to assess the overall pattern of cell death. Whole mount examination revealed abnormally high levels of LysoTracker staining at the pharyngeal arch apparatus and the frontonasal prominences (Figures 5A and E). A detailed histological analysis of these stained embryos demonstrated that the LysoTracker stained cells were concentrated near the caudal SHF and TEC (Figures 5B–D and F–H). Stained cells were also detected at the rostral SHF and OFT region. Collectively, these findings indicate effective ablation of Six2⁺ progenitors and their descendants in the double heterozygous mutants.

Figure 5. Global ablation of Six2⁺ progenitors causes severe cardiovascular and craniofacial malformations.

(A-H) Lysotracker® staining to show the cell death patterns of Six2^GC/+ control (A-D) and Six2^GC/+;R26R^DTA/+ mutant (E-H) embryos at e9.5. A and E, whole-mount views of positive staining (white dots) from the right lateral side of the embryos; B-D and F-H, sagittal sections from embryos shown in A and E. Red, Lysotracker® staining; blue, DAPI counter staining.

(I-L) RNA in situ hybridization of Six2^GC/+ (I and K) and Six2^GC/+;R26R^DTA/+ (J and L) hearts at e9.5 using Wnt11 (I and J) and Tbx5 (K and L) gene-specific probes (purple).

(M-T) Six2^GC/+ (M, O, Q and S) and Six2^GC/+;R26R^DTA/+ (N, P, R and T) embryos at e14.5 and e17.5 are visualized as whole mount (M and N), microdissected hearts (O-R) and four chamber view histological section (S and T). Note, atria were removed; *, ventricular septum defect.

The scale bar is 500μm.

See also Figures S6

AT, aortic trunk; CAT, common arterial trunk; CT, conus and truncus region; FB, forebrain; FNP, frontonasal prominences; Ht, heart tube; L-PA, left pulmonary artery; LV, left ventricle; Md, mandible; OFT, outflow tract; PT, pulmonary trunk; RV, right ventricle; SHF, second heart field.

The potential cardiovascular phenotype of embryos lacking Six2⁺ genetic lineages was examined at several later stages (Figure 5I–T). At e9.5, the developing RV and conotruncus were apparently smaller and hypoplastic in the mutants. Expression of Wnt11, a molecular marker that is strongly expressed in the OFT, was dramatically reduced in the mutants (Figures 5I and J). In contrast, the expression pattern of Tbx5, which is highly enriched in the developing LV, was not affected in the mutants (Figures 5K and L). A single common arterial trunk was observed in the mutants at e14.5 (Figure 5P). At the same stage, the separate aortic and pulmonary channels were clearly established in control embryos. Additionally, the RV and infundibulum were translucent and smaller in size in these mutant embryos (Figure 5M–P). At e17.5, the pulmonary arteries were formed but were inserted ectopically into the common arterial trunk (Figure 5R). The malformed arterial trunk lacked the genetic lineage tracer of the pulmonary trunk at e17.5 (Figure S6), suggesting a pulmonary atresia defect. A severe ventricular septation defect was also observed (Figure 5T). In addition, the double heterozygous mutants exhibited severe craniofacial malformations, including absence of mandibular and nasal structures and a potential brain anomaly (Figures 5M and N). Thus, ablation of Six2⁺ progenitors causes severe cardiovascular anomaly.

An early stage-specific ablation of Six2⁺ progenitors causes adult-onset heart dysfunction

Since new Six2⁺ progenitors are rapidly recruited and assigned, we therefore asked whether ablation of a subset of Six2⁺ progenitors might cause structure-specific cardiac defects. To determine how many cells would be ablated, we first used Six2^GCE;R26R^lacZ mice to estimate how many cells could be labeled under the experimental settings. Labeled cells were stained and analyzed at different stages after receiving a single dose of tamoxifen (100 μg/g, oral gavage) at e7.3. Most labeled cells were localized in the cephalic mesodermal regions and only a few of them were found in the developing heart tube at e8.5 (Figures 6A and B, n=8). A similar overall distribution pattern was observed at e9.5 (Figures 6C and D, n=12). Quantitative analysis of X-gal stained cells at these two stages revealed that there were approximately 4 labeled cells in the developing heart and ~100 at the pharyngeal region at e8.5 (Figure 6E). These numbers increased to 10 and 175 at e9.5, respectively, potentially due to proliferative expansion of the labeled cells. Many more labeled cells were found in the craniofacial regions at e14.5 (Figures 6F). In the developing cardiovascular system, labeled cells were dispersed throughout the RV, the infundibulum and the arterial trunks at e14.5 (Figures 6G, n=4). In addition, labeled cells were also detected in the aortic (~70 cells per heart) and pulmonary (~74 cells per heart) semilunar valves by e17.5 (Figures 6H–K, n=7). In summary, while the majority of Six2⁺ progenitors labeled at e7.3 is located at the craniofacial regions, a small number of them (~10 cells at e9.5) contribute directly to the RV and OFT.

