Overexpression of Bone Morphogenetic Protein 10 in Myocardium Disrupts Cardiac Postnatal Hypertrophic Growth*

Postnatal cardiac hypertrophies have traditionally been classified into physiological or pathological hypertrophies. Both of them are induced by hemodynamic load. Cardiac postnatal hypertrophic growth is regarded as a part of the cardiac maturation process that is independent of the cardiac working load. However, the functional significance of this biological event has not been determined, mainly because of the difficulty in creating an experimental condition for testing the growth potential of functioning heart in the absence of hemodynamic load. Recently, we generated a novel transgenic mouse model (αMHC-BMP10) in which the cardiac-specific growth factor bone morphogenetic protein 10 (BMP10) is overexpressed in postnatal myocardium. These αMHC-BMP10 mice appear to have normal cardiogenesis throughout embryogenesis, but develop to smaller hearts within 6 weeks after birth. αMHC-BMP10 hearts are about half the normal size with 100% penetrance. Detailed morphometric analysis of cardiomyocytes clearly indicated that the compromised cardiac growth in αMHC-BMP10 mice was solely because of defect in cardiomyocyte postnatal hypertrophic growth. Physiological analysis further demonstrated that the responses of these hearts to both physiological (e.g. exercise-induced hypertrophy) and pathological hypertrophic stimuli remain normal. In addition, the αMHC-BMP10 mice develop subaortic narrowing and concentric myocardial thickening without obstruction by four weeks of age. Systematic analysis of potential intracellular pathways further suggested a novel genetic pathway regulating this previously undefined cardiac postnatal hypertrophic growth event. This is the first demonstration that cardiac postnatal hypertrophic growth can be specifically modified genetically and dissected out from physiological and pathological hypertrophies.

Postnatal cardiac hypertrophies have traditionally been classified into physiological or pathological hypertrophies. Both of them are induced by hemodynamic load. Cardiac postnatal hypertrophic growth is regarded as a part of the cardiac maturation process that is independent of the cardiac working load. However, the functional significance of this biological event has not been determined, mainly because of the difficulty in creating an experimental condition for testing the growth potential of functioning heart in the absence of hemodynamic load. Recently, we generated a novel transgenic mouse model (␣MHC-BMP10) in which the cardiacspecific growth factor bone morphogenetic protein 10 (BMP10) is overexpressed in postnatal myocardium. These ␣MHC-BMP10 mice appear to have normal cardiogenesis throughout embryogenesis, but develop to smaller hearts within 6 weeks after birth. ␣MHC-BMP10 hearts are about half the normal size with 100% penetrance. Detailed morphometric analysis of cardiomyocytes clearly indicated that the compromised cardiac growth in ␣MHC-BMP10 mice was solely because of defect in cardiomyocyte postnatal hypertrophic growth. Physiological analysis further demonstrated that the responses of these hearts to both physiological (e.g. exercise-induced hypertrophy) and pathological hypertrophic stimuli remain normal. In addition, the ␣MHC-BMP10 mice develop subaortic narrowing and concentric myocardial thickening without obstruction by four weeks of age. Systematic analysis of potential intracellular pathways further suggested a novel genetic pathway regulating this previously undefined cardiac postnatal hypertrophic growth event. This is the first demonstration that cardiac postnatal hypertrophic growth can be specifically modified genetically and dissected out from physiological and pathological hypertrophies.
Hypertrophic growth of cardiomyocytes plays an important role in determining the size of the heart (1,2). Cardiomyocyte hypertrophy has usually been regarded as an adaptive response to hemodynamic load in postnatal hearts when cardiomyocytes irreversibly withdraw from cell cycle activity (3,4). As a part of the normal developmental process, the cell cycle activity of cardiomyocyte declines rapidly upon terminal differentiation and maturation. At birth, the majority of cardiomyocytes (97%) stops proliferation and remains in G 0 /G 1 phase (5,6). This cardiomyocyte cell cycle withdraw is considered a key phenomenon in the heart switching from hyperplastic growth to hypertrophic growth (3).
Cardiac hypertrophy has traditionally been classified as pathological or physiological (7). For example, persistent pressure or volume overload caused by disease conditions, such as hypertension, stenotic cardiac valve disorders, and genetic defects in contractile proteins, can induce cardiac pathological hypertrophy, which can be associated with diminished heart function and eventual heart failure (8,9). In contrast, physiological cardiac hypertrophy is a normal beneficial adaptive response of the heart to physiological stimuli (e.g. chronic swimming exercise), which is often accompanied by a decrease in resting and submaximal heart rate, an increase in ventricular contractile function, and an increase in chamber filling time and venous blood return (7, 10 -13). Despite an increase in cardiomyocyte size, this type of hypertrophic response maintains normal cardiac structure and cardiac gene expression and, most importantly, does not lead to heart failure. It is becoming clear now that intracellular signaling pathways regulating physiological cardiac hypertrophy are distinctively different from those that regulate pathological hypertrophy. The phosphoinositide 3-kinase (PI3K) 3 pathway and the Akt pathway have been implicated recently in the regulation of physiological hypertrophy (11)(12)(13)(14). Various genetically engineered mice with loss of function mutations in PI3K and Akt attenuate exerciseinduced hypertrophy, but not the pathological hypertrophy (11)(12)(13)(14). Consistently, transgenic mice overexpressing constitutive active Akt (caAkt) have larger hearts (12).
