Cardiac Expression of Tnnt1 Requires the GATA4-FOG2 Transcription Complex

Previous work by us and others has shown that the loss of interaction between GATA4 and FOG2 protein partners is embryonic lethal due to heart failure at embryonic day (E) 13.5; however, the role of this important protein duo in various cardiac compartments (e.g., myocardial, endocardial, or epicardial cells) remains to be understood. Although a dual role (both as an activator and a repressor) for the GATA4-FOG2 transcriptional complex has been put forward, the specific genes under GATA4-FOG2 control in the developing heart have remained largely elusive. Since the myocardial-restricted Fog2 re-expression in the Fog2 null embryos is sufficient to extend their life span, identification of GATA4-FOG2 target genes in cardiomyocytes could shed light on the molecular mechanism of GATA4-FOG2 action in these cells. We report here that cardiac expression of slow skeletal troponin T (Tnnt1) strictly depends on the physical interaction between GATA4-FOG2 in the myocardium of both atria and ventricles.


INTRODUCTION
The multitype, zinc-finger proteins of the FOG (Friend of GATA) family control biological activities of GATA-binding (GATA) transcription factors (for review, see [1,2]). The role for FOG2 (ZFPM2, Mouse Genome Informatics) protein in cardiac development has been firmly established. Initial characterization of the Fog2 gene revealed prominent expression in several developing organ systems (e.g., brain, heart, and gonads) [3,4,5]. Fog2 -/-(null) embryos die at mid-gestation (~E13.5), with a cardiac defect characterized by a thin ventricular myocardium, common atrioventricular (AV) canal, and the Tetralogy of Fallot malformation [6,7]. Importantly, Fog2 gene loss affects the development of cardiac vasculature [7]. While the formation of an intact epicardial layer and expression of epicardium-specific genes in Fog2 null mutants proceed apparently as normal, markers of cardiac vessel development are not detected [7,8]. Importantly, KDR expression is not detected in the epicardial layer of the Fog2 knockout mice. KDR (FLK1, VEGFR2), the major receptor for VEGF (vascular endothelial growth factor), is an important marker of vascular cells and is absolutely essential for vascular development (e.g., [9], see [10] for a review).
This earlier work drew attention to the role that FOG2 and GATA4 play in the development of the cardiac vascular and epicardial cells. The early demise of the Gata4 null embryos had limited the analysis of "Gata4-less" cardiac development to a narrow window between E7.0 and E9.0. Moreover, examination of the Gata4 null mutants did not reveal a substantial down-regulation of any prospective GATA4 target genes and the cardiac manifestation of the knockout (cardia bifida) was attributed to a nonautonomous effect [11,12]. Additionally, as Gata4 -/-ES cells could contribute to the developing heart and express a wide variety of cardiac markers [13], the significance of GATA4 expression in the cardiac compartment remained uncertain. Understanding the role for the GATA4-FOG2 complex (rather than for each protein separately) was facilitated by generating a Gata4 ki line of mice; "ki" is a V217G mutation in GATA4 that specifically cripples the interaction between GATA4 and FOG proteins [14]. Gata4 ki/ki mutants exhibited a similar (although not identical) phenotype to the Fog2 null mutants [14]; this work reinforced the importance of the GATA4 protein and the GATA4-FOG2 interaction for both cardiogenesis, and the development of the cardiac vascular and epicardial cells. Reexamination of the cardiac defects in Gata4 chimeric embryos [15], the recent generation of animals with cardiac-limited ablation of Gata4 that have distinct cardiac-specific defects [16,17], as well as a report connecting mutations in the GATA4 gene to congenital heart defects in humans [18], further confirmed the pivotal role for GATA4 in cardiogenesis.
The absence of the GATA4-FOG2 complex is embryonic lethal at E13.0-13.5 due to heart failure; however, the mechanism by which GATA4-FOG2 interaction loss translates into heart failure remains to be understood. Given that myocardial-restricted Fog2 re-expression is sufficient to rescue cardiac vascular development and extend the life span of the Fog2-null embryos [7,19], we reasoned that GATA4-FOG2 target genes should exist in the myocardium. We have now identified the skeletal troponin Tnnt1 gene as a myocardial target of the GATA4-FOG2 transcriptional complex. This finding was unexpected, since Tnnt1 is mostly expressed in skeletal muscle where TNNT1 forms a part of the skeletal troponintropomyosin complex. Now our data demonstrate that cardiac expression of Tnnt1 requires GATA4-FOG2 interaction.

