|Home | About | Journals | Submit | Contact Us | Français|
NDRG4 is a novel member of the NDRG family (N-myc downstream-regulated gene). The roles of NDRG4 in development have not previously been evaluated. We show that, during zebrafish embryonic development, ndrg4 is expressed exclusively in the embryonic heart, the central nervous system (CNS) and the sensory system. Knockdown ndrg4 in zebrafish embryos causes a marked reduction in proliferative myocytes and results in hypoplastic hearts. This growth defect is associated with cardiac phenotypes in morphogenesis and function, including abnormal heart looping, inefficient circulation and weak contractility. We reveal that ndrg4 is required for restricting the expression of versican and bmp4 to the developing atrioventricular canal. This constellation of ndrg4 cardiac defects phenocopies those seen in mutant hearts of heartstrings (hst), the tbx5 loss-of-function mutants in zebrafish. We further show that ndrg4 expression is significantly decreased in hearts with reduced tbx5 activities. Conversely, increased expression of tbx5 that is due to tbx20 knockdown leads to an increase in ndrg4 expression. Together, our studies reveal an essential role of ndrg4 in regulating proliferation and growth of cardiomyocytes, suggesting that ndrg4 may function downstream of tbx5 during heart development and growth.
NDRG4 belongs to the differentiation-related NDRG family, which includes four related members: NDRG1-4 (Qu et al., 2002; Zhou et al., 2001). NDRG1 was first isolated as a gene up-regulated in N-myc mutant mouse embryos and repressed by N-myc and C-myc (Shimono et al., 1999), although NDRG2 and NDRG3 expression is not under the control of N-myc or c-myc (Boulkroun et al., 2002; Okuda and Kondoh, 1999). The encoded proteins of this family are highly conserved in evolution. At the amino acid level, the four human NDRG4 members share 53–65% identity. While primarily found in the cytoplasm, the function of these proteins remains obscure with no obvious function predicted from protein structure. Each member contains an α/β hydrolase fold, which is common to a number of hydrolytic enzymes of widely differing phylogenetic origin and catalytic function (Qu et al., 2002). However, no hydrolytic catalytic site has been identified in this family (Shaw et al., 2002).
The existence of a gene family and its conservation through evolution often implies the important functions of its members. To date, studies of biological function have focused mainly on NDRG1 (also known as ndr1, drg1, RTP/rit42, GRP78/BiP, Cap43, Proxy-1). Accumulated data suggest that NDRG1 has at least two biological roles: it attenuates cell proliferation and promotes differentiation and is activated in response to various types of cell stress (Kokame et al., 1996; Nishimoto et al., 2003; Zhou et al., 1998 rev in Kovacevic and Ricardson, 2006). Interestingly, a nonsense mutation in the human NDRG1 gene is associated with hereditary motor and sensory neuropathy-Lom (HMSNL) (Kalaydjieva et al., 2000). HMSNL is a severe peripheral neuropathy characterized by Schwann cell dysfunction and progressive axonal loss in the peripheral nervous system (Kyuno et al., 2003). The importance of this nonsense mutation in disease causation is underscored by the Schwann cell dysfunction phenotypes in Ndrg1 knockout mice, which suggest that NDRG1 is essential for maintenance of myelin sheaths in peripheral nerves (Okuda et al., 2004). Over-expression of Ndrg1 in Xenopus laevis led to disrupted formation of the pronephric ducts and reduced size of pronephric tubules and somite disorganization, whereas Ndrg1 reduction by morpholino resulted in a failure of pronephric morphogenesis (Kyuno et al., 2003).
Much less is known about the functions of NDRG4 (also known as smap8, Bdm1) and no previous studies have investigated its role in cardiac development. All studies to date have utilized in vitro cell culture assays. NDRG4 protein has been shown to modulate activating protein-1 (AP-1) activity and may regulate neurite outgrowth in NGF-treated PC12 cells (Ohki et al., 2002). Most recently NDRG4 has been reported to enhance NGF-induced ERK signaling components and attenuates NGF-induced Elk-1 activation for transcription in PC12 cells, suggesting a possible role in neuronal differentiation (Hongo et al., 2006). Others have suggested a role for NDRG4 in amplification of the PDGF induced mitogenic response of aortic smooth muscle via the ERK signaling pathway in vitro (Nishimoto, et al., 2003). However, the in vivo function of NDRG4 has not been previously evaluated.
To investigate the developmental role of Ndrg4, we cloned ndrg4 from zebrafish. We demonstrate that zebrafish ndrg4 shows a highly restricted pattern of expression in CNS and the developing heart from 24 hours post-fertilization (hpf). We find that reduction of ndrg4 by morpholino antisense oligonucleotides attenuates myocardial proliferation and growth, and ultimately, limits heart function. Specifically, we show that the loss of Ndrg4 function led to a significant reduction of myocardial cell number in both heart chambers due to lower cell proliferation. In addition, we find that Ndrg4 is required for regulating versican and bmp4 expression levels correctly during the stages of chamber formation. Interestingly, the heart defects in ndrg4 morphant embryos are similar to that of the tbx5 (hst), including dysregulated versican and bmp4 expression in the developing atrio-ventricular canal of the heart (Garrity et al., 2002). Consistent with this observation, we find that ndrg4 expression in the developing heart is attenuated by hst mutants and tbx5 morphants. Further, we observed that tbx5 overexpression by tbx20 knock-down results in increased expression of ndrg4 in the heart, suggesting that ndrg4 may be downstream of tbx5 activation. This study provides the first demonstration of the crucial role for Ndrg4 in heart development and may shed new insight into the molecular mechanisms and genetic etiology of Holt-Oram syndrome (Bruneau et al., 2001).