Figure 6. An early temporal-specific ablation of Six2⁺ progenitors causes adult-onset heart dysfunction.

(A-E) X-gal staining of Six2^GCE/+;R26R^LacZ/+ embryos at e8.5 (A-B), e9.5 (C-D), e14.5 (F-G) and e17.5 (H-K) after receiving a single dose of TM at e7.3. Whole mount embryos (A, C and F) and micro-dissected hearts (B, D, G and H) and four-chamber views (I, J and K) are shown. Quantification of X-gal staining of e8.5 and e9.5 embryos is shown (E).

(L-P) Echocardiography to measure cardiac functions of 12–16 week-old Six2^iDel mice (n=7, black column) and their littermate controls (n=9, white column). Data are presented as mean ± SEM, *, p<0.05, Student t-test.

(Q) Whole mount micro-dissected hearts of 12-week-old littermate control and Six2^iDel mutant. (R and S) The four chamber view histological sections of hearts from 12-week-old littermate control and Six2^iDel mutant. *, thick RV free wall.

(T and U) Fast Green (non-collagenous protein) and Sirius Red (fibrillar collagen) staining of transverse heart sections.

(V) The heart weight to tibial length ratio of control (n=5) and Six2^iDel mice (n=6). Data are presented as mean ± SEM, *, p<0.05, Student t-test.

The scale bar is 1mm.

See also Figures S7 and Table S3 and S4

AT, aortic trunk; AV, aortic valve; d, diastolic; Ht, heart tube; HW, heart weight; IFT, inflow tract; LA, left atrium; LV, left ventricle; OFT, outflow tract; PA peak velocity, pulmonary artery peak velocity; PV, pulmonary valve; RA, right atrium; RV, right ventricle; RVEA, right ventricle empty area; RV FAC, right ventricle fraction area change; RV FS, right ventricle fraction shortening; RVID, right ventricular internal dimension; PT, pulmonary trunk; TL, tibia length; s, systolic.

We next ablated a similar subset of Six2⁺ progenitors with R26R^DTA mice using the same regimen, i.e. 100 μg/g of tamoxifen administered at e7.3 through oral gavage (Figure 6L–V). Six2^GCE;R26R^DTA embryos were treated with a single dose of tamoxifen at e7.3. These mice, Six2^GCE;R26R^DTA, were termed Six2^iDel to represent induced deletion of a subset of Six2⁺ progenitors. R26R^DTA heterozygote littermates that received the same tamoxifen treatment were used as negative controls. Six2^iDel mutants did not display any apparent gross morphological defects at birth. These mutants had similar growth curve and reproductive capacity to controls.

However, echocardiography identified functional defects of the mutants (Figures 6L–P, supplementary Table S3 and movie 1). Heart rate of controls was 659±13 (n=7) beats per minutes, which was similar to Six2^iDel mutants (646±8 BPM, n=9). LV dimensions and LV fractional shortening, a measure of systolic function, were comparable between mutant and control mice (Table S3). In contrast, echocardiographic measures of RV function(Cheng et al., 2014) were abnormal in Six2^iDel mutants. The RV wall of Six2^iDel mutants (n=8, 0.62±0.06 mm) was significantly thicker than control mice (n=6, 0.42±0.06 mm, Figure 6L and Table S4). RV internal diameter at end systole (RVID;s) but not end diastole (RVID;d) was increased and, consequently, RV fraction shortening (RVFS) was significantly reduced in Six2^iDel mutants (Figures 6M, N and Table S4). Analysis of RV areas from B-mode recordings corroborated these findings in Six2^iDel mutants (Figures 6O, P and Table S4). Histopathological and gross analysis of Six2^iDel hearts at 12–16 weeks of age reinforced these findings (Figure 6Q–V). The mutant hearts were significantly larger than the littermate controls. Morphological measurements on histological sections showed that the RV was dilated (Figure 6S). The free wall of the right but not the left ventricle was significantly thickened. We did not detect significant levels of tissue fibrosis, cellular hypertrophy or obvious structural anomalies of the semilunar valves (Figures 6T, U and S7). The heart weight to tibial length ratio in Six2^iDel mutants was significantly increased compared to controls (8.5 ± 1.17 mg/mm vs. 6.25 ± 0.16 mg/mm; P<0.05; Figure 6L–O). Collectively, these findings suggest that a temporal insult to cardiac progenitors during early embryogenesis may have a long-term impact on adult cardiac function.