Another physiological event is postnatal cardiac hypertrophic growth. This process is difficult to separate from physiological cardiac hypertrophy, mainly because of difficulty in creating an experimental condition for testing the growth potential of functioning heart in the absence of hemodynamic work load. The best demonstration was from a series of classic experiments by Bishop and co-workers (15,16) using an ectopic tissue culture system in which rat embryonic hearts were cultured in the anterior eye chamber (oculo) to examine the effect of hemodynamic unloading on the differentiation, maturation, and hypertrophic growth of myocardium or cardiomyocytes. They found that hypertrophic growth actually occurred in the absence of hemodynamic load during 3-5 weeks of culture (15), suggesting that cardiac hypertrophic growth could be regulated without directly associated with the cardiac working load. In addition, isoform switching from ␤-myosin heavy chain (␤MHC) to ␣MHC apparently occurred in oculo cultured heart as it does normally in postnatal heart (16). Although recent work suggested that the PI3K-Akt pathway may be partially responsible for this postnatal hypertrophic growth (14), the molecular and genetic cues have remained largely undefined as has the biological and physiological significance of this particular type of growth.
Bone morphogenetic protein 10 (BMP10) is a newly identified cardiac-specific peptide growth factor that belongs to the TGF-␤ superfamily (17,18). During cardiac development, BMP10 is expressed transiently in the ventricular trabecular myocardium from E9.0 to E13.5 (17,18), a critical time span when cardiac development shifts from patterning to myocardial growth and maturation. Ventricular development is characterized by a series of spatially and temporally coordinated events including cellular proliferation and maturation (3,4,19,20). Our initial characterization of the BMP10-deficient mice (17) and hANF-BMP10 transgenic mice (BMP10 overexpression in embryonic ventricle and atria throughout embryonic stage) (21) has demonstrated that BMP10 is an essential component in modulating cardiomyocyte proliferation and maturation during cardiac ventricular development. Interestingly, whereas BMP10 remains expressed in atria at E16.5-E18.5 and in only right atria in adult heart (17,21), BMP10 is rapidly down-regulated in ventricles after E14.5. Persistent elevation of BMP10 expression in the developing ventricular myocardium is found in several genetically manipulated mouse models, such as FK506-binding protein 12 (FKBP12)-deficient mice (17) and Nkx2.5-myocardial conditional knock-out mice (21). The common abnormal cardiac phenotype in these mutant mice is the development of severe congenital ventricular myocardial defects with hypertrabeculation and noncompaction, a myocardial morphogenetic defect that is likely caused by enhanced cell cycle activity in trabecular cardiomyocytes (17,21). These prior observations strongly suggest that the down-regulation of BMP10 in ventricular myocardium is a critical step in the normal development and maturation of the ventricular myocardium.
In an effort to examine the impact of ectopic and elevated BMP10 expression in postnatal myocardium, we generated transgenic mice in which BMP10 expression is driven by the cardiomyocyte-specific promoter, ␣MHC promoter. Although these unique ␣MHC-BMP10 transgenic mice have apparently normal cardiogenesis in utero, they give rise to smaller hearts during postnatal growth because of severely attenuated hypertrophic growth. Here, we report our assessment of the physiological impact of sustained BMP10 expression in postnatal cardiomyocytes on several aspects of cardiac hypertrophic growth and demonstrate for the first time that cardiac postnatal hypertrophic growth can be specifically modified genetically and has a significant role in postnatal cardiac function.

MATERIALS AND METHODS
Generation of ␣MHC-BMP10 Transgenic Mice-␣MHC promoter (a gift from Dr. Robbins, Cincinnati Children Hospital) was placed 5Ј of a mouse BMP10 cDNA fragment (coding region), followed by the SV40 early region transcription terminator/polyadenylation site. The procedures that generated the transgenic mice were carried out as previously described at the mouse transgenic core of Indiana University Cancer Center (21). In brief, the transgene insert (␣MHC-BMP10) was purified and microinjected into inbred C3HeB/FeJ (The Jackson Laboratories, Bar Harbor, ME) zygotes, which were then implanted into the oviducts of pseudopregnant Swiss Webster mice. The resulting pups were screened by diagnostic PCR. Five positive founders were identified from a total of 79 F 0 mice. Initially, these positive founders were bred to DBA/2J mice to minimize the silencing of the transgene caused by genomic methylation (22,23). We subsequently maintained ␣MHC-BMP10 mice in two genetic backgrounds, DBA/2J inbred strain and DBA/2J X C57B6 hybrid strain. Experiments were performed using the DBA/2J-C57B6 mixed strain in which age and sex matched non-transgenic littermates were used as controls.