We first predicted ndrg4 cDNA sequence according to mouse Ndrg4 cDNA, zebrafish htg data (AL772241 166073bp DNA HTG 26-SEP-2003 Danio rerio clone CH211-194M7, WORKING DRAFT SEQUENCE, 4 unordered pieces) and EST database using Blast and gene prediction tools (Gene finder, Genscan, GenomeScan, FGENESH). Following reverse transcriptase PCR (RT-PCT) with the specific primers (5′- TTGCTTTCCTGCATCCGATTGCAA-3′ and 5′ -GAGGTGAAGGCTAGAGGCAGCAAG-3′), we cloned the 1800 bp fragment into pCRII-TOPO vector (Invitrogen). Sequencing of three individual clones revealed identical, full-length cDNA sequence of zebrafish ndrg4, which has been deposited in GenBank with the accession number DQ649010.
All zebrafish and embryos were maintained at 28.5°C and staged as previously described (Kimmel et al., 1995). The tbx5 mutant hst was maintained by outcrossing to standard wild-type lines (Garrity et al., 2002). Wild-type, transgenic Tg(cmlc2:gfp) (Shin et al., 2005) or transgenic Tg(cmlc2:DsRed2-nuc) embryos (Mably et al., 2003) primarily at the 1-cell stage with chorion intact were injected with approximately 2 nl volume of morpholino oligonucleotide, mRNA or DNA in Danieau buffer (Nasevicius and Ekker, 2000).
Morpholino antisense oligonucleotides were purchased from Gene-Tools (Corvallis, OR). ndrg4-MO1 (5′-CCAGCACTCCGGCATGGTAGCTTAG-3′) was designed against the ndrg4 translational start site. ndrg4-MO2 (TGCATTCATCTTACCCTTGAGGCAT) is a splice-blocking MO, which targets the exon 3 and intron 3 splice junction of ndrg4 pre-mRNA. We used two sets of control MOs: 1) Standard control MO (5 ′-CCTCTTACCTCAGTTACAATTTATA-3′) and 5mis control MO, which is a 5 bp mismatch of ndrg4-MO2 (5′- TGgATTgATCTTAgCCTTcAGGgAT-3′, lower case letters indicate altered bases). To test the efficiency of the ndrg4-MO2, RT-PCR was performed using whole-embryo RNA from ~30 embryos at 48 hpf, using the primers, P1 (5′- ATGCCGGAGTGCTGGGATGG-3′ and P2 (5′- AGCCCCAACTCCAATTCCGAC-3′) to detect the end ndrg4 mRNA transcripts from MO treatment. The tbx5-MO 5′-GAAAGGTGTCTTCACTGTCCGCCAT-3′ (Garrity et al ., 2002) and the tbx20-MO 5′-CTCATGTGGAGAAGGGGTTTTGGAG-3′ (Szeto et al., 2002) have been previously described.
To test the efficiency of the ndrg4-MO1, a transgenic construct driving expression of an Ndrg4-EGYP fusion protein by the cmlc2 promoter was made (details available on request). Briefly, the cmlc2 minimal promoter (244bp) was cut from pGEM-210/34 (Huang et al., 2003) and the EYFP fragment was isolated from pEYFP-N1 (BD Biosciences Clontech). The cmlc2-Ndrg4-EGYP DNA fragment was inserted into the pWhere plasmid (InvivoGen) at the XhoI/NheI sites. Thus, H19 insulators flanked this entire fusion DNA fragment. 15pg of the linearized DNA was injected into wild-type embryos at one-cell stage with presence of ndrg4-MO1 or standard control MO.
For mRNA injection experiments, a 1029 bp PCR product (including initial codon ATG and stop codon TGA) from adult zebrafish heart cDNA with primers P1 and 5′-CGCAAGGGTTCAGCAGGACACT-3′ was cloned into pCRII-TOPO (named pCRII-zNDRG4) and then the ndrg4 insert was subcloned into pcGlobin2 (Ro et al., 2004) at EcoRI site. To generate capped mRNA, plasmid was linearized by ClaI and transcribed with T7 RNA polymerase using the mMESSAGE mMACHINE T7 in vitro transcription kit (Ambion) according to the manufacturer’s instructions.