DISCUSSION

There are four principal models concerning formation of the multiunit structures/organs depending on the nature of progenitor pool (homogenous or heterogeneous) and dynamics of the pool (permanent or changing) (Figure S1). Each model predicts the radically different lineage distribution patterns. This study identifies the temporally distinct heterogeneous population of Six2⁺ progenitors that contributes sequentially to components of the cardiovascular system (Figure 7). The chronological features of Six2⁺ progenitors detailed here provide insights into the mode of cardiogenesis and onset of congenital heart disease.

Figure 7. A proposed model of Six2⁺ progenitors contributing to mammalian heart development.

(A) Six2⁺ progenitors are located outside the embryonic heart tube. Six2 is expressed as early as e7.0 – e7.5. At e9.0, Six2 is restricted to a subdomain (bracket) of SHF, and its’ expression depends in part on the Shh signaling pathway. Descendants of Six2⁺ progenitors are sequentially recruited from the arterial pole to the heart tube. The early Six2⁺ progenitors become right ventricle (RV, red), infundibulum (If, yellow), and the later Six2⁺ progenitors form the arterial roots (green) and the pulmonary trunk (PT, blue). Unlike PT, the intrapericardial aortic trunk (AT, grey) smooth muscle is largely from the Wnt1-Cre⁺ cardiac neural crest (CNC) cells. The ductus arteriosus (DA, purple) originates from progenitors positive for both Six2 and Wnt1-Cre genetic lineage markers. LV, left ventricle.

(B) A model of building mammalian heart in vivo from pools of heterogeneous progenitors. Temporally distinct cardiac progenitors (large bi-color circles, upper panel) are rapidly recruited and assigned (or committed) between e6.0 and e11.0 of mouse embryos. These progenitors are deployed sequentially to the LV, RV, and lastly the OFT (small mono-color circles) of the heart.

See also Figures S1 and S2

Model of mammalian cardiogenesis

Six2⁺ progenitors contribute extensively to components of the cardiovascular structures including RV, IF, PT and DA. Early Six2 expression was detected by in situ hybridization as early as e7.0. However, the tamoxifen-induced lineage mapping results explicitly imply that Six2 is expressed even before e7.0 possibly at levels too low to be detected by in situ hybridization. Within the temporal resolution (as much as 24 hours) of tamoxifen-inducible system, results of the pulse-chase labeling and lineage distributions of Six2⁺ cardiac progenitors suggest several basic principles of cardiogenesis (Figure 7). For example, new structures are progressively added to existing structures, such that the RV forms first and is followed by infundibulum. The intrapericardial pulmonary trunk and ductus arteriosus develop thereafter. The RV myocardial progenitors are specified as early as e6.0-e7.0 during gastrulation. Descendants of early-labeled progenitors do not extend into the late structures. Whether two adjacent structures, such as the non-myocardial arterial root and the sub-arterial myocardium, come from the same multipotent progenitors remains to be determined. These findings are consistent with the previous observations, in which the labeled cardiac progenitors are progressively incorporated into the heart in chick (de la Cruz et al., 1977) and mouse embryos (Zaffran et al., 2004, Lescroart et al., 2010); the cardiac progenitors are rapidly assigned to discrete anatomic structures prior to or shortly after the initiation of gastrulation (Devine et al., 2014, Lescroart et al., 2014, Spater et al., 2013, Liang et al., 2013, Choquet et al., 2016); and the cardiac progenitors follow the proliferative mode instead of stem cell mode growth (Meilhac et al., 2003, Lescroart et al., 2012). While the presumptive common pool progenitors, such as the proposed cardiac founder cells (Meilhac et al., 2004, Xavier-Neto et al., 2012), might exist outside of Six2⁺ and/or Mesp1⁺ progenitor pools, our findings suggest that mammalian heart is built sequentially from several temporally distinct pools of progenitors (Figure 7B).

