INTRODUCTION

We have recently reported that ectopic expression of MEF2¹ proteins in responsive cells induces and is required for the specification and differentiation of the myogenic pathway through the induction of the bHLH myogenic gene family (1). Moreover, activation of skeletal muscle genes requires cooperative interaction between MEF2 and the myogenic bHLH proteins. Thus, skeletal myogenesis is mediated by two distinct families of interactive transcription factors either of which can initiate the developmental cascade.

Interestingly, in addition to its expression in muscle cell types (2, 3, 4), the MEF2 genes are also expressed in the nervous system (5, 6, 7), but their role in neurogenesis has not been established. Based on their pattern of expression and the similarities between myogenesis and neurogenesis (8), we propose a general model wherein the MEF2 family participates in the myogenic and neurogenic pathways by cooperative interaction with cell type-specific transcriptional regulators in these cell lineages. These observations raised the possibility that the cell type restricted MEF2 transcription factors might be the partners of the neuronal specific bHLH regulator MASH1 in a manner similar to the interaction between MEF2 and MyoD in myogenesis. We show here that, although different from MyoD in primary structure, the neural specific bHLH protein MASH1 physically interacts with MEF2A and the two together cooperatively activate gene expression.

EXPERIMENTAL PROCEDURES

Mouse teratocarcinoma cell line P19, COS, and the African green monkey kidney cell CV1 were obtained from American Type Cell Culture. P19 cells were maintained in Dulbecco's modified Eagle's medium with 7.5% calf serum and 2.5% fetal bovine serum and differentiated along a neural lineage by exposure to retinoic acid as previously described (9, 10, 11). CV1 and COS cells were cultured in 10% calf serum.

The anti-MASH1 monoclonal antibody was provided by Dr. David Anderson. The anti-MAP2 was gift from Dr. R. Vallee. Rabbit anti-MEF2A antibody was generated in the laboratory as described (1). Goat rhodamine-conjugated anti-rabbit or anti-mouse IgG and fluorescein-conjugated anti-mouse or anti-rabbit IgG were purchased from either Sigma or Boehringer Mannheim, respectively. Anti-GFAP was purchased from Sigma.

P19 cells were treated with retinoic acid and allowed to aggregated for 1 or 2 days. After culture for various days in the absence of retinoic acid, cells were either dispersed by trypsinization and cultured for 1 day, then fixed or fixed directly, and assayed by indirect immunofluorescence as described (5, 6).

P19 cells were differentiated as described above. Whole cell protein extracts were prepared as described elsewhere (5, 6) and Western blot analysis was performed with ECL (Amersham Corp.) reagents following the manufacturer's instruction.

Cells were grown to 40% confluence and transfected by calcium phosphate method as described previously (2). After transfection, cells were shocked with glycerol and generally maintained for 2 days before harvesting. CAT assays were performed as described elsewhere (2).

The in vivo binding assay was performed essentially as described elsewhere (12). COS cells were cotransfected with expression vectors PMT2/MEF2A and pRSV/MASH1. Whole cell protein extracts were prepared as described above. The protein extracts were incubated with anti-MASH1 antibody in NET buffer (50 mM Tris·HCl pH 8.0, 100 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40) in the presence of protein A beads. The immunoprecipitated complexes were then separated on a SDS-PAGE and probed with anti-MEF2A antibody on Western blot analysis as described above. For sequential immunoprecipitation, the experiment was carried as described elsewhere (12). The associated proteins from the initial immunoprecipitation, released by boiling SDS, was reprecipitated with secondary antibodies, and the final precipitates were analyzed by SDS-PAGE.

The in vitro binding assays were performed as previously described (1, 12). ³⁵S-MASH1 or MEF2A generated by in vitro translation was incubated with bacterially expressed GST MEF2A or GSTMASH1bHLH bound to glutathione-agarose beads, respectively, in NET buffer at 4 °C with rocking for 3 h. The protein complexes bound to the agarose beads were washed five times with the same buffer. The resulting precipitated complexes and supernatants were analyzed by SDS-PAGE, and the proteins were visualized by autoradiography.

