1946 RESEARCH ARTICLE

Development 140, 1946-1957 (2013) doi:10.1242/dev.089920 © 2013. Published by The Company of Biologists Ltd Fkbp1a controls ventricular myocardium trabeculation and compaction by regulating endocardial Notch1 activity Hanying Chen1,*,‡, Wenjun Zhang1,*, Xiaoxin Sun1, Momoko Yoshimoto2, Zhuang Chen1, Wuqiang Zhu1, Jijia Liu1,3, Yadan Shen1,3, Weidong Yong1, Deqiang Li1, Jin Zhang1, Yang Lin2, Baiyan Li1,4, Nathan J. VanDusen1, Paige Snider2, Robert J. Schwartz5, Simon J. Conway2, Loren J. Field1, Mervin C. Yoder2, Anthony B. Firulli1, Nadia Carlesso2, Jeffrey A. Towbin6 and Weinian Shou1,‡

SUMMARY Trabeculation and compaction of the embryonic myocardium are morphogenetic events crucial for the formation and function of the ventricular walls. Fkbp1a (FKBP12) is a ubiquitously expressed cis-trans peptidyl-prolyl isomerase. Fkbp1a-deficient mice develop ventricular hypertrabeculation and noncompaction. To determine the physiological function of Fkbp1a in regulating the intercellular and intracellular signaling pathways involved in ventricular trabeculation and compaction, we generated a series of Fkbp1a conditional knockouts. Surprisingly, cardiomyocyte-restricted ablation of Fkbp1a did not give rise to the ventricular developmental defect, whereas endothelial cell-restricted ablation of Fkbp1a recapitulated the ventricular hypertrabeculation and noncompaction observed in Fkbp1a systemically deficient mice, suggesting an important contribution of Fkbp1a within the developing endocardia in regulating the morphogenesis of ventricular trabeculation and compaction. Further analysis demonstrated that Fkbp1a is a novel negative modulator of activated Notch1. Activated Notch1 (N1ICD) was significantly upregulated in Fkbp1a-ablated endothelial cells in vivo and in vitro. Overexpression of Fkbp1a significantly reduced the stability of N1ICD and direct inhibition of Notch signaling significantly reduced hypertrabeculation in Fkbp1a-deficient mice. Our findings suggest that Fkbp1a-mediated regulation of Notch1 plays an important role in intercellular communication between endocardium and myocardium, which is crucial in controlling the formation of the ventricular walls.

KEY WORDS: FK506 binding protein 12, Ventricular hypertrabeculation/noncompaction, Notch1, Endocardial-myocardial signaling, Mouse

INTRODUCTION trabeculation is closely associated with myocardial growth arrest, The formation of the ventricular walls is essential to cardiogenesis whereas persistent trabeculation and/or a reduced level of and is required for normal cardiac function (Moorman and compaction are associated with left ventricular noncompaction Christoffels, 2003). Ventricular trabeculation and compaction are (LVNC) (Towbin, 2010) [also called left ventricular important morphogenetic events that are controlled by a precise hypertrabeculation (LVHT)] (Finsterer, 2009). LVNC balance between cardiomyocyte proliferation and differentiation (MIM300183) is a distinct form of inherited cardiomyopathy during ventricular chamber formation and wall maturation. The (Pignatelli et al., 2003; Sandhu et al., 2008). Although some LVNC trabeculation phase of cardiac development is initiated at the end patients survive into adulthood (Weiford et al., 2004), severe cases of cardiac looping (E9.0 to E9.5 in mouse) and is defined as the accompanied with other cardiac defects (e.g. Ebstein’s anomaly) growth of primitive cardiomyocytes to form the highly organized commonly die at a young age (Koh et al., 2009). muscular ridges that are lined by layers of invaginated endocardial The underlying molecular and cellular mechanisms that cells (Ben-Shachar et al., 1985). The newly formed trabeculae orchestrate ventricular trabeculation and compaction, as well as the gradually contribute to the formation of the papillary muscles, the pathogenesis of LVNC, are poorly understood due to the interventricular septum and cardiac conductive cells. Concomitant heterogeneity of the patients and the lack of suitable genetic models with ventricular septation, the trabeculae start to compact at their (Srivastava and Olson, 2000). Genetic screening of isolated LVNC base adjacent to the outer myocardium, adding substantially to patients has identified a handful of gene mutations in sarcomere ventricular wall thickness and coincidently establishing coronary proteins (Klaassen et al., 2008; Xing et al., 2006). However, the circulation (Wessels and Sedmera, 2003). A significant reduction in genes that are linked to the more severe cases of pediatric LVNC have yet to be identified. Previously, we showed that mutant mice deficient in FK506 binding protein 1a (Fkbp1a, also known as 1Riley Heart Research Center, Division of Pediatric Cardiology and 2Division of FKBP12) develop multiple abnormal cardiac structures that Neonatology, Herman B. Wells Center for Pediatric Research, Department of phenocopy severe cases of LVNC, including a characteristic Pediatrics, Indiana University School of Medicine, Indianapolis, IN 46202, USA. increase in the number and thickness of ventricular trabeculae (i.e. 3Department of Pediatric Cardiac Surgery, The Second Xiangya Hospital, Central South University, Changsha 410008, China. 4Department of Pharmacology, Harbin hypertrabeculation), deep intertrabecular recesses, lack of Medical University, Harbin 150081, China. 5Department of Biology and Biochemistry, compaction that leads to a thin left ventricular wall (i.e. 6 University of Houston, Houston, TX 77204, USA. The Heart Institute, Division of noncompaction) and a prominent ventricular septal defect (VSD) Pediatric Cardiology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229, USA. (Shou et al., 1998). Fkbp1a is a ubiquitously expressed 12 kDa peptidyl-prolyl *These authors contributed equally to this work isomerase (Bierer et al., 1990; Schreiber and Crabtree, 1995). It ‡Authors for correspondence ([email protected]; [email protected]) binds to FK506 and rapamycin and inhibits calcineurin and mTOR