Affymetrix Microarray Analysis of Gene Expression
Embryonic hearts were dissected from E12.5 wild-type and Gata4 ki/ki or Fog2 -/mutant embryos and transferred to RNAlater solution (Ambion). RNA was isolated with an RNeasy Mini Kit (Qiagen) by standard protocol and subsequently treated with DNAseI (Roche) to remove any possible DNA contamination. DNAseI was heat-inactivated for 15 min at 70 o C, and RNA was precipitated by standard protocol and diluted in 20 µl H 2 O. Affymetrix oligonucleotide arrays were used for RNA expression analysis [20,21]. The array experiment was performed by the Dartmouth Genomic and Microarray Laboratory, according to the standard protocol. The microarray data have been deposited at the GEO database GSE14906 and were analyzed using Gene Traffic Software (Iobion Informatics).

Transgenic Mice
A 5' 2.4-kb Tnnt1 mouse genomic fragment was obtained by PCR using primers tnnt1_2.4F 5'-AAGTTTGAGGGCTGAGCCAT-3' and tnnt1pR GGCTGGGTCCACAGATGCTGTA; the conserved fifth intron of the mouse Tnnt1 gene was similarly generated with primers tnnt1iF 5'-TTGAACTCATAGCAACTCTC-3' and tnnt1_2.4R 5'-TTAAGAGTTAAGGTTGGCTG-3'. To identify cis-regulatory elements responsible for cardiac and skeletal muscle expression of Tnnt1, we fused the 5' fragment upstream of an ATG codone of the lacZ reporter gene in pSDKlacZpA (a kind gift of Janet Rossant), while the intron sequence was subsequently inserted 3' to the LacZ-SV40 polyA sequences to generate pTnnt1-LacZpa. The fragment of pTnnt1-LacZpa containing the Tnnt1 regulatory sequences directing expression from the LacZ-pA reporter was isolated using standard methods [22]. Transgenic animals were generated by the Dartmouth Transgenic and Genetic Construct Shared Resource Center by pronuclear injection into fertilized eggs. F 0 embryos were collected and analyzed at E12.5 by an X-gal staining assay.

Whole Mount In Situ Hybridization
Embryos at various stages were removed from the uterus and their internal organs were removed to expose the heart; alternatively, whole embryonic hearts were dissected out. Embryonic tissues were fixed with 4% paraformaldehyde (PFA) in 1xPBS at 4 o C overnight. Further processing of embryos and in situ hybridization analysis were carried out essentially as described [23]. Tnnt1 and Tnnt2 dig-labeled RNA probes have been generated by RT-PCR from E12.5 hearts. To generate the Tnnt1 probe, we used primers tnnt1pF 5'-GGTCAAGGCAGAACAGAAGC-3' and tnnt1pR 5'-CTCCACACAGCAGGTCATGT-3'; Tnnt2 probe primers are tnnt2pF 5'-CGGAAGAGTGGGAAGAGACA and tnnt2pR 5'-AGCTAAGCCAGCTCCCACTA-3'. Hearts were photographed and images were processed and assembled as previously described [22,24].

