Transcriptional activity of MEF2 during mouse embryogenesis monitored with a MEF2-dependent transgene

The Myocyte Enhancer Factor-2 (MEF2) family of transcription factors is encoded by four genes, mef2-a, mef2-b, mef2-c and mef2-d, each of which gives rise to multiple proteins via alternative mRNA splicing (Pollock and Treisman, 1991; Yu et al., 1992; Chambers et al., 1992; Leifer et al., 1993; McDermott et al., 1993; Breitbart et al., 1993; Martin et al., 1993, 1994). MEF2 proteins, also referred to as RSRFs (Pollock and Treisman, 1991) (for Related to Serum Response Factors), share about 80% amino acid identity within a MADS (MCM1, Agamous, Deficiens, Serum Response Factor) domain that mediates DNA binding and dimerization, and an adjacent region, called the MEF2 domain, that influences DNA-binding affinity and cofactor interactions (reviewed in Black and Olson, 1998). MEF2 proteins bind the DNA sequence CTA(A/T)₄TAG/A as homodimers and heterodimers and are potent transactivators through that sequence (Gossett et al., 1989; Cserjesi and Olson, 1991). Whereas a single MEF2-binding site is transcriptionally inactive, when the MEF2 site is multimerized or combined with binding sites for heterologous transcription factors, it can direct high levels of transcription (Gossett et al., 1989; Kaushal et al., 1994).

MEF2 was originally defined as a muscle-specific DNA-binding activity that recognized an A/T-rich sequence in the control regions of numerous skeletal and cardiac muscle genes (Gossett et al., 1989). During mouse embryogenesis, MEF2 transcripts are expressed in the cardiac, skeletal and smooth muscle cell lineages, as well as in neural crest cells and specific regions of the brain (Edmondson et al., 1994; Lyons et al., 1995; Subramanian and Nadal-Ginard, 1996; Molkentin et al., 1996; Lin et al., 1998). MEF2B and MEF2C are initially expressed in the cardiac mesoderm at about embryonic day (E) 7.5, followed about a day later by expression of MEF2A and MEF2D. MEF2C transcripts are detected in the somites at E8.5, concomitant with the onset of myocyte differentiation in the myotome. Mice lacking MEF2C die at embryonic day 9.5 due to severe cardiovascular defects (Lin et al., 1997). Whereas many MEF2-dependent genes fail to be upregulated in the hearts of MEF2C null embryos, some cardiac genes known to be regulated by MEF2 are expressed in the absence of MEF2C. These results suggest that another MEF2 factor, probably MEF2B, can substitute for MEF2C to support expression of these genes.

While it is clear that MEF2 factors play important roles in muscle differentiation, there is also evidence for the involvement of MEF2 in the control of nonmuscle genes. In a majority of these cases, the MEF2 family has been shown to activate gene expression in response to mitogenic signaling. A MEF2 site has been identified, for example, in the c-jun promoter, where it mediates transcriptional activation via a MAP kinase signaling cascade (Han et al., 1992; Han and Prywes, 1995; Kato et al., 1997). Similarly, the orphan steroid receptor gene, Nur77, and the Epstein-Barr Virus lytic switch gene, BZLF1, both contain functional MEF2 sites within their promoters that respond to activated calcium signaling pathways (Woronicz et al., 1995; Liu et al., 1997).

MEF2 mRNAs are abundantly expressed in the developing and adult central nervous system (Leifer et al., 1993; Martin et al., 1993; McDermott et al., 1993; Lyons et al., 1995; Lin et al., 1996) arguing for the existence of downstream neuronal target genes. Recently, a functional MEF2-binding site was identified in the promoter of the NMDA (N-methyl-D-aspartate) receptor subunit 1 (NR1) gene (Krainc et al., 1998). MEF2C binds this site and synergistically interacts with Sp1 to activate NR1 expression in cerebrocortical neurons. A putative MEF2 site has also been identified in the promoter of the neurofilament light chain (NF-L) gene (Yaworsky et al., 1997). Moreover, studies have demonstrated that MEF2 proteins activate transcription in collaboration with neurogenic basic helix-loop-helix transcription factors (Black et al., 1996; Mao and Nadal-Ginard, 1996), suggesting an important role for the MEF2 family in neuronal differentiation.

