Hand1 regulates cardiomyocyte proliferation versus differentiation in the developing heart

Heart morphogenesis begins shortly after gastrulation with the specification of the cardiac precursors of the cardiac crescent in the anterior lateral mesoderm followed by the migration of these precursors to the ventral midline and their fusion to form the linear heart tube(Harvey, 2002). These cells comprise the so-called primary heart field that later contributes mainly to the left ventricle (Cai et al.,2003). Subsequent rapid elongation coincident with cardiac looping then occurs following the recruitment of cells to the growing poles of the heart (Kelly and Buckingham,2002). At the arterial pole, the heart tube is extended by the addition of extracardiac cells that originate from progenitors in the pharyngeal mesoderm, dubbed the anterior (or secondary) heart field (AHF)(Kelly et al., 2001; Mjaatvedt et al., 2001; Waldo et al., 2001). Cardiomyocytes arising from the AHF take on right ventricular and outflow tract (OFT) identities (Zaffran et al.,2004). More recently, the AHF itself has been further redefined as a subpopulation of a broader second lineage that not only contributes to the right ventricle and OFT at the arterial pole of the heart, but also to the atria and inflow regions at the venous pole(Meilhac et al., 2004). Inroads into deciphering transcriptional regulation in the cardiac lineages suggest a requirement for combinatorial activities of multiple transcription factors that have an impact on differentiation and deployment of myocardial precursor cells. However, factors that maintain a balance between proliferation and differentiation of different populations of myocardial precursors have not been identified.

Hand1 (previously known as eHAND, Hxt and Thing1), a member of the bHLH transcription factor family, is a candidate regulator of cardiac precursor cell fate. Hand1 is expressed in distinct regions of the linear heart tube and, post-looping, becomes localised to both primary heart field and second lineage derivatives, specifically in the outer curvature of the presumptive left ventricle and the developing OFT, and also at lower levels in the outer curvature of the right ventricle. The role of Hand1 in heart development has previously been explored in mice lacking Hand1, but detailed analysis is confounded by early embryonic lethality resulting from extra-embryonic defects (Firulli et al.,1998; Riley et al.,1998). Rescue of the extra-embryonic defects by tetraploid aggregation experiments yielded limited insight into the precise cellular role of Hand1 in the heart because embryos were only partially rescued, with failed looping morphogenesis, defective chamber septation and impaired ventricular development (Riley et al.,1998).

Controversy surrounds the specific function of Hand1 in the heart in terms of whether it is involved in regulating looping morphogenesis as a primary role that feeds back directly onto ventricular development, or whether it is restricted to ventricular specification and maturation coincident with looping. Conditional targeting of Hand1 has not resolved this issue,as the defects observed following heart-specific knockout of Hand1(McFadden et al., 2005) are less severe than would be predicted based on the reported phenotype of tetraploid-rescued conventional Hand1-null embryos(Riley et al., 1998). In the conditional targeting study by McFadden and colleagues(McFadden et al., 2005), Hand1 cardiac knockouts under the control of a Nkx2.5 Cre driver formed hearts that clearly looped and had a degree of left-ventricular expansion. These observations are significantly milder than previously described for the tetraploid-rescued mutants, which are characterised by a complete failure in looping morphogenesis and aberrant ventricular differentiation (Riley et al.,1998). Putative limitations with respect to the Cre approach taken are probably caused by mosaicism of the Cre expression under the described Nkx2.5 promoter (McFadden et al.,2005) and/or a temporal delay in efficient Cre excision resulting in Hand1 being inactivated at the heart tube stage, excluding an effect on earlier cells expressing Hand1(Buckingham et al., 2005). Nevertheless, the conditional, heart-specific Hand1 mutant embryos presented with left-ventricular hypoplasia, leading to the postulation that this specific anomaly could be the result of a proliferation defect in the primary heart field lineage (Buckingham et al., 2005). However, because of the early lethality of mouse embryos lacking Hand1 and the limited insight gained from the conditional knockout approach, it was necessary to develop a model that would enable us to express Hand1 in an appropriate spatially and temporally controlled manner, circumventing complications associated with extra-embryonic defects, which could then be used to assess its precise role during heart development. To achieve this, we engineered the tetracycline system into the endogenous Hand1 locus. The Tet-Off transactivator (tTA) was targeted specifically by homologous recombination and thus placed under the control of the endogenous Hand1 promoter and regulatory regions. The resulting mouse tTA knock-in strain (driver) was then crossed to multiple transgenic lines (responders) to regulate expression of a tet-responsive Hand1transgene. In light of the limitations of pre-existing loss-of-function mouse models, we sought to use the tet-system knock-in approach to induce overexpression of Hand1, restricted exclusively to Hand1-expressing regions, and to facilitate the analyses of early-and late-onset functions of Hand1 in the developing heart. At E9.5,overexpression of Hand1 disrupted cardiac morphogenesis and brought about an extension of the heart tube and extraneous looping caused by the elevated proliferation of cardiac precursors in the distal OFT. In complimentary studies employing in vitro differentiated embryonic stem (ES)cells, the complete loss of Hand1 resulted in elevated cardiomyocyte differentiation, whereas Hand1 gain of function maintained cardiomyocyte proliferation, via G1 progression and delayed cell-cycle exit. The combined in vivo and in vitro studies, therefore, indicate a cell-autonomous role for Hand1 in regulating the balance between cardiomyocyte differentiation and proliferation during cardiac development.