Quantitative RT-PCR Analysis
The hearts (ventricles and atriums) were dissected in PBS from E12.5 or E17.5 embryos, and transferred in RNAlater solution (Ambion). RNA was isolated with an RNeasy Mini Kit (Qiagen) in 30 µl of RNAsefree TE buffer. During isolation, RNA samples were treated with RNAse-Free DNAse set for 15 min on RNeasy columns (Qiagen), according to the manufacturer's instructions. Each sample was divided into two aliquots, one of which was reverse transcribed using the M-MLV reverse transcriptase (Invitrogen), following the manufacturer's instructions. The second aliquot was used as a control without reverse transcription to identify and discard samples with DNA contamination. All real-time PCR assays were carried out using SYBR Green Universal PCR Master Mix (Applied Biosystems). The PCR reactions contained 25 ng of cDNA and gene-specific primers at a final concentration of 1 µM each. The assays were run under standard SYBR Green conditions on the ABI 7500 instrument. A standard curve for each gene was generated using serial dilutions of cDNA. Relative expression levels for each sample were determined in the same run and were expressed as the ratio of the RNA amount (of interest) to the amount of control RNA (Gapdh). SYBR Green reactions were performed in duplicates and the experiments were repeated independently at least three times (for at least three samples). Gene-specific primers were designed using the Primer Express software (Perkin Elmer Life Sciences), namely: Gapdh-qRT_F 5'-GCTCACTGGCATGGCCTTCCGTG-3'; Gapdh-qRT_R 5'-TGGAAGAGTGGGAGTTGCTGTTGA-3'; Tnnt1-qRT_F 5'-GGTCAAGGCAGAACAGAAGC-3'; and Tnnt1-qRT_R 5'-GCGGTTGTAGAGCACATTGA-3'

ß-Galactosidase Assay
Embryos were fixed and stained using X-gal essentially as previously described [22]. The staining was continuously monitored until a satisfactory color development was achieved (2-5 h). Embryos were then fixed overnight in 4% PFA in PBS and photographed as previously described [22].

Microarray Analysis of RNA Expression in FOG2 and GATA4 Mutant Hearts
In order to identify the targets of GATA4-FOG2 action in mammalian heart development, we performed Affymetrix microarray comparisons of gene expression in normal and mutant hearts at E12.5. We compared RNA samples from both Fog2 null and Gata4 ki/ki mutant E12.5 hearts to the wild-type control E12.5 hearts. We reasoned that as the phenotypes of the Fog2 knockout and Gata4 ki/ki mutation are similar [7,14], we should expect to identify a similar set of differentially expressed genes in both experiments. As an additional control, we expected to find the Fog2 gene expression absent in the mutant (null) Fog2 cardiac sample, but not Gata4 ki/ki sample.
The microarray profiling yielded surprisingly few gene sets that were differentially represented (~2.5 times up-or down-regulated) in the mutant samples vs. controls. Importantly, the results were consistent between "Fog2" and "Gata4 ki " experiments (similar gene sets were recovered), with the exception of the gene set corresponding to the Fog2 gene that was absent in the Fog2 mutant sample, as we had predicted (Supplemental Table 1; see also [25]). The results of the microarray experiment are available at the GEO database (GSE14906).

Tnnt1 is a Target of the GATA4-FOG2 Transcription Complex
Microarray experiments have identified Tnnt1 as a target of GATA4-FOG2 activation in the heart. Based on the microarray data, the expression of Tnnt1 was down-regulated ~5 times in the Fog2-null sample and ~7 times in the Gata ki/ki sample; Tnnt1 was the "most down-regulated" gene in both mutants (see also Supplementary Table in Smagulova et al. [25]). Tnnt1 was also the only gene of the troponin group that was down-regulated in the Gata4 and Fog2 mutant hearts; other troponins (e.g., Tnni1, Tnnc1, or cardiacrestricted Tnnt2 and Tnni3) were expressed at a similar level in control and mutant GATA4/FOG2 samples (not shown). Given this strong dependence of cardiac Tnnt1 expression on the GATA4-FOG2 interaction, we decided to pursue the analysis of Tnnt1 expression further.

