Physiological functions of GABAB receptor-associated proteins

GABAB receptors (GBRs) play a crucial role in synaptic transmission (Gassmann and Bettler, 2012), and alterations of GBR levels and functions are associated with various neurological diseases (Evenseth et al., 2020; Heaney and Kinney, 2016; Kumar et al., 2013). The GBR is an obligate heterodimer composed of GB1 and GB2 subunits, comprising the core subunits of GBRs (Kaupmann et al., 1998; Schwenk et al., 2010). The GB1 subunit exists in two isoforms, GB1a and GB1b, localizing at pre-or postsynaptic sites, respectively (Vigot et al., 2006). The sole difference between the two isoforms is two sushi domains (SDs) located exclusively at the N-terminus of GB1a (Hawrot et al., 1998; Kaupmann et al., 1997). SDs are required for axonal localization and stabilization at the cell surface of GB1a/2 receptors (Biermann et al., 2010; Hannan et al., 2012). At the presynapse, GB1a/2 receptors inhibit neurotransmitter release by blocking voltage-gated Ca2+ (Cav) channels, and at the postsynapse, GB1b/2 receptors induce hyperpolarization of neurons through activating inward-rectifier K+ (Kir3) channels (Gassmann and Bettler, 2012). Native GBRs form macromolecular complexes with auxiliary subunits and various constituents, which impart distinct functional properties to GBRs (Schwenk et al., 2010; Schwenk et al., 2016).
Cytosolic potassium channel tetramerization domain (KCTD)-containing protein 8, 12, 12b, and 16 (hereafter collectively designated KCTDs) are auxiliary GBR subunits, influencing the GBR response (Schwenk et al., 2010). KCTDs comprise a T1 and an H1 domain with KCTD8 and -16 containing an additional C-terminal H2 domain (Schwenk et al., 2010). Homopentameric KCTDs interact through their T1 domain with the C-terminus of GB2 and accelerate GBR-mediated Kir3 channel responses (Schwenk et al., 2010). However, solely KCTD12 and 12b induce pronounced desensitization of GBR-mediated Kir3 currents and Cav channel inhibition by uncoupling Gβγ from the effector channel, suggesting that GBR/KCTD complexes can generate distinct functional properties (Schwenk et al., 2010; Turecek et al., 2014). Because of their overlapping expression patterns (Metz et al., 2011), it is conceivable that KCTDs also form hetero-oligomers interacting with GBRs and G proteins. In co-immunoprecipitation experiments, my colleagues identified KCTD12/KCTD16 hetero-oligomers in brain tissue that form a complex with GBRs. They further demonstrated that KCTD12 and KCTD16 retained their distinct regulatory properties in KCTD12/KCTD16 hetero-oligomers, resulting in intermediate GBR-mediated Kir3 current desensitization. They also revealed that KCTD12/KCTD16 hetero-oligomers produce slow deactivation kinetics of Kir3 currents that lead to an increase in the duration of GBR-mediated slow inhibitory postsynaptic currents (sIPSCs). However, it was yet unknown whether KCTD hetero-oligomers can interact with G proteins. My data showed that KCTD homo-and hetero-oligomers bind to the G protein in living cells, which contributed to the understanding of distinct functional properties of KCTD12/KCTD16 hetero-oligomers. Thus, my data complemented the findings of my colleagues and resulted in a co-authorship publication (Fritzius et al., 2017). Together our data show that KCTD12/KCTD16 hetero-oligomers regulate the fine-tuning of GBR-mediated Kir3 currents and enrich the molecular and functional repertoire of native GBRs.