At 24 hpf, 0.2 mM 1-phenyl-2-thio-urea (Sigma, St Luis, MO) was added to embryo media to prevent pigment formation. Embryos were collected at the appropriate stages and fixed in 4% paraformaldehyde (PFA), pH 7.0, in phosphate-buffered saline (PBS), overnight at 4°C. Fixed embryos were dechorionated, washed three times with PBS and stored in methanol at −20°C. Whole-mount in situ hybridization was performed using digoxigenin labeled antisense RNA probe and visualized using anti-digoxigenin Fab fragments conjugated with alkaline phosphatase (Roche Molecular Biochemicals). Riboprobes were made from DNA templates, which were linearized and transcribed with either SP6, T3 or T7 RNA polymerase. Embryos were processed and hybridized as described (Thisse et al., 1993). The antisense RNA probes used in this study were tbx5 (Begemann and Ingham, 2000), cardiac-myosin light chain-2 (cmlc2), ventricular myosin heavy chain (vmhc) (Yelon et al., 1999), atrial myosin heavy chain (amhc), nppa (anf) (Berdougo et al., 2003), versican (Walsh and Stainier, 2001), bmp4, gata4, Nkx2.5, hand2 (Yelon et al., 2002), and ndrg4. To synthesize an ndrg4 antisense riboprobe, linearized pCRII-zNdrg4 (containing a 1029 bp fragment of zebrafish ndrg4 cDNA) was used as a template.
For histology, embryos were fixed overnight in 4% PFA in PBS at 4 °C and embedded in JB-4 resin (Polysciences, Warrington, PA) according to manufacturer’s instructions and sectioned at 5 μm, using an ultramicrotome. Serial sections were cut in transverse orientation and stained with hematoxylin-eosin. For photography, living zebrafish embryos, plastic tissue sections and whole-mount in situ hybridized embryos were photographed on an Axioplan microscope (Zeiss) using a digital camera (Dage).
Cardiomyocyte number and cell size in ndrg4-MO2 injected and control transgenic embryos were measured as described (Mably, et al., 2003; Jia, et al., 2007). One Tg(cmlc2:DsRed2-nuc) line (expressing RFP in cardiomyocyte nuclei) was used to measure myocardial cell number at 48 hpf. The second Tg(cmlc2:gfp) line (expressing GFP throughout the cardiomyocyte) was used to determine myocardial cell size at 48 hpf and myocyte number at 24 hpf. The Tg(cmlc2:DsRed2-nuc) line was not used for quantifying myocyte number at 24 hpf because RFP positive cells are hardly detectable at the heart tube stage (Mably, et al., 2003). Briefly, myocyte number in the atrium and ventricle was determined by quantifying nuclei in 5 embryos from each group of ndrg4 morphants and control embryos. For clearly visualizing the margin of individual cell surface, embryos were flat mounted between two micro cover glasses and were subjected to confocal imaging using 40x planapochromat of Zeiss LSM 510 confocal microscope system. Cell sizes of 150 individual cardiomyocytes from each group of ndrg4 morphants and control embryos were measured using Image J. Numbers of cardiomyocytes from flat-mounted embryos at 24 hpf were counted in each confocal section using Pickpointer embedded in the Image J. Pickpointer permits a single user-defined mark to appear through z stacks of images, allowing the tracking of a single cell in overlapped z sections to avoid double counting.
Cell proliferation in the zebrafish heart was analyzed as previously described (Ribeiro et al., 2007). Briefly, the ndrg4-MO2 injected or control Tg(cmlc2:gfp) embryos at 31 hpf or 48 hpf were injected with 2 nl 100mM BrdU in the pericardiac cavity region of tricaine anesthetized embryos. These embryos were fixed after one hour incubation in medium containing BrdU at 28.5°C. The embryos were then treated with acetone for 10 minutes and digested in 0.25% trypsin in PBS without Mg2+ and Ca2+. Hydrochloric acid (HCl) was used to treat the embryos for 1 hour at room temperature. After extensive washing, double whole-mount immunofluorescence was performed using anti-BrdU antibody (1:400, Invitrogen) and anti-GFP antibody (1:400, Invitrogen) as primary antibodies, and goat anti-mouse Cy3 antibody (1:200, Invitrogen) and Alexa 488 donkey anti-rabbit antibody (1:200, Invitrogen) as secondary antibodies. The BrdU labeled embryos were then stripped of their head, oriented in solidifying 0.5% agarose and the heart region was confocal imaged using a Zeiss LSM510 microscope system.
In order to investigate the potential function of Ndrg4 in zebrafish heart development, we cloned a single full-length cDNA of ndrg4 from adult zebrafish heart. Zebrafish ndrg4 contains an open reading frame capable of encoding a 339 amino acid polypeptide, a 101-bp 5′-UTR, and a 679-bp 3′-UTR. Sequence analysis revealed that the amino acid sequence of zebrafish Ndrg4 protein shows a high degree of conservation (identities: 87%, positives: 94%) with mouse Ndrg4 (see supplemental data). There is a typical α/β hydrolase fold as in other NDRG proteins (Qu et al., 2002). Homology searches with the current zebrafish genomic database with the entire ndrg4 cDNA revealed one identical match on chromosome 25 (NC_007136.1) (data not shown), while no other significant alignment was observed with any other zebrafish genomic sequence. The zebrafish ndrg4 gene includes 12 exons and spans 26 kb.