Tnnt1 Expression in Cardiac Development
Expression of the Tnnt1 gene in the rodent heart has been previously documented [26,27] and is consistent with our data. The whole-mount in situ hybridization (WISH) experiment using anti-Tnnt1 RNA as a probe demonstrated that, at E9.5, the embryonic heart of a mouse is positive for Tnnt1, with expression visible in the outflow track and in the forming interventricular groove (Fig. 1A,B). At E10.5, Tnnt1 expression expands posteriorly towards the apex; the expression also appears in the left ventricle ( Fig. 1C-E). However, at E11.5, the expression in the outflow track is down-regulated and by E12, the expression is mostly confined to the ventral interventricular groove with some expression in the left ventricle; from about E12.0, the outflow track cells are negative for Tnnt1 (Fig. 1F). In the E12.5 heart, interventricular Tnnt1 expression expands laterally and by E14.5, the gene is expressed throughout the ventral side of the left ventricle; in the right ventricle, the expression is enhanced in the apical region, while the cells in the outflow track remain negative (Fig. 1G,H). The Tnnt1 atrial expression becomes prominent at E12.5, with the left atrium being more positive.

Tnnt1 RNA Expression in Fog2 Null and Gata4 ki/ki Hearts
Microarray analysis and qRT-PCR both reveal a dramatic down-regulation of Tnnt1 expression upon GATA4-FOG2 interaction loss. In accordance with microarray data, qRT-PCR demonstrated a significant down-regulation of Tnnt1 in E12.5 mutant hearts ( Fig. 2A). WISH corroborated this down-regulation in GATA4-FOG2 mutants (Fig. 2B-D). The residual expression in the Fog2 null E12.5 heart (Fig. 2C) resembles the earlier (~E9.5) wild-type pattern, with positive cells persisting in the outflow track and the apical portion of the interventricular groove. No residual Tnnt1 expression is apparent in the atria or ventricles of the Gata4 mutant (Gata4 ki/ki ) at E12.5 (Fig. 2D). Importantly, the noncardiac expression of Tnnt1 (e.g., in skeletal muscle) remains intact in both Fog2 and Gata4 mutants (data not shown).

Tnnt1 Expression is Increased in α α α αMhc-Fog2 Transgenic Animals
To validate Tnnt1 as a bona fide target of the GATA4-FOG2 complex, we performed additional experiments. Fog2 expression is decreased in the developing heart shortly after E16.5 [4]. Tnnt1 expression was reported to follow a similar trend [26,27]. If the GATA4-FOG2 complex is required for Tnnt1 activity, FOG2 concentration could be limiting and therefore responsible for Tnnt1 downregulation in the late gestation heart. In this case, cardiac Fog2 overexpression should be sufficient for increasing Tnnt1 levels in the heart. To test this possibility, we took advantage of the transgenic mice that express Fog2 under the control of regulatory sequences from the cardiac alpha myosin heavy chain (αMhc) promoter [7]. The αMhc promoter directs expression specifically to cardiomyocytes by E10.5 [28]; the αMhc-Fog2 transgenic animals have been previously described and were successfully used to rescue the lethality of Fog2 -/embryos at ~E14.5 from cardiac pathology [7,19].
As was shown previously [26,27], embryonic Tnnt1 RNA expression is transient in the murine heart and starts to decline after E16.5. Accordingly, WISH with control E17.5 hearts showed low levels of cardiac Tnnt1 expression (Fig. 3A). In contrast, in the E17.5 αMhc-Fog2 transgenics, the Tnnt1 expression level remains high in the atria, the known preferential site of αMhc expression at this stage (e.g., [29]) (Fig. 3B, arrows). qRT-PCR confirms that Tnnt1 levels are elevated in the αMhc-Fog2 neonatal animals compared to the controls (Fig. 3C). This demonstrates that elevating Fog2 levels in cardiomyocytes is sufficient for increasing Tnnt1 expression.

