Jumonji regulates cardiomyocyte proliferation via interaction with retinoblastoma protein.

Jumonji (JMJ) can function as a transcriptional repressor and plays critical roles in embryonic development including heart development in mice. Although JMJ has been suggested to play a role in cell growth, the molecular mechanisms have not been resolved. The present data demonstrate that JMJ interacts with the retinoblastoma protein (Rb), one of the master regulatory genes of cell cycle. JMJ potentiates the repression function of Rb on E2F activities, leading to reduced cell cycle progression. The transcriptional repression domain of JMJ is critical for the interaction with Rb as well as repression of cell cycle. The physiological relevance of the association between Rb and JMJ was assessed in cardiomyocytes. Primary cardiomyocytes cultured from homozygous jmj knock-out mouse embryos (jmj mutants) show increased cell mitosis in a cardiomyocyte-specific manner. Reporter gene analyses demonstrate that promoter activities of cyclin D1, cyclin D2, and Cdc2 are up-regulated in jmj mutant cardiomyocytes. These data suggest that JMJ down-regulates the cell growth via interaction with Rb, which would provide important insights into the cardiac defects observed in jmj mutant mice.

Jumonji (jmj) 1 plays important roles in mouse embryonic development including cardiovascular development, which was first identified by gene trap technologies (1,2). Homozygous jmj knock-out mice (hereafter referred to as jmj mutants) displayed severe cardiac morphological defects and altered heart-specific gene expression (for review, see Ref. 3). The jumonji protein (JMJ) is a member of the JMJ transcription factor family and belongs to an AT-rich interaction domain transcriptional factor family (4 -6). In vitro analyses on structural and functional relationships revealed that JMJ contains distinct functional domains for transcriptional repression, DNA binding, and nuclear localization (7). More recently, it has been reported that JMJ represses the expression of cardiac-specific genes such as the atrial natriuretic factor gene through inhibiting Nkx2. 5 and GATA4 activities that are important cardiac-restricted transcription factors (8). Therefore, it is possible that JMJ employs multiple regulatory mechanisms to regulate the target gene expression. Although the molecular basis of JMJ for transcriptional regulation in the heart is now emerging, the evidence underlying the causes of the cardiac defects observed in jmj mutants is still lacking.
JMJ contains the AT-rich interaction domain and jmj-like domain (jmjN/jmjC) that constitutes a jmj family of transcription factors (9,10). Interestingly, these domains are also conserved in retinoblastoma protein (Rb)-binding protein 1 (RBP1) and RBP2, which were originally identified as Rb pocket domain-binding proteins (11,12). Rb is known as a key regulator on cell cycle progression and differentiation partly via suppressing E2F activity (13)(14)(15), suggesting the role of JMJ in the regulation of cell proliferation. The inhibition of E2F activity leads to failure of activation of numerous genes required for S-phase entry or DNA synthesis including c-myc, N-myc, dihydrofolate reductase, and thymidine kinase (16 -18). In growtharrested cells, E2F activity is repressed by association with hypophosphorylated Rb. In response to mitogenic stimuli, Rb is phosphorylated at multiple sites by cyclins/cyclin-dependent kinases (Cdks), which releases E2F from Rb, allowing cell cycle progression into late G 1 -and S-phase. Rb and its relatives, such as p107 and p130, are known to interact with other binding partners through the conserved pocket domain such as E1A (19), histone deacetylase 1 (HDAC1), and HDAC2 (20,21).
Despite variable phenotypes observed in jmj mutants, partly due to different genetic backgrounds (3), the role of JMJ in cell proliferation has been also suggested by phenotype analyses. jmj mutant mice with a C57BL/6J ϫ 129SVJ (mixed) genetic background died upon birth and showed cardiac defects such as ventricular septal defects, double-outlet right ventricle, and thin ventricular wall at later embryonic stages (2). These cardiovascular defects mimic human congenital cardiovascular diseases (22,23). In addition to the thin ventricular wall (2), jmj mutant embryos with a pure BALB/c background showed deficient cell growth in the liver, thymus, and spleen (24), suggesting the role of JMJ in facilitating cell proliferation. In contrast, jmj mutant mice with a C3H/He genetic background died at embryonic day (E)11.5, which exhibited hyperplasia and increased cyclin D1 expression in the trabecular layer of the ventricle at E10.5 (25,26). The molecular mechanism by which JMJ represses the cyclin D1 promoter activity in the trabeculae remains largely unknown.