Distinct embryonic origins of the intrapericardial arterial trunks

One of the unexpected findings of this study is that Six2⁺ genetic lineages contribute predominantly to the pulmonary trunk but much less so to the aortic trunk (Figure 7A). The asymmetric distribution pattern of genetic lineages to the arterial trunks has been observed previously. For example, Tbx1⁺ genetic lineages contribute much more to the pulmonary trunk than the aortic trunk (Huynh et al., 2007). We found that Six2 is expressed in progenitors that are committed to the pulmonary trunk fate as early as e7.0 - e8.0. While we could not rule out the possibility that a common progenitor pool prior to onset of Six2 expression may contribute to both the aortic and pulmonary trunks, the presence of a presumptive Six2-negative “common progenitor”, therefore, would have to be before e8.0 - four days ahead of the morphological signs of aortic and pulmonary trunk differentiation. Cardiac neural crest cells (CNCs) marked by Wnt1-Cre⁺ genetic lineages also contribute to the arterial trunk smooth muscle and valve formation (Epstein et al., 2000, Jiang et al., 2000, Olaopa et al., 2011, Lewis et al., 2013). The original study noted that Wnt1-Cre⁺ genetic lineages contribute more abundantly to the aortic than pulmonary trunk (Jiang et al., 2000). This initial observation was further confirmed using an independent mouse line (Lewis et al., 2013, Huang et al., 2008). Unlike Six2⁺ genetic lineages, which contribute to the entire pulmonary trunk, Wnt1-Cre⁺ genetic lineages contribute only to a thin band of subendothelial pulmonary trunk. In contrast, Wnt1-Cre⁺ genetic lineages are enriched within the aortic trunk smooth muscle layer. The ductus arteriosus is positive for both Six2 and Wnt1-Cre genetic lineage markers. Therefore, there is a small intersection between the Six2⁺ and Wnt1-Cre⁺ genetic lineages. Together, these findings demonstrate that the aortic and pulmonary trunks originate from different genetic lineages, and the intrapericardial arterial trunks are prefigured before any morphological signs of anatomical separation.

In addition to the critical roles of the SHF in cardiovascular development, recent findings suggest that formation of the facial skeletal muscles is also linked to the SHF (Diogo et al., 2015, Meilhac et al., 2014, Lescroart et al., 2015). We have observed that Six2⁺ progenitors contribute extensively to the facial musculature. However, since Six2 is expressed in both SHF and craniofacial mesoderm, we could not rule out the possibility that they are from two clonally unrelated progenitors.

Insights on embryonic origins of the intrapericardial arterial trunks have important implications in the pathogenesis of congenital OFT defects. Cardiac OFT defects account for nearly 1/3 of all congenital heart diseases. The common arterial trunk phenotype is one of the most severe forms of OFT defects. Clinical presentations of the common arterial trunk vary significantly and, are often classified as one of the four major types of persistent truncus arteriosus (PTA)(Collett and Edwards, 1949). This classification implicitly suggests that the embryonic basis of the phenotype is the failure of OFT septation. However, underdevelopment of the pulmonary or aortic trunks could also lead to clinical presentations of common arterial trunk with either aortic or pulmonary dominance, respectively (Russell et al., 2011). We show that ablation of Six2⁺ progenitors causes a common arterial trunk phenotype. This is likely due to agenesis of the pulmonary trunk. Since Six2 is also expressed in the subset of neural crest cells that contribute to cardiovascular structures including ductus arteriosus, we could not rule out the possibility that ablation of Six2⁺ neural crest might also contribute to the OFT phenotypes. Interestingly, ablation of the Wnt1-Cre neural crest lineages also results in the common arterial trunk phenotype (Olaopa et al., 2011). In this case, however, it remains to be determined whether the phenotype is the aortic atresia, pulmonary atresia or PTA.

Lasting impacts of Six2⁺ cardiac progenitor cell deficiency

Consistent with the critical distribution pattern of Six2 genetic lineages, global ablation of Six2⁺ progenitors causes severe cardiac and craniofacial malformations. A stage-specific ablation of a small subset of Six2⁺ progenitors at the cardiac crescent stage causes no apparent structural defects at birth. This is expected because Six2⁺ progenitors are highly proliferative, and a slight reduction of these progenitors should be reasonably compensated and/or tolerated. Unexpectedly, these animals grow up developing adult cardiac hypertrophy and RV dysfunction. While the process leading to the adult-onset cardiac dysfunction remains to be determined, possible reasons likely include the insufficiency of Six2⁺ genetic lineages and/or the lack of Six2⁺ cells in the adult. Nonetheless, these observations strongly suggest that embryonic insults may manifest later in life and cause adult heart disease.