RESULTS

In initial studies we found that MASH1 and MEF2A were coordinately expressed during neural differentiation. When teratocarcinoma P19 cells are exposed to retinoic acid, they differentiate along a neural lineage, as indicated by expression of various neural markers and acquisition of a neuron-like phenotype (9, 10, 11). Early in differentiation, MASH1 mRNA was transiently expressed and first detected 2 days after retinoic acid addition to the cells (13).² To determine if MEF2A was also induced during P19 neural differentiation, we examined protein extracts for the presence of MEF2A by Western blot analysis using an isoform-specific antibody to MEF2A. Analysis of MEF2A expression by Western blot assay revealed that MEF2A protein was not present before differentiation but was induced concomitantly with MASH1, appearing around day 2 after retinoic acid stimulation (Fig. 1A). However, unlike the transient expression of MASH1, MEF2A showed a prolonged pattern of expression starting on day 2 and continuing at high levels 5 days after the exposure to the inducer. Since the induction of MEF2A coincides with that of MASH1 in P19 cell cultures, we next asked whether MEF2A and MASH1 were coexpressed in the same cells during the initial stages of neural differentiation. Double immunostaining of P19 cells with antibodies to MASH1 and MAP2 was performed at differentiation day 2. Indirect immunofluorescence clearly identified differentiating cells which express both MASH1 and MAP2 (Fig. 1B, panel c), confirming the neuronal identity of the MASH1-positive cells. Immunostaining of retinoic acid-treated cells with antibodies to MASH1 and MEF2A revealed cells positive for both proteins (Fig. 1B, panel a). Although there were also cells in culture which were not stained with either antibody throughout the differentiation process or expressed just one of the two markers (Fig. 1B, panel b), a higher percentage of MASH1-positive cells was found at the early stage of differentiation. At this stage, a high percentage of the MASH1-positive cells were also positive for MEF2A (70%) (Fig. 1B, panel a and Table I). The percentage of MASH1-positive cells declined as cells became more differentiated while the number of MEF2A-positive cells increased. At later stages of the process, only about 5% of MEF2A-positive cells were also MASH1-positive (Table I).

Table I. Expression of MASH1 and MEF2A during retinoic acid (RA)-induced P19 neuronal differentiation P19 cells were treated with RA and differentiated as described under ``Experimental Procedures.'' After different periods of initial RA treatment, cells were costained with anti-MASH1 and anti-MEF2A. The percentage of MEF2A-positive among MASH1-positive cells is indicated for day 3 and 4. For day 5 and 6, percentage of MASH1 positive among MEF2A positive cells is indicated. The number of positive cells counted are as follows: day 3, 193/277; day 4, 246/414; day 5, 58/326; day 6, 25/471. Days after initial RA treatment MEF2A/MASH1 MASH1/MEF2A 3 4 5 6 % % 70 60 18 5

The observation that MASH1 and MEF2A are coexpressed in differentiating P19 cells raised the possibility that neuronal specific MASH1 might cooperate with MEF2A to activate gene expression in a manner similar to the cooperative roles of MEF2 and MyoD gene families in myogenesis (1). To test this hypothesis, we examined MEF2A and MASH1 alone or in combination in a transient transfection assay for their ability to stimulate expression of a CAT reporter gene containing both a MEF2 and a MASH1 binding site (the latter termed as an E box) (14) within its regulatory region (Fig. 2A). Constructs with different basal promoters (TK or embryonic myosin heavy chain genes) were tested, and similar results were obtained (these same constructs have been used to test the cooperative interaction be- tween MEF2 and bHLH protein of the MyoD family) (1, 2). Cotransfection of CV1 cells with the TKCAT reporter construct and the expression constructs for either MEF2A or MASH1 produced no significant change in reporter gene activity (Fig. 2B). However, coexpression of both MEF2A and MASH1 together resulted in a significant increase in CAT reporter activity. That this induction was due to interaction of the respective transcription factors with their DNA binding sites was supported by the demonstration that point mutations in either the E box or the MEF2 binding site, which in DNA mobility shift assays disrupted MASH1 or MEF2A binding, respectively (data not shown), rendered the reporter gene unresponsive to both MASH1 and MEF2A individually or in combination (Fig. 2B).

The observed functional cooperativity between MASH1 and MEF2A raised the possibility that these factors might physically interact with each other in vivo. To test this possibility, we transfected COS cells with expression vectors encoding MASH1 and MEF2A and performed immune coprecipitation experiments (12) using anti-MASH1 antibodies followed by Western blot analysis using a MEF2A-specific antiserum. Antibodies against MASH1 specifically immunoprecipitated MEF2A as indicated by Western blots with anti-MEF2A antibody (Fig. 3A). MEF2A was not detected when control antibody was used for the immunoprecipitation step. Furthermore, of the two MEF2A species expressed in cells, only the faster migrating MEF2A band was coprecipitated with MASH1. The slower migrating MEF2A species, although present in the cellular protein extract prior to immunoprecipitation, was not detected in the MASH1 immunoprecipitation. In a reverse experiment, immunoprecipitation of ³⁵S-Met-labeled cell extracts was carried out with first either anti-MEF2A or preimmune serum and followed by sequential immunoprecipitation with anti-MASH1 antibodies. Only anti-MEF2A antibody specifically coimmunoprecipitated MASH1 (Fig. 3B). These findings provide evidence that MASH1 and MEF2A physically interact within cells and raise the possibility that a post-transcriptional modification of MEF2A regulates its interaction with MASH1.