For the driver targeting construct a 1 kb EcoRI-BamHI fragment containing the tTA transactivator was excised from pTet-Off(Clontech) and cloned into pKS+. A 0.8 kb rabbit β-globin poly(A)sequence (van Ooyen et al.,1979) and neomycin-resistance cassette from pPNT(Tybulewicz et al., 1991)flanked by loxP sites were cloned downstream of the tTA. This construct was then subcloned into a SalI-BglII digest of a Hand1-targeting vector described previously(Riley et al., 2000). For the responder transgenic construct, a full-length Hand1 cDNA was cloned downstream of the tetracycline response element (TRE) in pTRE2 (Clontech) and a neomycin-resistance cassette was inserted downstream in the opposing orientation. The driver and responder mouse lines were generated by genOway(Lyon, France). The driver targeted line was established by homologous recombination in 129sv ES cells. Correctly targeted ES-cell clones were identified by Southern blot analysis, using a BglII digest of genomic DNA and a 5′ probe that is a 2 kb EcoRI/KpnI fragment that hybridises to a 13 kb wild-type band and a 9 kb targeted mutant band. The responder transgenic strain was generated by `Safe DNA Transgenesis^'(http://www.genoway.com/). Positive ES-cell clones were again identified by Southern blot (see Fig. 1A-D). Transgene copy number was estimated by scanning densitometry using the Scion Image program(Scion Corporation; http://www.scioncorp.com/frames/fr_scion_products.htm). Positive clones were subsequently used for injection into C57BL/6-derived blastocysts for generation of chimeras using standard procedures, for both the targeted driver and transgenic responder lines.

Mice were maintained on a mixed background. Embryos were generated by mating heterozygous driver and responder mice. A total of 654 embryos were analysed and all were stage-matched to controls by somite count. The targeted driver allele was identified by a 690 bp product(S-5′-GGGGTGGGGTGGGATTAGAT-3′ and AS-5′-AGAAGGGCCCAGGGAAGACT-3′). The presence of the responder transgene was detected by a 2.2 kb product(S-5′-GCGCCTGGCTACCAGTTACA-3′ and AS-5′-GGGGTGGGGTGGGATTAGAT-3′).

ES cells containing the transgenic responder construct were co-transfected with pBIG3r (Strathdee et al.,1999) containing the tTA under the control of a TK promoter and pTK-Hyg (Clontech). Positive clones stably expressing the tTA were isolated following hygromycin selection and presence of the tTA was determined by a 150 bp PCR product (S-5′-CAGCGCATTAGAGCTGCTTA-3′ and AS-5′-ATCTCAATGGCTAAGGCGTC-3′). RT-PCR was performed to confirm that the positive clones expressed both the tTA (primers as above) and the Hand1 transgene, as detected by a 400 bp PCR product(S-5′-GATGGGACTGGAGAAGACCA-3′ and AS-5′-GAAGTCAGATGCTCAAGGGG-3′).

ES cells were cultured and differentiated in vitro to form embryoid bodies(EBs) derived from wild-type, control (transgenic responder construct alone), Hand1-overexpressing and Hand1 heterozygous or null lines,as previously described (Riley et al.,2000; Smart et al.,2002). Single cardiomyocytes were isolated from adherent EB cultures at days 10 and 14, according to the protocol described previously(Maltsev et al., 1993).

mRNA was isolated from whole embryos, cells and EBs using the Micro FastTrack 2.0 kit (Invitrogen), according to the manufacturer's instructions. The RNA was DNase treated before reverse transcription was performed using Superscript II RT (Invitrogen), according to the manufacturer's instructions. PCR was performed using standard conditions.

Optical projection tomography (OPT) was performed essentially as described(Sharpe et al., 2002) on whole-mount embryos. Analysis and visualisation of OPT data was performed with Improvision Volocity 3.1.

Embryos were fixed, dehydrated and wax embedded as previously described(Moorman et al., 2001). They were then serially sectioned at 10-15 μm. In situ sections were counterstained with 0.5% Eosin, as described previously(Smart et al., 2002).

RNA in situ hybridisation on whole-mount embryos and sections was performed as previously described (Moorman et al.,2001; Smart et al.,2002), using riboprobes specific for Hand1(Cserjesi et al., 1995), atrial natriuretic factor (Anf; Nppa - Mouse Genome Informatics) and Gata4 (Kuo et al.,1997), islet 1 (Isl1) and Wnt11(Cai et al., 2003), Nkx2.5 (Lints et al.,1993), Hand2(Srivastava et al., 1997), and myocyte enhancer factor 2c (Mef2c)(Edmondson et al., 1994).