Histological Analysis-Heart samples were harvested, fixed, sectioned, stained with hematoxylin and eosin (H&E) for cardiac histology, Masson's trichrome for potential cardiac fibrosis, and subjected to in situ hybridization and immunohistochemical analyses as previously described (17). Antibodies against pSmad 1 and pSmad 2 were purchased from Cell Signaling Technology.
Morphometric Measurement of Isolated Cardiomyocytes-Cardiomyocytes were enzymatically dissociated from the heart with 0.17% type I collagenase (Worthington Biochemical Inc.) and stained with Hoechst to demonstrate nuclei. Only typical rod-shaped cardiomyocytes were included in the morphometric measurement. Using ImagePro plus 5.1 software, the long axis, short axis, and cell area were measured and compared between transgenic and control mice. To count the total number of cardiomyocytes in adult hearts, formalin fixed hearts (the atria were removed) were digested with 50% KOH for 24 h (24). After careful washing with phosphate-buffered saline, rodshaped cells were counted using a hemacytometer.
Echocardiography-Transthoracic echocardiograms were performed as previously described (25) under 1.5-2.0% isoflurane. Two-dimensional short-axis images were obtained using a high resolution Vevo 660 and 770 Imaging System (Visualsonics Inc, Toronto) equipped with a 35 MHz scan probe. Left ventricular chamber dimension and wall thickness at systolic and diastolic phases were measured from M-mode recording. Heart rate was calculated from simultaneous electrocardiogram recording. Left ventricular volume, fractional shortening (FS%), and ejection fraction (EF%) were calculated using the Vevo Analysis program as described (25).
Swimming Training-Programmed conditioning using a swimming training protocol was performed as previously described by Kaplan et al. (26). In brief, 10-week-old male ␣MHC-BMP10 transgenic mice and their non-transgenic littermates were individually placed into a murine swimming pool (surface area 220 cm 2 , water depth 15-20 cm, water temperature 30 -32°C). On the first day, the mice swam for two 10-min cycles. On successive days, each swimming cycle was increased by 10-min increments until reaching to two 90-min cycles per day (this constituted the training period). Once trained, the mice swam for two 90-min cycles per day for a total of 28 days.
Insulin Treatment-3-month-old male ␣MHC-BMP10 mice and their non-transgenic siblings were fasted overnight and then treated with insulin (Eli Lilly Pharmaceutics) via IP injection (1.5 unit/kg). Cardiac tissues were harvested 10 min after injection for Western blot analysis.
Quantitative RT-PCR and Western Blot Analyses-Total RNA was extracted from the hearts of mice using TRIzol (Invitrogen). First strand cDNA was synthesized by the iScript cDNA synthesis kit (Bio-Rad) using 1 g of total RNA as a template according to the protocol provided by the manufacturer. Real time PCR was performed using iCycler iQ (Bio-Rad) with iQ SYBR Green supermix (Bio-Rad). The relative expression was normalized to the reference gene ribosomal protein L7 (RPL7) as previously described (28). The sequences of specific primers are listed in Table 1. Western blot analysis was performed as previously described (29). Antibodies against Akt, phospho-Akt (Ser 473 ), phospho-p38 MAP kinase (Thr 180 / Tyr 182 ), p38 MAP kinase, p44/42 MAP kinase, phospho-p44/42 MAP kinase (Thr 202 /Tyr 204 ), phospho-Smad1 (Ser 463/465 ), and phospho-Smad2 (Ser 465/467 ) were from Cell Signaling Technology, and antibodies against Smad 1 and Smad 2 were from Santa Cruz Biotechnology, Inc.