Accepted 18 February 2013 activity. Fkbp1a is associated with multiple intracellular protein DEVELOPMENT Fkbp1a in cardiac development RESEARCH ARTICLE 1947 complexes (Ozawa, 2008; Wang and Donahoe, 2004), such as recombination in these conditional genetic ablations, we first created BMP/Activin/TGFβ type I receptors, Ca2+-release channels, and Fkbp1a+/−/Cre+ mice followed by an additional intercross onto functionally regulates the voltage-gated Na+ channel (Maruyama et Fkbp1aflox/flox mice. For the most part, we used Fkbp1aflox/–/Cre+ as an flox/+ + flox/– – al., 2011). Previously, we revealed that Bmp10 is upregulated in experimental group and Fkbp1a /Cre and Fkbp1a /Cre as the Fkbp1a-deficient hearts. Bmp10-deficient mice die by E10.5 and control group. Animal protocols were approved by the Indiana University School of Medicine Institutional Animal Care and Research Advisory exhibit hypoplastic and thin ventricular walls and impaired Committee. ventricular trabeculation (Chen et al., 2004). Transgenic overexpression of Bmp10 in embryonic heart led to ventricular Histological, morphological, whole-mount and section in situ hypertrabeculation (Pashmforoush et al., 2004). The elevated level hybridization, and immunohistochemical analyses of Bmp10 observed in both Fkbp1a and Nkx2-5 knockout hearts Embryos were harvested by cesarean section. Embryos and isolated correlates strongly with the ventricular hypertrabeculation and embryonic hearts at specific stages were fixed with 4% paraformaldehyde noncompaction phenotypes displayed in these mutants (Chen et al., in PBS. The fixed embryos were paraffin embedded, sectioned (7 μm), and stained with Hematoxylin and Eosin. Whole-mount and section in situ 2009). However, the underlying mechanism by which Fkbp1a hybridization were performed as previously described (Franco et al., 2001). regulates Bmp10 expression and ventricular wall formation remains In brief, complementary RNA probes for various cardiac markers were elusive. labeled with digoxigenin (DIG)-UTP using the Roche DIG RNA Labeling Recently, it has been shown that endocardial Notch1 provides System according to the manufacturer’s guidelines. Immunohistochemical key spatial-temporal control of myocardial growth via regulation of staining was performed using the staining system from Vector Laboratories Bmp10 and neuregulin 1 (Nrg1) (Grego-Bessa et al., 2007). according to the manufacturer’s instructions. The primary antibodies used Endocardium is primarily made up of endothelial cells. Activated in the immunohistochemical analyses were: anti-Fkbp1a (FKBP12) Notch1 intracellular domain (N1ICD) was found to be more antibody (Thermo Scientific, PA1-026A), MF-20 anti-myosin heavy chain abundant in endocardial cells near the proximal end of the trabecular monoclonal antibody [Developmental Studies Hybridoma Bank (DSHB), University of Iowa], anti-Ki67 antibody (ab15580; Abcam), anti-CD31 myocardium, where trabeculation initiates, and was significantly (Pecam1) antibody (BD Biosciences, 553370) and anti-cleaved Notch1 less abundant in the endocardial cells at the distal end of the (N1ICD) antibody (Cell Signaling, 2421s). trabeculae (Grego-Bessa et al., 2007). Ablation of Notch1 or its transcriptional co-factor Rbpjk within endothelial cells results in Whole-mount immunofluorescence staining and confocal hypotrabeculation and, subsequently, early embryonic lethality (Del microscopy imaging Monte et al., 2007; Grego-Bessa et al., 2007). Interestingly, both Staining and microscopy procedures were as previously described (Chen et endocardially expressed Nrg1 and myocardially expressed Bmp10 al., 2004). Embryos were harvested and washed three times in ice-cold PBS and fixed for 10 minutes in pre-chilled acetone before being treated with were downregulated in endothelial-restricted Notch1 knockout blocking solution containing 3% non-fat dried milk (Bio-Rad) and 0.025% hearts (Grego-Bessa et al., 2007). Collectively, these findings Triton X-100 for 1 hour. Directly conjugated primary antibodies were then suggested a crucial role for endocardial Notch1 in regulating added to a final concentration of 1 μg/ml for 12-18 hours at 4°C. The anti- ventricular trabeculation. CD31 antibody (BD Biosciences) and the MF-20 monoclonal antibody To determine the cellular and molecular mechanism of Fkbp1a in (DSHB) were labeled with Alexa Fluor 488 and the anti-Fkbp1a antibody regulating ventricular trabeculation and compaction, and its (Thermo Scientific) was labeled with Alexa Fluor 647 using a monoclonal pathogenetic role in LVNC, we generated Fkbp1a conditional antibody labeling kit (Molecular Probes). Samples were analyzed using a knockouts using the Cre-loxP recombination system. Ablating Bio-Rad MRC 1024 laser scanning confocal microscope equipped with a Fkbp1a in cardiac progenitor cells via the use of Nkx2.5cre mice krypton-argon laser (488, 647 nm). z-series were obtained by imaging serial confocal planes at 512×512 pixel resolution with a Nikon 20× oil- (Moses et al., 2001), we were able to generate Fkbp1aflox/–:Nkx2.5cre immersion objective (2 μm intervals). mice that recapitulate the ventricular hypertrabeculation and noncompaction with full penetrance observed in systemic Fkbp1a Echocardiography null mice. By contrast, ablation of Fkbp1a using cardiomyocyte- Mice were gently anesthetized with 1.5% isoflurane. Two-dimensional specific Cre lines did not give rise to abnormal ventricular wall short-axis images were obtained with a high-resolution micro-ultrasound formation. Surprisingly, endothelial-restricted ablation of Fkbp1a system (Vevo 770, VisualSonics) equipped with a 40 MHz mechanical scan phenocopied the ventricular hypertrabeculation and noncompaction probe. Fractional shortening (FS), ejection fraction, left ventricular internal diameter (LVID) during systole, LVID during diastole, end-systolic volume, observed in Fkbp1a systemically deficient mice, suggesting that and end-diastolic volume were calculated with Vevo Analysis software endocardium plays an important role in regulating ventricular (version 2.2.3) as described previously (Zhu et al., 2009). LVID during trabeculation and compaction. Biochemical and molecular analyses systole and diastole were measured from M-mode recording at the level of demonstrated that Fkbp1a regulates Notch1-mediated signaling the mid-papillary muscle, whereas end-systolic and end-diastolic volumes within developing endocardial cells. An excess of activated Notch1 were measured with B-mode recording in a plane containing the aortic and is found in Fkbp1a-ablated endothelial cells and is the likely cause mitral valves. of the observed Fkbp1a mutant phenotypes. Treatment of Fkbp1a- Cell transfection and protein stability assay deficient mice with Notch inhibitors reduces the hypertrabeculation Flag-tagged human FKBP1A cDNA was subcloned into the BamHI and phenotype. Taken together, we have revealed a mechanism whereby EcoRI sites of the pCDNA3 vector (Invitrogen). N1ICD expression plasmid Fkbp1a modulates a critical level of Notch1 activity that is required (N1ICD-PCS2) is a generous gift from Dr Raphael Kopan (Washington for regulating ventricular chamber and wall formation. University). Western blots employed anti-Flag (Sigma, F1804), anti-N1ICD (Cell Signaling) and anti-tubulin (Sigma, T6199) antibodies. For the protein MATERIALS AND METHODS stability assay, the N1ICD-expressing plasmid was transfected into control Generation of Fkbp1a floxed and conditional knockout mice HEK293 cells and those stably expressing human FKBP1A tagged with The generation of Fkbp1a floxed mice (Fkbp1aflox) has been described Flag epitope (FKBP1A-HEK293) using FuGENE HD (Roche) according to previously (Maruyama et al., 2011). To achieve cell lineage-restricted the manufacturer’s instructions. Cycloheximide (50 µg/ml, Sigma) was deletion of Fkbp1a in the developing heart, Fkbp1aflox mice were crossed to added to the medium for the indicated length of time. To inhibit

various cell type-specific Cre mouse lines. To ensure efficient Cre-loxP proteasomes, transfected cells were treated with the proteasome inhibitor DEVELOPMENT 1948 RESEARCH ARTICLE Development 140 (9)

