BRAG1/IQSEC2 as a regulator of small GTPase-dependent trafficking

ABSTRACT

Precise trafficking events, such as those that underlie synaptic transmission and plasticity, require complex regulation. G-protein signaling plays an essential role in the regulation of membrane and protein trafficking. However, it is not well understood how small GTPases and their regulatory proteins coordinate such specific events. Our recent publication focused on a highly abundant synaptic GEF, BRAG1, whose physiologic relevance was unknown. We find that BRAG1s GEF activity is required for activity-dependent trafficking of AMPARs. Moreover, BRAG1 bidirectionally regulates synaptic transmission in a manner independent of this activity. In addition to the GEF domain, BRAG1 contains several functional domains whose roles are not yet understood but may mediate protein-protein interactions and regulatory effects necessary for its role in regulation of AMPAR trafficking. In this commentary, we explore the potential for BRAG1 to provide specificity of small GTPase signaling, coordinating activity-dependent activation of small GTPase activity with signaling and scaffolding molecules involved in trafficking through its GEF activity and other functional domains.

G-protein signaling is essential for many critical cellular functions, including protein trafficking between various cellular compartments. Some small GTPases, called monomeric G-proteins, are known as master regulators of membrane trafficking between cellular compartments, vesicles and the cell membrane.¹ For instance, Rab8 mediates the transport of trans-Golgi network-derived membranes into specific compartments of the plasma membrane.² A major gap in the current understanding of GTPase-mediated trafficking is how the specificity of their action is integrated with the trafficking machinery. Our recently published study describes the synaptic function of Brefeldin-A-Resistant-ArfGEF 1 (BRAG1), a multi-domain regulator of G-protein signaling belonging to the BRAG sub-family of the IQSEC class of ArfGEFs.³ We find that BRAG1, also referred to as IQSEC2, bidirectionally regulates synaptic transmission and that its GEF activity is required for the activity-dependent removal of a major neurotransmitter receptor in the central nervous system, the α-amino-3-hydroxyl-5-methyl-4-isoxazolepropionic acid (AMPA) receptor.⁴ Based on our results regarding the role of BRAG1-dependent regulation of G-protein signaling, along with other features of BRAG1 whose roles are yet to be fully explored, we can begin to understand how the specificity of GTPase signaling is provided through regulatory proteins like BRAG1.

Trafficking of synaptic receptors requires precise regulation and specificity. AMPARs undergo both constitutive and regulated trafficking into and out of the synaptic membrane.⁵ Regulated trafficking in response to neuronal activity can lead to persistent changes in synaptic strength, an underlying mechanism of learning and memory. An increase in synaptic strength through increased AMPAR numbers at the synapse is termed Long-Term Potentiation (LTP), while the reduction in synaptic strength through removal of AMPARs is referred to as Long-Term Depression (LTD).^6-9 These plasticity events require the integration of specific signals resulting from synaptic activity (e.g. N-methyl-D-aspartate receptor activation) with the trafficking machinery, involving G proteins, scaffolding proteins and the actin cytoskeleton. Many questions remain about the regulation of AMPAR trafficking but it is clear that G-protein signaling plays a vital role. Monomeric G-proteins are regulated by 2 opposing processes: 1) G-protein activating proteins (GAPS), which facilitate the hydrolysis of GTP to GDP, and 2) guanine nucleotide exchange factors (GEFs), which activate G-proteins by exchanging the bound GDP for GTP. Small GTPases can be regulated by multiple specific GEFs and GAPs, which shape the spatial and temporal profile of the resulting G-protein signaling. High levels of BRAG1 in the synapse indicate a critical role in synaptic function for this GEF and the GTPase that it regulates, Arf6.

An in vitro assay has identified Arf6 as the preferential target of BRAG1.¹⁰ Like other members of this sub-family, BRAG1 is characterized by the presence of both the GEF-mediating SEC7 domain and an IQ-like motif.^11,12 BRAG1 is particularly abundant in the post-synaptic density,^3,10 found at levels higher than some N-methyl-D-aspartate (NMDA) receptor subunits.¹³ In multiple families, mutations of BRAG1 have been identified as the underlying cause of X-Linked-Intellectual Disability (XLID), a heterogenous neurodevelopmental disorder resulting in decreased cognitive and adaptive abilities.^14-20 In several families, a missense mutation of BRAG1 affecting its GEF activity has been found, suggesting the importance of G-protein signaling regulation in synaptic function and human disease.¹⁵ Of the 4 point mutations originally identified by Shoubridge et al. (2010), 3 are found in the SEC7 domain.¹⁵ The forth is located in the IQ-like motif.¹⁵ BRAG1s GEF activity is reduced by each of these mutations.¹⁵ The role of BRAG1s GEF activity in XLID highlights the importance of understanding G-protein signaling and its regulation. GEF and GAPS regulating Arf6, as well as other G-proteins, can produce distinct outcomes, despite common targets. Our recent work investigating the physiologic function of BRAG1, along with key findings from the existing literature on G-protein signaling and synaptic function, provides some insight into how this specificity may be achieved.