Sections were dewaxed and rehydrated(Moorman et al., 2001), and stained with antibodies against MF20 (1/10; Developmental Studies Hybridoma),phospho-histone H3 (1/200; Upstate) and cleaved caspase 3 (1/100; Cell Signalling Technology). Sections were then counterstained with the nuclear marker bis-benzamide, as previously described(Hill and Riley, 2004; Riley et al., 2000).

Real-time qRT-PCR was carried out on RNA extracted from embryos at E9.5 and EBs following 14 days of differentiation. This was performed on an ABI 7000 Sequence Detector (Applied Biosystems) using SYBR Green (Absolute QPCR SYBR Green Mix, ABgene). Complementary DNA PCR primers were obtained from published sequences (Fijnvandraat et al.,2003a; Fijnvandraat et al.,2003b; Lekanne Deprez et al.,2002) or designed using Primer Express software (version 2.0,Applied Biosystems).

EB lysates were prepared in RIPA lysis buffer following 14 days of differentiation. Western blot was performed by standard methods using antibodies against cyclin D2 (1/500; Abcam), Cdk4 (1/500; Santa Cruz) and Gapdh (1/1000; Chemicon). Scanning densitometry was performed as described above.

The Tet-Off system was employed to investigate the effects of overexpressing Hand1 in the developing heart, specifically in Hand1-expressing cells. Two independent mouse strains were generated:the driver (tTA) was targeted to the endogenous Hand1 locus by homologous recombination in ES cells (Fig. 1A,B); and the responder (TRE2), was placed upstream of a full-length Hand1 cDNA and randomly inserted, as a transgene, into ES cells (Fig. 1C). Correctly targeted and transgenic ES-cell clones were identified by Southern blotting and PCR (Fig. 1D,E). Southern blotting was performed to confirm germline transmission of both the driver and responder strains. Incorporation of a single tTA allele into the Hand1 locus resulted in mice that were heterozygous for Hand1 and thus exhibited no discernable phenotype in advance of driver/responder matings (Riley et al.,1998).

To determine whether the tTA was expressed in Hand1-expressing lineages in the driver strain, we performed RNA in situ hybridisation analysis to detect tTA transcripts in whole-mount embryos and in sections at E9.5. The results confirmed that tTA expression from a gross whole-mount perspective colocalises with Hand1-expressing regions, notably in the branchial arches, lateral plate mesoderm, distal OFT and left ventricular (LV)myocardium at E9.5 (Fig. 2A,B). In histological sections, there were some differences between Hand1 and tTA expression, reflecting a degree of mosaicism in the tTA driver, notably ectopic expression in the first branchial arch(Fig. 2E,F). As we have taken the approach of knocking-in the tTA driver into the endogenous Hand1locus and we know from previous studies that Hand1 is not required for its own expression (Riley et al.,2000), it is reasonable to assume that the tTA is subject to all of the appropriate temporal and spatial control as endogenous Hand1. Any variation, therefore, between Hand1 and tTA expression should be minimal and inexplicable in light of our current understanding of the knock-in approach.

The fact that tTA expression, on the whole, recapitulated that of endogenous Hand1 confirmed the knock-in approach to be suitable for driving Hand1 overexpression exclusively in Hand1-expressing cells. In order to assess levels of Hand1 expression, compound heterozygote embryos were generated by crossing the driver and responder strains. Compound heterozygotes expressed the transgenic Hand1 cDNA by RT-PCR and northern blot (not shown), and, furthermore, had approximately twice as much Hand1 expression, as measured by real-time quantitative PCR, than control embryos (Fig. 1F). The single founder animal(Fig. 1F) and subsequent F1 and F2 generations were used throughout the study. The transgene copy number for the original founder was four and no correlation was observed between copy number and level of Hand1 expression (data not shown). The Hand1 responder transgene was not expressed in the absence of the tTA driver (data not shown), confirming that the system was not `leaky'.

The observation that Hand1 overexpression in developing trophoblasts promotes terminal differentiation to a giant cell fate(Cross et al., 1995) and the fact that tetraploid-rescued Hand1-null embryos have hypoplastic hearts with downregulation of differentiation markers such as Mlc2v(Myl2 - Mouse Genome Informatics)(Riley et al., 1998), suggest that Hand1 overexpression in the developing heart may promote cardiomyocyte differentiation in vivo. In compound driver/responder heterozygotes we observed, as early as E8.0, expansion of the primary heart tube and extension at the earliest stages of looping compared to littermate controls (embryos containing either only the driver targeted allele or the responder transgene), suggesting hyperplasia and a putative expansion of cell number (Fig. 3). Extension of the primary heart tube at E8.0-E8.5 (Fig. 3A-L) was associated with a corresponding temporal and spatial overexpression of Hand1, as confirmed by in situ analysis(Fig. 3M-P). At E9.5, Hand1-overexpressing embryos revealed a distinct cardiac phenotype(Figs 4, 5). Hearts in which overexpression occurred clearly looped to the right, suggesting no apparent defect in the initiation and onset of looping. However, looping was abnormal,with a significant overextension of the heart tube. Most notably, the OFT was elongated distally and physically displaced from the inflow region(Fig. 4D). Coincident with the abnormal looping was defective LV development (Figs 4, 5). The Hand1overexpression embryos appeared morphologically normal in other lineages, and the phenotype could be reversed with the addition of the tet derivative doxycycline to pregnant mothers (data not shown), indicating that the embryonic defects observed were specific and indeed caused by overexpression of Hand1.

