21 February 2006. Move over CREST and CREB; there are other calcium-activated factors that control the synaptic complement. The February 17 issue of Science draws attention to MEF2 transcription factors, while a recent paper in The Journal of Neuroscience reports that integrins reorganize dendritic spines. These findings may help scientists understand one of the earliest manifestations of Alzheimer disease (AD), synaptic loss (see, e.g., DeKosky et al., 1990). Not only do the MEF2 (myocyte enhancer factor-2) family members appear to modulate activity-dependent proliferation of synapses, but they have also been found fraternizing with other AD-related proteins, most notably, wild-type, but not mutant, amyloid-β precursor protein (AβPP) and the kinase Cdk5.
Activity-dependent control of neuronal synapses plays a major role in both long-term potentiation (LTP) and long-term depression (LTD), neuronal phenomena that underlie neuronal plasticity, learning, and memory. Researchers have known for some years that activation of the N-methyl-D-aspartate form of the glutamate receptor leads to an increase in the number of synapses. Recently, CREST (calcium-responsive transactivator) and CREB (cyclic AMP response element-binding protein) transcription factors were implicated in the process (see Redmond et al., 2002 and ARF related news story), suggesting that nuclear events, in addition to local axonal or dendritic events, contribute to synaptic plasticity. Now, reports that MEF2 family members are involved further emphasize the importance of the nucleus in regulating synapses.
The back-to-back papers in Science make a compelling case that activation of MEF-controlled genes puts a damper on synapses during development in the hippocampus and the cerebellum. Mechanistically, however, MEF2s appear to work in slightly different ways in the two parts of the brain. While Michel Greenberg and colleagues at Children’s Hospital, Boston, report that activation of MEF2 suppresses synapses in the hippocampus, Azad Bonni and colleagues at Harvard Medical School report that development of dendritic claws in the cerebellum is promoted by a suppressor form of MEF2.
Greenberg and colleagues noted that MEF family members MEF2A and MEF2D began to increase in rat brain during the first three weeks after birth. To determine what these proteins might be up to, joint first authors Steven Flavell and Christopher Cowen and their colleagues decided to knock down MEF using RNA interference. They transfected hippocampal neurons with an MEF2 short hairpin RNA just when synapses begin to form (5 days in vitro), then analyzed these neurons for the presynaptic marker synapsin-1 and the postsynaptic marker PSD-95 (postsynaptic density-95). They found that both were dramatically increased after MEF2s were silenced. The number of synaptic contacts, or puncta, that stained for both markers also increased by about twofold. The results were statistically significant.
To determine if these increases have any functional significance, the researchers measured miniature excitatory postsynaptic currents (mEPSCs) in hippocampal neurons treated with the RNAi vectors. Sure enough, they found that the frequency of these currents was almost threefold higher than in the untreated cell cultures. Both the amplitude and the waveform of the mEPSCs were unchanged, however, suggesting, in agreement with the marker data, that the quantity, not the quality, of synapses had increased.
Flavell and colleagues next tested if the MEF2 transcription factors might somehow dovetail with activity-dependent synaptic remodeling. Because activation of the N-methyl-D-aspartate (NMDA) glutamate receptor is thought to play a critical role in this process, the researchers added the NMDA antagonists AP5 (2-amino-5-phosphonopentanoic acid) and CNQX (6-cyano-7-nitroquinoxaline-2,3-dione) to cultures in which the MEF2 genes were silenced. Interestingly, under these conditions the synapse number did not increase, suggesting that NMDA activity is a prerequisite for synapse proliferation. But the researchers also found that glutamate is necessary for transcription of reporters driven by MEF2 regulatory elements and that this was blocked by the calcium channel blocker, nimodipine. The data suggested a scenario whereby glutamate-activated influx of calcium would somehow activate MEF2, which would turn on target genes in the nucleus that would then attenuate glutamate-driven proliferation of synapses. In other words, the MEF2 pathway would counterbalance any drive toward making new synapses.