Tnnt1 Expression is Restored in Hearts with Myocardial-Restricted Fog2
As Tnnt1 expression is increased in αMhc-Fog2 transgenics, we sought to determine whether myocardial FOG2 is sufficient to recapitulate (rescue) Tnnt1 cardiac expression in the otherwise Fog2-null fetuses.
During mid-gestation, αMhc promoter directs expression to the ventricular myocardium [28]; hence, the αMhc-Fog2/Fog2 -/fetuses express Fog2 cDNA driven by the αMhc promoter exclusively in the myocardium and are otherwise Fog2 null. In the E13.5 hearts from these "rescued" embryos, the Tnnt1 expression pattern is now restored (Fig. 3D) and appears indistinguishable from that in the contemporaneous wild-type hearts (compare Fig. 3D to Fig. 2B). We conclude that, although Fog2 is expressed in all three cardiac layers, restoring FOG2 function specifically in the myocardium is sufficient to "rescue" Tnnt1 expression.

Proximal DNA Elements are Sufficient to Direct both Skeletal and Cardiac Tnnt1 Expression
The cis-elements that are required to drive Tnnt1 expression in skeletal muscle have not been defined [30]; even less is known about the transcriptional regulation of this gene in the heart. The genomic organization of the human and mouse Tnnt1 gene has been reviewed [30]. The interesting feature of the Tnnt1 gene is its location in very close proximity to cardiac-restricted troponin I (Tnni3); the distance between the Tnnt1 and Tnni3 is only 2.4 kb in the mouse (2.6 kb in the human) [30]. Intriguingly, the gene downstream of Tnnt1 (14.7 kb; Ppp1r12c, Mbs85) is also highly expressed in the hearts of mice [31] and men [32]. However, both of these genetic neighbors are expressed normally in the Gata4-Fog2 mutants, excluding the possibility of coregulation (not shown).
Inspection of the Tnnt1 genomic locus using an ECR (evolutionary conserved region) browser [33] confirmed previously reported genomic organization of the locus [30]; however, outside of the TNNT1 coding sequence, we detected little conservation even between mammals, with the exception of the phylogenetically conserved fifth intron (Supplemental Fig. 1). This was unexpected, as a similar pattern and timing of cardiac expression was reported for a human and rat gene [34] and, hence, a better conservation of the regulatory sequences could have been expected. In order to identify the cis-regulatory elements that are responsible for cardiac-specific regulation of the Tnnt1 gene by the GATA4-FOG2 complex, we generated a LacZ (bacterial β-galactosidase) fusion transgenic construct (Fig. 4A). The construct contained the 2.4-kb region upstream of the Tnnt1 transcription start site with the 5' boundary delimited by the Tnni3 ORF; we also inserted the phylogenetically conserved fifth intron 3' to the LacZ-SV40polyA cassette to generate 2.4-Tnnt1-LacZpa-I5 (Fig. 4A). Transgenic construct was injected into fertilized eggs, and F 0 transgenic embryos were collected and analyzed at E12.5. The 2.4-Tnnt1-LacZpa-I5 sequences were fully sufficient to drive the expression of the LacZ reporter in the somites and developing skeletal muscle in all transgenic embryos (Fig. 4B). In addition to skeletal expression, we also observed cardiac-specific expression of the lacZ reporter in some (but not all) of these F 0 E12.5 embryos (Fig. 4C). While it is possible that other transcription factors are responsible for Tnnt1 regulation, we have not identified any myocardially expressed transcription factors in our microarray experiments. Hence, we reasoned that the GATA4-FOG2 complex is regulating Tnnt1 expression directly through one of the GATA/TATC sequences. Despite numerous attempts, a chromatin immunoprecipitation (ChIP) assay from the E12.5 embryonic hearts did not pull down DNA containing GATA sites within Tnnt1 DNA (not shown). Both antibodies used in this experiment (αGATA4 or αFOG2) were previously used successfully to isolate GATA-containing elements in the Lhx9 regulatory region [25]. Experiments are currently in progress to address the mechanism of the GATA4-FOG2-dependent regulation of the Tnnt1 by using the ES cell in vitro differentiation system.