MG-132 (10 μmol, EMD Chemicals) for various time periods. For detection were generated by infection with recombinant adenovirus expressing of ubiquitylated N1ICD, histidine-tagged ubiquitin (His-ubiquitin) Cre/eGFP (1×108 pfu). Cycloheximide (25 µg/ml) was added to the medium pCDNA3 plasmid was co-transfected with N1ICD expression plasmid into for the indicated length of time followed by western blot analysis. HEK293 or FKBP1A-HEK293 cells. After 30 hours, transfected cells were incubated with MG-132 (10 μmol) for 8 hours before being lysed in Ni- γ-Secretase inhibitor injection The γ-secretase inhibitor DBZ {(S,S)-2-[2-(3,5-difluorophenyl)acetylamino]- agarose lysis buffer [50 mM NaH2PO4, 300 mM NaCl, 5 mM imidazole, 0.05% Tween 20, 100 µg/ml N-ethylmaleimide, and Complete Protease N-(5-methyl-6-oxo-6,7-dihydro-5H-dibenzo[b,d]azepin-7-yl)propionamide} Inhibitor (Roche)]. His-ubiquitin-conjugated proteins were enriched using (EMD Chemicals, 565789) was given by intraperitoneal injection at 11.5 Ni-NTA-agarose (Qiagen) from cells co-transfected with N1ICD and His- days of pregnancy (10 μmol/kg) (Milano et al., 2004). Control mice were ubiquitin. The beads were washed ten times with Ni-agarose wash buffer administered with vehicle (0.5% hydroxypropylmethylcellulose in 0.1% Tween 80). Embryos were harvested at E13.5 and subjected to (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, 0.05% Tween 20). In order to reduce nonspecific binding to beads as much as possible, nickel- morphological, histological and in situ hybridization analyses as described binding proteins were resuspended in 2× sample buffer supplemented with above. 200 mM imidazole and heated for 10 minutes at 95°C before western blot Statistical analysis with anti-N1ICD antibody. All values are presented as mean ± s.e.m. Differences between groups were compared by Student’s t-test. P<0.05 was considered significant. Luciferase assay Hes1-pGL3 luciferase plasmid and full-length Notch1 and Dll4 expression plasmids have been described previously (McGill and McGlade, 2003). Fkbp1a knockout and control mouse embryonic fibroblasts (MEFs) and RESULTS endothelial cells were isolated as described (Weaver et al., 1997). For the Cardiac progenitor ablation of Fkbp1a luciferase assay, Hes1-pGL3 reporter plasmid was transfected into Fkbp1a To determine whether the ventricular hypertrabeculation and knockout or control MEFs and endothelial cells alone or with different noncompaction present in Fkbp1a-deficient mice is a primary combinations of Notch1, Dll4 and Fkbp1a expression plasmids. Renilla cardiac defect, we used Nkx2.5cre to ablate the Fkbp1a gene. luciferase plasmid was co-transfected as an internal control. Transfection Nkx2.5cre delivers Cre within cardiac progenitor cells in the cardiac efficiency was determined by co-transfection of pCDNA3-EGFP. Luciferase crescent, which give rise to cardiomyocytes, endocardial cells and activity was analyzed 36 hours after transfection with the Dual-Luciferase to a subpopulation of cells from the septum transversum that gives Assay System (Promega) according to the manufacturer’s instructions. rise to the epicardium, and within cardiac neural crest cells (Moses flox/– cre Isolation and culture of endothelial cells et al., 2001; Zhou et al., 2008). Fkbp1a :Nkx2.5 mice Murine endothelial cells were isolated from lung tissues of Fkbp1aflox/flox demonstrated fetal lethality, dying in utero between E14.5 and birth. neonates (day 10). The harvested lung tissue was minced to small pieces The mutants phenocopied Fkbp1a-deficient mice, with profound and then digested with 0.25% collagenase A (Stem Cell Technologies) for hypertrabeculation, noncompaction and VSDs with 100% 40 minutes followed by cell dissociation buffer (Life Technologies) in order penetrance (Fig. 1; Table 1) (Shou et al., 1998), indicating that the to produce a single-cell suspension. The cell suspension was filtered through ventricular hypertrabeculation and noncompaction within Fkbp1a a 70-μm strainer and stained with anti-CD31, anti-CD45 (Ptprc) and anti- systemic knockouts are primarily derived from defects in the Mac1 (Itgam) antibodies (all eBiosciences). The CD45– Mac1– CD31+ cardiogenic program, and are not secondary to defects in another endothelial cells were sorted on a FACS Aria (Becton Dickinson). The sorted endothelial cells were cultured in EGM2 (endothelial growth medium organ system. 2, Lonza) with 10% FBS, and 7 days after initiating culture the adherent cells were infected with recombinant retrovirus expressing human Specific ablation of Fkbp1a within myocardial, telomerase to maintain proliferation potential (Hahn et al., 1999). After 2 epicardial and neural crest cell-restricted cell weeks of further culture, the endothelial cells were reselected and enriched populations for CD45– Mac1– CD31+ cells by FACS sorting. The resulting endothelial To further determine the cell-autonomous contribution to the cells were analyzed by FACS and confirmed by immunofluorescence ventricular hypertrabeculation and noncompaction caused by staining for endothelial cell surface markers including CD31 (eBiosciences), Fkbp1a ablation, we crossbred Fkbp1aflox mice onto the Vcam1 (Santa Cruz) and Flk1 (Kdr) (eBiosciences) (see supplementary cardiomyocyte-specific sodium-calcium exchanger promoter H1- material Fig. S3A,B). To further validate the endothelial characteristics of cre (H1-Ncx-cre) (Müller et al., 2002). The Cre activity in H1-Ncx- the cells, we used an in vitro MatriGel tube-forming assay (see cre is observed in developing myocardium as early as E9.5 and in supplementary material Fig. S3C). As previously described, we plated Fkbp1aflox/flox cells on a thin layer of MatriGel (Stem Cell Technologies) at a cardiomyocyte-restricted manner throughout embryonic, fetal and 1×104 cells/well of a 96-well plate in ECM2 containing 10% FCS and adult life (supplementary material Fig. S1). To our surprise, flox/– allowed a tubular structure to form. Typical tube formation after 20 hours Fkbp1a :H1-Ncx-cre mice demonstrated normal ventricular in culture (supplementary material Fig. S3) suggested that the isolated chamber formation and survived to adulthood. Histologically, no Fkbp1aflox/flox cells maintained endothelial cell characteristics. To generate hypertrabeculation or noncompaction was found in Fkbp1aflox/–:H1- Fkbp1a-deficient endothelial cells, Fkbp1aflox/flox endothelial cells were Ncx-cre hearts (Fig. 2; Table 1). Western blot, qRT-PCR and infected with recombinant adenovirus expressing Cre/eGFP (Zhang et al., immunofluorescence analyses confirmed efficient ablation of 2004), in which the eGFP signal was used to evaluate transduction Fkbp1a within the embryonic myocardium (Fig. 2). This finding efficiency. Western blot and qRT-PCR analyses were used to confirm the suggests that cardiomyocytes are not the primary cell type expression levels of Fkbp1a in these cells. For plasmid transfection of contributing to the ventricular hypertrabeculation and endothelial cells, X-tremeGENE HP DNA transfection reagents (Roche noncompaction observed in mice with systemic Fkbp1a ablation. Diagnostics) were used according to the manufacturer’s instructions. flox To test Notch1 protein stability in endothelial cells, N1ICD-PCS2 was co- To validate this observation, we crossbred Fkbp1 mice to other transfected into Fkbp1aflox/flox endothelial cells with empty pCDNA3 cardiomyocyte-specific Cre transgenic lines, including cTnt-cre plasmid at a 10 to 1 ratio using FuGENE HD (Roche Applied Bioscience) (Table 1) (Jiao et al., 2003; Wang et al., 2000). Consistently, all of according to the manufacturer’s instructions. Cells stably expressing N1ICD these mutant mice exhibited normal ventricular chamber formation. were selected by neomycin resistance with G418 (Invitrogen, 400 ng/ml in cTnt-cre delivers Cre within the developing myocardium as early as cre

culture medium). N1ICD-expressing Fkbp1a-deficient endothelial cells E7.5 (Jiao et al., 2003), which is similar to that reported for Nxk2.5 . DEVELOPMENT Fkbp1a in cardiac development RESEARCH ARTICLE 1949

Fig. 1. Genetic ablation of Fkbp1a using Nkx2.5cre. (A) Comparison of gross morphology of Fkbp1aflox/–:Nkx2.5cre and control mouse embryos (top) and hearts (bottom) at E14.5. Fkbp1aflox/–:Nkx2.5cre embryos appear edematous (red arrow) and die in utero. Gray arrow indicates formation of the ventricular groove in the control heart. (B) Histological analysis of Fkbp1aflox/–:Nkx2.5cre and control hearts at E14.5. Arrows indicate the overgrowth of the trabecular myocardium in Fkbp1a mutant hearts. The width of the ventricular compact wall is indicated. The boxed regions are enlarged beneath. RA, right atrium; LA, left atrium; RV, right ventricle; LV, left ventricle; VSD, ventricular septal defect.

Despite this early Cre activity, Fkbp1aflox/–:cTnt-cre mice display Endothelial-specific ablation of Fkbp1a normal ventricular trabeculation and compaction (supplementary The developing endocardium has an important role in regulating material Fig. S2), further indicating that the abnormal ventricular ventricular wall formation (Brutsaert, 2003). To test the contribution development in Fkbp1a-deficient mice is not a direct function of of endocardium to the hypertrabeculation and noncompaction Fkbp1a within cardiomyocytes. Similarly, cardiac development was phenotypes observed with systemic Fkbp1a ablation, we studied the reported to be normal in studies using MHC-cre and Mck-cre to delete endothelial-specific deletion of Fkbp1a using Tie2-cre (Tie2 is also Fkbp1a (Maruyama et al., 2011; Tang et al., 2004). Together, these known as Tek) transgenic mice (Constien et al., 2001; Gitler et al., data rule out cardiomyocytes as the primary contributing cell lineage 2004; Kisanuki et al., 2001). Tie2-cre mice deliver Cre activity to to ventricular hypertrabeculation and noncompaction in Fkbp1a- endothelial cells, including endocardial cells, as early as E9.0 deficient mice, and instead imply another cell lineage or lineages as (Kisanuki et al., 2001). Fkbp1aflox/–:Tie2-cre mice phenocopied the the primary cause. ventricular hypertrabeculation and noncompaction seen in Fkbp1a- Epicardium has been implicated to play an important role in deficient hearts (Fig. 3; Table 1). Approximately 35% (15/45) of supporting ventricular wall growth and, potentially, compaction E13.5-14.5 Fkbp1aflox/–:Tie2-cre embryos were edematous and (Sucov et al., 2009). To test whether epicardial Fkbp1a contributes appeared to have failing hearts as evidenced by the absence of to hypertrabeculation or noncompaction, we ablated Fkbp1a from effective perfusion. Hearts isolated from these E13.5-14.5 epicardial cells using Tbx18cre mice (Cai et al., 2008). Tbx18cre mice Fkbp1aflox/–:Tie2-cre embryos were enlarged and typically deliver Cre activity largely to epicardium, with a low activity in ‘pumpkin-shaped’, lacking the normal ventricular groove, a strong cardiomyocytes refined in ventricular septum (Cai et al., 2008; indication of abnormal ventricle chamber development (Fig. 3). Christoffels et al., 2009). We found that Fkbp1aflox/–:Tbx18cre Histological analysis of these hearts confirmed multiple abnormal mutants had normal ventricular wall formation (Table 1; cardiac ventricular structures similar to those of Fkbp1a−/− hearts, supplementary material Fig. S2). Previously we showed that, despite including various degrees of hypertrabeculation, noncompaction low penetrance, Fkbp1a-deficient mice develop exencephaly (Shou and a thin compact wall. Prominent VSDs, including both et al., 1998), indicating potential involvement of cranial-facial membranous and muscular, were also observed in six of these 15 neural crest cells. We used Wnt1-cre to ablate Fkbp1a from neural mutant hearts. crest cell lineages (Jiang et al., 2000). Once again, we observed Although the majority of Fkbp1aflox/–:Tie2-cre mice (compared normal ventricular wall formation in Fkbp1aflox/–:Wnt1-cre mutants with just ~5% of Fkbp1a−/− mice) survived in utero, most of these (Table 1; supplementary material Fig. S2). When considering the survivors died within 8 weeks of birth, probably owing to cell lineages marked by Nkx2.5cre, the endocardium remained compromised cardiac function (see below) and defects in T cell untested. Therefore, we next sought to specifically ablate Fkbp1a function (data not shown); the surviving Fkbp1aflox/–:Tie2-cre mice from the endocardium. exhibited milder hypertrabeculation and noncompaction phenotypes

