Extensive GTPase crosstalk regulates Golgi trafficking and maturation

Abstract

Virtually all transport events at the Golgi complex are regulated by Arf and Rab family GTPases. Recent work has advanced our knowledge regarding the mechanisms controlling GTPase activity, and it has become clear that GTPases do not act in isolation but rather function in complex networks of crosstalk and feedback. Together with earlier findings, these recent studies indicate that communication between GTPases, their regulatory proteins, effectors, and lipids plays a pivotal role in Golgi transport and cisternal maturation.

Introduction

The Golgi complex is the central sorting station of the eukaryotic secretory pathway. It also serves as the primary site of post-translational modification, including glycosylation and proteolytic processing, of secreted proteins. Multiple trafficking pathways enter and exit the Golgi in both anterograde and retrograde directions. Accordingly, the Golgi is an incredibly dynamic organelle, with protein cargos arriving from multiple origins and sorting to multiple destinations, including the endoplasmic reticulum (ER), the endolysosomal system, and the plasma membrane (PM) [1].

Every trafficking pathway that arrives at or departs from the Golgi appears to be regulated by at least one Arf or Rab family GTPase. These GTPases function as molecular switches, cycling between an inactive GDP-bound state and an active GTP-bound state. GTPase activation is controlled by specific guanine-nucleotide exchange factors (GEFs) that trigger GTP binding. This results in a conformational change that anchors the GTPase to the surface of the Golgi membrane where it recruits effectors through direct protein-protein interactions. Inactivation of GTPases is catalyzed by GTPase-activating proteins (GAPs) that stimulate hydrolysis of the bound GTP to GDP, resulting in a conformational change that leads to release of the GTPase from the membrane surface. Arf and Rab GEFs and GAPs at the Golgi localize to specific sub-compartments, enabling the precisely localized activation of specific GTPases at the correct place and time [2–7].

Recent work in the field has led to important advances in our understanding of the mechanisms underlying the regulation of these GTPase pathways. An emerging theme is that distinct GTPase pathways are intimately connected through extensive crosstalk interactions, in which GTPases regulate the GEFs and GAPs of other GTPases. Other recent studies have provided significant insight regarding the mechanisms governing the Golgi maturation process, in which the composition of Golgi cisternae progress in an ordered series of stages, from the cis-Golgi to the trans-Golgi network (TGN) (Figure 1). Several key factors identified as being important for maintaining the stereotypical maturation timeline are GTPases and GTPase effectors. Therefore, it is likely that a key function of the interconnected network of Golgi GTPases is to maintain the coordinated maturation process.

Figure 1. Arf and Rab GTPases recruit effectors to drive Golgi maturation.

In the cisternal maturation model, early/cis-Golgi compartments form from cargo-containing, ER-derived COPII vesicles. These transient cisternae mature over time, driven by COPI-mediated recycling of resident Golgi proteins, and disperse as transport carriers at the TGN. Note that Golgi compartments are organized as stacks in many, but not all, organisms and cell types, despite high conservation of virtually all core factors. In cells without stacked Golgi compartments, the spatial separation shown in this illustration depicts a temporal progression. Membrane lipid composition also changes during Golgi maturation, with membranes becoming increasing anionic and saturated as compartments progress from cis to trans. GTPase effectors are critical mediators of Golgi dynamics and maturation; vesicle coats and adaptors are designated in green while motors and tethers are depicted in blue.