The cell type distribution of MEF2 activity has been controversial. Some studies have reported that MEF2 DNA-binding activity is restricted to muscle cells (Gossett et al., 1989; Cserjesi et al., 1991; Yu et al., 1992; Breitbart et al., 1993), whereas other studies have concluded that MEF2 factors are expressed at comparable levels in muscle and nonmuscle cells (Pollock and Treisman, 1991; Braun et al., 1994; Dodou et al., 1995). This confusion may be due in part to the complexity of MEF2 regulation. For example, it is clear that MEF2 mRNAs are widely expressed and there is evidence for translational regulation of MEF2 expression (Black et al., 1997). In addition, phosphorylation of the DNA-binding and transcription activation domains of MEF2 factors influence DNA binding and transcriptional activity, respectively (Molkentin et al., 1996; Han et al., 1997; Kato et al., 1997; Zhao et al., 1999).

In light of the discordance between MEF2 DNA-binding activity and MEF2 mRNA expression, we sought to determine the pattern of expression of transcriptionally active MEF2 proteins during mouse embryogenesis. Using a lacZ reporter gene under the control of tandem copies of the desmin MEF2 DNA-binding site in transgenic mice, we show MEF2 activity in the developing heart, skeletal and smooth muscle and specific regions of the brain. The pattern of expression of the MEF2-dependent transgene suggests that MEF2 proteins are transcriptionally active in early myogenic and neuronal cell lineages where they regulate tissue-specific gene expression.

Complementary oligonucleotides encompassing the MEF2 site from the mouse desmin enhancer (Kuisk et al., 1996) and human c-jun promoter (Han and Prywes, 1995), as well as a mutant MEF2 site with HindIII sites on both ends were synthesized and annealed. The sequences of the top strands (MEF2 site underlined, mutant MEF2 site in bold) were:

desmin: GGCCTTTCCTTCTCCTCTATAAATACCAGCTCTGG-TATTTCA

mutant MEF2: GGCCTTTCCTTCTCCTCTATAGCGACCAGC-TCTGGTATTTCA

c-jun: GGGGTGACATCATGGGCTATTTTTAGGGATTGACTG-GTAGCA

The oligonucleotides were ligated and cloned into the HindIII site of the vector hsp68-lacZ, which contains a lacZ reporter gene linked to the heat shock protein (hsp) 68 promoter (Kothary et al., 1989). The number of inserts in the resulting plasmids was determined by DNA sequencing. Transgenic mice were generated using constructs containing three tandem copies of each MEF2 site in a head- to-tail orientation. These constructs were referred to as des-MEF2-lacZ, mut-MEF2-lacZ and jun-MEF2-lacZ.

Transgenic mice were created by injection of transgenes into the male pronuclei of fertilized mouse oocytes, as described previously (Cheng et al., 1992, 1993). Briefly, DNA was gel purified and eluted using a QIAEXII kit (QIAGEN). Fertilized eggs from B6C3F1 or FVB female mice were collected for pronuclear injections. lacZ expression was assayed in F₀ embryos or in F₁ offspring that were shown by Southern analysis to carry the transgene.

Embryos were fixed in 2% paraformaldehyde/0.2% glutaraldehyde in PBS on ice for 30 to 90 minutes followed by washing twice in PBS. Embryos were then immersed in X-gal stain (5 mM ferrocyanide, 5 mM ferricyanide, 2 mM MgCl₂, 1 mg/ml X-gal in dimethylformamide) overnight at room temperature. Transgenic embryos were subsequently processed for histology. Brain sections were prepared by perfusing adult mice with 4% paraformaldehyde. Coronal sections were obtained by embedding dissected brains in a 50% chicken egg albumin/2.5% glutaraldehyde protein matrix (Herzog and Brosamle, 1997) and sectioned with a vibratome at 400 μm prior to staining. For detailed localization of β-galactosidase expression, brain slices were dehydrated, embedded in paraffin and sectioned at 5 μm.