Generation and Analysis of ␣MHC-BMP10 Transgenic Mice-
To study the impact of ectopic BMP10 expression in the postnatal ventricular myocardium, the ␣MHC promoter was used to target BMP10 to the heart (Fig. 1A). The ␣MHC promoter has a transient burst of activity in embryonic heart around E9.5-10.5. This cardiac-specific promoter is re-activated during early postnatal life and remains persistently high into adult (30). Five transgenic lines carrying the ␣MHC-BMP10 transgene were generated with similar BMP10 expression levels and identical cardiac abnormalities as described below. These ␣MHC-BMP10 transgenic mice appear to have normal embryonic cardiac development and survive to adulthood (over 12 months of age). Some lethality at 8 -10 months was evident for ␣MHC-BMP10 mice in a DBA/2J inbred background. We compared the expression profiles of transgene-encoded BMP10 and total BMP10 in the transgenic hearts during embryonic and postnatal development using RT-PCR (Fig. 1B, panel a). There is a brief window of transgenic ␣MHC-BMP10 expression from embryonic day E9.0 to E12.5, which coincides with the temporal pattern of endogenous BMP10 expression in the ventricle (17,18). To further confirm the level of BMP10 expression during the development, quantitative qRT-PCR was used to determine total BMP10 level in ␣MHC-BMP10 hearts at E10.5, E16.5, and neonatal day 8 (Fig. 1B, panel b). BMP10 expression was about 40 -50% higher in E10.5 ␣MHC-BMP10 hearts when compared with non-transgenic littermate controls. As expected, transgenic BMP10 expression was rather high by N8 and remained high thereafter. In situ hybridization (Fig. 1C, panels a and b) and quantitative qRT-PCR (Fig. 1C, panel c) further confirmed persistent BMP10 expression in the ventricles of the adult ␣MHC-BMP10 mice.
Interestingly, the level of transgenic ␣MHC-BMP10 expression in the right and left atrium is relatively low compared with ventricles in these ␣MHC-BMP10 mice (Fig. 1C, panel c). This phenomenon is different from the majority of transgenic mice using the ␣MHC promoter. The ␣MHC promoter typically drives constitutively high expression in atria (30). The lack of high atrial BMP10 expression suggests that some ␣MHC-BMP10 transgenic mice with higher expression levels may die in utero. Supporting evidence is from another transgenic mouse line in which BMP10 expression is driven by human atrial natriuretic peptide (hANF) promoter (21). hANF-BMP10 transgenic mice that have higher and persistent BMP10 expression in embryonic hearts are embryonic lethal because of the devel- ACATCATCCGGAGCTTCAAGAACG AACCGCAGTTCAGCCATGACG opment of hypertrabeculation and noncompaction (21). Significantly lower percent of ␣MHC-BMP10-positive F0 mice (6.2%, 5 out of 79 F0) obtained in our study is consistent with this notion. Therefore, the generated ␣MHC-BMP10 transgenic lines were likely preselected for lower expressing mice. It is not surprising that cardiac histological analysis revealed normal cardiac structure in these ␣MHC-BMP10 embryos (Fig.  1D) and newborns (Fig. 4A, panel a). Smaller Cardiomyocytes in ␣MHC-BMP10 Transgenic Hearts-Although ␣MHC-BMP10 transgenic mice exhibit normal embryonic cardiac development, adult transgenic mice (both male and female) exhibited a dramatic reduction in heart size as compared with their sexmatched non-transgenic siblings (Fig. 2). Systematic comparison of the heart weight, body weight, and ratio of heart weight to body weight of ␣MHC-BMP10 mice and their non-transgenic and sex-matched littermates was performed. There was no significant difference in body weight before 4 months of age in the MHC-BMP10 transgenic mice when compared with littermate controls, the heart weight and size were consistently smaller, and the heart weight versus body weight ratio remained notably lower in all age groups after 1 month of age (Fig. 2).
To determine whether the smaller hearts were attributable to compromised hypertrophic growth of the cardiomyocytes, we analyzed the surface area, long axis, short axis, and the ratio of long axis versus short axis of dispersed ventricular cardiomyocytes isolated from early postnatal (1-week-old) and adult (3-monthold) ␣MHC-BMP10 mice and their non-transgenic littermates. Although cardiomyocytes isolated from 1week-old ␣MHC-BMP10 mice had normal cell size (249 m 2 ϩ 36, n ϭ 3,500 cells from 5 mice) when compared with littermate controls (242 m 2 ϩ 27, n ϭ 2,800 cells from 4 mice, p Ͼ 0.05), our analysis demonstrated a marked reduction in cardiomyocyte size in adult ␣MHC-BMP10 transgenic hearts (Fig. 2C,  panel a). Interestingly, the majority of ␣MHC-BMP10 cardiomyocytes still exhibited the usual bi-nuclear and rod-shaped cardiomyocyte morphology, suggesting that the terminal differentiation of cardiomyocytes is not perturbed in ␣MHC-BMP10 hearts. To determine whether this alteration in cardiomyocyte hypertrophy is because of altered hemodynamic load, we compared minimal diameters of cardiomyocytes between right and left ventricles using H&E stained histological section of transgenic hearts (3-month-old) and compared with non-transgenic controls. The change of hemodynamic load would affect the cell size of cardiomyocytes in left ventricles when compared with that in the right ventricles. Similar to normal non-transgenic heart, we found no difference in minimal diameters of cardiomyocytes when we compared left and right ventricular myocardia in ␣MHC-BMP10 hearts (data not shown), suggesting that the smaller cardiomyocytes in ␣MHC-BMP10 hearts are because of the alteration in the intrinsic hypertrophic genetic program in cardiomyocyte.