β-amyloid precursor protein (APP), adherens junction-associated protein 1 (AJAP1), and PILRα-associated neural protein (PIANP) are single-spanning membrane proteins that interact with the N-terminal SD1 of GB1a (Dinamarca et al., 2019; Schwenk et al., 2016). Axonal GB1a/2 receptor trafficking is dependent on kinesin-1 (Valdes et al., 2012), but the SD1 required for axonal transport reside within the lumen of transport vesicles (Biermann et al., 2010; Vigot et al., 2006). Due to the interaction with the SD1, APP, AJAP1, and PIANP represent promising candidates for linking GB1a/2 receptors in transport vesicles to the kinesin motor. While APP, AJAP1, and PIANP share the ability to bind SD1 of presynaptic GB1a/2 receptors, only APP linked GB1a/2 receptors to the kinesin motor and mediated axonal trafficking of GB1a/2 receptors, as demonstrated by my colleagues. They further showed that the interaction of APP with GB1a/2 receptors resulted in mutual stabilization at the cell surface, which prevented GB1a/2 receptor internalization and reduced the proteolytic processing of APP in endosomes. However, whether the APP/GB1a/2 receptor complex formation is altered upon GB1a/2 receptor activation and whether APP modulates GB1a/2 receptor signaling remained unclear. I showed that the GB1a/2 receptor activation did not regulate the association or dissociation between APP and GB1a/2 receptors. My data further demonstrated that the co-expression of APP did not modulate GB1a/2 receptor signaling in heterologous cells. Thus, my data contributed to the characterization of the interaction of APP and GB1a/2 receptors and were integrated into a publication with me as a co-author (Dinamarca et al., 2019). Since proteolytic APP processing in the amyloidogenic pathway yields Aβ fragments, a hallmark of Alzheimer’s disease (AD) (Huang and Mucke, 2012; Muller et al., 2017; Selkoe and Hardy, 2016), and alterations in GB1 surface levels are observed in AD patients (Chu et al., 1987a, b; Iwakiri et al., 2005) and a mouse model of AD (Martin-Belmonte et al., 2020), our data reveal that APP/GB1a/2 receptor complex formation links presynaptic GB1a/2 trafficking to Aβ generation.
Proteolytic APP processing in the non-amyloidogenic pathway generates soluble APPα (sAPPα) fragments (Muller et al., 2017). sAPPα binds to SD1 of presynaptic GB1a/2 receptors (Dinamarca et al., 2019; Rice et al., 2019) and signals through G proteins (Fogel et al., 2014; Pasciuto et al., 2015). A recent publication reported that sAPPα induced presynaptic GB1a/2 receptor-mediated inhibition of neurotransmitter release in vivo (Rice et al., 2019). Likewise, a 17 residue long peptide composed of the APP sequence containing the SD1 binding motif, termed APP17, suppressed neurotransmitter release and neuronal transmission by activating endogenous GB1a/2 receptors at presynaptic sites (Rice et al., 2019). In contrast, my colleagues showed that sAPPα does not modulate GB1a/2 receptor-mediated G protein activation in heterologous cells (Dinamarca et al., 2019). Due to the controversial findings and the lack of data explaining the mechanism of GB1a/2 receptor modulation by sAPPα, I studied whether APP17 modulates recombinant GB1a/2 receptors. I confirmed binding of APP17 to GB1a/2 receptors expressed in HEK293T cells by displacing fluorescent APP17 peptides from receptors. However, my data demonstrated that APP17 does not modulate GB1a/2 receptor-mediated G protein activation or Gα signaling in heterologous cells. Using a very sensitive GB1a/2 receptor-induced firefly luciferase (FLuc) accumulation assay, my data further evidenced that APP17 does not exert subtle modulatory properties at recombinant GB1a/2 receptors. I further confirmed the absence of allosteric modulatory properties of co-expressed full-length APP at recombinant GB1a/2 receptors in G protein activation, Ga signaling and sensitive FLuc accumulation assays, confirming earlier data published in Dinamarca et al. (2019). My data further showed that the displacement of full-length APP by APP17 did not induce modulatory effects at recombinant GB1a/2 receptors. In addition, my colleagues observed that APP17 neither influenced GB1a/2 receptor-mediated Gβγ signaling in vitro nor changed GB1a/2 receptor-mediated inhibition of neurotransmitter release or neuronal transmission in vivo, using electrophysiological recordings. Hence, neither sAPPα nor APP17 modulates recombinant or native GB1a/2 receptor signaling. A manuscript reporting these findings with me as the first author is in preparation.