Formation of the mature four-chambered heart depends on the precise control of cell proliferation, cell migration, differentiation, programmed cell death, and coronary vasculature formation. As heart formation proceeds, cardiomyocyte proliferation is temporally and spatially regulated. In normal mouse embryos at around E9.5, myocardium begins to proliferate and is subdivided into an inner and outer side of the ventricular chamber called the trabecular and compact layer, respectively. Cardiomyocytes at the compact zone proliferate and ingress to form a ventricular septum. From E11.5 to 14.5, cardiomyocytes rapidly proliferate, leading to the expansion of the ventricular wall. Therefore, the compact zone is composed of replicating cardiomyocytes, whereas the forming trabeculae consist of less proliferating cardiomyocytes. During later stages, the proliferation rate gradually declines followed by irreversible withdrawal from the cell cycle shortly after birth. In addition to cell cycle regulators, numerous transcription factors have been suggested to play a role in cardiomyocyte proliferation, given that various knock-out mice showed a thin ventricular wall including retinoic acid receptor ␣, retinoid X receptor ␣, vascular cell adhesion molecule-1, Pax-3, and Fog-2 (27)(28)(29)(30)(31). Based on these observations, we hypothesized that JMJ binds to Rb to modulate the Rb function, which may lead to altered cell proliferation as a consequence. Here we demonstrate that JMJ enhances the suppressive function of Rb on E2F activities, which is mediated by interaction between the transcriptional repression (TR) domain of JMJ and the pocket domain of Rb. The lack of JMJ results in up-regulated cell proliferation of cardiomyocytes derived from jmj mutants, probably due to up-regulated promoter activities of cyclin D1, cyclin D2, and cdc2, which may lead to hyperphosphorylation of Rb. These results suggest that JMJ mediates suppression of the cardiomyocyte proliferation via interaction with Rb, which contribute to a trabecular hyperplasia phenotype of jmj mutant hearts.
Primary Culture of Embryonic Cardiomyocytes-The primary culture of mouse embryonic hearts was prepared as described previously with slight modifications (8,36). The dissected hearts were digested in Hanks' Balanced Salt Solution containing 0.1% collagenase, 0.002% trypsin, and 0.01% chick serum for 10 min at 37°C and inactivated by the addition of horse serum. After centrifugation, the cell pellet was resuspended and plated in DMEM supplemented with 10% fetal bovine serum (FBS). The next day, the medium was replaced by DMEM supplemented with 10% horse serum and 5% FBS and assayed after 24 h in culture. Approximately 40 -50% cultured cells were estimated to be a cardiomyocyte population by cardiac troponin I immunostaining. Heterozygous jmj knock-out mice in a C57BL/6J ϫ 129SVJ (mixed) genetic background were mated to obtain homozygous jmj knock-out embryos, which were genotyped by tail-genomic DNA as described previously (2). For LacZ staining, cells were fixed by 0.5% glutaraldehyde and then incubated with X-gal containing solution (1 mg/ml X-gal, 4 mM K 4 Fe(CN) 6 -3H 2 O, 4 mM K 3 Fe(CN) 6 , 2 mM MgCl 2 , 0.1% Nonidet P-40 in PBS) followed by conventional immunostaining procedure. For immunoblot analyses, cardiomyocyte extracts (50 g) were separated by SDS-PAGE and analyzed using antibodies to Rb, cyclins A, D1, D2, and E, and p27 (Upstate Biotechnology). Mouse embryonic fibroblasts (10T1/2) and human kidney cells (HEK293) were grown in DMEM supplemented with 10% FBS and 100 units/ml penicillin/streptomycin.
Cell Mitosis Assay-To assay BrdUrd incorporation by immunofluorescence, cardiomyocytes or 10T1/2 cells were labeled with 10 M BrdUrd (Sigma) for 1-24 h as indicated. Cells were then fixed with 4% formaldehyde for 15 min at room temperature followed by permeabilization with 0.2% Triton X-100 in PBS (PBST) for 5 min. Cells were denatured with 2 N HCl for 1 h, neutralized with 0.1 M Na 2 B 4 O 7 , washed with PBST, and blocked by 10% goat serum in PBST. Cells were then incubated with rat anti-BrdUrd antibody (Serotec, 1:200) and cardiac troponin I antibody (Advanced ImmunoChemical Inc., 1:300) in blocking solution for 1 h. Cells were washed with PBST followed by incubation for 1 h with goat anti-rat IgG-Texas Red (Jackson Immunoresearch Laboratories) and goat anti-mouse IgG-Alexa Fluor 488 (Molecular Probes) in 5% goat serum/PBST. After washing, cells were incubated with Hoechst dye 33342 (Molecular Probes, 1 g/ml) and mounted. For each sample, an average of 600 cells was counted, and mitotic indices were calculated by dividing the number of BrdUrd-positive nuclei by cardiac troponin I-and Hoechst dye 33342-positive cells. Slides were examined with a Zeiss AxioCam HR microscope (Carl Zeiss) equipped for fluorescence microscopy.