Flavell and colleagues delved deeper into this pathway, finding that MEF2A was dephosphorylated at several serine residues (Nos. 221, 255, and 408) in response to calcium influx. This was blocked by cyclosporin A and FK506, inhibitors of calcineurin, suggesting that this calcium-binding protein mediated the dephosphorylation. [In contrast, phosphorylation of MEF2 by wild-type AβPP (see Burton et al., 2002) and Cdk5 (see Tang et al., 2005) has been linked to neuronal survival.]
For their part, Bonni and colleagues reported similar results. Joint first authors Aryaman Shalizi and Brice Gaudillière and colleagues also found that depolarization of cerebellar neurons led to dephosphorylation of MEF2 at serine 408, and that this was blocked by nimodipine or cyclosporin A. But unlike in hippocampal cells, when these authors silenced MEF2A in young cerebellar slices, they found that synapse-laden dendritic claws, which are characteristic of granule cells, failed to develop normally. That result suggests that activation of MEF has the exact opposite effect in the cerebellum, stimulating rather than attenuating synapse formation. Opposing effects on hippocampal versus cerebellar synapses tend to be of interest for AD researchers because the former brain area degenerates early in AD, whereas the latter is somehow protected. However, it turns out that the story is more complex.
Shalizi and colleagues noticed that serine 408 of MEF2A lies near a conserved SUMO (small ubiquitin-like modifier) site that is centered around lysine 403. They also found that while an RNAi-resistant version of MEF2 could rescue dendritic claws when expressed in RNAi-treated granule cells, a sumoylation-deficient MEF2 failed in this respect, suggesting that sumoylation of MEF2 is essential for development of the claws. In support of this, the authors detected sumoylated MEF2A in granule cells. They also found that a constitutively active calcineurin all but abolished this modification, as did potassium chloride-induced depolarization of granule cells. These results suggest a slightly different role for MEF2 in the cerebellum. They suggest that calcium-influx leads to dephosphorylation of MEF2, which in turn leads to de-sumoylation. Not only that, but the same events lead to increased acetylation of MEF2 near the SUMO site, suggesting that a sumoylation/acetylation switch might control the impact of MEF2A in the cerebellum.
Sumoylation in Synapses
What would be the role for such a switch? Sumoylation has been shown to turn transcriptional activators into repressors (see Johnson, 2004) and in this case that seems to be exactly what is going on because Shalizi and colleagues found that a MEF2-SUMO fusion protein prevented wild-type MEF2A from inducing transcription. But perhaps more importantly, the authors found that the same MEF2-SUMO fusion increased the number of dendritic claws when transfected into cerebellar slices. All told, the results indicate that silencing of MEF2 in granule cells ablates dendritic claws because it abolishes a normally present repressor effect of sumoylated MEF2. In discussion, AD researchers sometimes mention sumoylation as a candidate post-translational modification that might help explain local effects of synaptic degeneration in subareas of the AD hippocampus, but there is barely any published literature on the topic yet (Li et al., 2003).
Together the two papers paint detailed though subtly different pictures. In both the hippocampus and the cerebellum, MEF2 transcription factors appear to play an important part in activity-dependent synaptic remodeling. In the hippocampus, synaptic activity leads to an influx of calcium, activation of calcineurin, dephosphorylation of MEF2A, and activation of MEF-responsive genes. The bottom line is a reduction in synapse number. In the cerebellum, the events are similar in that calcium influx inhibits dendritic claw development, but this is achieved by de-sumoylating MEF2 and thereby lifting repression off target genes. The major difference appears to be that synapses can form in hippocampal neurons in the absence of MEF2s, but in the cerebellum the repressor function of sumoylated MEF2 is crucial and dendritic claws cannot form without it. But in both cases, MEF2s appear to mediate activity-dependent suppression of synapses, which hints at a role for MEF2s in long-term depression. “Given that calcineurin is also required for the expression of certain forms of LTD, it is possible that MEF2 might also contribute to LTD,” write Flavell and colleagues.
