Flip-Flopping to the Membrane

Abstract

Excitatory synapses that use the neurotransmitter glutamate are highly dynamic, constantly changing their character in an activity-dependent manner. In this issue of Neuron, Penn et al. (2012) describe a novel mechanism that changes the fidelity of glutamate signaling to maintain homeostatic synaptic plasticity.

The hallmark of nervous systems—how we perceive, think, and evolve—is adaptability. The majority of synapses in the mammalian central nervous system use the excitatory neurotransmitter glutamate. Embedded in the postsynaptic membrane to detect these glutamate signals are ionotropic glutamate receptors, including the prototypical work-horse, the AMPA-type glutamate receptor (AMPAR). Numerous mechanisms have been identified that modify glutamatergic transmission in an activity-dependent manner with most focusing on the number (Anggono and Huganir, 2012) and subunit composition (Cull-Candy et al., 2006) of AMPARs at the synaptic membrane. While specific inputs may change, neuronal networks maintain an overall balance in excitability, a process termed homeostatic plasticity (Turrigiano, 2012). In this issue of Neuron, Penn et. al. (2012) present a novel means by which neurons regulate glutamatergic neurotransmission in an activity-dependent manner to maintain homeostatic plasticity— regulation of AMPAR subunit composition via the flip/flop splicing cassette. This work provides the first glimpse into mechanisms that regulate AMPAR assembly, and hence synaptic fidelity, at the level of the endoplasmic reticulum (ER) (Figure 1).

Figure 1. Schematic of the Biogenic Pathway of Ionotropic Glutamate Receptors from Synthesis to Membrane Expression in the CA1 Region of the Hippocampus Either under Normal Conditions (Left Half) or following Chronic Deprivation of Activity using TTX (Right Half).

Following TTX treatment, the ratio of flip to flop is decreased for both GluA1 and GluA2, but the splicing transition occurs more rapidly for GluA1, favoring the earlier appearance of GluA1o. Because of the slower splicing transition for GluA2 and its longer half-life in the endoplasmic reticulum (ER), GluA2i persists in the biosynthetic pathway. GluA1o also forms heteromers more readily than GluA1i with GluA2i. All of these factors favor the formation of GluA1o/Glu2Ai heteromers that display reduced desensitization and a faster recovery from desensitization. At CA1 synapses the fidelity of transmission is maintained better following TTX (red) than in the control (black) with high-frequency stimulation, presumably reflecting the presence of GluA1o/GluA2i at the synapse. Accessory proteins, such as TARPs or cornichons (green cylinder), apparently do not contribute to the phenotype.

AMPARs play a major role in determining the time course and magnitude of excitatory synaptic responses. AMPARs possess features of a detector of the glutamate transient during a synaptic event: its ion channel rapidly opens and closes, defining the “fast kinetics” that epitomizes glutamatergic signaling. Overlaying this fast detection process, AMPARs can also enter into a nonconducting or desensitized state in response to glutamate. This interplay between opening, closing and desensitization defines the fidelity of AMPAR-mediated signaling. It is dependent on AMPAR subunit composition (there are four subunits, GluA1–GluA4), alternative splicing, mRNA editing, post-translational modifications, and interactions with accessory proteins such as TARPs and cornichons (Traynelis et al., 2010; Jackson and Nicoll, 2011; Lu and Roche, 2012). AMPARs, like all ionotropic glutamate receptors, form functional, tetrameric receptors in the ER. They are preferential heteromers predominately composed of GluA1 and GluA2 subunits (Lu et al., 2009). Previous studies have shown that AMPAR subunits, alternative splicing in the ligand-binding domain (flip/flop cassette), editing at the R/G site upstream of the flip/flop cassette, and editing at the Q/R site in the pore of the ion channel can influence heteromerization and export of AMPAR complexes from the ER, thus potentially modulating synaptic transmission (Sukumaran et al., 2012). One way in which these factors affect heteromerization is by affecting the dwell time of specific variants in the ER. However, the significance of ER-assembly mechanisms for AMPARs in neurons (previous work had largely been done in recombinant receptors) and how they might impact synaptic transmission was unknown. Penn et al. (2012) provide evidence that alternative splicing facilitates the regulated assembly of AMPARs and directly modulates synaptic transmission in the CA1 region of the hippocampus.