For phosphohistone staining, embryonic hearts were fixed by 4% paraformaldehyde in PBS for 3 h and incubated in 30% sucrose/PBS. Hearts were frozen in tissue-freezing medium (Triangle Biomedical Sciences), and cryosectioned at 6 m. The sections were permeabilized with PBST and incubated with anti-phosphohistone H3 antibody (Upstate Biotechnology, 1:200) and MF20 antibody that recognizes sarcomeric myosin (Developmental Studies Hybridoma Bank, Iowa City, IA) overnight at 4°C. After washing with PBS, sections were incubated with anti-rabbit IgG-Texas Red (Amersham Biosciences, 1:100) and anti-mouse IgG-Alexa 488 (1:200) and counterstained by Hoechst dye 33342. To examine the apoptotic rate of the embryonic hearts, in situ apoptosis assays were performed on paraffin-embedded transverse sections of embryos using ApopTag TM Plus Kit (Oncor) according to manufacturer's recommendation.
Reporter Gene Assay-To examine the promoter activities in the absence or presence of JMJ, cardiomyocytes seeded on 24-well plates were transfected with 0.2 g of cyclin D1, cyclin D2, Cdc2, p21, or SV40 promoter (pGL2, Promega) luciferase reporter plasmids and 0.2 g of pCMV-␤-galactosidase using Lipofectamine 2000 (Invitrogen). 10T1/2 cells grown on 24-well plates at a density of 3 ϫ 10 4 cells/well were transfected with 0.2 g of each reporter gene with 0.05 g of pCMV-␤galactosidase and compared with cells cotransfected with 0.05 g of JMJ. To examine the effect of JMJ on the repression activity of Rb on E2F, 10T1/2 cells grown on 24-well plates at a density of 5 ϫ 10 4 cells/well were cotransfected with various combinations of 5 ng of E2F, 60 ng of Rb, and 30 ng of JMJ mutants by Lipofectamine Plus reagent. All of the transfectant was cotransfected with 50 ng of the E2F-dependent luciferase reporter plasmid and 50 ng of the pCMV-␤-galactosidase plasmid. After a 24-h incubation, luciferase activities were measured and normalized to ␤-galactosidase activities. The expression levels of cotransfected proteins were determined by Western blot using antibodies to E2F (Santa Cruz Biotechnology, Inc), Rb, and JMJ.
GST-pull-down Assay- 35 S-labeled JMJ wild type and its mutants were prepared in vitro by the T7 promoter according to manufacturer's instruction (TNT coupled reticulocyte lysate system, Promega). Statistics-Student's two-tailed t test was used for data comparison with a significant level of p Ͻ 0.05.

RESULTS
In Vivo Interaction of Rb with JMJ-An amino acid sequence alignment reveals that the C-terminal JMJ contains three domains that are conserved in various transcription factors: the jmjN; jmjC; and AT-rich interaction domains (see diagram in Fig. 3D) (6,10). Interestingly, this region of JMJ is homologous to RBP1 and RBP2 that interact with Rb, a master regulator of cell cycle (11,12). Because JMJ has been suggested to be involved in cell cycle regulation (2,25,26), we examined whether JMJ physically associated with Rb in vivo by coimmunoprecipitation assays (Fig. 1). After FLAG-tagged JMJ was cotransfected with Rb or Myc-tagged Rb into HEK293 cells, their expressions were confirmed by direct Western blot analyses using various antibodies as indicated (Fig. 1A, lanes 1 and  2). Cell extracts were immunoprecipitated with anti-Myc antibody followed by Western blot analyses using various antibodies (lanes 3 and 4). The JMJ band was detected when cells transfected with Myc-Rb and FLAG-JMJ were coimmunoprecipitated with anti-Myc antibody followed by Western blot with anti-FLAG or JMJ antibodies (lane 3), whereas no band was detected when cells were transfected with nontagged Rb and FLAG-JMJ (lane 4). To further confirm the protein interaction in vivo, reciprocal experiments were performed (Fig. 1B). The Rb or JMJ band was detected when cell lysates transfected with JMJ or FLAG-JMJ with Myc-Rb were subjected to direct Western blot analyses using various antibodies as indicated (lanes 1 and 2). The Rb band was detected when cell lysates expressing FLAG-JMJ and Myc-Rb were coimmunoprecipitated with anti-FLAG antibodies followed by Western blot us-ing anti-Rb or Myc antibodies as indicated (lane 4). However, the Rb band was not detected when cell lysates expressing JMJ and Myc-Rb were coimmunoprecipitated with anti-FLAG antibodies followed by Western blot using anti-Rb or Myc antibodies (lane 3), indicating specific interaction of JMJ with Rb. These data clearly demonstrate that JMJ physically interacts with Rb in vivo when they are coexpressed.