To monitor the expression of functional MEF2 factors during mouse embryogenesis, we created transgenic mouse lines harboring a lacZ reporter gene linked to the hsp 68 gene promoter and three tandem copies of the MEF2 site from the desmin enhancer in a head- to-tail orientiation (des-MEF2-lacZ) (Fig. 1). Of 20 independent transgenic lines that carried the integrated reporter gene, 3 expressed lacZ. Similar expression patterns were observed in these lines, except that two of the lines expressed lacZ in the neural tube, whereas one did not. Several independent F₀ founder transgenic mice showed the same expression pattern in muscle lineages as these stable lines. Subtle variations in expression among the lines were observed at certain developmental stages (not shown), which probably reflect position effects that depend on the sites of transgene integration. Consistent with previous results, the hsp-lacZ transgene lacking MEF2 sites was not expressed at detectable levels in embryos, neonates or adult mice (Kothary et al., 1989; data not shown). Moreover, two independent F₀ founder transgenic mice that carried the hsp-lacZ reporter gene harboring a mutant MEF2 site (mut-MEF2-lacZ) expressed lacZ in a non-specific fashion, demonstrating that des-MEF2-lacZ detects MEF2 activity.

MEF2B and MEF2C are the first members of the MEF2 family to be expressed during mouse embryogenesis, with transcripts appearing in cardiac progenitor cells of the anterior lateral plate mesoderm by day 7.5 p.c. (Edmondson et al., 1994; Molkentin et al., 1996). Expression of the des-MEF2-lacZ transgene was first detected at day 7.5 p.c. in these cardiogenic precursors (Fig. 2A), suggesting that MEF2B or MEF2C proteins are present in these cells and are responsible for activating expression of the transgene. Expression of the transgene was not homogeneous throughout the cardiac crescent, but rather, localized to clusters of cells near the medial region of the cardiac crescent. By day 8.5 p.c., des-MEF2-lacZ was expressed throughout the looping heart tube (Fig. 2B), consistent with the homogeneous expression of the four mef2 genes. Staining was evident in the aortic sac, conotruncus, the future right and left ventricles, and the developing atria (Fig. 2C). The expression pattern of the transgene continued to be superimposable over that of MEF2 mRNAs at this stage.

At day 9.5 p.c., the des-MEF2-lacZ transgene continued to be expressed uniformly throughout the developing heart (Fig. 3A) and, by day 11.5 p.c., expression was detected in both atria and ventricles of the heart. Transverse sections through X-gal-stained embryos showed restricted expression in the myocardium of all four chambers of the heart (Fig. 2D), with no expression in the endocardium. Expression in the heart continued through day 14.5 p.c. (Fig. 2E), after which time it began to decline to low but detectable levels into adulthood (not shown). Expression of the endogenous MEF2 genes also declines in the heart by this stage (Edmondson et al., 1994).

Expression of the MEF2-dependent lacZ transgene was detected in the rostral somites beginning at day 9.0-9.5 p.c. (Fig. 3A), which corresponds to the time in which MEF2C transcripts appear in these cells (Edmondson et al., 1994). The metameric expression pattern progressed caudally through day 11.5 p.c., contiguous with somite maturation (Fig. 3B). Transverse sections of E11.5 embryos showed specific lacZ expression throughout the myotomal compartment within the somite (Fig. 3D,E). As the somites matured, β-gal staining was detected in differentiated myocytes that extended longitudinally through the myotomes (Fig. 3F). There was no expression of lacZ in the sclerotomes or dermatomes (Fig. 3D). A similar expression pattern has been observed for MEF2C mRNA.

At day 11.5 p.c., lacZ expression was also observed in cells throughout the limb buds (see Fig. 3B). We are uncertain whether these cells represent a specific lineage. However, by day 12.5 p.c., expression of lacZ in the limb buds was restricted to differentiating skeletal muscle fibers, which suggests that the earlier expression in the limb buds marks skeletal muscle precursor cells that migrate from the ventrolateral region of the somite.

Expression of the transgene was demonstrable in mature skeletal muscle fibers throughout the embryo at day 14.5 p.c. MEF2 mRNAs are expressed at high levels in skeletal muscle at this stage (Edmondson et al., 1994). As shown in Fig. 3C, transgene expression could be readily detected in the trapezius, lattisimus dorsi and deltoid embryonic muscles. Sagittal sections through an embryo at day 15.5 p.c. revealed staining distributed homogeneously along the entire length of all muscle fibers (Fig. 4A). Expression persisted in neonatal (Fig. 4B) and adult skeletal muscle (not shown). Interestingly, staining of adult hindlimb muscle from a 2-month-old transgenic animal showed localized expression in a subset of muscle fibers. Whether, this transgene is being expressed in a fiber-type-specific manner is currently under investigation.

