Finding the Right Balance: AMPA Receptor Palmitoylation Regulates Network Excitability

Epilepsy, one of the most common neurological diseases, affects approximately 1 percent of the U.S. population and is characterized by recurrent, unprovoked seizures (Stafstrom and Carmant, 2015). During seizures, neurons synchronize in one region of the brain and the aberrant neural activity can spread to other regions (Staley, 2015). Thus, epilepsy can arise when the balance between neural excitation and neural inhibition (E/I balance) is disrupted. Current therapeutic options aim to either lower excitatory activity or increase inhibitory activity in the brain, to restore the E/I balance. Nevertheless, one-third of epileptic individuals will not respond to currently available treatments (Stafstrom and Carmant, 2015), indicating the need to understand the underlying mechanisms that lead to neuronal hyperexcitability.

Excitatory activity in the brain primarily relies on the neurotransmitter glutamate, a chemical signal that crosses the synapse and binds AMPA receptors (AMPARs) on the post-synaptic cell (Hollmann and Heinemann, 1994). AMPAR quantity is proportional to the strength of the excitatory synapse. Regulating AMPAR quantity is a highly dynamic process, where AMPARs are modified at the protein level to either be inserted into the synapse (becoming functional) or to remain inside the cell (remaining inactive) (Shepherd and Huganir, 2007). At the protein level, these modifications include: addition of a phosphate group to the AMPAR protein (phosphorylation) to deliver it to the synapse (Rakhade et al., 2012); or, addition of palmitic acid lipids to the AMPAR (palmitoylation) to remove it from the synapse (Hayashi et al., 2005). Elucidating the role of AMPAR palmitoylation in vivo can inform how AMPARs regulate network excitability and how this process is affected in hyperexcitable states such as epilepsy.

Itoh et al. (2018) seek to understand how AMPAR palmitoylation regulates neuronal physiology and synaptic plasticity, whereby activity-dependent changes strengthen or weaken synapses. While prior work in vitro suggests that palmitoylation regulates whether AMPARs are removed from the synapse, little is known in vivo. Itoh et al. generated mice with mutated AMPAR subunits (GluA1) that lacked the main palmitoylation site, which prevents palmitoylation at GluA1 subunits throughout the brain (Fig., 1C).

Itoh et al. find new roles for AMPARs in regulating neuronal excitation. To begin with, their results suggest that palmitoylation and phosphorylation normally work in opposition to regulate AMPAR levels in vivo. The CS mutant mice have decreased palmitoylation in the hippocampus as expected, but display increased phosphorylation of GluA1 subunits and increased total quantity of GluA1 subunits at synapses in the hippocampus, a key brain region for learning and memory (Fig. 2B, 2C). Thus, palmitoylation is a protein modification that normally helps to decrease AMPARs at the synapse in vivo, thereby keeping neural excitation under control.

Furthermore, Itoh et al. propose that synaptic plasticity is regulated by palmitoylation. One form of synaptic plasticity is long-term potentiation (LTP) in neurons, which governs strengthening of synapses by AMPAR insertion following a strong stimulation analogous to a seizure. The CS mutant mice show a drastic increase in the size of dendritic spines, which are small protrusions along dendrites where synapses occur (Fig. 4, 5C). Taken together, palmitoylation may normally be required to prevent excessive dendritic spine growth during LTP.

Moreover, the authors discover that palmitoylation is a key protein modification to decrease the likelihood of aberrant neural activity, as seen in seizures. When the inhibitory (GABA-A) receptors were blocked to increase neural excitation, CS mutants showed greater susceptibility to develop generalized convulsive seizures, compared to controls (Fig. 6E). These tonic-clonic seizures recur in the CS mutants and show a sustained response (Fig. 6F-H). When anticonvulsants are administered to the CS mutant mice, the mutant mice are more resistant to common antiseizure medications (Figure 9). Therefore, palmitoylation is normally required to maintain the E/I balance to prevent hyper-excitable states and epilepsy.

Itoh et al. (2018) investigated the role of palmitoylation at AMPAR GluA1 subunits in vivo, to better understand how such protein-level modifications can have significant effects on regulating the balance of excitation and inhibition in the brain. Mutating one of the key palmitoylation sites of GluA1 subunits resulted in greater AMPAR levels at synapses, increased dendritic spine volumes, and a lower threshold (greater likelihood) for seizure-inducing activity (Figure 1). While this work makes a substantial advance in our understanding of molecular-level changes that contribute to epilepsy, several questions remain. For instance, GluA1s have other palmitoylation sites that may regulate AMPARs differently, or may amplify the effects at the palmitoylation site studied in this paper. Also, it remains unknown how palmitoylation of other AMPAR subunits (i.e. GluA2/3) contributes to regulation of the receptor, as this study focused only on GluA1 subunits. Itoh et al. show that palmitoylation of GluA1 subunits is important for maintaining the excitation/inhibition balance in the brain that, if disrupted, leads to hyperexcitable states and epilepsy, which advances our understanding of the molecular basis of this disease.

Figure 1:  LTP normally promotes AMPAR insertion at the synapse; mutating the site of GluA1 palmitoylation prevents the removal of AMPAR, leading to neurons with greater sensitivity to seizure-inducing activity (adapted from Itoh et al., 2018 Figure 10).

Figure 1: LTP normally promotes AMPAR insertion at the synapse; mutating the site of GluA1 palmitoylation prevents the removal of AMPAR, leading to neurons with greater sensitivity to seizure-inducing activity (adapted from Itoh et al., 2018 Figure 10).

Works Cited:

·       Hayashi T, et al. (2005) Differential regulation of AMPA receptor subunit trafficking by palmitoylation of two distinct sites. Neuron 47:709-723.