The Effect of JMJ on E2F Activities-Because Rb is known to repress the cell cycle by sequestering E2F, it is possible that the association of Rb and JMJ may alter the suppressive activity of Rb. To examine whether the Rb activity on E2F is affected in the presence of JMJ, we performed cotransfection assays using the reporter gene containing the E2F-dependent promoter linked to a luciferase gene (Fig. 2). E2F activated the reporter gene and cotransfection of Rb with E2F repressed E2F-dependent activation by 50.1% as expected. When the wild type JMJ (J-Wt) was cotransfected with Rb and E2F, JMJ repressed E2F-dependent reporter activity by 71.7%, suggesting that JMJ potentiates the repression activity of Rb on E2F-dependent activation. To determine the JMJ domain that mediates this repression function, several JMJ mutants were cotransfected with E2F and Rb. The N-terminal JMJ (J-Nt, aa 1-528) containing the TR domain (aa 131-220) showed similar repression to J-Wt (78.2% reduction), whereas C-terminal JMJ (aa 529 -1234) completely lost its repression activity. The mutant lacking the TR domain (⌬TR) also lost its repression activity, whereas the DNA-binding domain (aa 529 -792) deletion mutant (⌬DBD) retained its repression activity (73.4% reduction). The cotransfected proteins were expressed at similar levels  2). The amount of input was 5% cell lysates. Cell lysates were immunoprecipitated (IP) with anti-Myc antibodies followed by immunoblotting (IB) using various antibodies as indicated (lanes 3 and 4). B, the reciprocal experiments were performed using cell lysates expressing Myc-Rb and FLAG-JMJ (lane 2) or nontagged JMJ (lane 1). Cell lysates were coimmunoprecipitated by FLAG antibodies followed by Western blot analysis using various antibodies (lanes 3 and 4). Reciprocal experiments were performed by GST-pull-down assays using in vitro translated [ 35 S]JMJ and GST-Rb fusion protein coupled to beads (Fig. 3C). We used GST-Rb-(379 -928) that contains the pocket domains, because the A/B pocket domain appears to be required for binding to JMJ (Fig. 3, A and  B). The expression of J-Wt-(1-1234) and J-Nt-(1-528) containing the TR domain is shown in the input lanes. Both J-Wt and J-Nt interacted with GST-Rb but not with GST alone, indicating the specific interaction between JMJ and Rb. To further map the interaction region, several more deletion mutants of JMJ were subjected to GST-pull-down assays (data not shown) and the results are summarized in Fig. 3D. We determined that the TR domain of JMJ (aa 131-222) interacted with Rb in vitro. These data indicate that the C-terminal A/B pocket domain of Rb interacts with the TR domain of JMJ in vitro. JMJ also interacted with other members of the pocket domain family such as p107 and p130 in vitro (data not shown), suggesting that JMJ may have a broad regulatory effect on the pocket domain proteins.