The nature of MEF2 target genes backs up the LTD idea. Both the Greenberg and Azad labs have already begun the work of identifying what lies downstream of MEF activators and repressors. The Greenberg lab used microarrays to look for genes that may be activated by the transcription factors. They mention two that appear to be particularly interesting—Arc (activity-regulated cytoskeletal associated protein) and synGAP (synaptic RAS guanosine trisphosphatase activating protein). On the other hand, the Bonni lab found that MEF2A occupied the Nur77 promoter in granule neurons and that expression of Nur77 is induced by activation of the calcineurin-MEF2 signaling pathway. “It is noteworthy that all of these MEF2 targets act as inhibitors of synaptic differentiation, thereby providing a mechanism for the regulation of synapse number through the MEF2-dependent transcriptional program,” write Asim Beg and Peter Scheiffele, Columbia University, New York, in an accompanying Science perspective. “The exciting new twist in this story is the identification of a calcium-dependent (and hence, activity-dependent) transcriptional program that inhibits rather than promotes the formation of synapses,” they add.
New Ground to Till for AD Researchers?
It is unclear whether the MEF pathways may have any bearing on the pathophysiology of AD or other dementias, but, of course, loss of synapses is the best correlate to cognitive decline (for a review, see Coleman et al., 2004), and the suppression of synapses by the MEF pathway does not appear to be limited to developmental scenarios. Flavell and colleagues demonstrated that in 20-day-old hippocampal cultures, constitutively active MEF2A significantly reduced synapses that had already formed.
“MEF2 target gene products may act differentially on existing synapses by preferentially destabilizing inactive synapses,” suggest Beg and Scheiffele. “Such a mechanism would be well suited for sculpting the connectivity pattern of the brain in response to sensory experience and might contribute to the extensive synapse elimination observed during postnatal life,” they add. In this respect, it is interesting that expression of the MEF2 target gene Arc is disrupted by expression of human amyloid-β precursor protein (see Palop et al., 2005). Arc is one of a number of synaptic proteins AD labs are beginning to explore in an effort to establish a broader panel of functional readouts of AD pathogenesis than the present amyloid-related ones.
As suggested by its name, activity-regulated cytoskeleton-associated protein, Arc interacts with actin/actin binding proteins, which provide the structural framework for synapses. In The Journal of Neuroscience, February 8, Yang Shi and Iryna Ethell, from the University of California, Riverside, report that integrins control dendritic spine plasticity by regulating calcium-activated actin reorganization. The paper adds yet another dimension to calcium-activated regulation of synapses.
Shi and Ethell found that the tripeptide arginine-glycine-aspartate (RGD), a well-characterized integrin ligand, induced elongation of dendritic spines and promoted formation of new filopodia in hippocampal neurons in vitro. These morphological changes were accompanied by reorganization of the actin cytoskeleton—F-actin, normally concentrated in dendritic spines, was mobilized to the dendritic shafts where the number of synapses increased. The data suggest that RGD treatment induces a redistribution of synapses from spines to the dendrite shafts.
The physiological relevance of this is unclear, but like MEF2s, the integrin ligand seems to depend on activation of NMDA receptors to be effective. Shi and Ethell found that RGD failed to remodel synapses if the glutamate receptor or the calcium/calmodulin-dependent protein kinase II (CaMKII) were blocked, once again putting NMDA-dependent calcium influx at the heart of synaptic remodeling.—Tom Fagan.
Flavell SW, Cowan CW, Kim T-K, Greer PL, Lin Y, Paradis S, Griffith EC, Hu LS, Chen C, Greenberg ME. Activity-dependent regulation of MEF2 transcription factors suppresses excitatory synapse number. Science. February 17, 2006;311:1008-1012. Abstract
Shalizi A, Gaudillière B, Yuan Z, Stegmüller J, Shirogane T, Ge Q, Tan Y, Schulman B, Harper JW, Bonni A. A calcium-regulated MEF2 sumoylation switch controls postsynaptic differentiation. Science. February 17, 2006;311:1012-1017. Abstract
Beg AA, Schieffele P. SUMO wrestles the synapse. Science. February 17, 2006;311:962-963. Abstract
Shi Y, Ethell IM. Integrins control dendritic spine plasticity in hippocampal neurons through NMDA receptor and Ca2+/calmodulin-dependent protein kinase II-mediated actin reorganization. J. Neurosci. February 8, 2006;26:1813-1822. Abstract