Whole mount in situ hybridization indicated an unexpected expression of ndrg4 in the developing zebrafish heart. For comparison, we used the pan-cardiac marker cmlc2 to identify cardiac tissues at various stages. As we and others (Yelon et al., 1999; Yelon, 2001) observed, cmlc2 is significantly expressed in the bilateral cardiac primordia at the 14-somite stage (Fig. 1A) and in the cardiac cone at the 21-somite stage (Fig. 1C). At these stages, ndrg4 expression is marked in the CNS but is almost undetectable in cardiac tissues (Fig. 1B, D). However, by 24 hpf, ndrg4 expression resembles cmlc2 in occurring uniformly throughout the heart tube (Fig. 1E,F). Subsequently, as the heart tube undergoes looping morphogenesis, robust ndrg4 expression continues throughout the heart with strong expression detected in the ventricle (Fig. 1G-L). The expression of ndrg4 persisted in the heart at 72 hpf (Fig. 1M), the later embryonic period and the adult heart (RT-PCR, data not shown). Thus, the temporally-regulated cardiac expression pattern suggests Ndrg4 is not required for early specification of cardiac progenitors, but may play a later role in chamber formation, organ growth and remodeling.
Initially, ndrg4 was expressed throughout CNS, from the 10-somite through 14-somite stages (Fig1.B; Fig. 2A). By the 21-somite stage, its expression in CNS was restricted to the telencephalon, ventral diencephalons and spinal chord (Fig. 2B). Subsequently (at 22–72 hpf), expression was accentuated in cranial ganglia (trigeminal ganglion, anterior lateral line ganglion and posterior lateral line ganglion), hindbrain neuron cells, tegmentum and cerebellum (Fig. 2C-I). By 33–72 hpf, prominent expression occurs in the retina but no expression within the retinal proliferative zone (Fig. 2H).
To determine the role of Ndrg4 in zebrafish development, we designed two morpholino antisense oligonucleotides: 1) one to block the translation of ndrg4 (ndrg4-MO1) and another to knockdown Ndrg4 function by blocking RNA splicing (ndrg4-MO2). To test the efficacy of ndrg4-MO1, we developed a transgenic construct driving expression of an Ndrg4-EYFP fusion protein under control of the cmlc2 promoter (Huang et al., 2003), which includes the hybridization homology region targeted by ndrg4-MO1. After injection of this linearized DNA with 6.4 ng of control MO into one-cell stage fish embryos, green fluorescent signal due to Ndrg4-EYFP expression in cardiomyocytes was observed in 7.9% of injected embryos (n=108) at 48 hpf. By contrast, injection of the Ndrg4-EYFP DNA in the presence of 6.4 ng of ndrg4-MO1 resulted in the absence of EYFP expression in all injected fish (n=155) (see Supplemental data). These experiments demonstrate that ndrg4-MO1 is able to specifically block the translation of the ndrg4 gene.
To further demonstrate that the morphant phenotypes were the result of specific loss of the ndrg4 gene function, we tested ndrg4-MO2, which binds to the exon 3 / intron 3 splice junction (Fig. 3A). Because ndrg4-MO2 could result in either exon deletion or intron insertion in ndrg4 mRNA, we harvested embryos injected with sequential concentrations of morpholino and checked mRNA by RT-PCR with specific primers as noted. As seen in Fig. 3B, gradual increase in MO2 resulted in a progressive loss of exon 3 (121 bp). The 224 bp band was confirmed by sequencing to directly join exon 2 and exon 4 (Fig. 3C). Injection of 6.4 ng of MO2 resulted in complete exclusion of exon 3 from the ndrg4 mRNA in the injected embryos, resulting in a shift of the reading frame and creation of an early stop codon. However, there was no inhibition of ndrg4 splicing in embryos injected with 6.4 ng of a 5 bp mismatch control ndrg4-MO. Thus, ndrg4-MO2 blocked the splicing of zebrafish ndrg4 in a dose-dependent and highly specific manner.
In our initial injection experiments, both morpholino oligonucleotides caused the same phenotype at a range of 3.2 to 6.4 ng with nonspecific effects observed at higher doses (data not shown). Since at the same high concentration (>6.4 ng), MO2 was less toxic than MO1, we used MO2 in the following experiments. At 6.4 ng/embryo of ndrg4-MO2, we consistently obtained morphants that had defects in the hearts and heads (89%, n=180) but overall normal development. Although the morphology of the initial heart tube of ndrg4 morphants at 24–33 hpf is indistinguishable from wild type, abnormal cardiac development was visible in ndrg4-MO2 injected zebrafish embryos after 36-hour development (data not shown). By 48 hpf and 3 days post-fertilization (dpf), the hearts of ndrg4 morphants showed pericardial edema, dilated atrium, looping defects, reduced circulation, and slower heart rate with weaker contraction compared to their control siblings (Fig. 4A-D; also see Movie 1 and 2 in Supplemental data). The heart rate of ndrg4 morphant embryos (101±38 beats/min, n=8) at 48 hpf is 75.1% of the control (135±27 beats/min, n=6). In addition, ndrg4 morphants also displayed defects in the development of their heads including small eyes and prominent edema in the hindbrain. Thus, the defects in the morphants are consistent with the expression pattern of ndrg4. To demonstrate that these results were not due to an injection artifact, we injected 1-cell embryos with a standard control MO or a 5 base-pair mismatch control MO of ndrg4 at a concentrations 10 ng/embryo. This caused non-specific abnormalities distinct from the typical defects seen in ndrg4 moprhants in less than 5% of the standard control and 8% of the mismatch control embryos.