Previously, we have demonstrated that BMP10 is a positive regulator for cell cycle activity during ventricular development (17,21). Because it was possible that the smaller cardiomyocytes in ␣MHC-BMP10 hearts could be a consequence of abnormal cardiac hyperplastic growth as seen in the c-Myc transgenic mice (31), we assessed the total number of cardiomyocytes in ␣MHC-BMP10 hearts (3-month-old) and compared this to non-transgenic littermate controls. We found no difference between ␣MHC-BMP10 hearts and littermate controls (Fig. 2C, panel b). This finding was also consistent with data indicating that there was no abnormal cell cycle activity observed in ␣MHC-BMP10 hearts because [ 3 H]thymidine labeling and immunoreactivity of antibody against phosphohistone 3 in ␣MHC-BMP10 hearts was not different from agematched littermate hearts (data not shown). In addition, we used the TUNEL assay and immunochemical staining for activated caspase-3 to determine whether there was any apoptosis in ␣MHC-BMP10 hearts. Our data indicated there were no unusual apoptotic events in adult ␣MHC-BMP10 hearts when compared with littermate controls (data not shown). Taken together, these results indicate that ectopic overexpression of BMP10 in postnatal myocardium does not re-activate cell cycle activity nor trigger apoptosis in cardiomyocytes, but rather, disrupts cardiomyocyte hypertrophic growth.
Response to Hypertrophic Stimuli Is Preserved in ␣MHC-BMP10 Hearts-Initially, we hypothesized that this inhibition of cardiac hypertrophy was caused by the lack of response to physiological hypertrophic stimuli (e.g. exercise-induced hypertrophy) as that was seen in dnPI3K (11,32) and dnAkt (12) transgenic mice. To test this idea, a swimming exercise test was performed on 3-month-old ␣MHC-BMP10 mice and agematched non-transgenic littermate controls as previously described (26). Surprisingly, ␣MHC-BMP10 mice subjected to chronic swimming exercise had a marked increase in cardiac mass (Fig. 3A). Because the ␣MHC-BMP10 cardiomyocytes were smaller to begin with, the heart size in ␣MHC-BMP10 mice remained small. However, a similar proportion of increase in cardiac mass (32.4 ϩ 6.5%) was seen in ␣MHC-BMP10 mice when compared with non-transgenic littermate controls (40.1 ϩ 3.9%, n ϭ 10, p Ͼ 0.05) after 4 weeks of intensive swimming exercise. Furthermore, the size of cardiomyocytes isolated from these exercised mice was also proportionally increased (Fig. 3D), suggesting that ␣MHC-BMP10 cardiomyocytes maintained their ability to respond to physiological stimuli.
Akt activation (phosphorylated form) has been shown to be associated with cardiac hypertrophy induced by exercise (11,12,14,32). Using Western blot analysis, we assessed the level of activated Akt in ␣MHC-BMP10 transgenic and littermate nontransgenic hearts at baseline and after exercise. Consistent with the observation above, Akt activation in response to exercise was normal in ␣MHC-BMP10 mice (Fig. 3B). Furthermore, we also assessed PI3K-mediated Akt activation in ␣MHC-BMP10 hearts by administering insulin to ␣MHC-BMP10 transgenic mice. Similar activation of Akt was found in ␣MHC-BMP10 hearts and littermate controls. Our data strongly indicate that BMP10 overexpression does not affect Akt activation, which further suggests that the BMP10-mediated pathway is not upstream of Akt-mediated pathway.
To determine whether overexpression of BMP10 altered the ability to respond to ␤-adrenergic agonist induced cardiac hypertrophy, 3-month-old ␣MHC-BMP10 mice and their nontransgenic littermates were treated with isoproterenol as previously described (27). This treatment typically results in an ϳ30 -40% increase in heart weight/body weight ratio, which is reflected by uniform hypertrophic cardiomyocyte growth (27). Control mice were treated with saline-filled minipumps. Similar to the response from the chronic swimming exercise, a marked and proportional increase in cardiac mass and cardiomyocyte size was apparent in both the non-transgenic and ␣MHC-BMP10 isoproterenol-treated mice (Fig. 3, C and D), indicating that BMP10 overexpression does not alter ␤-adrenergic signaling in postnatal myocardium. Taken together, our results strongly imply that BMP-10 overexpression mainly disrupts cardiac postnatal hypertrophic growth.