MEF2C transcripts are also expressed in vascular smooth muscle cells throughout development (Edmondson et al., 1994; Lin et al., 1998). Expression of des-MEF2-lacZ was detected in a subset of vessels in sectioned day 15.5 p.c. embryos (Fig. 4A) and in the adult (not shown). Specific staining was observed in arteries but not veins of des-MEF2-lacZ transgenic animals. The reason for this restricted expression in the vasculature is unclear. Finally, we failed to detect transgene expression in the vascular plexus of E9.5 yolk sacs (not shown) in contrast to previously reported expression of MEF2C mRNA in this region (Lin et al., 1998).

Numerous studies have documented the importance of DNA sequences flanking MEF2 sites for muscle gene activation (Yu et al., 1992; Black and Olson, 1998). DNA-binding site selection experiments have demonstrated that MEF2 proteins in neuronal nuclear extracts preferentially bind sequences immediately adjacent to and including the core MEF2 site whereas MEF2 DNA-binding activity from muscle extracts does not exhibit these sequence constraints (Andres et al., 1995). The activity of other MADS-box proteins is also dependent on sequences adjacent to their binding sites. To begin to determine whether the specific expression pattern of des-MEF2-lacZ was dependent solely on MEF2 factors or, alternatively, was also influenced by other factors that bound surrounding DNA sequences contained in the multimerized MEF2 site, we created a transgene, called jun-MEF2-lacZ. This transgene contained three tandem copies of the MEF2 site from the c-jun promoter, along with surrounding DNA sequence. When this transgene was used to generate F₀ transgenic mice, it showed the same expression pattern as des-MEF2-lacZ (compare Fig. 5 with Fig. 3B). Since desmin and c-jun are normally expressed in entirely different patterns, these results suggest that the expression of these transgenes is controlled by the levels of transcriptionally active MEF2 protein, not by gene-specific cofactors that recognize sequences flanking the MEF2 sites.

Beginning at about 11.5 p.c., MEF2 transcripts begin to be expressed in highly specific patterns throughout differentiating neurons in the brain (Leifer et al., 1993; Lyons et al., 1995). The temporospatial pattern of MEF2 gene expression is coincident with neuronal maturation and suggests an important role in development of the central nervous system. The identification of a functional MEF2 site in the NR1 gene (Krainc et al., 1998) supports the notion that MEF2 proteins are indeed involved in neuron-specific gene expression. A putative MEF2 site has also been identified in the promoter of the NF-L gene (Yaworsky et al., 1997). Furthermore, MEF2 has been shown to synergistically interact with members of the neurogenic family of basic helix-loop-helix transcription factors (Black et al., 1996; Mao and Nadal-Ginard, 1996). Consistent with the proposed role of MEF2 proteins in neuronal differentiation, the des-MEF2-lacZ transgene was expressed in a pattern correlating with that of endogenous MEF2 gene expression in the nervous system.

Coronal sections through the brain of a 2-month-old des-MEF2-lacZ transgenic animal revealed a striking and highly complex pattern of lacZ expression. Prominent expression was observed in the internal pyramidal cell layers V and VI of the cerebral cortex (Fig. 6A-C). In the postnatal murine brain, all four MEF2 mRNAs are expressed throughout the cerebral cortex with MEF2C gene transcripts being more abundant in layers V and VI (Lyons et al., 1995). Staining was also detected in the piriform cortex (Fig. 6A,B), as well as the optic nerve (Fig. 6A), optic chiasm (Fig. 6B) and optic tracts (Fig. 6C). Consistent with expression in optic neurons, the transgene was also expressed in the developing neural retina (not shown). Progressing caudally, expression became clearly evident in the dorsal and medial thalamic nuclei and the stria medularis (Fig. 6B). Cells within the amygdala, fasciculus retroflexus and the habenular nuclei of the diencephalon (Fig. 6C) also expressed the transgene. The most caudal region of the brain demonstrated expression in the vestibular nuclei of the lateral medulla and in cells within the granular cell layer of the cerebellum (Fig. 6D). No staining was evident in the molecular or Purkinje cell layers of the cerebellum.