Table 1. Phenotypes of cell lineage-restricted Fkbp1a mutants Ablation Cre line Phenotype n (mutant, control, litters)* Cardiac Nkx2.5cre 100% embryonic lethality with ventricular hypertrabeculation, 10, 12, 7 noncompaction and VSD Cardiomyocyte H1-Ncx-cre 100% viable and fertile, normal ventricular development 11, 35, 12 Endothelial Tie2-cre ~35% with ventricular hypertrabeculation, noncompaction and VSD 45, 22, 31 (E13.5 and E14.5) Epicardial Tbx18cre 100% viable and fertile, normal ventricular development 5, 5, 4 Neural crest cell Wnt1-cre 100% survive to postnatal, normal ventricular wall development, but 100% 25, 25, 17 die within 1 month of birth due to defect in thymus development *The number of mutant and control embryos analyzed from the indicated number of litters, respectively. VSD, ventricular septal defect. DEVELOPMENT 1950 RESEARCH ARTICLE Development 140 (9)

of Fkbp1a-deficient, Fkbp1aflox/–:Tie2-cre and wild-type control hearts. Using immunohistochemistry, we were able to confirm that nuclear N1ICD was robustly distributed within the endocardial cells close to the proximal end of the trabecular myocardium of control hearts (Fig. 5A). By contrast, this characteristic pattern of N1ICD was clearly altered in Fkbp1a mutant hearts, as N1ICD was found in endocardial cells throughout and alongside the trabecular myocardium (Fig. 5A), suggesting that Notch1 is activated ectopically within Fkbp1a-deficient and Fkbp1aflox/–:Tie2-cre endocardium. Furthermore, Hey1 and ephrin B2 (Efnb2), two well-known direct downstream targets of Notch1 that are expressed within endocardium (Grego-Bessa et al., 2007; Hainaud et al., 2006), were found significantly upregulated in the Fkbp1a-deficient endocardial cells, supporting the observation that there is increased Notch1 activity in Fkbp1a mutant endocardium (Fig. 5B,C). Interestingly, although the overall level of Hey2 expression did not appear to differ significantly in Fkbp1a mutant versus control heart (Fig. 5C), its expression pattern was significantly altered (Fig. 5B). In normal myocardium, Hey2 transcript is restricted to compact myocardium. We found that Hey2 expression is expanded into the trabecular myocardium in Fkbp1a mutant heart (Fig. 5B). Endocardial Nrg1 expression is downstream of Efnb2 and is downregulated in Notch1 mutant endocardium (Grego-Bessa et al., 2007). Both in situ hybridization and qRT-PCR analyses confirmed that Nrg1 was upregulated in Fkbp1aflox/–:Tie2-cre heart, as was Bmp10 (Fig. 5B,C). Taken together, our data demonstrate that Fkbp1a is an important regulator for endocardial Notch1-mediated signaling during ventricular wall formation.

Fkbp1a regulates N1ICD stability and turnover To further define the role of Fkbp1a in Notch1 activation, we isolated mouse embryonic fibroblasts (MEFs) from Fkbp1a- Fig. 2. Cardiomyocyte-restricted ablation of Fkbp1a using H1-Ncx- deficient and control embryos (E12.5) and analyzed Notch cre transgenic mice. (A) Dual immunofluorescence analyses confirm the activity. Notch-dependent Hes1-luciferase reporter plasmid was genetic ablation of Fkbp1a in developing myocardium at E12.5. transfected into Fkbp1a-deficient and wild-type control MEFs in Representative confocal images of heart sections co-stained with Alexa combination with Notch1 and delta-like 4 (Dll4) expression Fluor 647-conjugated anti-Fkbp1a antibody (red) and Alexa Fluor 488- plasmids. The level of luciferase activity reflects the level of conjugated anti-myosin heavy chain antibody (MF-20; green). There is a flox/– activated Notch1 (McGill and McGlade, 2003). From four significant reduction in Fkbp1a in Fkbp1a :H1-Ncx-cre myocardium. independent sets of experiments, we found that luciferase The boxed regions of the merge are enlarged to the right. (B) Western blot confirms the efficient removal of Fkbp1a in Fkbp1aflox/–:H1-Ncx-cre activities were significantly enhanced in both Dll4-transfected and heart. (C) Morphological and histological analysis of Fkbp1aflox/–:H1-Ncx- Dll4/Notch1 co-transfected Fkbp1a-deficient MEFs when cre and control hearts at E14.5. RA, right atrium; LA, left atrium; RV, right compared with transfected wild-type MEFs (Fig. 6A). Re- ventricle; LV, left ventricle. introduction of Fkbp1a into the Fkbp1a-deficient cells reduced luciferase activity to a level comparable to that in wild-type cells. Consistent with these findings, western blot analysis demonstrated (Fig. 4). Echocardiographic analysis demonstrated severely that the endogenous N1ICD level is significantly higher in compromised cardiac function in these mutants, consistent with the Fkbp1a-deficient than in wild-type MEFs (Fig. 6A). To validate histological findings (Fig. 4). Collectively, our observations indicate this finding in endothelial cells, we prepared endothelial cells from that abnormal endocardial function is likely to be the primary cause Fkbp1aflox/flox neonatal mice (supplementary material Fig. S3). of the hypertrabeculation and noncompaction phenotypes in Fkbp1a-deficient endothelial cells were generated by transducing Fkbp1a-deficient mutants. the cells with Ad-Cre/eGFP. Western blot confirmed the efficient ablation of Fkbp1a in Cre-transduced Fkbp1aflox/flox mouse Notch1 and downstream endocardial signaling are endothelial cells (Fkbp1aflox/flox/Cre) (Fig. 6B). Hes1-luciferase altered in Fkbp1a mutant hearts reporter assays were performed in Fkbp1aflox/flox and Recent studies have demonstrated that perturbation of Notch1 Fkbp1aflox/flox/Cre cells (Fig. 6B). Similar to the finding in MEFs, signaling within the endocardium impairs cardiomyocyte Bmp10 Notch-dependent luciferase activities were significantly higher in expression, ventricular trabeculation and wall formation (Grego- Fkbp1aflox/flox/Cre than in Fkbp1aflox/flox endothelial cells (Fig. 6B). Bessa et al., 2007). To define whether Notch1 activation within the We transfected activated Notch1 (N1ICD-PCS2) into Fkbp1aflox/flox endocardium is altered in Fkbp1a mutants, we compared the pattern endothelial cells in conjunction with pCDNA3-FKBP1A (human) or

of activated N1ICD in the developing ventricular endocardial cells control pCDNA3-eGFP vector. The N1ICD protein level in DEVELOPMENT Fkbp1a in cardiac development RESEARCH ARTICLE 1951

Fig. 3. Endothelial-restricted ablation of Fkbp1a using Tie2-cre transgenic mice. (A) Dual immunofluorescence and PCR analyses to confirm the genetic ablation of Fkbp1a in developing endocardial/endothelial cells at E12.5. Representative confocal images of heart sections are co- stained with Alexa Fluor 647-conjugated anti-Fkbp1a antibody (red) and Alexa Fluor 488-conjugated anti-CD31 antibody (green). There is a significant reduction of Fkbp1a in the CD31-positive endothelial cells in Fkbp1aflox/–:Tie2-cre hearts. Beneath is shown a diagnostic genomic PCR analysis of DNA samples isolated from E13.5 heart alongside a schematic of the Fkbp1aflox allele showing the location of diagnostic primers p1, p2 and p3; E, exon. A specific p3-p2 band represents the allele after Cre-loxP-mediated recombination. (B) Morphological and histological analysis of Fkbp1aflox/–:Tie2-cre and control embryos and hearts at E14.5. Fkbp1aflox/–:Tie2-cre embryos are edematous (red arrow). The mutant hearts lack a normal ventricular groove (yellow arrows) and demonstrate hypertrabeculation and noncompaction (black arrows). RA, right atrium; LA, left atrium; RV, right ventricle; LV, left ventricle; VSD, ventricular septal defect.

