Hand factor ablation causes defective left ventricular chamber development and compromised adult cardiac function

Coordinated cardiomyocyte growth, differentiation, and morphogenesis are essential for heart formation. We demonstrate that the bHLH transcription factors Hand1 and Hand2 play critical regulatory roles for left ventricle (LV) cardiomyocyte proliferation and morphogenesis. Using an LV-specific Cre allele (Hand1LV-Cre), we ablate Hand1-lineage cardiomyocytes, revealing that DTA-mediated cardiomyocyte death results in a hypoplastic LV by E10.5. Once Hand1-linage cells are removed from the LV, and Hand1 expression is switched off, embryonic hearts recover by E16.5. In contrast, conditional LV loss-of-function of both Hand1 and Hand2 results in aberrant trabeculation and thickened compact zone myocardium resulting from enhanced proliferation and a breakdown of compact zone/trabecular/ventricular septal identity. Surviving Hand1;Hand2 mutants display diminished cardiac function that is rescued by concurrent ablation of Hand-null cardiomyocytes. Collectively, we conclude that, within a mixed cardiomyocyte population, removal of defective myocardium and replacement with healthy endogenous cardiomyocytes may provide an effective strategy for cardiac repair.


Author summary
The left ventricle of the heart drives blood flow throughout the body. Impaired left ventricle function, associated either with heart failure or with certain, severe cardiac birth defects, constitutes a significant cause of mortality. Understanding how heart muscle grows is vital to developing improved treatments for these diseases. Unfortunately, genetic tools necessary to study the left ventricle have been lacking. Here we engineer the first mouse line to enable specific genetic study of the left ventricle. We show that, unlike in the adult heart, the embryonic left ventricle is remarkably tolerant of cell death, as remaining cells have the capacity to proliferate and to restore heart function. Conversely, disruption of two related genes, Hand1 and Hand2, within the left ventricle causes cells to a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
The left ventricle (LV) of the heart drives systemic circulation. Because the LV must be large enough to support adequate cardiac output but not hypertrophic or hypoplastic, such that it obstructs blood flow, congenital heart defects (CHDs) and acquired diseases that impact LV morphology or cell number present a significant cause of morbidity and mortality in the human population [1]. These CHDs include left ventricular noncompaction or hypertrabeculation (LVNC; OMIM: 604169), which is a cardiomyopathy characterized by prominent trabeculations occluding the ventricular lumen and associated with a high risk of heart failure and sudden death [2,3] and hypoplastic left heart syndrome (HLHS; OMIM: 614435), which presents an underdeveloped LV unable to sustain sufficient blood flow [4,5]. From the perspective of adult cardiac disease, especially cardiomyopathies, it is of particular importance to establish the capacity of proliferating cardiomyocytes to replace dead or dysfunctional cardiomyocytes [1]. Critical breakthroughs in patient outcomes demand a better understanding of the etiology of cardiac growth, differentiation and morphogenic patterning.
The embryonic heart forms from two distinct cardiomyocyte progenitor populations, termed the primary (PHF) and secondary (SHF) heart fields, which give rise to the LV and right ventricle (RV), respectively [6]. The bHLH transcription factor Hand1 is predominantly expressed within the LV myocardium, with minimal expression in SHF derivatives [7]. Interestingly, HAND1 mutations have been identified in HLHS patients [8]; however, cardiac-specific Hand1 ablation in mice does not recapitulate HLHS [9]. Previous studies suggest that the related bHLH transcription factor Hand2 can functionally cooperate with Hand1 during murine LV development [9]. Although Cre-loxP technology has been used to great effect in mice to genetically model SHF myocardium defects, the genetic tools to specifically interrogate genetic loss-of-function in LV cardiomyocytes have, thus far, not been available.
We have isolated a conserved 744 bp enhancer 5' to the Hand1 transcription start site that is sufficient to drive reporter and Cre recombinase gene expression specifically within the LV. Hand1 LV -Cre recapitulates endogenous Hand1 expression within an estimated 80-90% of LV cardiomyocytes between embryonic stages (E) E8.5-E13. 5. Ablation of the Hand1-lineage LV myocardium results in a markedly smaller LV by E10.5; however, LV chamber size and adult cardiac function is ultimately rescued via proliferation of non-Hand1-lineage LV cardiomyocytes.
