MEF2 structure and expression
Vertebrates have four different MEF2 proteins (MEF2A–D), all of which are composed of three functional domains: a highly conserved N-terminal MCM1-Agamous-Deficiens-Serum response factor (MADS) box domain, a conserved MEF2 domain and a highly divergent C-terminal transcriptional activation domain. The MADS box is primarily responsible for DNA binding to an A/T-rich domain in MEF2 target genes (Andres et al. 1995; Molkentin et al. 1996b), whereas the MEF2 domain regulates DNA-binding affinity and cofactor interactions (Black & Olson 1998; McKinsey et al. 2002) as well as heterodimerization and homodimerization with other MEF2 proteins (Molkentin et al. 1996a; Yu 1996). The C-terminal transcriptional activation domain differs among the MEF2 proteins and undergoes complex alternative splicing depending on the tissue (Martin et al. 1994; McDermott et al. 1993), resulting in wide functional diversity.
In both the developing and adult vertebrate brain, the four MEF2 proteins show distinct, but overlapping, expression patterns (Heidenreich & Linseman 2004; Ikeshima et al. 1995; Leifer et al. 1993, 1994; Lin et al. 1996; Lyons et al. 1995). MEF2 proteins are expressed throughout the adult brain, including regions involved in formation of different types of memory such as the cortex, hippocampus, amygdala and striatum. Expression of these MEF2 subtypes appears to be differentially regulated in a region- and cell-specific manner. For example, all four MEF2 subtypes are present in the dentate gyrus of the hippocampus, but only MEF2A, C and D are expressed in the thalamus and cerebellum (Lyons et al. 1995). Within the cerebellum, MEF2C mRNA appears to display strongest expression in Purkinje cells, whereas MEF2A and D mRNA are preferentially expressed in the granule cell layer. A summary of MEF2 expression in different brain regions involved in memory formation is shown in Table 1.
Table 1. Relative expression of MEF2 in brain regions involved in memory formation
|Hippocampus|| || || || |
|Striatum (incl. nucleus accumbens)||++||−||+||++|
|Cerebellum|| || || || |
|Purkinje cell layer||+||−||+||+|
|Granule cell layer||++||−||−||+++|
Regulation of MEF2 activity
The regulation of MEF2 activity is complex. MEF2-dependent gene transcription depends not only on the nature of extracellular stimuli (McKinsey et al. 2002) but also on the complement of MEF2 isoforms expressed within a cell (including alternatively spliced variants), isoform dimerization, regulated degradation and phosphorylation state (Ornatsky & McDermott 1996; Yin et al. 2012). Indeed, phosphorylation and dephosphorylation of MEF2 regulate a large number of processes related to MEF2 function, including modulation of DNA-binding affinity, association with transcriptional co-regulators or co-repressors, acetylation and sumoylation, nuclear and cytoplasmic trafficking and caspase-mediated degradation (Heidenreich & Linseman 2004; Li et al. 2001; Shalizi et al. 2006; Yin et al. 2012) (Fig. 1).
Figure 1. Regulation of MEF2 expression and activity. Schematic illustrating some of the ways in which MEF2 can be regulated in neurons. In blue are modulatory elements that act on MEF2 to increase MEF2 activity, while modulations that decrease MEF2 activity are shown in red. Increased or decreased MEF2 activity can be manifested through alterations in DNA-binding affinity (PKA, CKII and GSK3β), transcriptional activity (p38, ERK5 and CaMKIV), shuttling to the nucleus (PP2B action on MEF2C), regulation of the acetylation (Ac) or sumoylation (Sm) of a specific C-terminal residue (PP2B and cdk5) in MEF2A/D, lysosomal degradation (HSc70 binding) and caspase-mediated degradation. In addition, HDAC binding, which represses MEF2 activity, can be promoted or inhibited by PP1α or CaMKIIα, respectively.
Download figure to PowerPoint
There are eight reported phosphorylation sites in the transactivation domain of MEF2A, demonstrating the diverse regulation of MEF2 activity by various kinases and phosphatases (Shalizi & Bonni 2005). Phosphorylation in the transactivating domain by p38 MAP kinases (MEF2A and C) (Han et al. 1997; Wu et al. 2000a) or ERK5 (MEF2A, C and D) (Kato et al. 2000) increases the transcriptional activity of MEF2, whereas both GSKβ (MEF2D) and cdk5-mediated phosphorylation (MEF2A, C and D) reduce MEF2-dependent transcription (Gong et al. 2003; Wang et al. 2009). Notably, phosphorylation at Ser408/444 of MEF2A/D by cdk5 inhibits MEF2 transcriptional activity in HEK-293T cells (Pulipparacharuvil et al. 2008) and cultured neurons (Gong et al. 2003) and pMEF2 has been used as a marker for reduced MEF2 activity in the brain (discussed below). Also, cdk5 phosphorylation promotes sumoylation of MEF2 in vitro (Lys403/439, MEF2A/D) (Gregoire et al. 2006; Shalizi et al. 2006). Sumoylated MEF2A is a transcriptional repressor (Shalizi et al. 2006) and it has been suggested (Gregoire et al. 2006) that sumoylation may also increase the caspase-mediated degradation of MEF2A and D (but not MEF2C) (Tang et al. 2005). Outside of the transactivation domain, both casein kinase II and protein kinase A phosphorylation increase the DNA-binding affinity of MEF2 (Du et al. 2008; Molkentin et al. 1996b, Wang et al. 2005).
