November 2023 - Amanda Perozzo

Amanda Perozzo on GSG1L-containing AMPA receptor complexes are defined by their spatiotemporal expression, native interactome and allosteric sites


A PhD is a marathon, not a sprint; this is a phrase I fully embodied over the past six years as a graduate student. The phrase also evokes themes of patience and perseverance, which speak to the publication journey of Perozzo et al., 2023 in Nature Communications (full-text access here).

When I toed the start line and joined Dr. Derek Bowie’s research group at McGill in September 2017, I had just completed my BSc in Biopharmaceutical Science at the University of Ottawa. In my final few semesters, I elected to take some upper-level courses in animal and cellular physiology that piqued my interest in neuroscience. Although I had extensive undergraduate research experience across varied disciplines, my knowledge of ion channels and patch-clamp electrophysiology was rather limited. Driven by an inquisitiveness to learn more about the molecular machines at play in brain function, I undertook a project focusing on glutamate-gated AMPA receptor (AMPAR) signaling complexes, which mediate the majority of fast excitatory neurotransmission in the mammalian central nervous system (CNS). The overall aim of the research was to better understand how AMPAR auxiliary subunits fine-tune the properties of the core receptor from a structure-function perspective.

AMPARs are ion channels that assemble as multiprotein complexes with different families of auxiliary subunits. Association with these helper proteins diversifies AMPAR responsiveness at synapses by directly regulating their gating behaviour. Transmembrane AMPAR regulatory proteins (TARPs) and Germ cell-specific gene 1-like protein (GSG1L) are evolutionarily- and structurally-related AMPAR auxiliary subunits that modulate AMPAR function in opposing manners by unresolved mechanisms. In Perozzo et al., 2023 Nat Commun., we sought to not only answer this question, but also to provide insight into the native expression and interactome of GSG1L-containing AMPARs, which are relatively understudied compared to AMPAR-TARP complexes. This mission was important as GSG1L’s role in slowing recovery from AMPAR desensitization may set the refractory time course at select glutamatergic synapses.

As summarized in Fig. 1 below, our interdisciplinary study used a three-pronged approach to uncover several new pieces of information regarding GSG1L: First, we showed that GSG1L exhibits a peculiar spatiotemporal expression pattern in the rodent brain. For example, while GSG1L expression is persistent in the cerebellar granule layer across development, it appears in cortical layer 2/3 and the caudate putamen only in adulthood. Interestingly, in the cerebellum, GSG1L is localized strictly to the anterior lobules. Next, our proteomics data revealed that GSG1L constitutes about 5% of all AMPAR complexes in the mature rat brain. Surprisingly, it can integrate into AMPARs as the sole auxiliary subunit in contrast to TARPs, which often partner with other accessory proteins, such as Cornichons (CNIHs) and Cystine-knot AMPA receptor modulating proteins (CKAMPs), to co-decorate AMPARs. This notion is consistent with GSG1L-specific synapses. Finally, through careful electrophysiological characterization of isolated AMPAR complexes, we demonstrated that TARPs and GSG1L exert opposing modulatory effects by working through two distinct allosteric regulatory sites on the lower lobe of the AMPAR ligand-binding domain. Together, the study lays the foundation to interrogate the impact of GSG1L on AMPAR response fidelity, excitability, and neuronal circuits.

Though the outcome was certainly positive, the trek to the finish line involved recovering from a few tumbles. After my Candidacy Exam in late Spring 2019, we submitted a very different version of this manuscript to eLife, but the reviewer comments prompted us to re-think our narrative. In early 2020, I embarked on a molecular biology frenzy to generate DNA constructs in which the AMPAR subunit was covalently linked to GSG1L, as we were interested in studying the effect of GSG1L stoichiometry on AMPAR gating. Unfortunately, the PCR failed over and over, and so we turned to a biotechnology company to complete the task. Of course, ‘early 2020’ should instill a sense of impending doom. Upon returning to the lab after a three-month shutdown due to the relentless and devastating COVID-19 pandemic, I began testing the functionality of the new constructs by e-phys… only to observe some bizarre phenomena (eventually becoming supplementary figures in the final publication). These results sparked the pursuit to identify a GSG1L-mediated phenotype in the mouse brain that matched our recombinant data, and we thought we had a promising neuronal candidate. We crafted a manuscript that was reviewed at Neuron over the Summer of 2021, but the reviewers requested data from a GSG1L KO mouse to bolster our hypotheses. While acquiring and breeding this mouse line, we also performed biochemistry and molecular dynamics simulations, but the results were not entirely conclusive. Finally, the KO mouse line was at the stage to perform recordings; however, we quickly (and unexpectedly) observed that the data were similar to WT, and thus were unable to confirm our initial propositions regarding the expression and stoichiometry of native AMPAR-GSG1L complexes. At this point, I was concerned that my findings would never see the light of day beyond my thesis. In fact, the preface to my second thesis chapter states, “the electrophysiology contents of Chapter II have lived many lives.”

The tides turned in late 2022 thanks to some lucky conversations. In discussing our observations with Dr. Teru Nakagawa who had provided us with the GSG1L KO mouse line, we learned that a former PhD student had recently conducted a histological characterization of GSG1L expression in the rodent brain across development, which helped us to better understand our seemingly puzzling results from cortical layer 2/3 neurons. During a serendipitous dinner conversation with Dr. Jochen Schwenk at a Society for Neuroscience Meeting, we learned that the native GSG1L interactome is unique in a way that is consistent with GSG1L fulfilling a specialized signaling role. Through scientific curiosity and altruism, the third rendition of my study was conditionally accepted in Nature Communications - three days after I crossed the finish line of the Montreal Marathon (Fig. 2). I did trip and scrape both knees, but I got back up and kept going. I think there’s a metaphor in there somewhere.


Fig. 1: The definining features of GSG1L-containing AMPARs. Adapted from Perozzo et al., 2023 Nat Commun., CC BY 4.0 (

Fig. 2: Crossing that (PhD) finish line. I spent the Summer of 2023 writing my PhD thesis and training for the Montreal Marathon. This was my first full marathon, completed in 03:53:36.





















Back to top