In contrast to this LV cell ablation model, Hand1;Hand2 loss-of-function within the LV results in increased cardiomyocyte proliferation resulting in morphogenic defects that lead to the occlusion of the LV lumen. LV cardiomyocytes invade the LV chamber and show misexpression of compact zone, trabeculae, and intraventricular septum (IVS) restricted genes. We reveal a cooperative Hand factor function that is required to morphogenetically specify subpopulations of ventricular myocardium, and to thereby regulate cardiomyocyte growth. Functional analysis of surviving Hand1;Hand2 double conditional knockouts (CKOs) reveals impaired LV function that can be rescued by ablating the mutant Hand1 LV-lineage cells. Taken together, these data suggest that, due to its inherent proliferative capacity, the developing LV tolerates myocardial cell death, whereas developmentally abnormal cardiomyocytes adversely impact cardiac function.

Results
A distal Hand1 enhancer recapitulates gene expression within the LV We have identified a cis-regulatory element(s) that drives gene expression specifically within the LV, and not within other cardiac tissues (Vincentz and Firulli, manuscript in preparation). We used this Hand1 enhancer to make an LV-specific Cre driver. Lineage analysis using the Hand1 eGFPCre knock-in allele revealed that Hand1-lineage cells are restricted to the LV myocardium and to a ring of SHF-derived myocardium occupying the OFT, termed the myocardial cuff [7,10]. Using the 2.7kb Hand1 basal promoter to provide the eGFPCre with a transcriptional start site, we cloned the Hand1 LV-enhancer 5' (Fig 1A) and generated several F0 transgenic lines. In two of these lines, Cre activity, as assessed via the R26R lacZ reporter allele, was detectable within the forming LV at E9.0 ( Fig 1E; white arrow) specifically marking only the Hand1 LV cardiomyocyte-lineage ( Fig 1E, 1G, 1I, 1K and 1M). Hand1-lineage epicardium (black arrowhead; Fig 1H and 1I) and OFT tissues (Fig 1J and 1K; black arrows), detectable from the Hand1 eGFPCre allele, were not observed in the Hand1 LV -Cre transgenic. Immunohistochemistry using β-galactosidase to mark Cre-lineage cells and Mlc2v to mark ventricular cardiomyocytes showed total co-localization of these two markers (S1A, S1C, S1E and S1G Fig), whereas expression of β-galactosidase and PECAM, an endothelial/endocardial marker, appears mutually exclusive (S1A, S1C, S1E and S1G Fig). Together, these data validate that this Hand1 LV -Cre driver recombines specifically within LV cardiomyocytes.
Cellular ablation of the Hand1 LV -Cre embryonic LV myocardium results in hypoplastic LV at E10.5 that fully recovers in size and function To address the importance of the Hand1 LV lineage during cardiac development, we utilized the conditionally active Rosa26 (R26R) Diphtheria Toxin A chain (R26R DTA ) allele to conditionally ablate Hand1-expressing LV cardiomyocytes. Analysis of cell death via TUNEL on E9.5 sections revealed that the LVs of Hand1 LV -Cre; R26R lacZ/+ control embryos displayed few apoptotic cells (Fig 2A), whereas Hand1 LV -Cre; R26R lacZ/DTA LVs contained scattered TUNELpositive cells (Fig 2B, quantified in Fig 2C). However, given the robust LV expression of the Hand1 LV -Cre, fewer TUNEL-positive cells than would be predicted were detected in E9.5 R26R DTA -ablated LVs. Additional TUNEL staining, ( Fig 2D) and whole mount lysotracker staining at E10.5 revealed markedly increased cell death within the LV of Hand1 LV -Cre; R26R lacZ/DTA embryos when compared to controls (white arrows Fig 2E and 2F). At both time points, no difference in cell death was observed in the RV myocardium (Fig 2A-2D, 2G and 2H), and cell death in the pharyngeal arches served as a positive control (white arrowheads Fig  2A, 2B and 2E-2H).