In Hand1-overexpressing embryos, the distal OFT region was significantly extended and the heart tube displaced both ventrally away from the body wall and laterally to reveal extraneous rightward looping (Figs 4, 5). The abnormal looping of the heart is most probably a direct consequence of the OFT extension, as the heart tube might be expected to undergo additional turns/folds to become accommodated within a restricted pericardial region. In order to confirm the observation of OFT elongation, distal OFT length was measured in moderately(n=11) and severely (n=5) affected overexpression embryos and compared with transgene positive (n=9) controls(Fig. 5A-D). The distal OFT was defined as the region of the heart tube extending from the body wall to the apex of the so-called `dog leg bend'(Anderson et al., 2003). In severely affected embryos, the OFT was found to be approximately twice as long as controls. In the moderately affected embryos, OFT length was found to be approximately 50% longer than in controls. Variation in the cardiac phenotype between Hand1-overexpressing embryos at mid-gestation stages (compare Fig. 4D-F with 4G-H) is most likely to be attributable to a degree of mosaicism at the level of tTA expression despite the knock-in approach (data not shown). However, the mosaicism fortuitously enabled us to assess a range of hypermorphic cardiac phenotypes, with the most severe case illustrated in Fig. 4G-I and Fig. 5E-H. Furthermore,hypermorphic embryos provided a valuable insight into changes in gene expression profiles in the heart, even when the anatomical defects observed were mild to modest (see below).

In severely affected embryos, extraneous looping was accompanied by impaired LV differentiation and expansion leading to severe necrosis(Fig. 4G-I,M, Fig. 5H). In moderately affected embryos, the presumptive left ventricle was reduced in size but had otherwise undergone a degree of expansion and appeared to be developing normally (Fig. 4D-F,K). Sections through the hearts of the severely affected Hand1-overexpressing embryos revealed that the failure in ventricular expansion was associated with an almost complete absence of chamber lumen(Fig. 4M) caused by a combination of elevated myocyte density (control, 110±15(n=6); overexpression, 185±12 (n=6); P<0.05) and cellular hypertrophy. The LV phenotype in the severely affected embryos may be a primary effect of Hand1 overexpression in the LV myocardium itself, or could be a secondary effect caused by the impaired haemodynamics of the expanded OFT and extraneously looped heart tube.

In severely affected embryos, the extended OFT and impaired ventricular expansion was not compatible with survival beyond E10.5 (data not shown). Moderately and mildly affected embryos recover such that, at E12.5-14.5, heart development appears entirely normal (see Fig. S1 in the supplementary material) and these embryos survive to birth. Neither the mild nor moderate class of Hand1-overexpressing embryos revealed overtly abnormal cardiac morphogenesis. There appears to be a tolerance of variation in outflow-tract length, even amongst control somite-matched embryos (see Fig. 5D), which ensures normal development. The expansion of the OFT observed in the mildly and moderately affected mutants is presumably within this variance threshold such that the OFT is remodelled appropriately and LV maturation remains comparable with controls.

As Hand1 expression in the second lineage is restricted to the aortic sac mesothelium (Fig. 2)and is excluded from key areas such as the distal pharyngeal mesoderm and the mesodermal core of the branchial arches(Kelly et al., 2001), we deemed it unlikely that the OFT expansion would be attributed to elevated migration, or elevated number, of extracardiac cells from the anterior heart field. To confirm this hypothesis, we investigated the contribution of the AHF in the Hand1-overexpressing embryos by examining the expression levels of: the LIM-domain-containing transcription factor Isl1, which is involved in the differentiation of AHF cells(Cai et al., 2003); Mef2c, which is believed to be a target of Isl1(Dodou et al., 2004), and changes in which are thought to reflect defects in the formation of the AHF and its derivatives (Kelly and Buckingham,2002); and Hand2 (downstream of Mef2c in the same pathway) (Lin et al.,1997). In the case of all three markers - Isl1(Fig. 6A-F), Mef2c(Fig. 6G-L) and Hand2(data not shown) - there were no observable differences in expression between control and Hand1-overexpressing embryos, suggesting that the distal OFT expansion is not caused by an elevated deployment of AHF cells. This suggests that the effect of Hand1 overexpression was autonomous to the distal OFT and may have arisen as a result of an inherent expansion in cardiac precursors in this region. To test this hypothesis, we looked at markers of cardiomyocyte differentiation by in situ hybridisation and quantitative real-time PCR. Atrial natriuretic factor (ANF) is a marker of differentiated myocardium in the atrial and ventricular chambers, but has previously been reported to be undetectable in the OFT(Habets et al., 2002). Not only was Anf (Nppa - Mouse Genome Informatics) downregulated in the trabeculated myocardium of the LV in modestly affected embryos(Fig. 7A-D), it was also downregulated in the OFT and restricted exclusively to the proximal OFT,suggesting alternate differentiation status along the proximal and distal axes(Fig. 7E,F). To assess the differentiation status specifically of the extended distal OFT in the overexpression embryos, we examined the levels of Wnt11, a marker of the myocardium of the OFT (Cai et al.,2003) implicated in the induction of cardiogenesis and in the cardiomyocyte differentiation program(Koyanagi et al., 2005; Pandur et al., 2002). Wnt11 was expressed at high levels throughout the distal OFT in control embryos (Fig. 7G,I) but restricted to low-level expression in a subpopulation of cells in the myocardial wall of the OFT in overexpression embryos(Fig. 7H,J). This suggests a reduction in the level of myocardial differentiation coincident with distal OFT extension. Reduced myocardial differentiation in the primary heart tube and, notably, the LV was confirmed in the overexpression embryos by in situ hybridisation for the early cardiac markers Nkx2.5 and Gata4(Fig. 8). Whereas robust expression for both markers was observed in the primary heart tubes of control embryos (Fig. 8A,C,E,G), Nkx2.5 and Gata4 were significantly downregulated throughout the heart, most notably in the left ventricle of the overexpression embryos(Fig. 8B,D,F,H).