Golgi Maturation: Evidence and Mechanism

The Golgi complex is polarized: newly synthesized cargos from the ER arrive at the cis-Golgi, and secretory cargos depart from the TGN. In many cell types, the Golgi is arranged as a series or “stack” of flattened cisternae, with the cis-Golgi on one side and the TGN on the opposite side. In some cell types, stacks are less prominent or missing altogether. The budding yeast S. cerevisiae Golgi lacks stacks during the bulk of its cell cycle under normal growth conditions, but stacks appear during mitosis, in response to nutrient depravation, and in certain genetic mutants [8,9]. The dispersed nature of Golgi compartments in budding yeast has enabled researchers to observe the dynamics of individual Golgi cisterna and revealed that the composition and identity of the cisterna change over time, within a matter of minutes. Pioneering work in the Glick and Nakano labs used live-cell imaging of fluorescently-tagged Golgi proteins to observe rapid transitions of Golgi compartments labeled with a cis-Golgi maker to labeling with a trans-Golgi marker [10,11]. This temporal, rather than spatial, polarity of the Golgi in budding yeast has led to use of the terms “early” and “late” to describe Golgi compartments at the “cis” and “trans” stages, respectively. These studies provided strong evidence in favor of maturation as the model best able to explain the relationship between Golgi compartments [8], although other models have been supported by separate studies [12,13].

One caveat of the early studies from the Glick and Nakano labs was that secretory cargo was not visualized. This potential shortcoming has now been addressed through recent studies from the same groups, in which fluorescently-tagged cargos were visualized concurrently with Golgi resident and regulatory proteins [14,15]. These studies provided key evidence that cargo proteins remain within Golgi compartments during the maturation process. A surprising new finding in these reports arose from quantitation of the amount of cargo protein present during different stages of maturation. The amount of cargo appeared to increase as the compartments transitioned from the early to late stages. This was hypothesized to represent the recycling of cargo within vesicles that bud from the TGN and then fuse with the medial/trans-Golgi (Figure 1). The Glick group also provided strong evidence that AP-1, a TGN-localized clathrin adaptor complex, was responsible for these retrograde vesicles [14].

Interestingly, the Glick lab also recently reported that a significant portion of PM-derived endocytic traffic appears to fuse directly with the Golgi complex, at the medial/trans-Golgi, rather than first fusing with an endosome [16]. Therefore, multiple incoming pathways, including vesicles from the PM, endosomes, and the TGN, all appear to fuse with medial/trans-Golgi compartments during the early-to-late transitional period. This leads to the hypothesis that this incoming traffic plays a key role in driving the maturation process during the early-to-late transition.

GTPases play a prominent role in the maturation process. The AP-1 complex requires the GTPase Arf1 for its recruitment to the TGN. Another vesicle coat, the COPI complex, is also Arf1-dependent and is required for normal maturation of the early-Golgi [17,15]. The Segev lab demonstrated that the Rab GTPases Ypt1 (Rab1) and Ypt31/32 (Rab11) are critical for the dynamics of Golgi maturation [18]. Therefore, complete comprehension of Golgi maturation and trafficking requires understanding how activation of Golgi GTPases is coordinated.

Golgi GTPase activation is controlled by extensive crosstalk

Fidelity in membrane transport pathways requires that Arf and Rab GTPases are activated and inactivated in a regulated manner. GTPase activity in the secretory pathway is controlled by GTPase networks, in which the localization and/or activity of specific GEFs and GAPs is directed by other GTPases [19,20]. Early evidence for the existence of GTPase networks came from the discovery that the Rab Ypt31/32 recruits the GEF for the later acting Rab Sec4 [21]. This led Novick and colleagues to propose a cascade model for Rab activation, in which early acting Rabs recruit GEFs to activate later acting Rabs in a pathway, thereby templating recruitment of downstream Rabs (Figure 2A). This GEF cascade mechanism was found to be conserved for numerous GTPases, including other Golgi Rabs as well as Arf and Arf-like (Arl) GTPases [22–26]. Additionally, GEF cascades have been found to occur in which a Rab recruits an ArfGEF, or vice versa, thus connecting Rab and Arf GTPase activation [27–29]. Recently, an interesting bifurcated GTPase cascade was reported in which activation of the ARFRP1 protein drives parallel activation of both the ARL1 and ARL5 GTPases [30]. GEF cascades enforce directionality of transport pathways and therefore provide a potential mechanism for driving organelle maturation.