Calmodulin-dependent protein kinases (CaMK) also modulate MEF2 function, both directly and indirectly. While MEF2D appears to be directly phosphorylated in vitro by CaMKIV, resulting in increased transcriptional activity (Blaeser et al. 2000), there is an indirect role for a CaMK-mediated increase of MEF2 activity through modulation of HDACs. HDACs can bind to MEF2 and repress MEF2 activity. CaMK-mediated phosphorylation of HDACs (and possibly binding to HDACs directly) results in dissociation of HDACs from MEF2 (McKinsey et al. 2000a), allowing HDAC to be exported from the nucleus through interaction with 14-3-3 protein (Grozinger & Schreiber 2000; McKinsey et al. 2000b). The newly liberated MEF2 is then free to interact with transcriptional co-activators and promote MEF2-dependent gene transcription. Interestingly, CaMKI and CaMKIV have been implicated in MEF2 derepression in muscle cells (Kao et al. 2001; Lu et al. 2000), whereas in cultured neurons, evidence suggests the involvement of αCaMKII (Linseman et al. 2003), a kinase critical for synaptic plasticity and memory formation, in MEF2-dependent transcription. In addition to multiple phosphorylation sites, MEF2 may also be differentially regulated by phosphorylation state depending on the tissue- or cell-type. In T lymphocytes, p38-mediated phosphorylation of three residues, T293, T300 and T387, is necessary for MEF2C action (Han et al. 1997). However, phosphorylation of only T293 by p38 is required for MEF2C activation in differentiating myocytes (Wu et al. 2000b).
MEF2 activity is also regulated by several phosphatases. Protein phosphatase 2B (PP2B) or calcineurin is a calcium-dependent serine/threonine phosphatase implicated in some forms of synaptic depression (Wang & Kelly 1997) and which negatively regulates memory formation (Malleret et al. 2001; Mansuy et al. 1998). While calcineurin interacts with multiple transcription factors, one mechanism by which calcineurin may negatively regulate memory is via dephosphorylation of MEF2 (Ser408/444, MEF2A/D) (Flavell et al. 2006; Gregoire et al. 2006; Shalizi et al. 2006). Calcineurin can increase MEF2 binding to DNA and MEF2-dependent transcription (Cohen & Greenberg 2008; Greer & Greenberg 2008; Mao & Wiedmann 1999). Dephosphorylation of MEF2 by calcineurin also triggers a switch from sumoylation to acetylation of Lys403 in MEF2A (Shalizi et al. 2006). The increase in MEF2-dependent transcription reduces synapse number in cultured hippocampal neurons (Flavell et al. 2006) and inhibits dendritic morphogenesis in cultured cerebellar granule cells (Shalizi et al. 2006). Interestingly, MEF2C appears to be regulated by calcineurin in a distinct manner from MEF2A/D as calcineurin action on a residue specific to MEF2C (Ser412) results in increased nuclear accumulation (Lynch et al. 2005). Protein phosphatase 1α (PP1α), a phosphatase involved in long-term synaptic potentiation and memory formation, binds to MEF2 as well and recruits class II histone deacetylases (HDAC4 isoform) to MEF2 transcription complexes (Perry et al. 2009). In this way, PP1α can override calcineurin modulation and repress MEF2 activity.
In addition to phosphorylation, MEF2 function is regulated by decreasing levels of MEF2 protein itself. MEF2 proteins may be degraded by chaperone-mediated autophagy (CMA) (Yang et al. 2009). In a dopaminergic progenitor neuronal cell line, shuttling of MEF2D to the cytoplasm (which occurs under basal conditions) and subsequent interaction with the chaperone heat shock protein HSc70 promotes MEF2D lysosomal degradation. Inhibition of this CMA process results in accumulation of inactive MEF2D in the cytoplasm and increased cell death. Notably, a reduction of MEF2D-Hsc70 binding and increased cell death also occurs by increasing either wild-type α-synuclein or a mutated α-synuclein (A53T mutant) that is implicated in Parkinson's disease (PD)-associated neurodegeneration. Furthermore, increased levels of MEF2D in the brain are found in transgenic mouse model of PD (A53T α-synuclein mutant) as well as patients with PD (Yang et al. 2009). Therefore, the regulation of MEF2 activity is complex and depends on multiple factors.
Diversity of MEF2 signaling
The complexity of upstream signaling of MEF2 is mirrored by an equally complex downstream function; MEF2 stimulates the transcription of a wide array of genes. Flavell et al. (2008) used an elegant genome-wide targeting strategy and identified more than 180 activity-dependent MEF2 targets, including a large complement of genes important for synapse development and function. Interestingly, this screen detected genes thought to contribute to both synapse weakening (e.g. Homer 1a, kcna1 and kcna4 potassium channels) as well as synapse strengthening (e.g. bdnf and adenylyl cyclase 8). MEF2 targets also included genes that encode proteins involved in both excitatory and inhibitory synapses as well as neurotransmitter release. Collectively, this diverse array of genes that are involved in seemingly distinct neuronal processes suggest a sophisticated and multifaceted role for MEF2 that likely depends on factors such as age (development vs. adulthood), behavioral context (learning vs. basal conditions) as well as the MEF2 subtype involved. An example of this complexity comes from a study that showed differential regulation of BDNF expression depending on the MEF2 subtype involved. Specifically, a MEF2C splice variant lacking the γ-domain (which is particularly sensitive to membrane depolarization in rat cortical neurons in culture) was shown to selectively regulate increased expression of bdnf exon IV, whereas MEF2D limited depolarization-induced expression of bdnf exon I (Lyons et al. 2012).