Fkbp1aflox/flox FKBP1A-overexpressing endothelial cells was The rate of N1ICD degradation was significantly greater in significantly lower than in Fkbp1aflox/flox eGFP-expressing cells FKBP1A-HEK293 cells (Fig. 7Aa,b). To validate this finding in (Fig. 6C, top). As expected, overexpression of FKBP1A reduced endothelial cells, we generated stable N1ICD-overexpressing Notch-dependent Hes1-luciferase activity (Fig. 6C, bottom). Fkbp1aflox/flox cells (N1ICD-Fkbp1aflox/flox) and compared their To determine whether Fkbp1a could affect endogenous Notch1 N1ICD degradation rate with that of N1ICD-Fkbp1aflox/flox/Cre expression in endothelial cells, we compared Notch1 mRNA levels cells. N1ICD-Fkbp1aflox/flox/Cre cells had a significantly slower in Fkbp1a-deficient and human FKBP1A-overexpressing degradation rate than that of N1ICD-Fkbp1aflox/flox cells endothelial cells (Fig. 6D). Notch1 mRNA levels were not altered (Fig. 7Ac,d). in Fkbp1a-deficient or FKBP1A-overexpressing endothelial cells. We also evaluated the rate of N1ICD translation. There is no However, the mRNA levels of the Notch1 downstream targets noticeable difference in N1ICD protein synthesis between Efnb2, Hey1 and Hey2 were upregulated in Fkbp1a-deficient and FKBP1A-HEK293 cells and controls (Fig. 7B; supplementary downregulated in FKBP1A-overexpressing endothelial cells, material Fig. S5). We next blocked ubiquitylation-mediated protein consistent with the notion that Notch1 activities are impacted degradation using the proteasome inhibitor MG-132, and assessed directly by the intracellular levels of Fkbp1a. the amount of ubiquitylated N1ICD in FKBP1A-HEK293 and Previously, we established HEK293 cell lines with control cells. The amount of ubiquitylated N1ICD was significantly overexpression of Flag-tagged human FKBP1A (FKBP1A- higher in FKBP1A-HEK293 cells than in HEK293 control cells HEK293). When we attempted to overexpress N1ICD in HEK293 (Fig. 7C). Taken together, these data indicate that Fkbp1a regulates and FKBP1A-HEK293 cells, we found that the N1ICD protein level the level of activated Notch1 by controlling its stability via was significantly reduced in FKBP1A-HEK293 cells when ubiquitin-mediated protein turnover. compared with control HEK293 cells (supplementary material Fig. S4). Given that Notch activity is closely associated with The γ-secretase inhibitor DBZ reduces ubiquitylation-mediated degradation, this finding, along with the hypertrabeculation in Fkbp1a-deficient hearts observation that there is no change in Notch1 mRNA levels and that Clearly, elevated Notch signaling is directly associated with N1ICD protein is significantly higher in Fkbp1a-deficient cells, abnormal trabeculation in Fkbp1a mutant embryos. We therefore suggested that, mechanistically, Fkbp1a regulates Notch1 activity reasoned that reducing Notch1 activation in vivo would be able to by modulating N1ICD stability. To test this hypothesis, we rescue the ventricular morphological defects in Fkbp1a mutant compared the N1ICD protein degradation rate in human FKBP1A- hearts, thereby validating the hypothesized mechanism. We applied

overexpressing (i.e. FKBP1A-HEK293) and control HEK293 cells. the γ-secretase inhibitor DBZ (Milano et al., 2004) to Fkbp1a DEVELOPMENT 1952 RESEARCH ARTICLE Development 140 (9)

Fig. 4. Histological and functional analyses of adult Fkbp1aflox/–:Tie2-cre mice. (A) (Left) Restricted growth of surviving Fkbp1aflox/–:Tie2-cre mice. (Right) Abnormal ventricular wall structure in Fkbp1aflox/–:Tie2-cre hearts, with hypertrabeculation and noncompaction. Arrows indicate abnormal trabecular myocardia and ventricular septum. (B) M-mode echocardiographic analysis of 2-month-old adult males. One-second traces are shown. Both fractional shortening and ejection fraction are severely compromised in Fkbp1aflox/–:Tie2-cre mutants. Values indicate mean ± s.e.m.; n=12; *P<0.05. LVIDd, left ventricular internal diameter diastolic; LVIDs, left ventricular internal diameter systolic; FS%, fractional shortening; LV vol d, left ventricular volume diastolic; LV vol s, left ventricular volume systolic; EF%, ejection fraction; HR, heart rate.

mutant embryos at E11.5, a stage prior to the appearance of the administration of DBZ significantly diminished the hypertrabeculation and noncompaction phenotypes (Chen et al., hypertrabeculation phenotypes in Fkbp1a-deficient hearts when 2009). The pregnant Fkbp1a+/− females that were bred with compared with vehicle-treated controls (Fig. 8A,Ba). The ratio of Fkbp1a+/− males were administrated with DBZ (10 μM/kg via trabeculae thickness to compact wall thickness approached a normal intraperitoneal injection, n=6 females) and vehicle (100 μl, n=6 morphological distribution (Fig. 8Bc). Remarkably, ventricular females). The embryos were harvested at E13.5 and processed for septum formation was also greatly improved in all DBZ-treated morphological and histological examination. Results show that Fkbp1a-deficient hearts.

Fig. 5. Assessment of Notch1-mediated signaling in the developing endocardium of Fkbp1a-deficient and Fkbp1aflox/–:Tie2-cre mouse hearts. (A) Immunohistological analysis using antibody specific to activated Notch1 (N1ICD). In control heart (WT), nuclear N1ICD is mainly located in endothelial cells at the proximal end of trabeculae (arrows in aЈ). By contrast, N1ICD was found throughout endothelial cells in both Fkbp1a-deficient and Fkbp1aflox/–:Tie2-cre mutant hearts (arrows in bЈ and cЈ). The boxed regions in a-c are magnified in aЈ-cЈ. (B) In situ hybridization of downstream targets of Notch1 in Fkbp1aflox/–:Tie2-cre and control hearts at E12.5. Ephrin B2 (Efnb2), neuregulin 1 (Nrg1) and Hey1 are upregulated in endocardial cells in Fkbp1aflox/–:Tie2-cre hearts. Bmp10 is upregulated in both trabecular and compact myocardium. Interestingly, Hey2 expression is upregulated in trabecular myocardium compared with controls. Arrows indicate positive signals. (C) qRT-PCR confirms the expression levels of Notch1, Efnb2, Nrg1, Hey1, Hey2 and Bmp10 in Fkbp1aflox/–:Tie2-cre hearts at E13.5. Interestingly, despite the altered expression pattern, the overall expression level of Hey2 is not altered. Error bars indicate s.e.m. LV, left ventricle. DEVELOPMENT Fkbp1a in cardiac development RESEARCH ARTICLE 1953

Fig. 6. Biochemical analyses of altered Notch1 activity in Fkbp1a mutant cells. (A) (Left) Luciferase assay to determine Notch1 activity in Fkbp1a- deficient and control mouse embryonic fibroblasts (MEFs) in which Hes1-luciferase reporter was transfected together with Notch1, Dll4, Fkbp1a and eGFP as indicated. Fkbp1a-deficient cells maintain significantly higher luciferase activity, which is reduced by Fkbp1a re-introduction. (Right) Western blot analysis of endogenous N1ICD levels in Fkbp1a-deficient and wild-type MEFs. (B) (Left) Luciferase assay to determine Notch1 activity in Fkbp1a- deficient and control mouse endothelial cells. (Right) Western blot shows that the N1ICD protein level is significantly higher in Fkbp1a-deficient cells. (C) (Top) Western blot showing that the N1ICD protein level is downregulated in mouse endothelial cells overexpressing human FKBP1A. (Bottom) Luciferase assay to determine Notch1 activity in FKBP1A-overexpressing and control endothelial cells. (D) qRT-PCR analysis shows that Notch1 mRNA levels are not altered in Fkbp1a-deficient or FKBP1A-overexpressing endothelial cells. However, Efnb2, Hey1 and Hey2 expression levels are altered. Error bars indicate s.e.m.