Physical interactions between JMJ and Rb were further confirmed by GST-pull-down assays using cell lysate from cultured cardiomyocytes (Fig. 3, E and F). 293 cell lysates transfected with JMJ showed the JMJ band (Fig. 3E, lane 1) that was absent in control cell lysates (lane 2) by Western blot using anti-JMJ antibodies, indicating the specific JMJ band. When cardiomyocyte (CM) extracts from E13.5 embryos were subjected to Western blot using anti-JMJ antibodies, the endogenous JMJ protein was not detected (lane 3) under the condition we employed. When CM extracts were incubated with GST-Rb-(379 -792) beads, the JMJ band was detected (lane 4) but not with GST beads (lane 5), indicating specific interaction between JMJ expressed in cardiomyocytes and the pocket domain of Rb. The reciprocal experiments were performed using GST-JMJ-(1-222) (Fig. 3F). The endogenous Rb proteins were detected as two bands that represent the hyperphosphorylated Rb (pRb) and the underphosphorylated Rb in CM extracts from E13.5 and 17.5 by Western blot analyses using anti-Rb antibody (lanes 1 and 2, respectively). When CM extracts at E13.5 (lane 3) and at E17.5 (lane 5) were incubated with GST-JMJ, the Rb band was detected in both embryonic stages but not with GST beads (lanes 4 and 6). It is interesting that only the underphosphorylated form of Rb bound to JMJ. It should be noted that the level of underphosphorylated Rb seems to be increased in cardiomyocytes from E17.5 compared with E13.5, which probably reflects the cell cycle withdrawal as the heart matures. Together with the results in Figs. 1 and 2, these data demonstrate that JMJ and Rb interact with each other, which probably mediates the regulation of the Rb function on cell cycle progression.
Regulation of Cell Proliferation by JMJ-We sought to investigate whether the repression activity of JMJ on E2F activities via interaction with Rb resulted in the regulation of cardiomyocyte cell proliferation. The cardiomyocytes in the developing heart are subjected to complex growth signals including those from the epicardium and endocardium (38). Therefore, we employed the primary culture system, which minimizes the multiple growth signals originating from adjacent cell lineages of the developing heart or from the whole body. Primary cardiomyocytes cultured from jmj mutants with a mixed genetic background, which showed severe cardiac defects and died right after birth, were used throughout this study (2).
To examine whether cell mitosis is affected in jmj mutants, we performed BrdUrd incorporation assays using primary cardiomyocytes cultured from embryonic hearts. The primary cardiomyocytes grown on coverslips were incubated with BrdUrd followed by immunostaining using anti-BrdUrd antibodies (Fig. 4, A and B). To identify cardiomyocytes, we performed coimmunostaining experiments using antibodies against cardiac troponin I. The number of BrdUrd-positive cells per ϳ600 cardiomyocytes was counted. Interestingly, jmj mutant cardiomyocytes exhibited a 114% increase in BrdUrd uptake compared with wild type, indicating that cell growth is up-regulated in mutants (Fig. 4A). To examine whether cell proliferation of cardiac fibroblasts was also affected in jmj mutants, we assessed the cell proliferation index of primary cardiac fibroblasts. The noncardiomyocyte population that was not stained by cardiac troponin I antibodies was regarded as cardiac fibroblasts, although other cell populations such as endothelial cells and smooth muscle cells may exist as a minor population. As shown in Fig. 4B, cell proliferation indices were not altered in cardiac fibroblasts cultured from mutant hearts as compared with wild type. Therefore, cell proliferation of cardiomyocytes, but not cardiac fibroblasts, was up-regulated in jmj mutant hearts when cells were cultured.
Although it has been shown that jmj is widely expressed in the developing mouse heart, the types of cells that express jmj have not been examined in the primary culture. To determine the expression pattern of jmj in cultured cardiomyocytes, we performed LacZ staining on cells isolated from jmj mutant hearts. Jmj mutant cells contain a lacZ cDNA at the jmj locus, which recapitulates endogenous expression patterns of jmj (2). As shown in Fig. 4C, LacZ staining was readily observed in a cardiomyocyte population that was stained by cardiac troponin I antibodies but not observed in cardiac fibroblasts. These results indicate that jmj is expressed at a higher level in cardiomyocytes than cardiac fibroblasts. Therefore, the increased mitotic index of jmj mutant cardiomyocytes, but not fibroblasts, may be attributed to the fact that jmj does not seem to be expressed in fibroblasts when cultured. Taken together, the increase in cell proliferation in jmj mutant cardiomyocytes suggests an inhibitory role of JMJ on cell proliferation.