We then bred the R26R lacZ reporter onto the R26R DTA allele to monitor the loss of Hand1expressing LV cardiomyocytes in our ablation model. At E9.5, the Hand1-LV lineage is largely absent from the heart, but the LV is not grossly hypoplastic ( Fig 2J). By E10.5, the lack of Hand1-LV lineage cells results in a markedly hypoplastic LV (Fig 2L). At E12.5, the LVs of Hand1 LV -Cre; R26R lacZ/DTA embryos were visibly smaller than Hand1 LV -Cre; R26R lacZ/+ controls (Fig 2M and 2N) and showed few X-gal-stained cells when compared to littermates that do not carry the DTA allele. Interestingly, these mid-gestation LVs, from which the Hand1lineage has been ablated, do not display perturbed expression of the LV markers Nppa and Gja5 (S2 Fig). Thus, ablation of Hand1-lineage LV myocardium between E8.5, when the Hand1 LV enhancer is first upregulated, and E12.5, when the Hand1 LV enhancer begins downregulation, results in a hypoplastic LV.
https://doi.org/10.1371/journal.pgen.1006922.g001 . E-H) Whole mount lysotracker staining detecting cell death confirms that conditional activation of R26R eGFPDTA/+ via intercross to Hand1 LV -Cre causes pronounced LV cardiomyocyte death as assayed at E10.5 (white arrow; n = 3). White arrows denote TUNEL or lysotrackerpositive cells in the LV. White arrowheads denote TUNEL or lysotracker-positive cells in the pharyngeal arches. I-P) X-Gal staining to detect the R26R lacZ reporter allele demonstrates that activation of the R26R eGFPDTA allele nearly completely ablates the Hand1 LV -Cre lineage (black arrow) by E9.5 (I, J; n = 5); however, LV hypoplasia is not apparent until E10.5 (K, L; n = 3) and persists at E12.5 (M, N; n = 4). Despite DTA-mediated ablation of the Hand1 LV -Cre lineage, by E16.5, the initially hypoplastic LV shows a pronounced recovery, and LV size is comparable to controls (O, P; n = 3). Q-T) Immunohistochemistry for pHH3 at E12.5 (Q, R) and E14.5 (S, T) in Hand1 LV -Cre(+);R26R +/+ control (Q, S) and Hand1 LV -Cre(+);R26R +/DTA embryos (R, T). U) Quantification of pHH3 + cells relative to the number of DAPI+ pixels show that proliferation is not altered at E12.5, but is elevated specifically within the LV at E14.5. Data are represented as mean ± standard error of mean. Asterisks denote significance (p 0.05) as determined by student's t-test. These hearts exhibited cardiac function, as assayed by ejection fraction (EF), fractional shortening (FS), and other echocardiographic parameters, that was indistinguishable from Cre-negative controls (S3D-S3O Fig), and within the normal range for adult mice under isoflurane anesthesia [17]. Taken together, these data demonstrate that embryonic ablation of the Hand1-lineage is not sufficient to permanently disrupt LV development.