Another important criteria in determining the specificity of any antisense approach is rescue of observed defects by expression of gene specific RNA. Therefore, we simultaneously injected capped ndrg4 mRNA along with ndrg4-MO2. In rescue experiments, when 100 pg/embryo of ndrg4 mRNA was injected with 6.4 ng ndrg4-MO2 into 1-cell zebrafish embryos, about 50% (78/156) of the injected embryos displayed no or less severe cardiac defects at 48 hpf (Fig. 4E) compared to that for ndrg4-MO2 only, in which 92% (228/248) of injected embryos showed the typical cardiac phenotype described above. Injection of ndrg4 mRNA (100 pg) alone did not induce heart defects. Thus, the cardiac phenotype in ndrg4 morphants could be ameliorated in about 45.5% of the co-injected embryos. The eye size and head defects in MO injected embryos was also partially rescued by mRNA injection (Fig. 4E). Therefore, we conclude that the specific morphant phenotypes produced by ndrg4-MO2 result primarily from the selective inhibition of Ndrg4.
As described above, one of the prominent morphological defects in the ndrg4 morphant heart is failure to complete looping. To further define abnormalities in looping of the heart, we injected transgenic Tg(cmlc2:gfp) embryos with ndrg4-MO2 and evaluated heart looping at sequential stages of development. At 36 hpf, the heart tube of control zebrafish embryos loops to the right, whereas the heart tube in ndrg4 morphants remains central and linear in a tubular structure (data not shown). By 48 hpf and 3 dpf, there was no obvious perturbation in cmlc2-driven GFP expression in ndrg4 morphants when compared to the control, but there was an obvious defect in looping as the heart continued to retain its linear configuration, midline orientation, and circulation was depressed (Fig. 5). In histological sections of ndrg4-MO2 injected embryos, we found that at 2 dpf (data not shown) and at 3 dpf, the overall size of ventricle and atrium was markedly smaller than age-matched controls (Fig. 6A,B). In addition, sections showed that regions of local cell proliferation around the outflow tract and atrioventricular (AV) junction was less extensive in morphants than wild-type. The extracellular matrix between myocardium and endocardium in the atrium was enlarged compared to age-matched wild-type embryos.
The smaller overall size of the cardiac chambers in ndrg4 morphants suggests a possible primary defect in proliferation and / or size of cardiomyocytes. To address this question directly, we determined myocyte cell number and size using confocal microscopic analysis after injection of the transgenic Tg(cmlc2:DsRed2-nuc) and Tg(cmlc2:gfp) embryos with ndrg4-MO2. We decided to quantify cell number at 48 hpf, the earliest stage in which cardiac morphology was overtly different from wild-type. As shown in Fig. 6C-E, ndrg4-MO2 injected Tg(cmlc2:gfp) embryos at 2 dpf showed a significant reduction in the number of cardiomyocytes in both chambers (atria and ventricle) when compared to controls: myocyte number in atrium and ventricle of injected fish was reduced by 36.1% (97±11 vs 62±6) and 18.5% (173±11 vs 141±14) respectively (Fig. 6E). To determine the potential effect of Ndrg4 inhibition on myocyte size, we treated Tg(cmlc2:gfp) embryos with ndrg4-MO2 and measured their cell size at 48 hpf with confocal imaging. We found only a slight reduction (10.3%) in myocyte size (area) in ndrg4 morphants in comparison with control (193.2±8.5μM2 vs 215.3±4.5μM2, P>0.05) (Fig. 6F-H).
We questioned whether the number of cardiomyocytes in ndrg4 morphants was decreased at even earlier stages and thus reduction in the myocyte number at 48 hpf might reflect a decrease in initial myocyte differentiation. Therefore, we injected Tg(cmlc2:gfp) embryos with ndrg4-MO2 and measured myocyte number in control and morphant embryos at 24 hpf. We found no attenuation in myocardial cell number between control (182±16; n=5;Fig. 6I) and ndrg4-MO2 (177±12; n=5;Fig. 6J) injected Tg(cmlc2:gfp) embryos based on the number of GFP positive cells in the heart tube. Thus, we conclude that attenuation of Ndrg4 did not result in a decrease in early myocyte differentiation.