Additional regions of expression in the adult brain included the olfactory bulbs, superior colliculus, raphe nuclei and pontine nuclei (data not shown). However, we failed to detect any transgene expression in the hippocampus or dentate gyrus (Fig. 6B,C), in contrast to previously reported MEF2 mRNA expression in these regions of the adult brain (Lyons et al., 1995). At earlier developmental stages, we observed expression in scattered neurons within the mesencephalon/ midbrain (Fig. 3A), in the trigeminal ganglia (Fig. 3A,B) and within the ventral region of the neural tube that gives rise to motor neurons (Fig. 3E), demonstrating that MEF2 is trancriptionally active even at the earliest time points of neuronal differentiation. In conclusion, our results illustrate that MEF2 factors are transcriptionally active in the developing and adult brain and are likely performing important functions in neuronal maintenance.

We stained a variety of adult tissues from des-MEF2-lacZ transgenic mice for lacZ expression. In contrast to the high level of lacZ expression seen in developing and adult skeletal muscle and brain, there was no appreciable β-gal staining in other adult tissues, including the heart (data not shown).

The results of this study demonstrate that a multimerized MEF2 site linked to lacZ is a sensitive and specific indicator of transcriptionally active MEF2 factors during embryogenesis and provide several insights into the regulation of MEF2 expression and function. (1) In the embryo, expression of the MEF2-dependent transgene and, by inference, transcriptionally active MEF2 protein, is restricted primarily to developing muscle cell lineages and the brain, correlating with sites of highest MEF2 mRNA expression (Edmondson et al., 1994; Lyons et al 1995). However, during late fetal development, when MEF2 mRNAs become widely expressed, the MEF2-dependent transgene continues to show a restricted expression pattern, suggesting that post-transcriptional mechanisms limit the expression of transcriptionally active MEF2 protein to specific lineages during embryogenesis. (2) A MEF2 site from the desmin gene, which is expressed only in muscle cells, can, when multimerized, direct expression in developing brain as well as muscle and a MEF2 site from the c-jun promoter can direct expression in a similar fashion. Thus, these MEF2 sites and immediately adjacent sequences do not discriminate between different cell types that express MEF2 factors. (3) Since the expression patterns of the transgenes do not correlate with any specific MEF2 isoform, our results indicate that the desmin and c-jun MEF2 sites do not discriminate between the various MEF2 proteins and are equally capable of serving as platforms for transcriptional activation by all MEF2 family members in vivo.

MEF2 sites are contained in the promoters of muscle-specific genes that show lineage-restricted and temporal-specific expression patterns (Black and Olson, 1998). For example, desmin is expressed in cardiac, skeletal and smooth muscle cells. Expression of desmin in skeletal and cardiac muscle is dependent on a muscle-specific enhancer located at about −1.0 kb, which contains a single, essential MEF2-binding site (Li et al., 1993; Kuisk et al., 1996). Desmin expression in smooth muscle cells is controlled by a separate enhancer that does not contain a MEF2 site. Myogenin is expressed exclusively in the skeletal muscle lineage beginning at E8.5. The myogenin promoter contains an essential MEF2 site shown to mediate a positive feedback loop between myogenin and MEF2 (Edmondson et al., 1992; Cheng et al., 1993; Yee and Rigby, 1993). The MRF4 gene is also expressed specifically in skeletal muscle and contains a MEF2 site in its promoter (Black et al., 1995; Naidu et al., 1995). MRF4 is expressed late relative to myogenin. Transcription of the MCK and MLC2v genes in skeletal and cardiac muscle, respectively, is controlled by muscle-specific enhancers that also contain MEF2 sites (Navankasattusas et al., 1992; Amacher et al., 1993; Lee et al., 1994). Thus, MEF2 binding alone cannot account for different expression patterns of muscle genes.