The Effect of JMJ on Cell Cycle-related Components-Cell cycle progression is controlled by various cyclins/Cdks. From G 1 -to S-phase, D-type cyclins and their cognate kinases are activated followed by successive activation of other cyclins, including cyclins E and A and Cdc2, which are constrained by Cdk inhibitors such as p16 and p21. These coordinated regulations determine the phosphorylation of Rb that is required for cell cycle entry. Because jmj mutants exhibit the defective regulation of cell growth, it is plausible that cell cycle machinery is affected by the absence of JMJ. To investigate whether the expression levels of critical components of cell cycle regulators are affected in the absence of JMJ, we performed luciferase reporter gene assays using several promoters of cell cycle regulators in primary cardiomyocytes (Fig. 5A). When each reporter gene was transfected into primary cardiomyocytes derived from embryonic hearts at E12.5-16.5, the promoter activities of cyclin D1, cyclin D2, and Cdc2 in mutant cardiomyocytes were increased by 1.9-, 1.7-, and 1.4-fold, respectively, compared with those of wild type. However, no significant differences in promoter activities were observed with the p21 or   FIG. 4. A and B, cell proliferation is up-regulated in jmj mutant cardiomyocytes. Mouse primary cardiomyocytes from wild type (ϩ/ϩ) and mutant (Ϫ/Ϫ) hearts were cultured at E13.5-15.5. After incubation with 10 M BrdUrd for 24 (A) or 1 h (B), cells were immunostained using anti-BrdUrd antibodies followed by goat anti-rat IgG coupled to Texas Red. To identify cardiomyocytes, cells were coimmunostained with cardiac troponin I antibodies as described in Fig. 4C. Percentages of cells incorporating BrdUrd were calculated in cardiomyocytes (A) and fibroblasts (B). Bars represent the means with S.E. from 3-5 experiments. *, p Ͻ 0.05. C, cardiomyocyte-restricted expression of JMJ. Mouse primary cardiomyocytes from jmj mutant were double-stained by cardiac troponin I antibodies followed by goat anti-mouse-Alexa Fluor 488 (green, left) and X-gal (blue, right) for detecting cardiomyocytes and jmj expression, respectively. Note that jmj is strongly expressed in cardiomyocytes (arrowhead), not in cardiac fibroblasts. control promoter (pGL2). These data suggest a possible role of JMJ in repressing cyclin D1, cyclin D2, and Cdc2 promoter activities, which may lead to inhibition of cell cycle progression.
To examine whether the expression levels of cell cycle proteins were affected in mutants, Western blot analyses were performed. In agreement with the results from reporter gene assays, cardiomyocytes from mutant hearts displayed increased protein levels of cyclins D1 and D2. In contrast, the protein levels of cyclin A, cyclin E, and p27 in mutant cardiomyocytes were not altered compared with wild type (Fig. 5B). The mutant showed more of the hyperphosphorylated Rb (pRb) than the underphosphorylated Rb, which is well correlated with the increased promoter activities, leading to an elevated expression level of cyclin D protein in mutant cardiomyocytes. These observations suggest that the increased promoter activities of cyclin D1, cyclin D2, and Cdc2 in jmj mutant cardiomyocytes result in phosphorylation of Rb to the hyperphosphorylated Rb, which in turn facilitates cell cycle progression or delays cell cycle withdrawal.
To confirm the repression function of JMJ on these promoters, we examined the promoter activities of the cell cycle regulators when JMJ was overexpressed in 10T1/2 cells (Fig. 5C). When cotransfected with JMJ, the promoter activities of cyclins D1 and D2 were markedly reduced by 54.5 and 62.5%, respectively, compared with the activity of the reporter gene alone. The promoter activities of Cdc2 and p21 did not seem to be repressed by JMJ, because the control promoter was also slightly repressed by JMJ. Together with the data in Fig. 5A, these results indicate that JMJ represses promoter activities of cyclins D1 and D2 and perhaps Cdc2 but not p21. To examine whether repression activities of JMJ on cyclin expression lead to the decreased cell proliferation, the BrdUrd uptake study was performed using 10T1/2 cells transfected with GFP-J-Wt, GFP-J-Nt, or GFP vector. J-Wt markedly reduced BrdUrd up-take by 88.1% compared with vector-transfected groups (Fig.  5D). J-Nt-(1-528) showed a similar repression activity (86.0%) as J-Wt, suggesting that the N terminus containing the TR domain is sufficient to down-regulate cell cycle progression.
These results indicate that the cyclin D1, cyclin D2, and Cdc2 promoters are negatively regulated by JMJ, which may lead to the decreased cell proliferation in wild type embryonic cardiomyocytes. Further investigation would be necessary to determine the molecular mechanisms by which JMJ regulates these promoters.