Myocardial deletion of Hand1 and Hand2 within the Hand1 LV -Cre lineage results in proliferative, morphological, and molecular abnormalities that result in a hypoplastic LV lumen As we have established that the Hand1-LV myocardial lineage is not required for cardiogenesis, we next investigated whether the expression of Hand1 and Hand2 within the embryonic LV is required for normal heart development. Although early embryonic expression analysis shows that the majority of Hand2 expression within the heart is restricted to the endocardium, epicardium, and SHF derived myocardium [10,18,19], at later embryonic stages, Hand2 mRNA becomes detectable within E11.5 LV myocardium in a pattern overlapping with Hand1 (S4C and S4D Fig). We subsequently generated compound heterozygous Hand1 LV -Cre;Hand1 fx/+ ; Hand2 fx/+ male mice and crossed them to Hand1 fx/fx ;Hand2 fx/fx females to generate Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx offspring (Fig 3). Hand1 LV -Cre;Hand1 fx/+ ;Hand2 fx/+ (Fig 3A-3D) embryos undergo normal cardiac development and are indistinguishable from wild type controls. Similarly, embryos that delete Hand2 from the LV but retain a single copy of Hand1 (Hand1 LV -Cre;Hand1 fx/+ ;Hand2 fx/fx ) also exhibit largely normal cardiac development (Fig 3E-3H). Interestingly, consistent with previous studies [9], deletion of Hand1 from the LV (Hand1 LV -Cre;Hand1 fx/fx ) results in a morphologically disorganized LV, wherein Hand1-lineage cells appear to localize more to trabeculations compared to the compact zone (Fig 3I-3L). At E17.5, abnormal cardiomyocytes localize within the LV lumen ( Fig 3L) Table 1. These data suggest that a Hand factor loss-of-function within the LV myocardium alters chamber morphology via a Hand gene dosage dependent mechanism, in which Hand1lineage cells are found less frequently in the compact zone at the expense of increased cells within the LV lumen representing trabecular and papillary muscle cardiomyocytes.  (Fig 4), whereas other measures of LV morphology and function, such as chamber dimensions and wall thickness, were not significantly altered (S7 Fig).   [20,21]. Section in situ hybridization of E11.5 hearts showed that Tbx20 and Hey2 are both expressed throughout the presumptive compact myocardium and excluded from the trabecular myocardium in control hearts ( Fig  5A and 5B). This sharp delineation between compact and trabecular myocardium is lost in the LVs of Hand1 LV -Cre(+); Hand1 fx/fx ;Hand2 fx/+ hearts (Fig 5E and 5F; asterisks). In Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx hearts, most of the trabeculae within the LV ectopically express compact myocardium markers, whereas the RV trabeculae do not (Fig 5I and 5J). Conversely, the trabecular markers Bmp10 and Nppb show robust expression throughout the trabeculae, and expression is largely excluded from the compact myocardium in control hearts (Fig 5C and  5D). In Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx hearts, Bmp10 and Nppb expression is downregulated within the LV trabeculae when compared to RV trabeculae (Fig 5K and 5L; asterisks). We conclude that Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx hearts display ectopic compact myocardium marker expression and reduced trabecular marker expression within the LV.

Surviving myocardial
Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx hearts display abnormal trabecular proliferation in the LV By Carnegie Stage 16 in the human embryo (equivalent to E11.5 in mice) proliferation in the ventricular trabeculae has declined [22]. We next sought to correlate differences in the proliferative capacity of compact and trabecular myocardium with the overgrowth seen in Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx LVs. Mki67 marks cells actively undergoing cell cycle, but is excluded from cells in G 0 [23]. Mki67 immunohistochemistry revealed that, in control embryos, nuclear Mki67 is robustly detected within compact myocardium and the IVS, but is largely excluded from trabecular myocardium (Fig 5M-5P). In contrast, Hand1 LV -Cre;  [24]. We reasoned that the abnormal proliferation and marker expression seen in Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx hearts may reflect aberrant ventricular septogenesis. Section in situ hybridization of E11.5 hearts showed that expression of the chemokine Cxcl12 is strong in the free walls of the ventricles, but is largely excluded from the IVS (Fig 6A) expression is excluded specifically from the trabecular myocardium, in addition to the left side of the ventricular septum (Fig 6D and 6G). The secreted Wnt inhibitor Dkk3 and the transcription factor Irx2 are both markers of the IVS [25]. Expression of both Dkk3 (Fig 6H; black arrowhead) and Irx2 (Fig 6I;