EXPERIMENTAL PROCEDURES

Mice

All animal studies were performed based on protocols reviewed and approved by the Institutional Animal Care And Use Committee at Boston Children’s Hospital. Mouse lines, including Six2^GCE, Six2^GC mice were reported(Wang et al., 2013). Wnt1-Cre (JAX 022137), Shh^GC (JAX 005622), R26R^mTmG (JAX 007676), R26R^LacZ (JAX 003309), R26R^DTA (JAX 009669) and R26R^tdT (JAX 007908) mice were purchased from the Jackson Laboratory. All mice were maintained on mixed genetic backgrounds.

Timed mating

We used a 2-hour timed mating strategy to obtain timed pregnancy. Briefly, male and females were put together when females are in oestrus. Two hours later, females were separated from the male. Appearance of a cervical mucus plug was examined 24 hours later, and was counted as embryonic day 1 (e1.0). Eight hours was considered 0.3 day, 16 hours 0.6 day and 24 hours 1.0 day.

Tamoxifen oral gavage

We used tamoxifen to induce Cre activity in Six2^GCE/+;R26R^LacZ/+ or Six2^GCE/+;R26R^DTA embryos. Tamoxifen (Sigma, T5648) was dissolved in ethanol and sesame oil mixture (1:9). The final stock concentration is 10 μg/ml. A single dose of tamoxifen (from 10 μg/g to 100 μg/g of body weight) was administered through oral gavage. Briefly, tamoxifen solution was directly administered into the lower esophagus or stomach using a feeding needle introduced into the mouth and threaded down the esophagus. Reusable curved feeding needles (22-gauge, Braintree Scientific) with the ball tip were used.

X-gal staining and Histology

X-gal staining was performed essentially as previously described (Guo et al., 2011, Wang et al., 2013). These methods are detailed in the Supplemental Experimental Procedures.

Immunofluorescence, in situ hybridization, cell death and collagen staining

Immunofluorescence, in situ hybridization and cell death assays were carried out essentially as described (Guo et al., 2011, Sun et al., 2012, Wang et al., 2011). Immunofluorescence staining performed as detailed in the Supplemental Experimental Procedures. Sirius red and fast green (FG&SR, Sigma) staining was described (Guo et al., 2015, Ding et al., 2015). The following antibodies were used for immunohistochemistry: SMA (1:200, Sigma), PECAM-1 (1:50, anti-CD31), Islet1 (1:100, DSHB, U of lowa), pHH3 (1:200, Upstate), Six2 (1:500, Proteintech) and GFP (1:200, Rockland). All secondary antibodies were purchased form Jackson ImmunoResearch Laboratories. Tbx5 and Wnt11 RNA in situ probes were described (Guo et al., 2011). LysoTracker® (Invitrogen) was used to detect cell death. Images were taken using Zeiss Axioplan 2 microscope.

To analyze the rostral and caudal subdomains of the SHF separately at e9.0 (Figure 4M), we divided the SHF as followings: serial cross sections were collected from every embryo (control and mutant). A total number of sections were counted from OFT to IFT and divided into the rostral and caudal half. Results from the rostral and caudal sections were counted and analyzed independently as shown in Figure 4I–M.

High resolution episcopic microscope (HREM) and 3D reconstruction

HREM was performed as described (Huang et al., 2016, Mohun and Weninger, 2012). These methods are detailed in the Supplemental Experimental Procedures.

Echocardiography

LV and RV functions were measured by transthoracic echocardiography using a Visual Sonics Vevo® 2100 Imaging System (Visual Sonics) with a 40 MHz MicroScan transducer (model MS-550D) as described (Cheng et al., 2014, Ding et al., 2015). This method is detailed in the Supplemental Experimental Procedures.

Statistics

Values are presented as mean ± SEM. Unpaired 2-tailed Student’s t-test and Fisher’s exact test were used to analyze the data. P<0.05 was considered statistically significant.

Highlights.

• Six2 marks temporally distinct subsets of second heart field progenitors

• Descendants of Six2-progenitors are allocated successively to discrete domains of the heart

• The Shh signaling pathway regulates development of Six2-positive genetic lineages

• Stage-specific ablation of Six2- progenitors causes late-onset of heart disease