The reduced myocyte number in ndrg4 morphants prompted us to investigate myocyte cell proliferation and apoptosis during 1–2 dpf. At 33 hpf or 48 hpf, there was not a significant number of apoptotic cells observed in the heart of control or ndrg4-MO2 treated embryos by TUNEL assay, although there was a clear increase in apoptosis within the developing CNS in ndrg4-MO2 treated embryos at 48 hpf (data not shown). To determine myocyte proliferation in the early developing heart we gave a one-hour pulse of BrdU by injecting a BrdU solution in the pericardiac cavity (Ribeiro, et al., 2007). As shown in Fig.7, at 31 hpf the number of BrdU positive cells in ndrg4-MO2 treated embryonic heart (21.0±3.0; n=5) is comparable to control (23.1±3.5; n=5). This result is consistent with the normal heart development of ndrg4 morphants at this stage. However, at 48 hpf ndrg4-MO2 injected embryos (15.8±2.9; n=5) display lower number of proliferating cells than control (35.1±5.3; n=5) in the developing heart. Because, the total number of cardiomyocytes (Fig.6) in ndrg4 morphants (203±9) at 48 hpf is lower than control (260±11), we calculated the proliferation index (the average of the total number of BrdU positive cells per 100 cardiomyocytes). Hearts of ndrg4 morphants at 48 hpf had a significantly lower proliferative index than control embryos (7.8±2.1 vs 13.5±3.9; P<0.01). Therefore, attenuation of Ndrg4 resulted in a decrease in both myocyte proliferation and myocyte cell size at 48 hpf that was not detected during early cardiac development at 24 hpf. Taken together, these data indicate an essential requirement for Ndrg4 in regulation of myocardial proliferation and cardiac growth during early cardiogenesis suggesting that cell growth and cell division are tightly coupled in the embryonic heart.
ndrg4 morphants formed an overtly normal heart tube, despite expression of ndrg4 in the heart already at this stage. Moreover, at 24 hpf, expression of cmlc2, nkx2.5, hand2 and gata4 expression in the developing heart tube of ndrg4-MO2 injected embryos was indistinguishable from un-injected embryos (Fig. 8A, B and data not shown). By 33 hpf, ndrg4 morphants also showed normal expression of tbx5, nppa and cmlc2 (Fig. 8C-H), indicating that myocardial differentiation proceeded normally in ndrg4-morphant embryos. To examine cardiac chamber formation in ndrg4 morphants, we analyzed the expression of two pan-cardiac markers cmlc2 and nppa, as well as the ventricular marker vmhc (Yelon et al., 1999), and the atrial marker amhc (Berdougo et al., 2003). We detected no abnormalities in the expression levels of these chamber specific molecular in the ndrg4 MO knockdown embryos (Fig. 8). These data suggest that atrial and ventricular specification occurred normally in ndrg4-MO2 injected embryos.
To investigate whether cardiomyocytes are properly differentiated, we analyzed two genes (versican and bmp4) which are initially expressed broadly in the wild-type ventricle but later (by 48 hpf) become highly restricted to the AV junction (Walsh and Stainier, 2001). In contrast to wild-type (Fig. 8O), ndrg4-MO2 injected embryos (41 of 46 morphants examined at 48 hfp) showed dispersed expression of versican throughout the ventricle, similar to what is ordinarily observed in 33 hpf wild-type hearts (Fig. 8P). In contrast, the punctate expression of versican in the otoliths is characteristic of embryos at 48 hpf (Garrity et al, 2002), and indicates that morphants are not subject to a general developmental delay. Similarly, bmp4 marked the developing AV canal in 48 hpf wild-type zebrafish embryos (Fig. 8S), whereas ndrg4-MO2 treated embryos (Fig. 8T) showed diffusely distributed bmp4 in the AV canal and ventricular regions. Thus, ndrg4 appears to be required for molecular patterning of the atrio-ventricular canal.
The phenotype of our ndrg4 morphants was remarkably similar to that described for the hst mutants (normal myocyte specification, abnormal looping, weak contractility, abnormal restriction of versican and bmp4 expression) which lacks functional Tbx5 (Garrity et al., 2002). In addition, expression of tbx5 (Begemann and Ingham, 2000) precedes the expression of ndrg4 in the heart field. Because tbx5 is a key transcription factor that regulates cardiac morphogenesis and gene expression within the developing heart (Basson et al., 1999; Horb and Thomsen, 1999; Bruneau et al., 2001, Garrity et al., 2002), we wished to determine the relationship between ndrg4 and tbx5 in regulation of versican and bmp4 in heart. We first injected wild type 1-cell embryos with tbx5-MO and checked ndrg4 expression with whole mount in situ hybridization. In tbx5 morphants at 35 hpf and 48 hpf, we observed that ndrg4 expression in the embryonic heart was significantly decreased (Fig. 9B, E), with no change in ndrg4 expression in the CNS and sensory system where ndrg4, but not tbx5, is expressed (data not shown). We then examined ndrg4 expression in hst embryos to confirm this effect of loss of Tbx5 function. As can be seen in Fig. 9, at 35 hpf and 48 hpf, there appears to be a marked attenuation of ndrg4 expression in the heart of hst mutant embryos.