AJAP1 and PIANP form distinct complexes with GB1a/2 receptors and are not involved in axonal trafficking of GB1a/2 receptors (Dinamarca et al., 2019). I demonstrated that the formation of AJAP1/GB1a/2 receptor and PIANP/GB1a/2 receptor complexes in cis neither stabilized GB1a/2 receptors at the cell surface nor modulated GB1a/2 receptor-mediated G protein activation in heterologous cells. The observation of my colleagues that AJAP1 and PIANP are located in the somatodendritic compartment, together with their observation that mice genetically lacking AJAP1 or PIANP showed a deficit in presynaptic GB1a/2 receptor-mediated inhibition of neurotransmitter release (S. Früh and T. Lalanne, personal communications) (Winkler et al., 2019), suggested a trans-synaptic interaction between GB1a/2 receptors and AJAP1 or PIANP. Indeed, I demonstrated that AJAP1 and PIANP recruit and cluster transcellular GB1a/2 receptors. I further showed that the formation of AJAP1/GB1a/2 receptor and PIANP/GB1a/2 receptor complexes in trans resulted in a stabilization and negative allosteric modulation of GB1a/2 receptors. However, the maximum efficacy of transcellular GB1a/2 receptor signaling was not affected by the interaction with AJAP1 or PIANP, resulting in an increased dynamic range of receptor activity. My data further demonstrated that the negative allosteric properties exerted by AJAP1 and PIANP at transcellular GB1a/2 receptors required anchorage into the cell membrane because soluble AJAP1 (sAJAP1), which is composed exclusively of the extracellular domain of AJAP1, did not allosterically modulate GB1a/2 receptors. Earlier my colleagues identified different affinities for SD1 binding in the rank order AJAP1>PIANP>>APP (Dinamarca et al., 2019). Thus, my data support a model in which APP traffics GB1a/2 receptors to axon terminals, where they are transferred to postsynaptic AJAP1 or PIANP that precisely localize the receptor and increase its dynamic range. My data will be part of a future publication from the lab with me as a co-author.
Elucidating the physiological functions of PIANP necessitates the analysis of mice genetically lacking PIANP and compare them to WT littermates. Since no PIANP-KO mice were available at the beginning of my Ph.D., I generated PIANP-KO mice using the clustered regularly interspaced short palindromic repeats (CRISPR) / CRISPR-associated (Cas) 9 system in collaboration with the Centre for Transgenic Models (CTM) of the University of Basel. Generating PIANP-KO mice using the CRISPR/Cas9 system requires the electroporation of one-cell embryos with ribonucleoprotein (RNP) that consist of the specific guide ribonucleic acid (RNA), the trans-activating RNA, and the CRISPR/Cas9. In the first step, I identified two guide RNAs. The CTM team pre-validated the two guide RNAs, electroporated CL57B/6 one-cell embryos with the RNP, transferred the surviving embryos into pseudopregnant females, and sampled biopsies of the offspring. The biopsies were transferred to me, and I validated the genomic alterations induced by the CRISPR/Cas9 system and used the PIANP-KO offspring to establish the PIANP-KO mouse line in the lab. I further confirmed the loss of endogenous PIANP protein in the PIANP-KO mouse line that my colleagues used for electrophysiological recordings published in Dinamarca et al. (2019) and Winkler et al. (2019), publications on which I am a co-author. In the Winkler et al. (2019) publication, the electrophysiological recordings of my colleagues showed that PIANP-KO mice exerted deficits in presynaptic GB1a/2 receptor-mediated inhibition of glutamate release. Thus, these data support that PIANP deficiency results in incorrect localization and function of presynaptic GB1a/2 receptors (Winkler et al., 2019).