Altered Cardiac Function in ␣MHC-BMP10 Mice-The ␣MHC-BMP10 transgenic mice appear to be an excellent model for assessing the physiological impact of disrupted postnatal hypertrophic growth on cardiac function. Although the majority of ␣MHC-BMP10 mice could survive to 12 months, histological analysis revealed abnormal cardiac structures in the ␣MHC-BMP10 mice as early as 4-week-old with 100% penetrance (Fig. 4A), in addition to smaller hearts. These abnor-  malities included a greatly reduced ventricular chamber volume, thickening of the ventricular wall, and narrowing in the subaortic region. The latter is reminiscent of subvalular aortic stenosis (SAS) (Fig. 4A). However, there was no evidence of accelerated blood flow velocity across the left ventricular outflow tract of 3-month-old ␣MHC-BMP10 mice compared with non-transgenic littermates when measured by Doppler echocardiography to suggest subaortic obstruction (data not shown). The ratio of right ventricular wall thickness to left ventricular wall thickness (RV/LV) was about 45% less in ␣MHC-BMP10 hearts (37.35% ϩ 8.9, n ϭ 15) when compared with littermate controls (66.36% ϩ 7.7, n ϭ 12, p Ͻ 0.01), indicating a disproportionate thickening of the left ventricular wall. We consistently found that the ventricular septum in ␣MHC-BMP10 hearts was equally thickened, indicating that the ventricular thickening was concentric and not asymmetric. The thickening of the ␣MHC-BMP10 ventricular wall could be resulted from abnormal myocardium modeling during the course of abnormal myocardial maturation (i.e. blocking hypertrophic growth).
Using echocardiography, we analyzed cardiac structure and function in the ␣MHC-BMP10 mice and compared it to littermate non-transgenic controls (Fig. 4B). This functional assessment revealed a dramatic reduction in both the diastolic and systolic ventricular chamber volumes associated with abnormal cardiac function in the ␣MHC-BMP10 mice, including significant increases in fraction shortening (FS) and ejection fraction (EF) (Fig. 4B). Some of these abnormal physiological features are similar to hearts in human patients with hypertrophic cardiomyopathy. However, there was no evidence of myofiber disorganization typically found in the hearts from these patients (33). This is also in sharp contrast to the largely normal cardiac function found in mutant mice that have defects in physiological hypertrophic response (11,12,32).
To determine whether this altered cardiac histology and function was associated with abnormal cardiac gene expression, we analyzed the expression level of several cardiac markers such as atrial natriuretic peptide (ANP), ␣MHC, ␤MHC, skeletal ␣-actin (acta1), sarcoplasmic reticulum Ca 2ϩ ATPase 2a (SERCA2a), and phospholamban (pln) (34)(35)(36). Using quantitative RT-PCR, we found that the expression of all these cardiac genes was normal in 1-week-old ␣MHC-BMP10 transgenic hearts when compared with littermate controls (data not shown). However, at 4 weeks of age when the ␣MHC-BMP10 mice are beginning to show abnormal cardiac histological structure, ␣MHC expression was comparable between ␣MHC-BMP10 transgenic hearts and littermate controls. ANF, ␤MHC, acta1, SERCA2a, and pln expression were all modestly up-regulated in ␣MHC-BMP10 hearts (Fig. 4C). This altered cardiac gene expression is clearly different from known altered cardiac gene expression profile in pathological hypertrophic hearts. ANF, ␤MHC, and acta1 are only present in embryonic hearts and are re-activated to persistently higher expression levels in hypertropic hearts (35). SERCA2a and pln are involved in regulating intracellular Ca 2ϩ homeostasis and contractilefunctionofcardiomyocytesandarenormallyfounddownregulatedinhypertrophicandfailinghearts (34,36).Thisuniquealteration in cardiac gene expression in ␣MHC-BMP10 hearts further reflects a novel mechanism that is likely associated with this particular cardiac physiology because of the overexpression of BMP10 and sub-sequent progression. Taken together, these findings indicate that cardiac postnatal hypertrophic growth is an important biological event that is essential to normal cardiac function.
Activation of Smad 1/5/8 in ␣MHC-BMP10 Heart-Our data suggested that overexpression of BMP10 disrupts normal cardiac postnatal hypertrophic growth, a biological process that is different from exercise-induced hypertrophy and pathological hypertrophy. The molecular basis for postnatal hypertrophic growth is currently unknown. ␣MHC-BMP10 mice provide a unique experimental model system to study the potential molecular mechanism. As members of TGF-␤ superfamily, BMPs signal through the heterotetrameric complex of type I and type II serine/threonine kinase receptors and via the activation of one or both of two distinct intracellular pathways, namely Smad-mediated pathway (37,38) and TGF-␤-activated kinase-1 (TAK-1)-MAPK-mediated pathway (38,39). In general, there are two sets of receptor-specific Smads (rSmad). Smad1/5/8 mediates BMP signaling, while TGF-␤s/activins activate Smad2/3. It has been shown that TAK-1 is able to mediate both BMPs and TGF-␤ signaling and functions as an upstream modulator of MKK6/p38MAPK and MKK7/JNK pathways (38,39). However, it is still not clear how ligandbound receptors activate TAK-1 and whether TAK-1 activation is Smad-dependent and/or -independent.