Quantitative real-time PCR analyses for an array of cardiac markers were performed on RNA extracted from control versus Hand1 overexpression embryos at E9.5. The results supported the downregulation of Anf,Nkx2.5 and Gata4 observed by in situ hybridisation and further revealed downregulation of chisel (smpx - Mouse Genome Informatics)as an additional early marker of cardiomyocyte differentiation(Fig. 8Q). By contrast, the expression of markers of gap junctions, such as Cx40 (Gja5 -Mouse Genome Informatics) and Cx43 (Gja1 - Mouse Genome Informatics), was unchanged, as was expression of the ventricle-specific homeobox gene Irx4 (Fig. 8Q). Interestingly, neither expression of Mef2c nor Hand2 was altered in the overexpression embryos(Fig. 8Q), consistent with the in situ hybridisation data (Fig. 6) and confirming that the effect of Hand1 overexpression is independent of an enhanced recruitment of extracardiac cells from the AHF.

As there appears to be reduced cardiomyocyte differentiation in the Hand1 overexpression embryos, we determined whether there were altered levels of apoptosis or elevated cell proliferation by immunostaining with antibodies against cleaved caspase 3 (CC3) and phospho-histone H3 (PH3),respectively. We observed no difference in expression of CC3 between wild-type and overexpressing embryos, with very few cells throughout the developing heart undergoing apoptosis at this stage (data not shown). By contrast, there were significantly more proliferating cells located in the distal OFT region of overexpressing hearts compared with wild type, with large numbers of PH3 positive cells, as an indicator of cells undergoing mitosis, situated in the wall and lumen of the OFT [control, 125±10 (n=6);overexpression, 238±14 (n=6) P<0.01](Fig. 9A-H). The observation of an over-proliferation of cardioblasts in the OFT of overexpression embryos is further supported by the downregulation of Wnt11 described above(Fig. 7G-J).

Taken together, a lack of change in markers of the AHF, the downregulation of key genes associated with promoting cardiomyocyte differentiation and elevated levels of proliferation of cells in the distal OFT suggest that Hand1 does not affect deployment of cells into the OFT from the AHF, but rather that, once the extracardiac cells arrive in the OFT, they respond to Hand1 gain of function by continuing to proliferate as opposed to committing to a terminally differentiated cell fate.

To confirm the in vivo findings and enable us to compare directly Hand1 gain of function with a loss-of-function model, we characterised Hand1 activity in vitro using ES cells differentiated into EBs over a period of 14 days in floating culture. EB-derived cardiomyocytes represent a faithful model of the in vivo situation: not only do they exhibit characteristics of early chamber myocardium but a substantial proportion are reminiscent of those observed in embryonic OFT(Fijnvandraat et al., 2003a; Fijnvandraat et al., 2003b). ES cells generated during the derivation of the responder mouse strain containing the Tre2-Hand1 transgene responder construct were stably transfected with a tTA driver under the control of a TK promoter to generate a line that precociously overexpressed Hand1 (see Fig. S2A in the supplementary material). Hand1 expression is first detected in wild-type EBs at day 4 of differentiation(Smart et al., 2002). The Hand1-overexpressing line expressed the Hand1 responder transgene precociously, prior to endogenous Hand1 expression, between day 0 and 4 of differentiation (see Fig. S2B, inset i, in the supplementary material). Overexpression was maintained throughout differentiation up to day 14 (see Fig. S2B in the supplementary material), at which stage Hand1was expressed five- to sixfold higher in the Hand1-overexpressing cell line than in the control line (see Fig. S2B, inset ii, in the supplementary material).