Since tbx5 expression is down regulated by Tbx20 in early development of the zebrafish heart (Szeto et al., 2002), we injected wild-type embryos with tbx20-MO to induce tbx5 overexpression. As showed in Fig. 10, consistent with the previous report (Szeto et al., 2002), tbx5 expression in tbx20-MO injected embryos at 33 hpf or at 48 hpf is significantly higher than control. As we expected, compared to control embryos, ndrg4 expression in tbx20 morphants is correspondingly increased at 33 hpf or at 48 hpf. These data are consistent with the hypothesis that Tbx5 acts to regulate the levels of ndrg4 expression during heart development and suggest that ndrg4 may be a down-stream target of tbx5 activation.
We have cloned and characterized ndrg4, a novel member of the ndrg family from zebrafish. Homology analysis reveals that the NDRG4 protein is highly conserved across vertebrates. During development, spatial and temporal regulation confines the ndrg4 expression to the forming heart and nervous system. Specifically, ndrg4 is expressed in the developing heart from 24 hpf with increased expression in the later period and in the adult heart (XQu and HSB, unpublished). Once the heart tube is formed, several key cardiogenic events occur in rapid succession (Stainier et al., 1993; Stainier, 2001), including looping morphogenesis by 30 hpf, delineation of visibly distinct atrial and ventricular chambers by approximately 36 hpf, formation of the cardiac cushions, or AV valve analogue at approximately 48 hpf, and myocardial proliferative growth from 48 hpf. Thus, ndrg4 is expressed during the later critical stages of cardiac morphogenesis and proliferative growth (Srivastavia and Olson, 2000; Yelon et al., 1999; Stainier, 2001). Interestingly, in mouse, unlike the rather ubiquitous expression of Ndrg1–3, Ndrg4 is expressed predominantly in the myocardium of the developing heart and CNS with increased expression in the postnatal period and the adult (Okuda et al., 1999; Qu et al., 2002; and data not shown). The conservation of ndrg4 gene expression in the developing hearts may suggest an evolutionarily conserved function as well.
Prior to the onset of cardiac looping, cardiogenesis progresses normally in ndrg4 morphants, indicating that there is no essential requirement for Ndrg4 function in the very early stages of heart development. Although the dysmorphic hearts in the ndrg4 morphants appeared to form two distinct chambers, the heart remained as a linear tube and did not form the characteristically looped structure, resulting in a weakly contractile heart with a slower rhythm.
Reduction of Ndrg4 results in a significant decreased number of cardiomyocytes by 48 hpf. Because ndrg4 is not expressed during early myocardial differentiation from the mesoderm, the reduction in myocyte number is unlikely to result from a defect in cardiomyocyte commitment. The normal expression of cardiac markers cmlc2, nppa and tbx5 at 33–48 hpf indicates that a degree of myocardial differentiation and chamber maturation does occur in ndrg4 morphants, although continued differentiation is compromised and heart function deteriorates with time.
While we found no significant changes in apoptosis at these early stages of myocardial development, we were able to detect a decrease in myocardial cell number at 48 hpf. In addition, we documented that at 48 hpf ndrg4-MO2 injected embryos display significantly lower number of proliferating cells than control in the heart, although the number of BrdU positive cells in ndrg4-MO2 treated embryonic heart at 31 hpf is indistinguishable from control. Consistently, ndrg4 expression in zebrafish heart is not significant before 24 hpf and abnormal cardiac development in ndrg4-MO2 injected zebrafish embryos was only visible after 36 hpf. Therefore, the decrease in myocyte number in ndrg4 morphants is primarily due to the reduced myocyte proliferation suggesting that Ndrg4 is required for maintenance of normal cell proliferation after formation of the heart tube.
Of note, expression of NDRG2, which is most closely related to NDRG4, as well as NDRG1, has been documented to reduce cell growth in vitro (Kurdistani et al., 1998; Guan et al., 2000; Deng et al., 2003). Recently, NDRG1/Rit42 protein was shown to associate with the microtubules in centrosomes and to participate in the spindle checkpoint (Kim et al., 2004). Interestingly, NDRG4 may regulate neural cell proliferation and differentiation via the Ras-ERK pathway (Ohki et al., 2002; Hongo et al., 2006). The human NDRG4 (smap8) was suggested to play a role in PDGF-induced mitogenesis of a rat aortic smooth muscle cell line by amplification of the ERK (extracellular signal-regulated kinase) signal (Nishimoto et al., 2003). Therefore, these data from cell culture assays are consistent with our in vivo data suggesting a potential role of Ndrg4 in proliferation. It is noteworthy that vertebrate NDRG4 shows high of amino acid sequence conservation (56%) to Drosophila MESK2 (misexpression suppressor of dominant-negative KSR, kinase suppressor of Ras), a gene product which may represent a new component of the Ras pathway or of other signaling pathways that can modulate signaling by Ras (Huang and Rubin, 2000). The involvement of the Ras pathway in heart development has been well documented. Ras is implicated in hypertrophic growth of the myocardium (Clerk and Sugden, 2000; Sugden, 2003) and in valve formation (Lakkis and Epstein, 1998; rev in Yutzey, et al., 2005). Thus, it is tempting to speculate that Ndrg4 might regulate heart development via Ras pathway but clearly further investigation is warranted.