DTA ablation rescues the lethality and cardiac dysfunction seen in Hand1;Hand2 conditional knockouts
Given that the embryonic heart can recover from DTA-mediated ablation of Hand1-lineage cardiomyocytes, we tested whether ablation of Hand1;Hand2-null cardiomyocytes can rescue the phenotypes associated with their loss-of-function. The R26R DTA allele was bred onto a Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx background. E14.5 section in situ hybridization of control hearts (Fig 7A-7C) showed expected expression patterns for Bmp10, Irx2, and Dkk3 within the trabeculae and IVS respectively. As expected Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx hearts showed reduced Bmp10 LV expression along with expanded Dkk3-and Irx2 LV cardiomyocyte expression (Fig 7E-7G). These changes in gene expression are reversed in Hand1 fx/fx ;Hand2 fx/fx ; Hand1 LV -Cre(+);R26R DTA embryos (Fig 7I-7K). Morphological examination of X-gal-stained bisected P56 hearts revealed heart morphology indistinguishable from controls (Fig 7D, S8A  Fig), 7L, S8C and S8D Fig). Lineage tracing reveals that the majority of lacZ-positive cells are ablated in these hearts; however, consistent with DTA-ablation embryos (Fig 2P), small populations of lacZ-positive cells were sometimes detectable within these hearts (S8D Fig), which  we interpret as cells that have recombined only the lacZ reporter, but not the DTA allele. Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx ;R26R DTA pups survive at Mendelian ratios (Table 3) and display restored ejection fraction ( Fig 7M) and fractional shortening (Fig 7N). Again, other measures of cardiac function were not significantly altered (S8E- S8J Fig).Together, these data demonstrate that ablation of mutant cardiomyocytes from the developing LV restores cardiac function.

Discussion
CHDs that alter the LV exhibit poor clinical outcomes [5,26]. The inability to interrogate LV gene expression in isolation has limited the ability to understand the molecular mechanisms that drive LV morphogenesis. This study reports the generation of a novel LV-restricted Cre driver line that allows for such focused interrogation. First, we ablated Hand1-lineage cardiomyocytes within an E9.0-E13.5 developmental window. As expected, LV size was greatly reduced by E10.5, and this reduced size remains clear at E14.5 (Fig 2). Hand1 cardiac expression is downregulated by E13.5 [27]. We observed a significant increase in LV cardiomyocyte cell proliferation at E14.5 ( Fig 2U). By E16.5, LV size is indistinguishable from control hearts (Fig 2O and 2P). These findings support published data showing that embryonic and up to 7-day postnatal cardiomyocytes retain regenerative potential [14][15][16], and further demonstrate that the Hand1-lineage can be ablated from the developing LV, and the heart can nonetheless undergo full regenerative repair. It is also clear that these replacement cardiomyocytes do not express Hand1-if they did, Cre would also be expressed, thereby activating DTA expression and killing the replacement cell. That said, conclusive identification of the origin(s) of the cells that replace the ablated Hand1-lieage cells will require further study employing, for example, dual lineage tracing systems. A candidate for this progenitor population is the SHF. One of the most well studied markers of the SHF is the Isl1-Cre [28]. Although it is excluded from the majority of LV cardiomyocytes, Isl1-Cre-lineage cells do appear in the LV [28][29][30][31], and it is possible that this relatively minor population expands in the absence of Hand1-lineage cells. Indeed, this would explain why early stage DTA-ablation embryos, despite almost completely lacking Cre-lineage cells, are grossly similar to non-ablated littermates (Fig 2I and 2J). If this were the case, it would indicate that SHF-derived cardiomyocytes are sufficient to replace PHF-derived cardiomyocytes. Regardless of potential contributions from other cardiomyocyte lineages, including the SHF, these observations indicate that Hand1-lineage cells are not required for cardiac development, and that, through elevated proliferation, a Hand1-negative fated cardiomyocyte population within the developing heart is sufficient to generate a functional LV. In contrast, Hand1 and Hand2 are required for normal LV morphogenesis. By E11.5, both Hand genes are expressed within the LV myocardium (S4 Fig). Loss of Hand1, and, more  [32]. It would be informative to test whether IVS markers also expand into luminal domains in such models.