To identify the domains of the MASH1 and MEF2A proteins that are involved in the protein-protein interaction, we first established an in vitro binding assay using bacterially expressed glutathione S-transferase (GST) fusion protein (Fig. 4A) (12). ³⁵S-MASH1 generated by in vitro translation, when incubated with GSTMEF2A full-length fusion protein bound to agarose beads, bound specifically to the fusion protein but not to GST alone. The lack of binding to GST was not due to the degradation of ³⁵S-MASH1 because the supernatant still contained the intact protein. Previous studies have shown that the bHLH domain mediates protein-protein interactions involving bHLH-containing transcription factors (15, 16). Therefore, we tested the bHLH region of MASH1 for its ability to interact with MEF2A. A GST-MASH1 fusion protein containing only the bHLH region from MASH1 (residues 127-170) (GSTMASH1bHLH) was able to bind MEF2A (Fig. 2B). The interaction with the MASH1 bHLH domain appears to require the MADS region of MEF2A, as MEF2A mutants retaining the MADS domain, but lacking an internal portion including the MEF2 domain (DM57-322), were still capable of binding MASH1 (Fig. 4C). Together, these experiments demonstrated that the physical interaction between MASH1 and MEF2A is mediated through the bHLH motif of MASH1 and MADS domain of MEF2A (Fig. 4D).

DISCUSSION

We show here that MASH1 and MEF2A are coordinately induced and coexpressed in P19 cells during their differentiation along a neuronal lineage. In transient transfection assays, MASH1 and MEF2A cooperatively activate gene expression. The basis of the cooperativity appears to be a specific physical interaction between MASH1 and MEF2A mediated through their conserved bHLH motif and MADS domain, respectively. These findings suggest that these two classes of transcription factors may function as a complex to regulate transcription of genes important for neurogenesis.

These results raise questions regarding the natural gene targets of MASH and MEF2A factors during neurogenesis. Putative MASH1 and MEF2 binding sites are present in the promoters of neural specific genes such as neurological SCL1 and -2, brain-specific type II sodium channel, and neural cell adhesion molecule NCAM (17, 18, 19, 20, 21). Among them, it is particularly interesting to note that a MEF2 site and an E box are present in the regulatory region of NSCL1 and -2 genes, making them candidate targets of the cooperative effects of MASH1 and MEF2A.

The observation that MEF2A and MASH1 interact both functionally and physically may have broad implications for understanding aspects of neural development and differentiation. Several neural specific bHLH proteins and different isoforms of MEF2 are present in various regions of the central and peripheral nervous systems during very early stages of mouse development and also in the adult mouse brain (5, 6, 7, 22, 23). Taken together with the findings described in this report, these observations raise the possibility that neural specific bHLH proteins may interact with MEF2 factors at various stages of neural development. The MASH1 and MEF2A interaction may be an indication of a more general interaction between various members of these two families of transcription factors. Consistent with this idea is the finding that the highly conserved MADS and bHLH domains are required for the interaction between MASH1 and MEF2A and the observation that other neuronal specific bHLH proteins can also specifically interact with different isoforms of MEF2s.² Therefore, it is possible that different combinations of neural specific bHLH proteins and MADS domain-containing proteins may cooperate to regulate gene expression at different stages of development. Such a combinatorial mechanism would allow the production of a large number of different and specific arrays of proteins beginning with a relatively small number of transcription factors.

Given that an interaction between MEF2A and the bHLH protein MyoD has recently been shown to be required for muscle development, our observations predict that certain aspects of the basic mechanisms of transcriptional control of lineage-specific determination may be conserved during development of the mammalian muscle and neuronal system. In addition, normal myogenesis and neurogenesis are dependent upon the presence of functional retinoblastoma protein Rb (24, 25, 26). Taken together, all of these observations suggest a general model for myogenic and neurogenic gene expression as well as for cell lineage determination which requires the interaction of at least three families of regulatory factors with increasingly restricted patterns of gene expression. In this scheme Rb is the universal permissive factor, MEF2 with its multiple gene products is restricted to a limited and well defined set of cell types, while the myogenic and neurogenic bHLH proteins provide the required cell-type specificity.