Consistent with the histological findings, in situ hybridization DISCUSSION demonstrated that the levels of endocardial Efnb2 and Nrg1, as well Ventricular hypertrabeculation and noncompaction are congenital as myocardial Bmp10, were significantly reduced in DBZ-treated myocardial anomalies that result from disrupted development of the as compared with vehicle-treated Fkbp1a−/− hearts (Fig. 8C). As ventricular wall and have been increasingly recognized in the clinic further validation, Ki67 immunohistological staining showed that (Chen et al., 2009). Genetic analysis of patients, mostly with isolated DBZ treatment resulted in strong suppression of cardiomyocyte forms of ventricular noncompaction without CHDs (congenital heart proliferative activities in both trabecular and compact myocardium defects), has demonstrated a broad genetic heterogeneity, suggesting within Fkbp1a−/− hearts, showing a more wild-type growth profile a complex and/or multifactorial network contributing to the etiology (Fig. 8C; supplementary material Fig. S6). However, despite and pathogenesis (Chen et al., 2009). Currently, surviving patients significant reductions in cardiomyocyte hyperproliferation and the carrying FKBP1A mutations have not been reported. Given the myocardial hypertrabeculation phenotype of DBZ-treated mutant pivotal role of Fkbp1a in ventricular wall formation and the hearts, the thickness of the compact wall (ventricular compaction) embryonic lethal phenotype of Fkbp1a-deficient mice, patients with did not seem to show significant improvement upon DBZ treatment, a germline FKBP1A loss-of-function mutation would seem to have suggesting that a Notch-independent signaling pathway might little chance of survival in utero. However, mouse genetic contribute in part to the regulation of ventricular wall compaction. manipulation provides an excellent opportunity to determine the cause Taken together, these findings define an important role for Fkbp1a and the pathogenetic network that contributes to the ventricular in regulating Notch1-mediated signaling during ventricular wall hypertrabeculation and noncompaction, allowing for the identification

development. of genetic pathways that regulate normal ventricular development. DEVELOPMENT 1954 RESEARCH ARTICLE Development 140 (9)

Fig. 7. Biochemical analysis of N1ICD stability and degradation. (A) Assessment of the N1ICD degradation rate in HEK293 versus FKBP1A-HEK293 cells and in N1ICD- Fkbp1aflox/flox versus N1ICD-Fkbp1aflox/flox/Cre mouse endothelial cells. Cells were treated with the indicated concentrations of cycloheximide (CHX). Representative western blots show the N1ICD protein level at different time points after CHX treatment in HEK293 cells and endothelial cells (EC) of different genotype (a and c), and representative analyses of N1ICD degradation in FKBP1A-HEK293 cells versus control HEK293 cells (b) and N1ICD-Fkbp1aflox/flox versus N1ICD-Fkbp1aflox/flox/Cre endothelial cells (d). (B) Representative western blot of three independent experiments showing that N1ICD protein synthesis is not affected in FKBP1A-HEK293 cells. (C) Ubiquitylated N1ICD is significantly more abundant in FKBP1A-HEK293 cells. To detect the ubiquitylated N1ICD protein, 6× histidine-tagged ubiquitin (His-ubiquitin) plasmid was co-transfected with N1ICD expression plasmid into HEK293 or FKBP1A-HEK293 cells. Transfected cells were incubated with the proteasome inhibitor MG-132 (10 μmol) for 8 hours before cell lysis and Ni-NTA pulldown assay. Error bars indicate s.e.m.

As an isomerase, Fkbp1a presumably functions as a chaperone proteasome-dependent degradation of N1ICD is thought to be in maintaining the appropriate conformation of its substrate mediated by interactions of its PEST (proline, glutamic acid, proteins/peptides, which could be relevant to the regulation of serine and threonine) domain and a specific E3 ubiquitin ligase, their functional state and/or stability (Lu et al., 2007). Our findings Fbxw7 (Oberg et al., 2001). As a known cis-trans peptidyl-prolyl suggest that Fkbp1a has a role in regulating endocardial Notch1 isomerase, one attractive hypothesis is that Fkbp1a regulates the activity, which is crucial to ventricular wall formation. Our conformation of N1ICD via the proline residues contained within findings are entirely consistent with recent reports by Mysliwiec its PEST domain. To test this hypothesis, we used co- and colleagues that upregulation of Notch1 in endocardial cells immunoprecipitation to determine whether Fkbp1a is able to leads to ventricular hypertrabeculation and noncompaction directly bind N1ICD. However, we have not been able to (Mysliwiec et al., 2011; Mysliwiec et al., 2012). Most recently, consistently detect a direct molecular interaction between N1ICD Yang and colleagues demonstrated that upregulated Notch2 and Fkbp1a (data not shown). This negative result could be due to activity in Numb/Numb-like compound deficient mutants technical difficulties in capturing transient hit-and-run type contributes to Bmp10 upregulation and ventricular interactions between Fkbp1a and N1ICD. Further study hypertrabeculation and noncompaction phenotypes (Yang et al., employing systematic biochemical analyses will be necessary to 2012), further validating our conclusion. However, it should be determine whether Fkbp1a directly interacts with N1ICD to noted that Tie2-Cre is not endocardial specific, but targets to all facilitate its ubiquitylation, or operates via an indirect mechanism. endothelial cells including coronary endothelial cells. Nfatc1-Cre In addition, as Fkbp1a interacts with several other protein expression is initially restricted to the endocardium, and complexes, our findings do not exclude the potential involvement subsequently marks cardiac vascular endothelial cells as the of other signaling pathways in contributing to the ventricular wall coronary vasculature system develops (Wu et al., 2012). Owing defects in Fkbp1a mutant hearts. Indeed, although the Notch to the spatial-temporal pattern of Cre activity within these lines, inhibitor DBZ profoundly reduced cellular hyperproliferation and we currently cannot distinguish the relative contribution of the hypertrabeculation, the noncompaction phenotype was barely different cardiac endothelial cell subtypes. Despite this limitation, affected by DBZ (Fig. 8). This could simply reflect the complex it is nonetheless clear that Fkbp1a signaling in cardiac endothelial dynamics of ventricular compaction, which involves multiple cells plays a major role in regulating ventricular trabeculation and Notch-mediated signaling pathways in multiple cardiac cell compaction. lineages, or there could be an additional unknown parallel The biochemical mechanism by which Fkbp1a regulates signaling pathway(s) that synergistically controls ventricular Notch1 activity is likely to involve controlling the stability of compaction. N1ICD. Indeed, biochemical analyses in endothelial cells, MEFs Endothelial cell-restricted ablation of Notch1 and Rbpjk and HEK293 cells indicate that Fkbp1a can modulate the stability downregulates the expression of both Efnb2 and Nrg1 within of activated Notch1. It is well-known that Notch signaling is endothelial cells and Bmp10 within cardiomyocytes (Grego-Bessa regulated by post-translational modification events. Activated et al., 2007). Both Bmp10 and Nrg1 have been suggested to have N1ICD is cleared and thereby inactivated within the cell by a major functional role in the spatial-temporal control of

ubiquitylation and subsequent proteolysis via the proteasome. This cardiomyocyte proliferation and ventricular trabeculation (Grego- DEVELOPMENT Fkbp1a in cardiac development RESEARCH ARTICLE 1955

Fig. 8. Rescue of Fkbp1a-deficient mice with the γ- secretase inhibitor DBZ. (A) Representative images of E13.5 hearts isolated from DBZ-treated and control mouse embryos. (B) Quantification of the thickness of trabeculae (a), the thickness of the compact wall (b), the ratio of the thickness of trabeculae and compact wall (c), and the ratio of overall trabecular area versus the length of the compact wall (d). The data demonstrate that the thickness of ventricular trabeculae is reduced in DBZ-treated Fkbp1a- deficient hearts, and the ratio of trabecular thickness versus compact wall thickness is close to normal when compared with controls. Error bars indicate s.e.m. (C) Representative images of three independent sets of Efnb2, Nrg1 and Bmp10 in situ hybridization and Ki67 immunohistological staining of E13.5 hearts from DBZ-treated or vehicle-treated Fkbp1a-deficient and control Fkbp1a heterozygous embryos. LV, left ventricle; RV, right ventricle.