Figure 2. Extensive crosstalk and long-range cascades regulate GTPase activity during Golgi maturation.

A. Schematic depicting simple GEF and GAP cascades. In a GEF cascade, the early acting GTPase “A” recruits the GEF to activate the subsequent GTPase “B”, thus driving the pathway forward. In a GAP cascade, GTPase B recruits the GAP to inactivate A, thereby preventing GTPase overlap and establishing the distinct “Domain B”. Multiple cascades may be linked together to drive organelle maturation. B. Summary of currently known GEF, GAP, and GTPase interactions at the Golgi, with findings from studies using yeast or mammalian cells combined. Black arrows designate effector recruitment interactions, green arrows denote GEF-mediated activation, and red lines depict GAP-mediated inactivation. Rabs, RabGEFs, and RabGAPs are denoted in blue, and Arf/Arl GTPases, ArfGEFs, and ArfGAPs are depicted in orange. Note that Ypt1 is the closest homolog of mammalian Rab33.

Following the discovery that GEF cascades drive trafficking pathways, Novick and colleagues hypothesized that counter-current GAP cascades might terminate transport steps. In the GAP cascade model, later acting GTPases recruit GAPs to inactivate preceding GTPases in a pathway (Figure 2A). In support of this model, the Rab Ypt32 was found to recruit the GAP to inactivate a preceding Rab, Ypt1 [31]. The GAP cascade mechanism was subsequently shown to regulate inactivation of other GTPases, and also involves crosstalk between Rab and Arf GTPase pathways [32–35]. In combination with GEF cascades, GAP cascades drive organelle maturation by enforcing Rab conversion and directionality of transport pathways. Moreover, GAP cascades promote organelle identity and the establishment of specific GTPase domains by limiting the overlap of distinct GTPases.

In addition to linear GEF and GAP cascades, GTPase activity can be controlled by both positive and negative feedback loops [36–41]. These feedback loops can be relatively simple, for example newly-activated GTPases can potentiate or inhibit activity of their cognate GEFs. Feedback loops may also be more complex and involve additional factors, for example the Vps21 (Rab5) effector BLOC-1 recruits the RabGAP Msb3, thereby enabling Rab5 to influence its own lifetime on membranes [36]. Another recent example from our laboratory is the bi-directional positive feedback loop involving the ArfGEF Sec7, Arf1, the RabGEF TRAPPII, and the yeast Rab11 paralogs Ypt31/32 [41].

Moreover, in addition to direct recruitment by GTPases, GEF and GAP localization can be mediated though effector-GEF, GEF-GEF, or GEF-GAP interactions. For example, the Rab11 GEF TRAPPII was recently shown to recruit the Rab6 GAP Gyp6 [42]. Gyp6 is also a Rab11 effector [35], indicating that the TRAPPII-Gyp6 interaction serves to reinforce Ypt6 inactivation at the late Golgi. This example highlights that the mechanisms regulating GTPase activity are often nuanced and involve input from multiple factors. Moreover, discrete cascade or feedback events do not occur in isolation, but rather are integrated in complex crosstalk networks (Figure 2B) that are likely coupled intimately with organelle maturation.

Membrane composition regulates GTPase localization and activity

While GTPases are largely responsible for controlling vesicle formation, transport, and tethering, lipids also play an essential role in membrane trafficking. Organelle membranes have specific lipid compositions, and phosphorylated derivatives of phosphatidylinositol (PIPs) play an analogous role to Rab and Arf GTPases in defining organelle identity. The composition of Golgi membranes is modified during cisternal maturation, with membranes becoming increasingly saturated and anionic during the transition from early/cis to late/trans (Figure 1) [43,44]. This membrane maturation is driven in part by Arf1, which recruits multiple lipid modifying enzymes and lipid transfer proteins to alter the composition of the Golgi membrane [45].