In addition to LVNC, these Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx mutants share certain phenotypic similarities with HLHS patients; however, they lack the aortic valve phenotypes characteristic of HLHS (S5M Fig). A recently published mouse model of HLHS [33] posits a digenic etiology of HLHS, in which dysfunction of one gene disrupts cardiomyocyte proliferation and differentiation to cause LV hypoplasia, while a second mutation causes aortic valve abnormalities. Importantly, these findings provide evidence that the aortic phenotypes are not secondary to the LV phenotypes, and therefore, the lack of Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx aortic defects is not surprising. As mentioned, a subset of HLHS patients displays HAND1 mutations [8]. Hand1 is expressed in the OFT [27] as well as the LV, and as such could have pleiotropic functions in both aortic valve and LV development; however, ablation of Hand1 in the neural crest progenitors of the aortic valves is not pathogenic [34]. Given that this study reports the unexpected finding that Hand1 and Hand2 have overlapping roles in LV development, it would be of interest to reevaluate Hand1 function in the OFT considering potential functional redundancy with Hand2.
Increased cardiomyocyte proliferation can negatively impact specification [14,15]. The Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx LV hyperplasia occludes so much of the LV lumen that it results, in severe cases, in a single ventricle phenotype (Fig 3T, S5N Fig). In spite of the significant LV hyperplasia, 50% of Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx mutants survive to birth and live to adulthood. These surviving individuals exhibit compromised systolic function (Fig 4). It is clear that both Hand1 and Hand2 contribute to this phenotype, and that Hand1 plays a more significant role. Hand1 LV -Cre;Hand1 fx/+ ;Hand2 fx/fx mutants display no observable cardiac phenotypes, whereas Hand1 LV -Cre;Hand1 fx/fx ;Hand2 +/+ and Hand1 LV -Cre;Hand1 fx/x ;Hand2 fx/+ mutants show both morphological and molecular changes in gene expression. Given that Hand2 LV expression is not detectable until E11.5, and its LV expression is not dependent upon Hand1, Hand1 mutants are less severe likely due to the later expression of Hand2.
Finally, we observe a functional rescue of Hand1 LV -Cre;Hand1 fx/fx ;Hand2 fx/fx mutants when mutant cells are ablated via co-activation of DTA expression (Fig 7). Given that heart development does not require the Hand1-lineage (Fig 2) this may not be surprising. Nevertheless, this observation suggests that removal of molecularly abnormal cardiomyocytes from a developing heart could be beneficial if molecularly normal cells are present and are still proliferative.

Materials and methods
All relevant data are within the manuscript and its Supporting Information files.

Transgenic mice
The Indiana University Transgenic and Knock-Out Mouse Core generated the Hand1 LV -Cre transgenic mouse line on a C3HeB/FeJ background. Genotyping of the Hand1 tm2Eno , Hand2 tm1Cse , Gt(ROSA)26Sor tm1(DTA)Jpmb , and Gt(ROSA)26Sor tm1Sor alleles has previously been described [9,10,35,36]. These mice were maintained on a mixed C57Bl/6;129S background. Embryos were not selected for sex, and were evaluated blindly for all analyses. Mice and other reagents are available from the authors upon request.

Lysotracker and TUNEL
Cell death analysis on control and mutant embryos was performed as described [39]. Lysotracker (Life Technologies) was incubated with embryos as per the manufacturer's instructions. Embryos were imaged in a well slide on a Leica DM5000 B compound florescent microscope. TUNEL analyses were performed upon sectioned embryos using the ApopTag Plus Fluorescein in situ Apoptosis detection kit (S7111 Chemicon International) as per manufacturer's instructions. TUNEL-positive cells occupying the free wall of the LV and RV (including the myocardial outflow tract) were counted every other section. Significance was determined by student's t-test.

Echocardiography
Mice were lightly anesthetized with mixture of 1% to 1.5% isoflurane and 100% oxygen while supine on a heated platform. The heart rate were stabilized at 400 to 500 beats per minute before image recording. Images were obtained with a high resolution Micro-Ultrasound system (Vevo 2100, VisualSonics Inc, Toronto, Canada) equipped with a 40-MHz mechanical scan probe. Two-dimensional images were recorded in the parasternal long and short-axis to guide M-mode recordings in the mid-ventricular level. LV systolic function was computed from M-mode measurement according to the recommendations of American Society of Echocardiography committee [43].