·       Hollmann M, Heinemann S (1994) Cloned glutamate receptors. Annu Rev Neurosci 17:31-108.

·       Itoh M, et al. (2018) Deficiency of AMPAR-Palmitoylation Aggravates Seizure Susceptibility. J Neurosci 38:10220-10235.

·       Rakhade SN et al. (2012) Glutamate receptor 1 phosphorylation at serine 831 and 845 modulates seizure susceptibility and hippocampal hyperexcitability after early life seizures. J Neurosci 32:17800-17812.

·       Shepherd JD and Huganir RL (2007) The cell biology of synaptic plasticity: AMPA receptor trafficking. Annu Rev Cell Dev Biol 23:613-643.

·       Stafstrom CE and Carmant L (2015) Seizures and Epilepsy: An Overview for Neuroscientists. Cold Spring Harb Perspect Med 5: 1-18.

·       Staley K (2015) Molecular mechanisms of epilepsy. Nat Neurosci 18:367-372.

How does the eye talk to the brain?

Two eyes are better than one! Our brains integrate two slightly different images coming from our two eyes to give us depth perception. How does this happen? To answer this, we will look at where the neural pathways from the two eyes converge. Retinal ganglion cells (RGCs) are neurons in the eye that send visual information to the brain. RGCs are divided into many subtypes based on the shapes of the cells. The lateral geniculate nucleus (LGN) is the primary brain region where RGCs first deliver visual information to brain cells. Previous studies have suggested that some LGN cells receive input from only one eye (Chen and Regehr, 2000; Sincich et al., 2007), while others receive input from both eyes (Hammer et al., 2015; Morgan et al., 2016). To fully understand how the brain integrates the images coming from two eyes, we need to understand the pattern of connectivity between RGCs and LGN cells at the level of individual neuronal connections.

To reveal this pattern in mice, a recent study by Rompani et al. (2017) identified exactly one LGN cell from each mouse and looked at all the RGCs that synapsed with that one LGN cell. They wanted to look at (1) how many of those RGCs came from only one eye (monocular) vs. both (binocular) and (2) whether those RGCs belonged to few or many subtypes. To see all the RGCs that synapsed to a single LGN cell, the authors used a viral tracing method to label LGN cells by making them glow. Using the label as a guide, the authors introduced a modified rabies virus into a single LGN cell. The modified rabies virus infects any cell that synapses onto the LGN cell and therefore the modified rabies virus infected all the RGCs that have input to that LGN cell. Finally, the rabies virus was modified such that all infected cells get labeled. Thus, all the RGCs that synapse onto that single LGN cell were labeled.

When the authors looked at the labeled RGCs, they found that the RGCs formed circular clusters in the retina with no more than one cluster per retina (Figure 1D). This is consistent with the long-standing idea that RGCs map onto the LGN depending on where they are spatially in the retina. They found three types of input patterns: relay mode, combination mode, and binocular combination mode. In relay mode (28% of LGN cells), LGN cells received input from only the contralateral eye and from the same or mostly the same RGC subtype (Figure 3D). In combination mode (32% of LGN cells), LGN cells received input from only the contralateral eye and RGCs of many subtypes (Figure 3H). In binocular combination mode (40% of LGN cells), LGN cells received input from both eyes and RGCs of many subtypes (Figure 4A). For the first time, we can see a pattern by which RGCs are connected to individual LGN cells!


This study identified important characteristics about how visual information from two different eyes converge in the brain. By labeling all the RGCs that synapse to a single LGN cell, the authors were able to describe RGC subtype input patterns and monocular vs. binocular input. These findings have important implications for functional relevance, as relay mode in other sensory systems suggest innate connections while combination mode is associated with learning and plasticity. High resolution data like this is needed to understand how the brain integrates visual information from two different eyes to generate image formation and depth perception.


Chen, Chinfei, and Wade G. Regehr. "Developmental remodeling of the retinogeniculate synapse." Neuron 28, no. 3 (2000): 955-966.

Hammer, Sarah, Aboozar Monavarfeshani, Tyler Lemon, Jianmin Su, and Michael Andrew Fox. "Multiple retinal axons converge onto relay cells in the adult mouse thalamus." Cell reports 12, no. 10 (2015): 1575-1583.

Morgan, Josh Lyskowski, Daniel Raimund Berger, Arthur Willis Wetzel, and Jeff William Lichtman. "The fuzzy logic of network connectivity in mouse visual thalamus." Cell 165, no. 1 (2016): 192-206.

Rompani, Santiago B., Fiona E. Müllner, Adrian Wanner, Chi Zhang, Chiara N. Roth, Keisuke Yonehara, and Botond Roska. "Different modes of visual integration in the lateral geniculate nucleus revealed by single-cell-initiated transsynaptic tracing." Neuron 93, no. 4 (2017): 767-776.

Sincich, Lawrence C., Daniel L. Adams, John R. Economides, and Jonathan C. Horton. "Transmission of spike trains at the retinogeniculate synapse." Journal of Neuroscience 27, no. 10 (2007): 2683-2692.

The cerebellum – the brain’s built-in thought editor?

The cerebellum – the brain’s built-in thought editor?

The choices we make have a massive impact on almost every aspect of our lives, and poor decision-making is a common feature across many neurologic and psychiatric diseases. So it makes sense that neuroscientists have been fascinated by decision-making for a long time, and have sought to undercover what influences our choices, which parts of the brain contribute to a decision, and how different aspects of a choice (for example, the evidence in favor of one option over another, or the value of each potential outcome) are reflected in neural activity.

Read More