The Hand1-overexpressing cells were then compared with either control or Hand1-null ES-cell lines(Riley et al., 2000) in terms of their inherent ability to form cardiomyocyte contractile foci and real-time expression of cardiac markers. Hand1-overexpressing EBs revealed a significant delay and reduction in the formation of foci compared with wild type, whereas Hand1-null EBs formed a much higher proportion of contractile foci, up to fivefold higher at day 10 of differentiation(Fig. 10A). To demonstrate that the beating foci observed were cardiomyocytes, and to confirm that the Hand1-overexpressing line was deficient in forming cardiomyocytes,cells were dispersed and isolated from EBs at day 14 and immunostained for sarcomeric myosin using the MF20 monoclonal antibody. Whereas cardiomyocytes associated with beating foci were readily identified in Hand1-null,wild-type and control cell lines (see Fig. S2C in the supplementary material),no cardiomyocytes were identified in the Hand1-overexpressing line,confirming the deficiency in cardiomyocyte differentiation in the gain-of-function line.

The levels of contractile foci were supported by the gene-expression profiles for each cell line (Fig. 10B). Commitment to a cardiac fate was evident in Hand1-overexpressing EBs from the corresponding expression of Mef2c, Cx40 and Irx4, and Mlc2v as markers of chamber myocardium. Moreover, expression of Irx4 and Mlc2vat levels comparable to control were indicative of progression towards ventricular-like differentiation (Fig. 10B). However, the significant reduction in the expression of Nkx2.5, Gata4 and the pan-cardiac marker α-cardiac actin (Fig. 10B) suggests an overall reduction in cardiomyocyte differentiation. As the ventricular-specific markers (Irx4 and Mlc2v) appeared normal, this suggests that there are no significant changes in EB-derived cardiomyocytes that resemble early chamber myocardium and that the reduced differentiation is possibly attributable to the OFT component of the EB-derived cardiomyocytes (Fijnvandraat et al., 2003a). This is consistent with the identification of independent populations of EB-derived cardiomyocytes, the gene expression profiles and electrophysiological activity of which are characteristic of either embryonic OFT or early chamber myocardium(Fijnvandraat et al., 2003a; Fijnvandraat et al.,2003b).

The observation of Nkx2.5 inhibition following Hand1 gain of function suggests that, although Nkx2.5 precedes Hand1expression in the developing heart in vivo and is thought to be higher up the transcriptional hierarchy (Biben and Harvey, 1997), Hand1 is able to feedback onto Nkx2.5 and modulate not only its expression levels but subsequent induced myocardial differentiation. Hand1-null EBs, by contrast, revealed significantly elevated Nkx2.5 levels; in fact, all of the cardiac genes examined(with the exception of Cx40) were highly upregulated in response to Hand1 loss of function (Fig. 10B). This is consistent with the observation of significantly increased contractile cardiomyocyte foci in Hand1-null EBs(Fig. 10A). An overall observation that Hand1 overexpression affected a restricted cohort of the cardiac markers examined compared to loss of Hand1 is almost certainly explained by the fact that the two parent ES-cell lines differed considerably in their profiles for the markers in question (data not shown). The differences in cardiomyocyte differentiation between the Hand1-null and Hand1-overexpressing cell lines indicate a non-redundant role for Hand1 in maintaining equilibrium between cardiomyocyte proliferation versus differentiation.

The altered levels and rates of differentiation in the Hand1-null and Hand1-overexpressing cell lines suggested that these lines may have associated alterations in cell-proliferation levels. Furthermore, as there is elevated proliferation in the OFT in Hand1-overexpressing embryos, accompanied by a decrease in differentiation, it follows that, in the event of Hand1 overexpression, there may be a reduction in exit from the cell cycle, resulting in continued proliferation at the expense of differentiation.

To provide further evidence for elevated levels of cell proliferation in a Hand1 gain-of-function background and mechanistic insight into how Hand1 might promote increased cell division, we examined levels of cyclin D2 and the associated cyclin-dependent kinase, Cdk4, in Hand1-overexpressing EB-derived cardiomyocytes. As cardiomyocytes develop, they lose the ability to proliferate and there appears to be a block in the gap phases of the cell cycle, most notably at G1, consistent with the terminal differentiation of these cells (Brooks et al.,1998). Progression through G1 is dependent upon the D-type cyclins and, although many different stimuli induce cell proliferation through different signalling pathways, activation of one or more of the D-type cyclins is necessary for most, if not all, of these pathways(Busk et al., 2005). In Hand1-overexpressing EBs, we observed significantly elevated protein levels of both cyclin D2 and Cdk4 compared with wild type; this was reciprocated by reduced levels for both proteins in EBs that were either heterozygous or homozygous mutant for Hand1(Fig. 10C,D). The differences in cyclin D2 and Cdk4 levels observed between Hand1 gain- and loss-of-function EBs appears to reflect the significant differences in numbers of differentiated cardiomyocytes/beating foci observed between the two populations (5% in the Hand1-overexpressing EBs versus 90% in the Hand1-null EBs) following 14 days of differentiation(Fig. 10A). Therefore, Hand1 activity influences cyclin D expression and associated CDK activity, both of which are hallmarks of cell-cycle entry and pivotal for continued cell proliferation.