Dr. Hsueh and co-workers (40) recently demonstrated that both type I receptors Alk 3 and Alk 6, not other type 1 receptors, were able to transduce BMP10 signaling in MC3T3 cells. Using quantitative RT-PCR, we evaluated Alk 3 and Alk 6 expression in postnatal heart. Our data demonstrate that Alk 3 is the major BMP type I receptor in postnatal heart (Fig. 5A). Alk 3 expression is over 1,000 fold higher than Alk 6, suggesting that ALK3 is the best candidate for the BMP10 receptor in postnatal heart. It is consistent with the notion that Alk3-deficient mice develop severe defects in ventricular myocardium (41)(42)(43), whereas Alk6-deficient mice do not appear to develop cardiac defects (43,44).
To determine the downstream pathway, we used immunohistochemical staining and Western blot analyses to evaluate which of these two signaling pathways (i.e. Smad and/or TAK-1-MAPK) is altered in ␣MHC-BMP10 transgenic hearts. A series of specific antibodies against the activated (phosphorylated) form of Smads (pSmads), P38, and ERK1/2 were used in this analysis. The staining intensity of nuclear pSmad1/5/8 (activated Smad1/5/8) is significantly higher in ␣MHC-BMP10 transgenic hearts when compared with non-transgenic controls (Fig. 5B). In contrast, nuclear pSmad2/3 appears no different between ␣MHC-BMP10 transgenic and non-transgenic hearts. Western blot analysis further confirmed these findings (Fig. 5C). As a control, we also analyzed Smad activation in previously generated ␣MHC-TGF-␤1 mice (45). Smad2/3, not Smad1/5/8, is activated in ␣MHC-TGF-␤1 hearts. These observations are consistent with our general understanding of that BMP and TGF-␤ are engaged in different physiological functions via different intracellular cell signaling pathways. Indeed, ␣MHC-TGF-␤1 mice had completely different cardiac functional defects (45). In contrast to the activation of Smad1/ 5/8, the TAK1-MAPK-mediated pathway appears to be unaf-fected by overexpression of BMP10, as the levels of pP38 and pERK1/2 were normal in ␣MHC-BMP10 hearts when compared with non-transgenic control hearts (Fig. 5D), suggesting that BMP10 specifically activates Smad1/5/8-mediated signaling in ␣MHC-BMP10 transgenic hearts.
In summary, our data demonstrate that down-regulation of BMP10 expression in late development (after E14.5) is a critical step for cardiomyocytes to undergo normal developmental hypertrophic growth in early postnatal life. Smad1/5/8-mediated signaling, but not TAK-MAPK, is involved in regulating cardiac postnatal hypertrophic growth.

DISCUSSION
In general, adult cardiac hypertrophy is an adaptive response to hemodynamic work load. Depending on the nature of the stimulus and the subsequent physiological con-sequences, these adaptive responses are categorized as physiological hypertrophy or pathological hypertrophy. Although the interplay between these two load-induced hypertrophic responses remains unclear, emerging evidence supports that physiological hypertrophy and pathological hypertrophy are regulated by different intracellular signaling pathways (11,12,32). Our study suggests that there is a novel hypertrophic pathway, namely postnatal hypertrophic growth, which is independent, or less dependent, on the imposed load and is an important part of the developmental and maturation process of ventricular myocardium in early postnatal life.
Previously, by culturing rat embryonic heart in an environment without hemodynamic load (adult rat anterior eye chamber), Bishop et al. (15) described this cardiac hypertrophy phenomenon. They demonstrated that cardiomyocytes are capable Alk 3 mRNA level is over 1,000-fold higher than Alk 6 in the heart. B, immunohistochemical staining analysis of ␣MHC-BMP10 hearts (6-week-old) using anti-pSmad1 antibody (note, this antibody also cross-reacts with pSmad5 and pSmad8) and anti-pSmad2 antibody (note this antibody also cross-reacts with pSmad3). Arrows point to positive staining signals. pSmad1/5/8 is dramatically activated in ␣MHC-BMP10 hearts. C and D, Western blot analysis of Smads and TAK1-MAPK activation in ␣MHC-BMP10 hearts (1-month-old). BMP10 specifically activates Smad1/5/8. Each lane represents a different cardiac sample. Genotypes of these cardiac samples are as indicated in the figures.
of undergoing typical hypertrophic growth in the absence of hemodynamic load (15). As they suggested, this cardiac hypertrophic growth could be the late stage scenario of normal myocyte maturation bridging from neonatal to postnatal life. However, the lack of a convincing genetic approach to further delineate this phenomenon weakened this original observation as it was not clear whether this hypertrophic growth was a unique phenomenon in the oculo culture system. The ␣MHC-BMP10 transgenic mouse is the first genetically manipulated mouse model in which cardiac postnatal hypertrophic growth is severely compromised without affecting either the physiological or pathological hypertrophy.