Arf6 is a member of the ADP-Ribosylating Factor (Arf) family of G-proteins that regulate actin dynamics and membrane trafficking, and is the only Arf with a role in trafficking between the plasma membrane and endocytic compartments.^21-23 In the synapse, Arf6 was shown to be involved in the recruitment of 2 critical components of endocytosis, clathrin and AP2, to the synaptic membrane.²⁴ A later study demonstrated a requirement for Arf6 in the endocytosis of telencephalin, leading to maturation of filopodia into functional dendritic spines.²⁵ Additional studies have supported and elaborated on the role of ARf6 in the development, maturation and structural regulation of dendritic spines.^23,26 More recently, it was shown that Arf6-mediated recruitment of AP2 and clathrin is required for AMPAR endocytosis during LTD.²⁷ This study found that a member of the BRAG sub-family, BRAG2, can induce AMPAR internalization through Arf6 activation.²⁷ This activity, which requires the interaction between BRAG2 and an AMPAR subunit, is necessary for 2 forms of LTD, one dependent on metabotropic glutamate receptor (mGluR) activation and another induced by NMDAR receptor activation.²⁷ In addition, they showed that the loss of BRAG2 prevented NMDAR-dependent LTD, presumably due to disrupted Arf6 activation.²⁷ Our recent work shows that either loss of endogenous BRAG1 or disruption of its GEF activity was sufficient to block NMDAR-dependent LTD. These results, including the effects of a dominant negative mutation of BRAG’1s GEF activity (BRAG1-E849K) and an XLID-associated mutation that significantly reduces GEF activity (BRAG1-Q801P), are summarized in (Table 1). Interestingly, Arf6 has also been shown to be activated during chemically-induced LTP and is negatively regulated in this context by a GTPase activating protein, AGAP3.²⁸ Together, these studies illustrate how differential regulation of G-protein signaling through different GAP and GEF proteins can produce distinct outcomes and exemplifies one mechanism of providing the specificity of monomeric G-protein function in different circumstances.

Table 1.

Effect of BRAG1 Constructs on NMDAR-Dependent LTD. Overexpression of GFP-tagged BRAG1 in CA1 pyramidal neurons of organotypic hippocampal slices was achieved via biolistics transfection. Constructs encoded either wild-type BRAG1 sequence or a point mutation resulting in the indicated amino acid substitutions (BRAG1-Q801P and BRAG1-E849K) or a premature stop codon resulting in expression of BRAG1 lacking the terminal 9 amino acids. Expression of an RNAi sequence targeting BRAG1 untranslated regions, restricting activity to endogenous BRAG1, resulted in a significant reduction of endogenous BRAG1 protein. Biolistics transfection also allowed expression of BRAG1 RNAi (along with GFP to mark RNAi-expressing cells), as well as co-expression of BRAG1 RNAi with RNAi-resistant BRAG1 wild-type or truncated mutant construct. Electrophysiology was performed on pairs of untransfected (control) neurons and nearby transfected neurons, identified by expression of GFP. NMDAR-dependent LTD is induced in whole-cell patch-clamped cells by pairing low-frequency presynaptic stimulation with postsynaptic depolarization. Data summarized from Brown et al., 2016.

BRAG1 Construct(s)	Effect on LTD
BRAG1 WT	Normal LTD
BRAG1 RNAi	Blocked
BRAG1-Q801P	Blocked
BRAG1-E849K	Blocked
BRAG1 WT + RNAi	Normal LTD
BRAG1-ΔC-term. + RNAi	Blocked

