Summary Mutations in MECP2 underlie the neurodevelopmental disorder Rett (RTT) syndrome. One hallmark of RTT is relatively normal development followed by a later onset of symptoms. Growing evidence suggests an etiology of disrupted synaptic function, yet it is unclear how these abnormalities explain the clinical presentation of RTT. Here we investigate synapse maturation in Mecp2-deficient mice at a circuit with distinct developmental phases– the retinogeniculate synapse. We find that synapse development in mutants is comparable to that of WT littermates between postnatal days 9–21, indicating that initial phases of synapse formation, elimination and strengthening are not significantly affected by MeCP2 absence. However, during the subsequent experience-dependent phase of synapse remodeling, the circuit becomes abnormal in mutants as retinal innervation of relay neurons increases and retinal inputs fail to strengthen further. Moreover, synaptic plasticity in response to visual deprivation is disrupted in mutants. These results suggest a crucial role for Mecp2 in experience-dependent refinement of synaptic circuits.

In the developing visual system, correlated presynaptic activity between neighboring retinal ganglion cells (RGC) stabilizes retinotopic synapses via a postsynaptic NMDAR (N-methyl-D-aspartate receptor)-dependent mechanism. Blocking NMDARs makes individual axonal arbors larger, which underlies an unsharpened map, and also increases branch turnover, as if a stabilizing factor from the postsynaptic partner is no longer released. Arachidonic acid (AA), a candidate retrograde stabilizing factor, is released by cytoplasmic phospholipase A2 (cPLA2) after Ca(2+) entry through activated NMDARs, and can activate presynaptic protein kinase C to phosphorylate various substrates such as GAP43 to regulate cytoskeletal dynamics. To test the role of cPLA2 in the retinotectal system of developing zebrafish, we first used PED6, a fluorescent reporter of cPLA2 activity, to show that 1-3 min of strobe flashes activated tectal cPLA2 by an NMDAR-dependent mechanism. Second, we imaged the dynamic growth of retinal arbors during both local inhibition of tectal cPLA2 by a pharmacological inhibitor, arachidonic tri-fluoromethylketone, and its suppression by antisense oligonucleotides (both injected intraventricularly). Both methods produced larger arbors and faster branch dynamics as occurs with blocking NMDARs. In contrast, intraocular suppression of retinal cPLA2 with large doses of antisense oligos produced none of the effects of tectal cPLA2 inhibition. Finally, if AA is the retrograde messenger, the application of exogenous AA to the tectum should reverse the increased branch turnover caused by blocking either NMDARs or cPLA2. In both cases, intraventricular injection of AA stabilized the overall branch dynamics, bringing rates down below the normal values. The results suggest that AA generated postsynaptically by cPLA2 downstream of Ca(2+) entry through NMDARs acts as a retrograde signal to regulate the dynamic growth of retinal arbors.

Visual activity acts via NMDA Receptors to refine developing retinotectal maps by shaping retinal arbors. Retinal axons add and delete transient branches, and the dynamic rates increase when MK801 blocks NMDARs, as if this prevents release of a stabilizing signal. Ca(++) entry through NMDARs activates phospholipase A2 (cPLA2) to release arachidonic acid (AA), which taps into a presynaptic growth control mechanism. NCAM, L1, N-cadherin, and FGF all stimulate axon growth via AA activation of protein kinase C to phosphorylate GAP43 and polymerize/stabilize F-actin. Our previous results show that blocking cPLA2 mimics NMDAR blockers, whereas exogenous AA reverses the increased dynamics, and PKC inhibitors also arrest growth. To test whether this activity-driven F-actin control mechanism shapes retinotectal arbors in zebrafish, we used the alpha-1-tubulin promoter to express GAP43-GFP fusion proteins in retinal ganglion cells, and imaged arbors in time-lapse to test for effects of GAP43 levels and its phosphorylation. Overexpressing wildtype GAP43 gave faster growth and larger arbors (#branches, spatial extent, total length of branches) at three days and especially four days. Surprisingly, the N-terminal 20 amino acid segment alone caused the same increase in branching, but no increase in growth. Earlier studies implicate this region in activating G(o) resulting in collapse of growth cones, which is now known to precede branch initiation. In contrast, GAP43 with ser41 mutated to ala (S41A) to prevent phosphorylation did not increase either branching or growth but resulted in immature, elongated arbors even at four to five days. In support of this atrophic effect, only half of brain/spinal neurons expressing S41A successfully initiated axonal outgrowth (vs. nearly 100% for wtGAP43). These results suggest that the region around the ser41 phosphorylation site, which binds CaM and PIP2, promotes growth only when phosphorylated, and also activates the branching control region in the first 10-20 amino acids. Whereas phosphorylation introduces a bulky negative charge group, mutation of serine to arginine introduces a bulky positive charge. But this also produced the same growth and branching as phosphorylation, suggesting that the effect of phosphorylation is through hydrophilic bulk rather than negative charge, in agreement with other IQ motifs. The results implicate the cPLA2-AA-PKC-GAP43 pathway as part of an F-actin based mechanism that both stabilizes new synapses and initiates new branches near effective synapses.

Visual activity refines developing retinotectal maps and shapes individual retinal arbors via an NMDA receptor-dependent mechanism. As retinal axons grow into tectum, they slow markedly and emit many transient side branches behind the tip, assuming a "bottlebrush" morphology. Some branches are stabilized and branch further, giving rise to a compact arbor. The dynamic rate of branch addition and deletion is increased twofold when MK801 is used to block NMDA receptors, as if this prevents release of a stabilizing signal such as arachidonic acid (AA) from the postsynaptic neuron. In optic tract, AA mediates NCAM and L1 stimulation of axon growth by activating presynaptic protein kinase C (PKC) to phosphorylate GAP-43 and stabilize F-actin, and, if present in tectum, this growth control pathway could be modulated by postsynaptic activation. To test for the effects on arbor morphology of blocking PKC or AA release, we examined DiO-labeled retinal axons of larval zebrafish with time-lapse videomicroscopy. Bath application of the selective PKC inhibitor bisindolylmaleimide from 2 or 3 days onward doubled the rate at which side branches were added and deleted, as seen with MK801, and also prevented maturation of the arbor so that it retained a "bottlebrush" morphology. In order to selectively block the PKC being transported to retinal terminals, we injected the irreversible inhibitor calphostin C into the eye from which the ganglion cells were labeled, and this produced both effects seen with bath application. In contrast, there were no effects of control injections, which included Ringers into the same eye and the same dose into the opposite eye (actually much closer to the tectum of interest), to rule out the possibility that the inhibitor leaked from the eye to act on tectal cells. For comparison, we examined arbors treated with the NMDA blocker MK801 at half-hour time-lapse intervals, and detected the twofold rise in rates of branch addition and deletion previously reported in Xenopus larvae, but not the structural effect seen with the PKC inhibitors. In addition, we could produce both effects seen with PKC inhibitors by using RHC80267 to block AA release from DAG lipase, indicating that AA is the main drive for PKC activation. Thus, the results show a distinct role of AA and presynaptic PKC in both maturation of arbor structure and in the dynamic control of branching. The effects on branch dynamics were present regardless of the level of maturity of arbor structure. The fact that they mimicked those of MK801 suggests that presynaptic PKC may be involved in the NMDA receptor-driven stabilization of developing retinal arbors.