Bessa et al., 2007; Shi et al., 2003). Interestingly, Bmp10 cardiomyocyte proliferation and trabeculation in the highly expression can be specifically upregulated by Nrg1 (Kang and trabeculated zebrafish heart. Sucov, 2005). In addition, Nrg1 hypomorphic mutants have a In summary, our study demonstrates a novel genetic and significantly reduced Bmp10 expression level (Lai et al., 2010). biochemical role for Fkbp1a in regulating Notch1 activity within These findings, both in vitro and in vivo, suggest the interesting the endocardium that in part facilitates the intercellular interplay idea that Bmp10 is likely to be a downstream target of Nrg1. between endocardium and myocardium necessary to define the Endocardial-derived Nrg1 is of particular interest as it signals normal levels of ventricular trabeculation and compaction in the through its receptor complex, ErbB2/4, which is present only in mammalian heart. adjacent cardiomyocytes (Lemmens et al., 2006), providing a unique mechanism for intercellular interactions between two Acknowledgements We thank Dr Shaoliang Jing and William Carter of Indiana University mouse different cardiac cell populations. In zebrafish, the heart is highly core for their expert assistance. trabeculated. A recent study using zebrafish heart as a model system demonstrated that a Nrg1-ErbB2/4 signaling cascade has Funding a dual function: in addition to promoting cardiomyocyte This work was supported in part by the National Institutes of Health [grants proliferation, it regulates cardiomyocyte delamination and HL81092 to W.S. and HL85098 to L.J.F., A.B.F., S.J.C. and W.S.]; and by Riley Children’s foundation (to L.J.F., A.B.F., S.J.C. and W.S.). Deposited in PMC for migration (Liu et al., 2010). This dual activity ultimately controls release after 12 months. trabeculation in zebrafish. Interestingly, analysis of endocardial- restricted Rbpjk knockouts has shown that exogenous Bmp10 Competing interests statement rescues the observed myocardial proliferative defect, whereas The authors declare no competing financial interests. Nrg1 does not (Grego-Bessa et al., 2007). Apparently, the Nrg1- Supplementary material ErbB cascade is likely to be part of the evolutionarily conserved Supplementary material available online at mechanism for the regulation of cardiomyocyte delamination that http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.089920/-/DC1 allows for ventricular trabeculation (Liu et al., 2010) or for the regulation of cardiomyocyte differentiation via its pro- References Ben-Shachar, G., Arcilla, R. A., Lucas, R. V. and Manasek, F. J. (1985). differentiation activity (Lai et al., 2010). It would be interesting to Ventricular trabeculations in the chick embryo heart and their contribution to

determine whether Bmp10 has a similar function in regulating ventricular and muscular septal development. Circ. Res. 57, 759-766. DEVELOPMENT 1956 RESEARCH ARTICLE Development 140 (9)