There are numerous examples of MADS-box proteins that interact with cofactors to control specific patterns of gene expression. One of the best characterized examples is serum response factor (SRF), which recruits several cofactors to the c-fos promoter through protein-protein interactions (Dalton and Treisman, 1992; Grueneberg et al., 1992). In many cases, these cofactors bind sequences immediately adjacent to the CArG box. Because the sequences surrounding the MEF2 sites in the above muscle genes are distinct, we investigated whether these sequences in collaboration with MEF2 might interact with unique sets of cofactors that conferred the distinct expression patterns to these genes. Our results show that tandem copies of MEF2 sites from muscle and non-muscle genes with different expression patterns direct the same expression pattern of a linked lacZ transgene. Our interpretation of these findings is that these reporters monitor the levels of transcriptionally active MEF2 factors in cells and do not rely on tissue-specific DNA-binding cofactors for expression. However, we cannot formally rule out the possibility that MEF2 proteins interact with non-DNA-binding cofactors to augment and perhaps restrict transcriptional activity in muscle cells and in neurons. The behavior of these transgenes differs from that of native muscle gene control regions, which contain single MEF2 sites. In those cases, MEF2 is important for expression, but other factors are also required.

Interestingly, the expression patterns of our transgenes differ from that reported by Ross et al. (1996) in which two copies of the MEF2 site from the MLC2v gene along with flanking sequences were reported to direct expression of a lacZ reporter to the ventricles of the developing heart. The flanking sequences described in that study contain binding sites for widely expressed positive and negative trans-acting factors (Navankassattusas et al., 1992, 1994; Lee et al., 1994). Furthermore, the core MEF2 site may also harbor sequences that bind an additional tissue-restricted nuclear factor (Zhu et al., 1993). It is likely that combinatorial interactions between these broadly expressed factors and MEF2 proteins influence the overall expression pattern of their transgene. It is noteworthy that the sequences flanking the MEF2 sites from the desmin and c-jun promoters do not show any substantial homology with each other or with sequences surrounding the MLC2v MEF2 site.

Previous studies have demonstrated that the four mef2 genes are abundantly expressed in the developing and adult nervous system (Leifer et al., 1993; Martin et al., 1993; McDermott et al., 1993; Lyons et al., 1995; Lin et al., 1996). We have extended those observations by showing that the des-MEF2-lacZ transgene is expressed in a highly complex fashion in the developing and adult brain. These results demonstrate for the first time that MEF2 factors are indeed transcriptionally active in mature neuronal cells in vivo. Cooperative interaction between MEF2 proteins and members of the neurogenic family of basic helix-loop-helix transcription factors suggests that MEF2 is a regulator of neuron-specific gene expression (Black et al., 1996; Mao and Nadal-Ginard, 1996). The identification of a functional MEF2 site in the promoter of the NR1 gene (Krainc et al., 1998) validates the involvement of MEF2 in neuron-specific gene expression. A putative MEF2 site has also been identified in the promoter of the NF-L gene (Yaworsky et al., 1997). However, the biological significance of that site remains to be determined. Thus, it is becoming increasingly apparent that MEF2 proteins are important components in neuronal differentiation. Our results reinforce the notion that members of the MEF2 family of transcription factors are likely performing critical functions in neuronal development and maintenance; hence, the role of MEF2 in the brain warrants further investigation.

While the expression of our MEF2-dependent transgenes reveals high levels of transcriptionally active MEF2 protein in developing muscle and neural cell lineages, it is well established that MEF2 factors are expressed in nonmuscle cell types, where they mediate gene activation in response to MAP kinase signaling (Han and Prywes, 1995; Kato et al., 1997; Zhao et al., 1999). As a consequence, our MEF2-dependent transgenes reveal cell types with high levels of MEF2 activity, but the sensitivity of the assay and/or transgene reporter does not detect those cell types that express lower, basal levels of MEF2. It will be of particular interest to use these MEF2-lacZ mice to monitor post-translational regulation of MEF2 activity in vivo in response to various stimuli.

There are numerous examples in which transcription factors are rendered inactive by post-translational mechanisms. Thus, the expression of a transcription factor in a particular cell does not necessarily indicate that the factor is transcriptionally active. The use of transgenes linked to multimerized binding sites such as those reported here may be generally useful for detecting transcription factors that are transcriptionally active in the embryo.

We thank Alisha Tizenor for graphics, Robert Webb and John Shelton for histology, Brian Mercer for pronuclear injections, and members of the Olson lab for valuable input during the course of this work. Supported by grants from the NIH, the Muscular Dystrophy Association, the American Heart Association and The Robert A. Welch Foundation to E. N. O and by PO1 HL49953. F. J. N. was supported by a NIH postdoctoral fellowship.