DISCUSSION
Interaction between GATA4 and FOG2 is required in normal cardiac development; however, the genetic mechanism of GATA4 and/or FOG2 action in the heart, and specifically in the myocardium, is not well understood. It has even been proposed that the myocardial defects in Gata4 null hearts may be secondary to GATA4 loss in the proepicardium [15].
We now identify the Tnnt1 gene as a target of the GATA4-FOG2 complex in the myocardium. The cis-acting elements that are required to drive this gene's expression in skeletal muscle have not been defined [30]; we now show that the 2.4 kb and sequences from intron 5 are sufficient to direct musclespecific expression during mouse embryonic development. While all transgenic embryos express the transgene in the skeletal muscle, cardiac expression was observed in two out of five embryos, suggesting that while additional elements are required for consistent expression in the heart, sequences necessary for Tnnt1 cardiac expression are present within the transgene's regulatory elements. Despite our repeated attempts, the cardiac ChIP assay could not detect a GATA4-FOG2 complex bound to GATA/TATC elements within these regulatory sequences. In light of these negative results, we conclude that Tnnt1 is unlikely to be directly regulated by the GATA4-FOG2 complex in the heart.
A 2.4Tnnt1-LacZpA-I5 reporter we have generated is robustly expressed in embryonic skeletal muscle, thus indicating that the 2.4-kb fragment and intron 5 contain all the elements necessary for skeletal muscle expression. As skeletal muscle does not express FOG molecules, Tnnt1 expression in this tissue has to be independent of GATA-FOG interaction; correspondingly, Tnnt1 is expressed normally in skeletal muscle of the Fog2 null and Gata4 ki/ki mutants. Outside of skeletal muscle, Tnnt1 RNA has been detected in several other tissues; the significance of this extraskeletal Tnnt1 gene expression and its transcriptional regulation are not understood. In the murine and human heart, Tnnt1 RNA expression was previously described; this expression is transient during embryogenesis and starts to decline in mice after E16.5 [26,27]. Expression of TNNT1 was also reported in several examined human embryonic stem cell lines (where it is lost upon differentiation [35]) and in aging hearts [36,37]. Furthermore, Tnnt1 expression is dramatically induced in brains (neurons) of mice treated with ketamine [38]. Expression of Tnnt1 outside of muscle tissue hints at alternative function other than its conventional structural role in the sarcomere.
In humans, intact TNNT1 in skeletal musculature is required to support life: a nonsense mutation in TNNT1 causes an autosomal-recessive Amish Nemaline Myopathy (ANM). The children affected by ANM die of respiratory insufficiency, usually in their second year of life. The mutation (a stop codon) in exon 11 results in a deletion of the last 83 amino acids of the protein, removing the protein-binding modules that are necessary for TNNT1's structural function [39]. Although congestive heart failure has been commonly observed for the ANM patients, no evidence of primary cardiac involvement was reported [39].
While elucidating the function of Tnnt1 (sTnT) in mammalian development and cardiogenesis will have to await this gene's deletion in mice, targeted disruption of its cardiac homologue Tnnt2 (cTnT) is embryonic lethal at around E10 [40,41]. It has been reported that, while nearly lacking a heartbeat, a minor twitching consisting of a few cardiomyocytes was observed in all of E10 and about half of E9 cTnT -/-(Tnnt2 -/-) embryos, suggesting that some developmentally regulated mechanism compensated partially for the lack of TNNT2 [41]. Interestingly, the beating cells were observed in the outflow tract in these embryos, which is the zone of Tnnt1 expression ( Fig. 2A), in line with the author's hypothesis that Tnnt1 is able to compensate for Tnnt2 at this stage.