Bierer, B. E., Mattila, P. S., Standaert, R. F., Herzenberg, L. A., Burakoff, S. J., heart rhythm and the cardiac voltage-gated sodium current in mice. Circ. Res. Crabtree, G. and Schreiber, S. L. (1990). Two distinct signal transmission 108, 1042-1052. pathways in T lymphocytes are inhibited by complexes formed between an McGill, M. A. and McGlade, C. J. (2003). Mammalian numb proteins promote immunophilin and either FK506 or rapamycin. Proc. Natl. Acad. Sci. USA 87, Notch1 receptor ubiquitination and degradation of the Notch1 intracellular 9231-9235. domain. J. Biol. Chem. 278, 23196-23203. Brutsaert, D. L. (2003). Cardiac endothelial-myocardial signaling: its role in Milano, J., McKay, J., Dagenais, C., Foster-Brown, L., Pognan, F., Gadient, R., cardiac growth, contractile performance, and rhythmicity. Physiol. Rev. 83, 59- Jacobs, R. T., Zacco, A., Greenberg, B. and Ciaccio, P. J. (2004). Modulation 115. of notch processing by gamma-secretase inhibitors causes intestinal goblet Cai, C. L., Martin, J. C., Sun, Y., Cui, L., Wang, L., Ouyang, K., Yang, L., Bu, L., cell metaplasia and induction of genes known to specify gut secretory lineage Liang, X., Zhang, X. et al. (2008). A myocardial lineage derives from Tbx18 differentiation. Toxicol. Sci. 82, 341-358. epicardial cells. Nature 454, 104-108. Moorman, A. F. and Christoffels, V. M. (2003). Cardiac chamber formation: Chen, H., Shi, S., Acosta, L., Li, W., Lu, J., Bao, S., Chen, Z., Yang, Z., development, genes, and evolution. Physiol. Rev. 83, 1223-1267. Schneider, M. D., Chien, K. R. et al. (2004). BMP10 is essential for maintaining Moses, K. A., DeMayo, F., Braun, R. M., Reecy, J. L. and Schwartz, R. J. (2001). cardiac growth during murine cardiogenesis. Development 131, 2219-2231. Embryonic expression of an Nkx2-5/Cre gene using ROSA26 reporter mice. Chen, H., Zhang, W., Li, D., Cordes, T. M., Mark Payne, R. and Shou, W. (2009). Genesis 31, 176-180. Analysis of ventricular hypertrabeculation and noncompaction using Müller, J. G., Isomatsu, Y., Koushik, S. V., O’Quinn, M., Xu, L., Kappler, C. S., genetically engineered mouse models. Pediatr. Cardiol. 30, 626-634. Hapke, E., Zile, M. R., Conway, S. J. and Menick, D. R. (2002). Cardiac- Christoffels, V. M., Grieskamp, T., Norden, J., Mommersteeg, M. T., Rudat, C. specific expression and hypertrophic upregulation of the feline Na(+)-Ca(2+) and Kispert, A. (2009). Tbx18 and the fate of epicardial progenitors. Nature exchanger gene H1-promoter in a transgenic mouse model. Circ. Res. 90, 158- 458, E8-E9. 164. Constien, R., Forde, A., Liliensiek, B., Gröne, H. J., Nawroth, P., Hämmerling, Mysliwiec, M. R., Bresnick, E. H. and Lee, Y. (2011). Endothelial Jarid2/Jumonji G. and Arnold, B. (2001). Characterization of a novel EGFP reporter mouse to is required for normal cardiac development and proper Notch1 expression. J. monitor Cre recombination as demonstrated by a Tie2 Cre mouse line. Genesis Biol. Chem. 286, 17193-17204. 30, 36-44. Mysliwiec, M. R., Carlson, C. D., Tietjen, J., Hung, H., Ansari, A. Z. and Lee, Y. Del Monte, G., Grego-Bessa, J., González-Rajal, A., Bolós, V. and De La (2012). Jarid2 (Jumonji, AT rich interactive domain 2) regulates NOTCH1 Pompa, J. L. (2007). Monitoring Notch1 activity in development: evidence for expression via histone modification in the developing heart. J. Biol. Chem. 287, a feedback regulatory loop. Dev. Dyn. 236, 2594-2614. 1235-1241. Finsterer, J. (2009). Cardiogenetics, neurogenetics, and pathogenetics of left Oberg, C., Li, J., Pauley, A., Wolf, E., Gurney, M. and Lendahl, U. (2001). The ventricular hypertrabeculation/noncompaction. Pediatr. Cardiol. 30, 659-681. Notch intracellular domain is ubiquitinated and negatively regulated by the Franco, D., de Boer, P. A., de Gier-de Vries, C., Lamers, W. H. and Moorman, mammalian Sel-10 homolog. J. Biol. Chem. 276, 35847-35853. A. F. (2001). Methods on in situ hybridization, immunohistochemistry and Ozawa, T. (2008). Effects of FK506 on Ca release channels. Perspect. Medicin. beta-galactosidase reporter gene detection. Eur. J. Morphol. 39, 169-191. Chem. 2, 51-55. Gitler, A. D., Kong, Y., Choi, J. K., Zhu, Y., Pear, W. S. and Epstein, J. A. (2004). Pashmforoush, M., Lu, J. T., Chen, H., Amand, T. S., Kondo, R., Pradervand, Tie2-Cre-induced inactivation of a conditional mutant Nf1 allele in mouse S., Evans, S. M., Clark, B., Feramisco, J. R., Giles, W. et al. (2004). Nkx2-5 results in a myeloproliferative disorder that models juvenile myelomonocytic pathways and congenital heart disease; loss of ventricular myocyte lineage leukemia. Pediatr. Res. 55, 581-584. specification leads to progressive cardiomyopathy and complete heart block. Grego-Bessa, J., Luna-Zurita, L., del Monte, G., Bolós, V., Melgar, P., Cell 117, 373-386. Arandilla, A., Garratt, A. N., Zang, H., Mukouyama, Y. S., Chen, H. et al. Pignatelli, R. H., McMahon, C. J., Dreyer, W. J., Denfield, S. W., Price, J., (2007). Notch signaling is essential for ventricular chamber development. Dev. Belmont, J. W., Craigen, W. J., Wu, J., El Said, H., Bezold, L. I. et al. (2003). Cell 12, 415-429. Clinical characterization of left ventricular noncompaction in children: a Hahn, W. C., Counter, C. M., Lundberg, A. S., Beijersbergen, R. L., Brooks, M. relatively common form of cardiomyopathy. Circulation 108, 2672-2678. W. and Weinberg, R. A. (1999). Creation of human tumour cells with defined Sandhu, R., Finkelhor, R. S., Gunawardena, D. R. and Bahler, R. C. (2008). genetic elements. Nature 400, 464-468. Prevalence and characteristics of left ventricular noncompaction in a Hainaud, P., Contrerès, J. O., Villemain, A., Liu, L. X., Plouët, J., Tobelem, G. community hospital cohort of patients with systolic dysfunction. and Dupuy, E. (2006). The role of the vascular endothelial growth factor- Echocardiography 25, 8-12. Delta-like 4 ligand/Notch4-ephrin B2 cascade in tumor vessel remodeling and Schreiber, S. L. and Crabtree, G. R. (1995). Immunophilins, ligands, and the endothelial cell functions. Cancer Res. 66, 8501-8510. control of signal transduction. Harvey Lect. 91, 99-114. Jiang, X., Rowitch, D. H., Soriano, P., McMahon, A. P. and Sucov, H. M. Shi, W., Chen, H., Sun, J., Buckley, S., Zhao, J., Anderson, K. D., Williams, R. (2000). Fate of the mammalian cardiac neural crest. Development 127, 1607- G. and Warburton, D. (2003). TACE is required for fetal murine cardiac 1616. development and modeling. Dev. Biol. 261, 371-380. Jiao, K., Kulessa, H., Tompkins, K., Zhou, Y., Batts, L., Baldwin, H. S. and Shou, W., Aghdasi, B., Armstrong, D. L., Guo, Q., Bao, S., Charng, M. J., Hogan, B. L. (2003). An essential role of Bmp4 in the atrioventricular septation Mathews, L. M., Schneider, M. D., Hamilton, S. L. and Matzuk, M. M. of the mouse heart. Genes Dev. 17, 2362-2367. (1998). Cardiac defects and altered ryanodine receptor function in mice Kang, J. O. and Sucov, H. M. (2005). Convergent proliferative response and lacking FKBP12. Nature 391, 489-492. divergent morphogenic pathways induced by epicardial and endocardial Srivastava, D. and Olson, E. N. (2000). A genetic blueprint for cardiac signaling in fetal heart development. Mech. Dev. 122, 57-65. development. Nature 407, 221-226. Kisanuki, Y. Y., Hammer, R. E., Miyazaki, J., Williams, S. C., Richardson, J. A. Sucov, H. M., Gu, Y., Thomas, S., Li, P. and Pashmforoush, M. (2009). Epicardial and Yanagisawa, M. (2001). Tie2-Cre transgenic mice: a new model for control of myocardial proliferation and morphogenesis. Pediatr. Cardiol. 30, endothelial cell-lineage analysis in vivo. Dev. Biol. 230, 230-242. 617-625. Klaassen, S., Probst, S., Oechslin, E., Gerull, B., Krings, G., Schuler, P., Tang, W., Ingalls, C. P., Durham, W. J., Snider, J., Reid, M. B., Wu, G., Greutmann, M., Hürlimann, D., Yegitbasi, M., Pons, L. et al. (2008). Matzuk, M. M. and Hamilton, S. L. (2004). Altered excitation-contraction Mutations in sarcomere protein genes in left ventricular noncompaction. coupling with skeletal muscle specific FKBP12 deficiency. FASEB J. 18, 1597- Circulation 117, 2893-2901. 1599. Koh, C., Lee, P. W., Yung, T. C., Lun, K. S. and Cheung, Y. F. (2009). Left Towbin, J. A. (2010). Left ventricular noncompaction: a new form of heart failure. ventricular noncompaction in children. Congenit. Heart Dis. 4, 288-294. Heart Fail. Clin. 6, 453-469. Lai, D., Liu, X., Forrai, A., Wolstein, O., Michalicek, J., Ahmed, I., Garratt, A. Wang, T. and Donahoe, P. K. (2004). The immunophilin FKBP12: a molecular N., Birchmeier, C., Zhou, M., Hartley, L. et al. (2010). Neuregulin 1 sustains guardian of the TGF-beta family type I receptors. Front. Biosci. 9, 619-631. the gene regulatory network in both trabecular and nontrabecular Wang, Q., Sigmund, C. D. and Lin, J. J. (2000). Identification of cis elements in myocardium. Circ. Res. 107, 715-727. the cardiac troponin T gene conferring specific expression in cardiac muscle of Lemmens, K., Segers, V. F., Demolder, M. and De Keulenaer, G. W. (2006). transgenic mice. Circ. Res. 86, 478-484. Role of neuregulin-1/ErbB2 signaling in endothelium-cardiomyocyte cross- Weaver, A. M., Hussaini, I. M., Mazar, A., Henkin, J. and Gonias, S. L. (1997). talk. J. Biol. Chem. 281, 19469-19477. Embryonic fibroblasts that are genetically deficient in low density lipoprotein Liu, J., Bressan, M., Hassel, D., Huisken, J., Staudt, D., Kikuchi, K., Poss, K. D., receptor-related protein demonstrate increased activity of the urokinase Mikawa, T. and Stainier, D. Y. (2010). A dual role for ErbB2 signaling in cardiac receptor system and accelerated migration on vitronectin. J. Biol. Chem. 272, trabeculation. Development 137, 3867-3875. 14372-14379. Lu, K. P., Finn, G., Lee, T. H. and Nicholson, L. K. (2007). Prolyl cis-trans Weiford, B. C., Subbarao, V. D. and Mulhern, K. M. (2004). Noncompaction of isomerization as a molecular timer. Nat. Chem. Biol. 3, 619-629. the ventricular myocardium. Circulation 109, 2965-2971. Maruyama, M., Li, B. Y., Chen, H., Xu, X., Song, L. S., Guatimosim, S., Zhu, W., Wessels, A. and Sedmera, D. (2003). Developmental anatomy of the heart: a

Yong, W., Zhang, W., Bu, G. et al. (2011). FKBP12 is a critical regulator of the tale of mice and man. Physiol. Genomics 15, 165-176. DEVELOPMENT Fkbp1a in cardiac development RESEARCH ARTICLE 1957

Wu, B., Zhang, Z., Lui, W., Chen, X., Wang, Y., Chamberlain, A. A., Moreno- Zhang, S. S., Wei, J., Qin, H., Zhang, L., Xie, B., Hui, P., Deisseroth, A., Rodriguez, R. A., Markwald, R. R., O’Rourke, B. P., Sharp, D. J. et al. (2012). Barnstable, C. J. and Fu, X. Y. (2004). STAT3-mediated signaling in the Endocardial cells form the coronary arteries by angiogenesis through determination of rod photoreceptor cell fate in mouse retina. Invest. myocardial-endocardial VEGF signaling. Cell 151, 1083-1096. Ophthalmol. Vis. Sci. 45, 2407-2412. Xing, Y., Ichida, F., Matsuoka, T., Isobe, T., Ikemoto, Y., Higaki, T., Tsuji, T., Zhou, B., von Gise, A., Ma, Q., Rivera-Feliciano, J. and Pu, W. T. (2008). Nkx2- Haneda, N., Kuwabara, A., Chen, R. et al. (2006). Genetic analysis in patients 5- and Isl1-expressing cardiac progenitors contribute to proepicardium. with left ventricular noncompaction and evidence for genetic heterogeneity. Mol. Genet. Metab. 88, 71-77. Biochem. Biophys. Res. Commun. 375, 450-453. Yang, J., Bücker, S., Jungblut, B., Böttger, T., Cinnamon, Y., Tchorz, J., Müller, Zhu, W., Soonpaa, M. H., Chen, H., Shen, W., Payne, R. M., Liechty, E. A., M., Bettler, B., Harvey, R., Sun, Q. Y. et al. (2012). Inhibition of Notch2 by Caldwell, R. L., Shou, W. and Field, L. J. (2009). Acute doxorubicin Numb/Numblike controls myocardial compaction in the heart. Cardiovasc. Res. cardiotoxicity is associated with p53-induced inhibition of the mammalian 96, 276-285. target of rapamycin pathway. Circulation 119, 99-106. DEVELOPMENT