This Week in The Journal

Astrocytic FMRP Regulates Neuronal Phenotypes

Haruki Higashimori, Christina S. Schin, Ming Sum R. Chiang, Lydie Morel, Temitope A. Shoneye, et al.

Fragile X syndrome is the most common inherited form of intellectual disability, and it is frequently accompanied by autism. It is caused by a trinucleotide expansion that silences transcription of FMR1, resulting in loss of fragile X mental retardation protein (FMRP). FMRP is an mRNA-binding protein that suppresses translation of numerous transcripts, many of which encode synaptic proteins. Its loss increases basal levels of translation, precluding activity-dependent regulation. This is thought to underlie the increased metabotropic-glutamate-receptor-dependent long-term depression and increased density of long, immature dendritic spines found in fmr1-null mice. FMRP also binds directly to a regulatory subunit of voltage- and calcium-dependent potassium (BK) channels, increasing calcium sensitivity and thus narrowing action potential width. Loss of this function leads to action potential broadening and thus increased neurotransmitter release, which may contribute to hyperexcitability in cortical and hippocampal circuits (Contractor, et al. 2015 Neuron 87: 699).

Dendritic spine density in cortical layer 2/3 pyramidal neurons is elevated in fmr1-null mice (top and third panel). Restoring fmr1 selectively in astrocytes (second and fourth panel) rescues this phenotype. Top two panels are apical dendrites; bottom two panels are basal dendrites. See Higashimori et al. for details.

Although most studies of FMRP function focus on neurons, loss of FMRP also affects astrocytes. For example, cortical levels of the astrocytic glutamate transporter GLT1 are reduced in fmr1-null mice, and consequently, glutamate uptake is impaired. This likely exacerbates hyperexcitability and may thus contribute to neuronal phenotypes. In fact, Higashimori et al. demonstrate that this is the case. Deleting fmr1 selectively in astrocytes replicated several phenotypes found in fmr1-null mice. Cortical GLT1 levels and glutamate uptake decreased, as expected, but more surprisingly, neuronal protein synthesis, the length and density of dendritic spines, and neuronal excitability also increased. In contrast, restoring fmr1 expression selectively in astrocytes rescued these phenotypes in fmr1-null mice. These neuronal phenotypes were also rescued by pharmacologically enhancing surface expression of GLT1 in fmr1-deficient mice.

Together, these results suggest that loss of FMRP in astrocytes results in decreased GLT1 expression and reduced glutamate uptake. This likely exacerbates the effects of excessive glutamate release resulting from action potential broadening. Together, these may increase activation of metabotropic glutamate receptors, thus promoting long-term depression. The discovery that enhancing GLT1 expression rescues some of these deficits suggests that GLT1 may be an effective therapeutic target for minimizing the neurological effects of FMR1 mutations.

Spontaneous Vesicle Release Drives Synaptic Scaling

Miguel Angel Garcia-Bereguiain, Carlos Gonzalez-Islas, Casie Lindsly, and Peter Wenner

(see pages 7268–7282)

Increasing neural network activity for prolonged periods engages mechanisms that reduce neuronal excitability and/or decrease the strength of excitatory synaptic inputs to glutamatergic neurons. Conversely, prolonged neuronal silence leads to increases in excitatory synaptic strength and/or neuronal excitability. Such homeostatic plasticity has been proposed to ensure that circuits can continue to communicate information through changes in neuronal firing rate.

Much work has suggested that homeostatic synaptic scaling is induced by changes in firing rate and the resulting changes in spike-dependent calcium influx, which lead to changes in surface expression of AMPA receptors. But in most of these experiments, spiking was modulated throughout neuronal populations, raising the possibility that changes in neurotransmission, rather than spiking, drives homeostatic plasticity. Indeed, homeostatic increases in synaptic strength occur when AMPA receptors are blocked, even if spiking continues (Fong, et al. 2015 Nat Comm 6:1). Therefore, the triggers and functions of synaptic scaling continue to be debated.

Garcia-Bereguiain et al. provide additional evidence that upward synaptic scaling can be induced by increased activation of neurotransmitter receptors in the absence of spiking. Previous work in chick embryonic spinal cord showed that blocking spontaneous network activity or blocking GABA_A receptors (which are depolarizing at this stage) increased the amplitude of glutamatergic and GABAergic currents in motor neurons. To disentangle the effects of synaptic transmission and spiking on this synaptic scaling, Garcia-Bereguiain et al. altered the frequency of spontaneous GABA release by treating embryos for 2 d with nicotine (which enhances spontaneous release by acting on presynaptic acetylcholine receptors) or the nicotinic antagonist DHβE (which reduces spontaneous release). Nicotine treatment decreased the amplitude of AMPAergic and GABAergic miniature PSCs (mPSCs) in motor neurons, whereas DHβE increased mPSC amplitude. Importantly, increasing spontaneous GABA release reversed the upward scaling of mPSC amplitude produced by blocking action potentials. In fact, in the absence of spiking, nicotine continued to trigger downscaling at GABAergic synapses.

These results clearly demonstrate that synaptic scaling can be triggered by changes in spontaneous neurotransmitter release in the absence of spiking. This adds to mounting evidence that spontaneous release is not “noise” but rather has important functions, including regulating homeostatic plasticity. The results also have clinical implications, demonstrating that nicotine exposure can affect plasticity in developing circuits.