One layer of specificity in G-protein activation through guanine nucleotide exchange factors lies in the stimuli that triggers a GEF to activate its target G-protein. Because BRAG1s GEF activity is required for activity-dependent AMPAR removal, its GEF activity is likely to be triggered by synaptic activity. In particular, the GEF activity of BRAG1 is likely to be downstream of NMDAR activation, since it is required for the NMDAR-dependent form of LTD,²⁹ or downstream of other activity-dependent events occurring during LTD induction, such as Ca²⁺ influx from Ca²⁺ channels. As mentioned above, BRAG1 contains an IQ-like motif. This motif typically acts as a site for a Ca²⁺-dependent interaction with calmodulin (CaM) and, in some GEFs, CaM binding can enhance enzymatic activity.³⁰ While the interaction between CaM and BRAG1 has not been extensively characterized, a pull-down assay performed with BRAG1 and CaM exogenously expressed in HeLa cells found that BRAG1 preferentially binds to apo-CaM.³¹ Moreover, a mutation that blocks CaM binding to the IQ motif was found to result in enhanced Arf6 activation,³¹ suggesting that the interaction between BRAG1 and CaM attenuates its GEF activity. Together, these findings point to a Ca²⁺-dependent activation of BRAG1s GEF activity, providing a regulatory mechanism for its activation downstream of NMDARs. NMDAR activation leads to an influx of Ca²⁺ into the synapse, which can trigger LTP at high levels of Ca²⁺ or LTD with chronic low levels of Ca²⁺. CaM contains 4 Ca²⁺-binding EF-hand motifs that result in cooperative high-affinity binding between CaM and Ca²⁺,³⁰ suggesting that the Ca²⁺ influx could cause CaM to dissociate from BRAG1 and bind to Ca²⁺. This release of CaM may cause a conformational change and free BRAG1 from inhibition, allowing Arf6 activation. Moreover, the interaction between BRAG1 and NMDAR subunits may itself influence BRAG1s Ca²⁺-dependent GEF activity. These ideas are supported by a recent study of BRAG1 and NMDAR subunits exogenously expressed in HEK 293 cells. Elagabani and colleagues find that BRAG1 interacts with the c-terminus of NMDAR subunits and that glutamate binding to NMDA receptors in the presence of BRAG1 and Ca²⁺ is able to enhance Arf6 activation in this system.³²

In addition to regulating initiation of Arf6 activity, BRAG1 has multiple functional domains that may provide a link to its upstream regulation, through NMDAR signaling, as well as the trafficking machinery necessary to execute its specific function of AMPAR internalization. BRAG1 contains a Type-I PDZ-binding domain at its c-terminus.³ These domains are common in synaptic proteins and provide a means of interacting with other synaptic proteins containing a PDZ domain. PDZ domains are frequently found in scaffolding proteins which regulate stability of AMPA and NMDA receptors and organize signaling complexes at the post-synaptic density (PSD).³³ Within the PSD, interactions between scaffolding proteins, receptors and signaling molecules undergo frequent activity-dependent changes, altering the stability of AMPA and NMDA receptors and the downstream signaling initiated by their activation. Dynamic PDZ interactions are important for synaptic plasticity and the maintenance of synaptic transmission. The interactions of BRAG1s PDZ domain and its function have not been thoroughly characterized. However, it is known that through its PDZ-binding domain, BRAG1 interacts with a classical example of a scaffolding protein which has multiple essential roles in synaptic function, PSD-95¹⁰. PSD-95 provides several examples of how PDZ interactions can provide a physical link for key signaling and scaffolding proteins involved in synaptic plasticity. PSD-95 interacts with AMPAR subunits indirectly, through transmembrane AMPAR regulatory proteins (TARPs) like stargazin, as well as with other scaffolding and signaling molecules,^13,34 such as A-kinase anchoring protein (AKAP).³⁵ This interaction links the PSD to calcineurin, an AKAP-binding phosphatase necessary for dephosphorylation events that destabilize AMPARs during LTD.³⁵ PSD-95 dissociation from the post synaptic density due to PSD-95 phosphorylation is also involved in AMPAR destabilization during LTD.³⁶ The physiologic relevance of BRAG1s interaction with PSD-95 has not been explored but could contribute to the organization of the PSD scaffold, influence AMPAR stability, and/or localize Arf6 activation, and therefore recruitment of endocytic machinery, near AMPARs to be internalized.

BRAG1 was also shown to co-immunoprecipitate with 2 other scaffolding proteins, PSD-93 and SAP-97, likely through the PDZ-dependent interaction.¹⁰ Like PSD-95, these scaffolding proteins are known to have important roles in the organization of the PSD and synaptic plasticity. One of these proteins may mediate an indirect interaction between BRAG1 and NMDAR subunits, which also immunoprecipitated with BRAG1 but lack PDZ domains¹⁰ and as such, cannot bind directly to BRAG1. This may be essential for localizing BRAG1 near NMDARs, situating it to respond to Ca²⁺ influx and carry out its function in NMDAR-dependent LTD. Interestingly, we found that deficits in LTD caused by knockdown of endogenous BRAG1 levels could not be rescued by co-expression of an RNAi-resistant mutant of BRAG1 lacking the PDZ-binding domain.⁴ This strongly supports the idea that BRAG1s PDZ interactions are essential for its ability to induce GEF-dependent AMPAR internalization, whether by targeting it to an area where it is able to respond to activity-dependent enzymatic activation, linking it to proteins involved in AMPAR destabilizing and endocytosis, or both. While the role of these interactions requires further investigation, the potential for linking BRAG1 to trafficking machinery, scaffolding proteins and AMPARs suggest that the PDZ-dependent interactions may contribute to the specificity of BRAG1-mediated Arf6 signaling.