The Effect of JMJ on Cardiomyocyte Proliferation in Situ-Our results so far indicate that JMJ is involved in the suppression of cardiomyocyte proliferation when embryonic cardiomyocytes with a mixed background were grown in culture. This seems paradoxical, because the jmj mutant heart with a mixed background appeared hypocellular in the ventricular wall and septum during later embryonic stages (2). However, the cell proliferation rate in jmj mutant hearts has not been examined in situ at later embryonic stages. Therefore, we investigated mitotic activities of the different regions of the jmj mutant hearts with a mixed background between E12.5 and 16.5, because these mutants are alive until birth (Fig. 6). Cryosectioned heart was coimmunostained with phosphohistone H3 (P-H3) antibodies that mark proliferating cells and MF20 antibodies to identify cardiomyocytes (Fig. 6A). Interestingly, P-H3-positive cardiomyocytes were increased by 78.3% in the trabecular layer of mutant hearts (Fig. 6B). This result correlates well with our data indicating the increased mitotic rate, cyclin promoter activities, and hyperphosphorylated Rb in cultured mutant cardiomyocytes (see Figs. 4A and 5, A and B). In contrast, no significant changes were observed in the mutant ventricular wall compared with wild type (Fig. 6B). In situ terminal deoxynucleotidyltransferase-mediated dUTP nick end-labeling assays were performed to examine whether in- creased apoptosis was responsible for decreased cellularity in the mutant hearts. However, the apoptotic rate in the mutant heart did not seem to be changed compared with wild type (data not shown).
Therefore, it is plausible that JMJ acts as a regulatory factor of cardiomyocyte proliferation in a more complex manner than previously suggested, in a temporal and spatialdependent manner. It should be noted that the increased mitotic index of jmj mutant cardiomyocytes in our culture system (Figs. 4A, 5A, and 5B) probably reflects the increased cell proliferation in the trabecular layer of mutants. However, because of the lack of lineage markers that differentiate trabecular cardiomyocytes from compact layer one, we cannot currently assess whether our primary culture constitutes mainly trabecular cardiomyocytes. DISCUSSION Our study provides the first evidence for a role of JMJ in the regulation of cardiomyocyte proliferation in an Rb-dependent fashion. The interaction of JMJ with Rb results in further enhancing the inhibitory function of Rb on E2F activities, leading to the suppression of cell proliferation. The increased activities of the cyclin D1, cyclin D2, and Cdc2 promoters in cultured jmj mutant cardiomyocytes probably mediate up-regulation of the hyperphosphorylated form of Rb, which in turn contributes to an increase in cell proliferation in the absence of JMJ. Therefore, our present data indicate the role of JMJ in suppressing cell proliferation in developing cardiomyocytes. These results are further supported by the increased cell proliferation of the trabecular cardiomyocytes in jmj mutant hearts with a mixed background at later embryonic stages (E12.5-16.5) (see Fig. 6) and the reported hyperplastic trabeculae with a C3H/He background at early embryonic stage (E10.5) (25,26).
It is believed that phosphorylation of Rb determines its ability to interact with its binding partner proteins. Our data indicated that JMJ preferentially interacted with the underphosphorylated form of Rb. This seems to be a common char-acteristic of Rb-binding proteins that are involved in the regulation of cell cycle such as large T antigen, E7 viral antigen, E2F, and Id2 (39 -42). These findings strongly support the physiological role of JMJ in mediating cell proliferation via interaction with Rb in the developing hearts. Based on these results, Fig. 7 shows our working model. In quiescent cells, Rb is typically underphosphorylated, which can interact with E2F to inactivate it. In the presence of JMJ, JMJ is recruited to the underphosphorylated Rb to potentiate the repression function of Rb on E2F activities, perhaps by stabilizing Rb/E2F complex or by inhibiting phosphorylation of Rb, although the exact molecular basis remains to be elucidated. In proliferating cells, Rb is phosphorylated by cyclins/Cdks, leading to the disruption of association with E2F, thereby allowing S-phase genes to be activated. In addition to direct binding to Rb, JMJ down-regulates transcriptional activities of D-type cyclins and the level of hyperphosphorylated Rb and, as a consequence, allows further suppression of cell growth.