We describe a previously unknown role for Hand1 as a significant cardiac transcription factor in regulating cardiomyocyte exit from the cell cycle to finely control the balance between cell proliferation and differentiation. The finding that Hand1 can promote proliferation in sub-regions of the developing heart while still allowing adjacent myocardial precursors to differentiate is novel and supports the idea that proliferation and differentiation are intrinsically linked and are not, as previously thought, mutually exclusive processes. Moreover, this study also suggests that Hand1 is able to promote either cell proliferation or differentiation in embryologically distinct lineages. Hand1 is both necessary and sufficient for the terminal differentiation of trophoblast stem cells to a giant cell fate(Cross et al., 1995; Hemberger et al., 2004),whereas, in the heart, overexpression results in expansion of the proliferative pool of undifferentiated cardiomyocyte precursors.

An important potential caveat to the findings of this study resides in the interpretation of the result of overexpression of Hand1 as being either a gain-of-function or a dominant-negative effect, taking into account the reported role of Hand1 as part of heterodimeric complex(Firulli et al., 2000; Scott et al., 2000). Moreover,an additional consideration is whether, at high concentrations, specificity of Hand1 downstream effects may be lost resulting in an impact on non-endogenous Hand1 target genes. Although Hand1 functions as a heterodimer with E-factors in vitro, according to the classic bHLH paradigm(Firulli et al., 2000; Scott et al., 2000), it can also function as a homodimer (Hill and Riley, 2004; Scott et al.,2000) and, indeed, a tethered Hand1 homodimer allele, generated by homologous recombination in ES cells, can rescue the Hand1-null phenotype(Hu et al., 2006). Therefore,a dominant-negative effect on heterodimer formation can be ruled out as almost certainly lacking functional significance in vivo. The specificity issue is one that can be levied at any loss- or gain-of-function model, as the effects can be wide ranging and are often indirect. In this case, the Hand1expression is actually only twofold above controls(Fig. 1). Therefore, we are working throughout with concentrations of Hand1 where significant non-specific effects are unlikely; this is borne out by restriction of the phenotype exclusively to areas of Hand1 expression.

Hand1 overexpression, which is restricted to Hand1-expressing cells in the OFT, demonstrates how OFT myocardium following recruitment from the AHF is sensitised to Hand1, resulting in cell-cycle exit delay and maintenance of a proliferative myocardial progenitor population. Hand1 is expressed neither in distal pharyngeal mesoderm nor in pharyngeal endoderm and, as such, is unlikely to be involved in regulating proliferation/differentiation of the AHF or second lineage. Our results demonstrate that the OFT extension and elongation of the heart tube in Hand1-overexpressing embryos occurs independently of second lineage deployment as the expression of Isl1, Mef2c and Hand2 is unchanged in Hand1-overexpression embryos. The cells that migrate from the AHF into the embryonic heart are undifferentiated, and are thought to terminally differentiate into cardiomyocytes once they arrive at their destination in the developing heart (Waldo et al., 2001). It is, therefore, plausible that Hand1 crucially controls this differentiation process in the OFT myocardium, balancing levels of cell proliferation and differentiation to maintain appropriate OFT remodelling. In the event of Hand1 overexpression in the OFT, the balance shifts towards continued proliferation, resulting in increased cell number, abnormal OFT development and subsequent aberrant morphogenesis of the heart. Interestingly, a shortening of the OFT was noted in the cardiac-specific knockout of Hand1(McFadden et al., 2005) and,although this was not investigated further, it has been suggested that the OFT defect may be secondary to an observed LV hypoplasia(Buckingham et al., 2005; McFadden et al., 2005). Our results indicate that Hand1 directly affects the development of the OFT and suggest that shortening of the OFT in cardiac-specific Hand1-null embryos may well be a primary defect of reduced cardiomyocyte proliferation.

Hand1 gain of function acts to promote continued proliferation of existing precursors in the distal OFT, enabling them to maintain an early phenotype whereby cardiomyocytes proliferate even as they begin to differentiate, and subsequently prevents further differentiation associated with cell-cycle exit. Conversely, in a Hand1-null background the heart tube fails to loop, the OFT is hypoplastic and the presumptive ventricle is thin walled (Riley et al.,1998). Moreover, Hand1-null ES cells differentiated in vitro give rise to a significantly higher proportion of beating foci compared with wild type and this is associated with a significant upregulation in markers of cardiomyocyte differentiation. Therefore, upstream regulation of Hand1 levels in the expanding heart tube represents a switch between cardiomyocyte proliferation and differentiation during organogenesis, and Hand1 itself can regulate the progenitor pool in the developing OFT in a manner entirely distinct from signals intrinsic to and emanating from the AHF.