While GTPases influence membrane composition, lipids play a reciprocal role in controlling GTPase localization and activity. All currently established Golgi GEFs and GAPs are peripheral membrane proteins, which in many cases rely on membrane charge, interaction with specific lipids, or membrane curvature for proper targeting [28,37,40,41,46]. The ArfGEF GBF1 was recently found to interact with specific PIPs through its HDS1 domain, which likely functions analogously to a pleckstrin homology (PH) domain to target GBF1 to the Golgi [47]. Additionally, exchange activity of the ArfGEF Brag2, which functions in endocytosis, is significantly accelerated by the addition of PI(4,5)P₂-containing membranes. Molecular dynamic simulations indicate that Brag2 binds and clusters multiple PI(4,5)P₂ lipids, resulting in close apposition of the GEF to the membrane surface which likely enhances catalytic efficiency [48].

Lipid composition also directly affects association of GTPases with membranes. A recent biochemical study found that multiple Rab GTPases preferentially associate with membranes containing lipid packing defects, which likely facilitates insertion of their hydrophobic C-terminal prenyl modifications [49]. Lipids can also function with GTPases in cascade or feedback-loop mechanisms to drive organelle maturation. For example, the lipid kinase Vps34 generates PI(3)P on endosomes, which was recently found to recruit a Rab5 GAP, thereby enforcing directionality of endosome maturation [50]. Similar mechanisms are likely to be discovered operating at the Golgi complex.

Phosphorylation tunes Rab GTPase function

Phosphorylation is widely used as a signaling mechanism and many Rab GTPases have been found to undergo phosphorylation. However, the effects of this modification on Rab function remain incompletely understood. Several recent studies indicate that phosphorylation can impair Rab function by preventing interaction with effectors, GEFs, GAPs, or the guanine nucleotide dissociation inhibitor (GDI) that binds and solubilizes the inactive form of the Rab. Rab8 (Sec4 in yeast) controls delivery of secretory vesicles to the plasma membrane through its effector the exocyst complex. Sec4 is dynamically phosphorylated by the polo-like kinase Cdc5, which blocks interaction with exocyst to halt membrane delivery during cytokinesis [51,52]. Phosphorylation has also been shown to negatively regulate Rab8 by preventing activation by its cognate GEF [53].

A set of recent studies from the Pfeffer and Alessi groups found that at least 10 Rab GTPases are substrates for the kinase LRRK2, which is commonly mutated in Parkinson’s disease [54–56]. Targeted Rabs were phosphorylated at their switch II regions, impairing association with multiple regulatory proteins including the GDI. Phosphorylated Rabs were stabilized on membranes, indicating that LRRK2 impairs Rab function by preventing GDI-mediated recycling. A similar regulatory mechanism was described in an independent study, in which phosphorylation by TGF-β activated kinase 1 was found to stabilize Rab1 on membranes by preventing recognition by the GDI [57]. Intriguingly, Rab29 recruits LRRK2 to the Golgi to phosphorylate Rab10, suggesting a novel type of cascade in which a GTPase recruits a kinase to phosphorylate downstream GTPases [54].

Conclusions

Arf and Rab family GTPases are critical regulators of Golgi maturation, as they recruit effectors responsible for driving cisternal progression. Moreover, the cascade and feedback mechanisms that control GTPase activity likely contribute to Golgi maturation by enforcing directionality of transport pathways. In order to more fully understand Golgi trafficking and function, it is critical that we continue to define the mechanisms coordinating GTPase activity. To this end, GEFs, GAPs, effectors, and other novel regulators of GTPases can be identified using MitoID, a proximity labeling technique recently developed by the Munro lab [58]. Furthermore, the Novick lab has recently used a “rewiring” approach to provide additional support for the role of Rabs in establishing membrane identity [59]. Such an approach, using chimeric GTPases that are activated ectopically from their endogenous pathways, could be extended to further test the function of GTPase networks in Golgi dynamics.