The T-box transcription factor tbx5 has emerged as a key gene regulating heart development from amphibians to mammals (Horb and Thomsen, 1999; Liberatore et al., 2000; Bruneau et al., 2001). Mutations in human TBX5 are the cause of Holt-Oram syndrome, a haploinsufficient syndrome characterized by heart and forelimb defects (Basson et al., 1997; Li et al., 1997), and dominant negative interference of Xenopus tbx5 leads to near-total absence of differentiated heart tissue and early cardiac markers (Horb and Thomsen, 1999). Similar defects have been found in mice lacking one copy of the Tbx5 gene (Bruneau et al., 2001). Mice lacking both copies of Tbx5 have more severe defects, including an absence of heart looping and alterations in a subset of cardiac-specific genes, although the early steps in cardiac development appear normal (Bruneau et al., 2001).
We were impressed that loss-of-function of Ndrg4 by MO knockdown in zebrafish closely phenocopies the cardiac defects in hst mutant embryos (tbx5 mutants). Although tbx5 is expressed in the developing heart earlier (Begemann and Ingham, 2000) than ndrg4, the heart of hst mutant embryos appears to form and function normally through the early heart tube stage, manifesting only a slight bradycardia compared with wild-type siblings (Garrity et al., 2002). The predominant cardiac defects of hst mutant embryos become evident well after the development of discrete morphological chambers and after the onset of circulation. The heart fails to loop and then progressively deteriorates, a process affecting the ventricle as well as the atrium.
An intriguing observation in this study is that ndrg4 expression is specifically reduced in the hearts of tbx5 mutants or morphants while ndrg4 expression in other regions of mutant embryos are unaffected. It has been reported that other cardiac markers (nkx2.5, hand2, bmp4, cmlc2 and vhmc) are normal in hst mutant embryos at the 15-somite stage. Both vmhc and amhc are expressed normally in the heart in hst mutant embryos at 26 and 48 hpf (before and after overt chamber formation). These data suggest that atrial and ventricular fates are assigned properly in hst mutant embryos (Garrity et al, 2002). Therefore, reduction of ndrg4 expression in tbx5 mutant hearts is not just due to a general myocardial differentiation defect in tbx5 mutant myocardium. Moreover, attenuation of either Tbx5 or Ndrg4 resulted in abnormal upregulation of versican and bmp4 in zebrafish heart at 2 dpf. Unfortunately, multiple attempts at tbx5 mRNA injection resulted in severe embryonic toxicity or lethality preventing analysis of subsequent alterations in the expression of Ndrg4. However, tbx5 expression is down-regulated by Tbx20 in early development zebrafish heart (Szeto et al., 2002), therefore we investigated if tbx5 over-expression in heart by Tbx20 reduction could induce ndrg4 expression. Indeed, we observed that ndrg4 expression in tbx20 morphant heart is increased at 33 hpf or at 48 hpf. These data further suggest that Tbx5 acts to regulate the level of ndrg4 expression during the heart development, and that ndrg4 is potentially a novel downstream target of tbx5 activation and may be at least partially responsible for the defects observed in hearts of tbx5 null mutant animals. Since Tbx5 is not expressed in the CNS, ndrg4 regulation in neural tissues presumably occurs by alternative mechanisms.
To date, only a few cardiac specific genes have been identified as Tbx5 target genes (Plageman and Yutzey, 2005; Stennard and Harvey, 2005) while a myriad of genes have been recently identified by microarray analysis to be affected by alteration in Tbx5 dosage (Mori, et al., 2006; Plageman and Yutzer, 2006). Interestingly, Ndrg4 has not been identified as a downstream target in any of these previous studies. As the mechanism by which TBX5 haploinsufficiency causes cardiac abnormalities seen in Holt-Oram syndrome is still not well understood, it will be of great interest to determine if there is a genetic link between NDRG4 gene and congenital heart disease. The observation that NDRG4 expression is induced by homocysteine (Nishimoto, et al., 2003) leads to further speculation that it might be involved in the increased incidence of congenital heart disease in offspring of mothers with defects in homcysteine metabolism (Boot, et al., 2003; Hobbs, et al., 2005).
In summary, this study is the first to demonstrate a role for ndrg4 in cardiac development and provides the first functional study of Ndrg4 in a vertebrate animal species. While not required for myocardial specification and early differentiation, ndrg4 appears to play an important role in regulating myocyte cell proliferation and is required for normal cardiac development in the fish. Further delineation of its potential role in downstream mediation of Tbx5 signal transduction and its role in atrioventricular canal morphogenesis will be the subject of continuing investigation.
We thank Huai-Jen Tsai for the pGEM-210/34 plasmid; and Hae Chul Park, Karen McFarland and Jennifer Miller for help and advice during this project. X. Q. was supported by an AHA post-doctoral fellowship (0425408B). This research has been supported in part by a zebrafish initiative funded by the Vanderbilt University Academic Venture Capital Fund.