Overexpression of Hand1 in the LV appears to promote further proliferation, thus preventing the outer curvature from expanding and ballooning ventrally as secondary myocardium to form the ventricular chamber. Secondary myocardium has a distinct molecular profile to that of the primary myocardium of the linear heart tube. Once this program is initiated, secondary myocardium is thought to balloon out from the outer curvature of the primary heart tube to form the four chambers of the heart(Christoffels et al., 2000). The downregulation of markers of secondary myocardium, such as Anfand chisel (Christoffels et al.,2000), in Hand1-overexpressing embryos may, therefore,explain the failure of the ventricle to balloon out, as differentiation into secondary myocardium is simply not initiated. In this case, Hand1 may be controlling the balance between differentiation and proliferation in the ventricular chamber such that, in the event of Hand1 overexpression in the presumptive LV, as in the OFT, the balance shifts towards proliferation, resulting in an increased number of cells and reduced cardiomyocyte differentiation with the associated down-regulation of differentiation markers. Conversely, in tetraploid-rescued Hand1-null embryos, the thin myocardial wall of the LV in the absence of Hand1 can now be explained by a shift in balance towards decreased proliferation, rather than by secondary effects, as previously thought(Riley et al., 1998; Riley et al., 2000). The fact that the LV is a derivative of the primary heart field suggests that Hand1 uniquely impacts on derivatives of both the primary (LV) and secondary (OFT)lineages in a cell-autonomous manner. This is further supported by the extension of the heart tube observed in Hand1-overexpression embryos as early as E8.0, which suggests that elevated Hand1 levels exert a proliferative effect on the cells arising from the primary heart field. It remains to be seen, however, whether the failure in the ballooning and lack of ventricular lumen is primarily linked to an over-proliferation of ventricular myocardium or is secondary to the elevated haemodynamic load and increased fluid pressure arising from the extended heart tube.

In Hand1-overexpressing EBs, reduced and delayed cardiomyocyte differentiation was associated with increased levels of cyclin D2 and Cdk4,which is indicative of sustained entry into, and progression through, the cell cycle and, consequently, of enhanced proliferation. The opposite was true in the Hand1-null EBs, where there appeared to be down-regulation of cyclin D2 and Cdk4, indicating premature cell cycle exit, lower levels of proliferation and enhanced differentiation of cardiomyocytes. Moreover, these observations suggest that cyclin D2 and Cdk4 are downstream of Hand1;although it remains to be elucidated whether Hand1 directly, or indirectly,regulates cyclin D2/Cdk4 expression.

The observation that Hand1 lies upstream of cyclin D2 is important in light of recent studies implicating both genes in adult-onset cardiovascular disease. Adult transgenic mice expressing cyclin D2 in the myocardium show increased levels of basal cardiomyocyte cell-cycle activity(Pasumarthi and Field, 2002)and this increase in activity appears to result in a progressive reduction of infarct size and improved cardiac function(Rubart and Field, 2006). Additionally, it has been demonstrated that cyclin D2 represses hypertrophy by forcing cardiomyocytes through the cell cycle, resulting in cell proliferation. This suggests that cyclin D2 levels directly determine whether cells grow by hypertrophy or proliferation(Busk et al., 2005). Hand1 has been shown to be downregulated in human ischemic and dilated cardiomyopathies (Natarajan et al., 2001), and also following the induction of cardiac hypertrophy in adult rodent hearts(Thattaliyath et al., 2002). Furthermore, a study on cardiac lineage protein 1 (Clp1; Hexim1 - Mouse Genome Informatics)-null mice suggested that an observed downregulation of Hand1 in the hearts of Clp1embryos directly promoted a foetal form of cardiac hypertrophy(Huang et al., 2004). The downregulation of Hand1 during cardiac hypertrophy may, therefore,result in the downregulation of cyclin D2, which in turn is a likely contributing factor towards hypertrophic growth. This suggests that Hand1 and cyclin D2 are capable of modifying cardiac responses in the myocardium to external stimuli and that, collectively, they may play an important role during hypertrophic signalling.

Disruption in the balance of cell number during heart development, via over-proliferation and associated failure in differentiation, clearly leads to an inability to maintain appropriate morphogenesis. Conversely, failure to maintain an adequate pool of undifferentiated myocyte precursors results in organ hypoplasia, as is observed in mice lacking Hand1, and almost certainly contributes to the aetiology of congenital heart disease (CHD). OFT anomalies account for ∼30% of all cases of CHD(Kelly and Buckingham, 2002). Reduced OFT length is associated with defective alignment with the atrioventricular junction and abnormal rotation of OFT myocardium, which in turn leads to abnormal ventricular septation and OFT anomalies such as double outlet right ventricle and persistent truncus arteriosus. Autonomous,Hand1-induced proliferation of cardiomyocyte precursors during heart-tube elongation provides insight into directing progenitor cells along a path towards integrated myocardium, which is of direct relevance to cell-based therapies for cardiovascular disease and cell-autonomous strategies for cardiac regenerative medicine.

Two potential approaches for deriving cell-based therapies are the transplantation of stem cells from an external source and the activation of resident stem cells within the heart. Clearly, for these to be viable options there is a need for further understanding of the precise molecular pathways involved in directing cardiomyocyte proliferation and differentiation, and how to generate greater numbers of functional cardiomyocytes. These studies collectively reveal that Hand1 is involved in maintaining the correct balance between proliferation and differentiation in the developing heart, suggesting that manipulation of Hand1 levels could potentially affect both of these avenues of therapeutic research.

We thank P. Scambler and R. Anderson for critical discussions; E. Morrisey and S. Evans for providing the Anf, Isl1 and Wnt11 in situ probes,respectively; and genOway for the derivation of targeted and transgenic mice. This work was supported by grants from the Medical Research Council and British Heart Foundation.