Rb contains the A/B pocket domains that are highly conserved from plants to humans. Various proteins have been identified to physically associate with Rb including cyclin D1, HDAC-1, HDAC-2, BRG1, and oncogenes such as E1A, SV40 antigen, and human papilloma virus E7 (14,(43)(44)(45)(46). These proteins utilize a conserved Leu-X-Cys-X-Glu (LXCXE) motif or its variants that allows them to bind the pocket domain of Rb with high affinity and modulate the function of Rb (47). Using in vitro GST-pull-down assays, we demonstrate that the Rb binding region in JMJ is localized in the TR domain and that this interaction primarily involves the A/B pocket domain of Rb. An examination of the TR domain sequence revealed that it does not contain LXCXE-like sequence, suggesting a LXCXEindependent binding of JMJ to Rb. Although the LXCXE motif is well characterized as a binding site to the pocket domain and important for the repressive function of Rb, there are known Rb-binding proteins lacking this motif (37). Such an LXCXEindependent Rb binding is found in E2F, HDAC3, and plasminogen activator inhibitor-2 (47)(48)(49)(50). The present data indicate that the TR domain of JMJ binds to the pocket domain of Rb and is also critical for modulating the Rb function. The functional importance of the TR domain is underscored by the recent observation that this domain was responsible for transcriptional repression function, although the underlying mechanism remains unknown (7). It has been reported that, during embryonic and neonatal periods, cardiomyocytes utilize different members of the pocket domain family for regulating proliferation (51). Therefore, it would be interesting to examine whether JMJ has differential specificity for these proteins during embryonic development.
Cyclins and their cognate kinases regulate the cell cycle progression in response to mitogenic signals. Among them, FIG. 6. Cell proliferation in jmj mutant hearts is altered in a spatial-dependent manner. A, cryosections of wild type and mutant hearts at E12.5 were subjected to double immunostaining using MF20 and P-H3 antibodies. Nuclei were counterstained with Hoechst dye 33342. P-H3-positive cells were shown at a higher magnification in a bottom panel. B, the number of mitotic cardiomyocytes (positive for both P-H3 and MF20 staining) per section was counted separately in the trabecular layer and the compact layer. Six slides per hearts from seven wild type (ϩ/ϩ) and seven jmj mutant (Ϫ/Ϫ) hearts were stained. **, p Ͻ 0.01. FIG. 7. A schematic diagram delineating the mechanism of the cell cycle control by JMJ. In the resting state, underphosphorylated Rb associates with E2F, resulting in cell cycle withdrawal. In the presence of JMJ, JMJ binds to underphosphorylated Rb, which in turn further represses the E2F activity. In the absence of JMJ, increased level of cyclin promoter activities up-regulate phosphorylation of Rb, which leads to the release of E2F, and enter the cell cycle. D-type cyclins are the earliest checkpoint step upon extracellular signals for G 1 /S transition by phosphorylating Rb. We provide evidence that increased activities of cyclin D1, cyclin D2, and Cdc2 may mediate up-regulation of cell growth partly by increasing the level of hyperphosphorylated Rb in cultured jmj mutant cardiomyocytes with a mixed background. These data correlate well with the up-regulated mitosis in the ventricular trabeculae by phosphohistone immunostaining of jmj mutant heart sections (Fig. 6). In contrast, cell mitosis in the compact layer does not seem to show statistically significant differences between wild types and mutants. It seems paradoxical, because the mutant mice with a mixed background showed the thin ventricular wall (2), suggesting decreased cell growth in the compact layer. However, the hypocellularity in the ventricular wall of mutant hearts was not accompanied by corresponding defects in mitotic index. Hence, appropriate caution should be exercised before concluding a direct role of the factors in cardiac myocyte proliferation. Appealing possibilities are that cell migration from the compact layer to trabeculae is abnormally accelerated or that terminal differentiation/cell cycle withdrawal is delayed in the trabecular layer, resulting in increased cell proliferation in the trabeculae and thin ventricular wall in jmj mutants. The molecular mechanism underlying noncompaction of the ventricular wall in jmj mutants remains to be elucidated.
Previous studies have suggested that many forms of cooperation occur between the myocardium and adjacent layers contributing to cardiomyocyte proliferation (35,52). Mice lacking retinoid X receptor ␣, vascular cell adhesion molecule, ␣-integrin, or Fog-2 displayed thin ventricular walls due to the partially disrupted epicardium-derived signal (28,29,31,53,54). Indeed, cardiomyocyte trophic factors secreted from endocardium and epicardium have been identified such as neuregulin, retinoic acid, and erythropoietin (55)(56)(57). Therefore, we cannot rule out the possibility of the indirect effect of epicardium-and endocardium-originated defects on myocardium development in jmj mutants. Although the present results indicate that JMJ plays a regulatory role in cardiomyocyte proliferation via interaction with Rb, the diversity of defects exhibited in jmj mutant embryos suggests multiple roles of JMJ in cell cycle regulation or in other target gene regulation in a temporal and spatial-dependent manner. The fact that JMJ contains the jmjC domain that is conserved in chromatinassociated proteins (10) provides a possible role of JMJ in chromatin remodeling, which may lead to the modulation of multiple target gene expression.