The flip/flop cassette was identified soon after the initial cloning of AMPAR subunits and all AMPAR subunits undergo this alternative splicing (Sommer et al., 1990). Flip/flop has numerous effects on receptor function including the extent and degree of desensitization, though the specific effect depends on the specific subunit and subunit combinations (Dingledine et al., 1999). In the present study, the authors investigated the role of the flip/flop cassette in the hippocampus and found that chronic deprivation of activity by the Na⁺ channel blocker tetrodotoxin (TTX) decreased the ratio between flip/flop splice variants for GluA1 and GluA2 in the CA1 but not CA3 regions. These effects were reversed upon removal of TTX highlighting the dynamic nature of these actions. Importantly, the authors also found a difference in the subunit-specific turnover rate from flip-to-flop with the rate being more rapid for GluA1 (τ = 2.4 hr) than for GluA2 (τ = 4 hr). The relatively fast increase in GluA1o subunits combined with a longer dwell time of GluA2i subunits in the ER (Greger et al., 2002) contributed to the formation of more GluA1o/GluA2i receptor complexes. Further the authors show that the GluA1o isoform more readily recruits GluA2i to form heteromeric complexes than that of GluA1i. Hence, because of differential rates of alternative splicing, the longer dwell time of GluA2 in the ER, and the preferential assembly of specific subunit variants in the ER into heteromers, GluA1o/GluA2i becomes a more prominent AMPAR in CA1 pyramidal neurons with activity depravation.

But what makes GluA1o/GluA2i heteromeric receptors so distinctive? GluA1o/GluA2i heteromers show less desensitization and recover faster from desensitization than that of other GluA1/GluA2 splice variant combinations. The authors show that following TTX treatment, surface AMPARs from CA1 pyramidal neurons showed properties consistent with a GluA1o/GluA2i composition, an effect apparently not dependent on accessory proteins. Of course, the coup de grace is that the authors demonstrate that synaptic inputs to CA1 pyramidal neurons show greater fidelity in response to high frequency stimulation—presumably due to the reduced desensitization properties of GluA1o/GluA2i. Together, this work highlights the role of activity in regulating alternative splicing. However, the novel mechanistic insights are the dynamics of events in the endoplasmic reticulum—how residency times and preferential assembly of specific subunits ultimately impact surface expression and hence synaptic dynamics, including homeostatic plasticity.

Of course, with every new insight intriguing and unanswered questions arise. One question is how changes in neuronal activity alter alternative splicing at the flip/flop cassette. The current work presents a tantalizing clue suggesting that the activity-dependent regulated splicing at the flip/flop cassette is dependent on L-type voltage-gated Ca²⁺ channels. However, the molecular pathway to the nucleus and the targets regulating splicing remain unknown. Another intriguing question stemming from the current findings concerns how AMPARs and ionotropic glutamate receptors in general, as well as any multimeric protein, are assembled in the ER. What are the rules governing heteromeric assembly? Although mRNA synthesis and stability will affect the availability of subunits, others factors besides simple mass action are important. For ionotropic glutamate receptors, interactions at the level of the extracellularly located aminoterminal domain can affect preferential assembly (Kumar and Mayer, 2012; Sukumaran et al., 2012) as might the transmembrane domain, including Q/R-site editing (Greger et al., 2002) and the M4 transmembrane segment (Salussolia et al., 2011). The fact that the flip/flop cassette can affect preferential assembly coupled with the differential dwell times of subunits in the ER—why does GluA2 linger longer than GluA1—further complicates this picture. A related issue regarding AMPAR assembly concerns the oligomeric status of the subunits within the ER. Do AMPAR subunits available for mixing and matching exist as monomers, dimers or tetramers? Further, how do heteromeric AMPARs assemble: initially as homodimers or as heterodimers? Given the present results demonstrating the importance of dynamics of assembly in the ER to homeostatic regulation, further defining these rules will be critical to clarifying mechanisms underlying synaptic function. These results also highlight the limitation of measuring mRNA levels alone. Although this approach has proven extremely useful in terms of identifying gene expression profiles, it does not reveal, as has been long recognized, the actual composition of functional receptors in the membrane.

Other unanswered questions are more network or brain related. The authors found that activity-dependent changes in flip/flop ratios occurred in the CA1 region but not in the CA3 region. Hence, it is not a universal, all encompassing strategy but unique to distinct brain subregions. Further, although the authors investigated the effects of changes in synaptic composition on short-term synaptic plasticity, how such changes might affect long-term forms of synaptic plasticity and activity in local networks remains untested.

Penn et al. (2012), take a rigorous approach to address the composition of AMPAR complexes at synapses. There remain however great challenges in relating molecular events inside the cell to synaptic outcomes. Numerous genetic and optical approaches are needed to address the subunit-specific composition of receptor complexes not only at synapses but also within the biosynthetic and secretory pathways. Optical approaches aimed at determining subunit composition of synaptic iGluRs are being developed. For example, the use of single particle tracking photoactivation localization microscopy in concert with viral glycoproteins has begun to redefine our understanding of membrane receptor dynamics and their movement trajectories within the cell (Hoze et al., 2012). However, these techniques at present do not allow subunit/splice variant composition of AMPARs to be defined. Development of quantitative imaging and biochemical techniques will be required to discern the oligomerization processes and the factors that regulate their dynamics. Further, these techniques would allow us to better understand the role of endocytosis in synaptic transmission and perhaps whether recycling endosomes represent a secondary level of subunit-specific processing. These issues are critical to resolve because, unlike in politics, “flip-flopping” appears to be a good thing in neurons.