In addition, BRAG1 contains several other domains with the potential to mediate protein-protein interactions or interactions with cellular organelles, which may be important for providing specificity of G-protein signaling. A pleckstrin homology (PH) domain lies downstream of BRAG1s SEC7 domain.³ In other SEC7-containing GEFs, including the closely-related BRAG2, interactions between the PH domains and phosphoinositides regulate GEF activity, either suppressing or enhancing the enzymatic activity.^11,37,38 Another possibility is that interaction with a phosphoinositide could link Arf6 activation to an important downstream effector. Arf6 activation enhances the activation of 4-phosphate-5-kinase type Iγ (PIPKIγ).²⁴ This leads to formation of PIP₂ on synaptic membranes, which is necessary for the recruitment of components of the endocytic machinery, AP-2 and clathrin.²⁴ Thus, potential interactions between BRAG1 and phosphoinositides could facilitate Arf6-mediated formation of PIP₂, an essential step in initiation of endocytosis.³⁹ It is unknown whether BRAG1 interacts with any phosphoinositides through its PH domain, but this may provide a mechanism for regulation of its GEF activity or link BRAG1-dependent Arf6 activation to a downstream target important for endocytosis.

The coiled-coil (CC) domain of BRAG1 lies near its N-terminus and has been reported to allow the formation of BRAG1 homo-oligomers in response to NMDAR activation.³¹ Disruption of oligomer formation also disrupted the Ca²⁺-dependence of its interaction with CaM, allowing indiscriminate interaction with Ca²⁺-CaM or apo-CaM.³¹ Homo-oligomer formation is likely to disrupt its interaction with PDZ-containing proteins, such as PSD-95, possibly altering the structural organization of the PSD and AMPAR stability. Truncation of BRAG1 to remove the N-terminal CC domain also altered its ability to concentrate at the PSD,³¹ which can be expected to greatly impact its function. Interestingly, CC domains in other SEC7 family members can mediate interactions with the Golgi complex and regulate enzymatic activity. CC-dependent interactions of one SEC7-containing GEF, cytohesin-2, regulates both its GEF activity and its translocation between the cytosol and plasma membrane.⁴⁰ The functional significance of BRAG1s CC domain requires further exploration to determine its significance in regulation of BRAG1-dependent Arf6 activation and AMPAR trafficking.

A final domain of interest in BRAG1 is the proline-rich region. The proline-rich region is characteristic of the BRAG sub-family, as well as the EFA6 sub-family, of GEFs.^11,37 A proline-rich region is found near the C-terminus of BRAG1, upstream of its PDZ-binding domain.³ The BRAG1 proline-rich region interacts with the scaffolding protein IRsp53.⁴¹ IRsp53 itself is an effector of the monomeric G-proteins Rac1 and Cdc42 and plays a role in regulation of NMDARs and spine morphology.^42-44 Interestingly, Rac1 is downstream of Arf6,²³ possibly providing a physical link between the 2 and facilitating Arf6-dependent Rac1 activation. This interaction, along with those of the functional domains discussed above, requires further study to fully understand their contributions to BRAG1 function and the coordination of BRAG1s regulatory stimuli with the specific function of its target, Arf6.

Many questions remain regarding the regulation of G-proteins signaling specificity, links to trafficking machinery and integration of cellular signals with membrane trafficking. Our recent findings show that BRAG1 regulates AMPAR trafficking, with both its GEF activity and PDZ-binding domain being required for activity-dependent AMPAR endocytosis.⁴ This suggests initiation of its enzymatic activity in response to specific signals (e.g., Ca²⁺) as well as an ability to link this activation to Arf6 and the trafficking machinery. The multiple functional domains of BRAG1 provide a means of responding to synaptic activity to regulate its GEF activity, localizing it to the necessary area of the synapse for activity-dependent regulation, as well as linking it to potential downstream effectors of Arf6 involved in trafficking (summarized in Figure 1). Further work focused on the regulation of BRAG1s GEF activity and the role of its PDZ-interactions and other functional domains will provide insight into how regulators of G-protein signaling influence the specific actions of a small GTPase in response to signals, such as synaptic activity.

Figure 1.

Known and putative functions of BRAG1. The functional domains of BRAG1 are shown in the schematic. Known binding partners and functions for each domain are indicated above. Below, putative binding partners and functions with potential roles in GTPase-dependent trafficking events are suggested.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by grants from US. National Institute on Aging (AG032320) to N.Z.G.