The molecular mechanism that underlies developmental hypertrophic growth has not been determined, but it is reasonable to postulate that this process is part of one a more of the pathways involved in ventricular myocardial development. During embryonic development, growth of the heart primarily relies on cardiomyocyte proliferation. Soon after birth, the growing heart shifts from hyperplastic growth to hypertrophic growth. It is still not entirely clear what regulates this developmental switch. Down-regulation of positive cell cycle regulators (e.g. cyclins and Cdks) and up-regulation of negative cell cycle regulators (e.g. Cdk inhibitors) are believed to be the intrinsic mechanism for cardiac cell cycle withdraw (4). Previously, we had shown that BMP10 is a potential modifier of this process, possibly via its genetic interaction with the cell cycle machinery (17). Elevated BMP10 expression in embryonic heart maintained embryonic cardiomyocytes at a higher level of cell cycle activity and lead to abnormal ventricular wall development such as hypertrabeculation and noncompaction (17,21). Therefore, the down-regulation of BMP10 in developing ventricles at later embryonic stages is a necessary step for normal myocardial development.
In this current work, we generated a transgenic mouse model in which BMP10 expression is re-activated in the early postnatal stage. Apparently, the re-activation of BMP10 expression in the postnatal heart is not able to re-activate the cardiomyocyte cell cycle activity in post-mitotic cardiomyocytes, which is consistent with the idea that BMP10 is not the pivotal cell cycle "initiator" in myocardium, but rather a "modifier". The greatest impact that BMP10 re-activation had on the postnatal heart was its effect on cardiac hypertrophic growth. Although we cannot entirely dismiss the possibility that earlier transient expression of transgenic BMP10 around E10.5 may predispose the developing myocardium to maturation defect in cardiomyocytes at later developmental and neonatal stages, the observation of normal cardiac histology in embryonic/neonatal ␣MHC-BMP10 hearts and significantly smaller hearts and abnormal cardiac histology in ␣MHC-BMP10 hearts after 6 weeks of age suggests that the compromised hypertrophic growth occurred at early postnatal period. Further study will be using an inducible transgenic system to determine what the critical time window is for this novel postnatal hypertrophic growth and for cardiomyocytes to remain responsive to BMP10.
Importantly, physiological assessments indicated that ␣MHC-BMP10 hearts maintained their normal ability to respond to exercise and ␤-adrenergic stimulation. This suggested that the compromised cardiac hypertrophy in ␣MHC-BMP10 transgenic mice was related to the inhibition of postnatal hypertrophic growth. Therefore, BMP10-mediated signaling pathways are not only important in cardiogenesis during embryonic development (17,21), but are also a critical component in regulating postnatal myocardial growth and maturation. Two major intracellular signaling pathways are involved in conveying BMP signaling in the cell, the Smad-mediated pathway (37) and the TAK1-MAPK-mediated pathway (39). Our results clearly indicate that the Smad pathway is linked to cardiac postnatal hypertrophic growth.
Earlier studies by Izumo and co-workers (11) using transgenic mice overexpressing constitutively active and dominant negative forms of the p110␣ catalytic subunit of class I A PI3K (caPI3K and dnPI3K) and Akt (caAkt and dnAkt) have indicated that PI3K and Akt are critical to cardiac physiological hypertrophy. Recently, they further demonstrated that mice deficient in the p85␣ and p85␤ regulatory subunits of PI3K in myocardium have defects in both postnatal cardiac hypertrophic growth and physiological hypertrophy (14). Although there are some similar features between the abnormal cardiac phenotypes of p85␣/p85␤ mutant mice and ␣MHC-BMP10 transgenic mice, there are several key differences. These differences include that ␣MHC-BMP10 transgenic mouse hearts have a more severe growth defect and altered cardiac function, whereas p85␣/p85␤ mutant mice have less severe growth defects and normal cardiac function as assessed by echocardiogram and histology (14). In addition, Akt mediated signaling is altered in p85␣/p85␤ mutant mice (14), whereas ␣MHC-BMP10 transgenic mice are apparently normal in Akt signaling. Taken together, these findings strongly suggest that PI3K-Akt is not the downstream pathway of the BMP-Smad pathway. BMP-Smad signaling is likely to be either a parallel pathway to PI3K-Akt pathway, or is downstream of the PI3K-Akt pathway in cardiac postnatal hypertrophic growth. It would be interesting to determine whether BMP-Smad is activated in p85␣/p85␤ mutant mice and if there is cross-talk between the PI3K-Akt pathway and the BMP-Smad pathway in cardiac hypertrophic growth. Further studies will also be conducted to dissect events downstream of activated Smad1/5/8 in postnatal hearts.