Small GTP-binding proteins of the Rab and Arf (ADP-ribosylation factor) families play a central role in the membrane trafficking pathways of eukaryotic cells. The prototypical members of the Arf family are Arf1–Arf6 and Sar1, which have well-characterized roles in membrane traffic or cytoskeletal reorganization. However, eukaryotic genomes encode additional proteins, which share the characteristic structural features of the Arf family, but the role of these ‘Arf-like’ (Arl) proteins is less well understood. This review discusses Arl1, a GTPase that is widely conserved in evolution, and which is localized to the Golgi in all species so far examined. The best-characterized effectors of Arl1 are coiled-coil proteins which share a C-terminal GRIP domain, but other apparent effectors include the GARP (Golgi-associated retrograde protein)/VFT (Vps fifty-three) vesicle-tethering complex and Arfaptin 2. As least some of these proteins are believed to have a role in membrane traffic. Genetic analysis in a number of species has shown that Arl1 is not essential for exocytosis, but rather suggest that it is required for traffic from endosomes to the Golgi.
- GRIP domain
The Arf family of GTPases was originally described as a cytosolic cofactor activity required for cholera toxin to ADP-ribosylate the α-subunit of the Gs heterotrimeric G-protein. Subsequent characterization and cloning revealed that there are five Arf proteins in humans. Arf1 is required for the recruitment of COPI and AP1 (adaptor protein 1)/clathrin coats to the Golgi, and the activation of phospholipase D . Arf6 acts on the plasma membrane to control synthesis of PtdIns(4,5)P2, and hence endocytosis and reorganization of the actin cytoskeleton . The Arf GTPases are members of the Ras superfamily of GTP-binding proteins, but are distinguished from the other members of the superfamily by an N-terminal extension that comprises a site for N-myristoylation followed by an amphipathic helix [3,4]. In addition, in all Ras superfamily members, GTP binding rearranges the two ‘switch’ regions which recognize effectors, but in the Arf family GTP binding also induces the movement of an ‘interswitch toggle’ of a pair of β-strands. In the GTP-bound form, the toggle displaces the N-terminal amphipathic helix from a surface pocket, thereby forcing it to bind to any adjacent bilayer and hence GTP binding stabilizes the membrane association of the GTPase . Such a structural change is also seen in the Sar1 GTPase, which recruits COPII coats to the endoplasmic reticulum, although in this case the N-terminal amphipathic helix is not myristoylated [5,6]. In addition to Sar and the Arfs, there are a number of other ‘Arf-like’ (Arl) proteins whose sequences suggest that they would share this characteristic structural cycle, and in those cases where structures are available this has been confirmed [4,7–9]. There are 17 Arls in humans, some of which are widely conserved in evolution [10,11].
The discovery of Arl1
Arl1 was originally identified by serendipity when a genomic walk around the brahma locus in Drosophila melanogaster stumbled over an adjacent gene that encoded a protein related to mammalian Arfs . This protein bound and hydrolysed GTP, but lacked ‘Arf’ activity in the Gs ADP-ribosylation assay, although such activity could be detected in Drosophila cytosol. Tamkun et al.  concluded that this protein was not a Drosophila homologue of Arfs, and so they termed it ‘Arf-like’ or ‘arl’ for short. This conclusion transpired to be prescient, as homologues of mammalian Arfs were subsequently found in flies . Moreover, several laboratories then used degenerate PCR to look for homologues of Arf-like in mammalian cells. This revealed that there was in fact a family of Arf-related proteins, and so Arf-like was numbered as the first, and hence renamed Arl1 [14,15].
Properties of Arl1
A single homologue of Arl1 is present in the genome of most eukaryotes so far examined, including mammals, budding yeast, plants and even protozoa such as Trypanosomes and Dictyostelium. There appear to be a few exceptions including the primitive protozoa Giardia and, perhaps surprisingly, the fission yeast Schizosaccharomyces pombe. Nonetheless, Arl1 is one of the best-conserved Arls, presumably reflecting a fundamental role in eukaryotic cells [4,11].
Arl1 has all of the typical features of an Arf-family GTPase, including an amphipathic N-terminal helix, and a consensus site for N-myristoylation. Moreover, both human and Saccharomyces cerevisiae Arl1 are substrates for yeast N-myristoyltransferase, and the latter Arl1 has been confirmed to be modified in vivo . Arl1s from a number of species have been shown to bind and slowly hydrolyse GTP [12,16].
Arl1 is recruited to Golgi membranes
Antibodies raised against mammalian (rat) Arl1 stained the Golgi apparatus of rat tissue culture cells by immunofluorescence, and immunoelectron microscopy showed that the protein was restricted to the trans-side of the Golgi [15,17]. Antibodies to yeast Arl1 showed a punctate pattern when used for immunofluorescence, but identification of the corresponding compartment by fractionation was precluded by the protein being entirely cytosolic after cell lysis . However, GFP (green fluorescent protein)-tagged yeast Arl1 was found to co-localize with the trans-Golgi marker Sec7 . Finally, epitope-tagged trypanosomal Arl1 showed a Golgi localization by immunofluorescence . In both mammals and Trypanosomes, mutation of Gly2 results in delocalization, indicating that, as for other members of the Arf family, myristoylation is required for membrane recruitment [17,19].
The presence of Arl1 on the trans-Golgi presumably reflects the action of a Golgi-localized GEF (guanine nucleotide-exchange factor). Genetic studies in yeast have shown that Golgi recruitment of Arl1 requires a second Arf-like GTPase, Arl3 (ARFRP1 in humans), and a small four-transmembrane protein Sys1 [18,20–22]. Arl3 is an atypical Arf family member, in that it lacks a myristoylation site and instead is N-terminally acetylated, with this modification being required for recruitment of the protein to the Golgi by Sys1 (Figure 1). Presumably, Arl3-GTP is required for the recruitment or activity of the GEF for Arl1. Neither the GEF nor the GAP (GTPase-activating protein) for Arl1 has been reported, although it is possible that the Golgi-localized GEFs or GAPs for Arf1 could also be active on Arl1. However, the cytohesin family of Arf GEFs, which share a Sec7 domain with the Golgi GEFs, and act on both Arf1 and Arf6, are not active on Arl1 .
The function of Arl1
The overall function of Arl1 is, like other Arfs and Rabs, presumably to recruit specific effectors to a specific membrane in the cell, in this case the trans-Golgi. Thus its function has been pursued both by searching for effectors and by examining the consequences of removing Arl1 or its effectors from cells. I will consider first the identification of Arl1 effectors, and then the genetic studies on Arl1 and its effectors.
Proteins that bind specifically to the GTP-bound form of Arl1 have been identified by Y2H screens (yeast two-hybrid screens), yeast genetics and affinity chromatography. Y2H screens and yeast genetics revealed that GTP-bound Arl1 binds to the GRIP domain, a conserved sequence of 45–55 residues that is found at the C-terminus of a number of Golgi-localized large coiled-coil proteins [18,22,24,25]. Mammals have four GRIP domain proteins (golgin-245, golgin-97, GCC88 and GCC185), and such Golgi-localized coiled-coil proteins have been proposed to act as vesicle tethers, or contribute to the structural organization of the Golgi [26,27]. Mammalian Arl1 has been reported to interact with the GRIP domains of all four proteins in vitro [9,24], although this is not universally agreed upon . Yeast has a single GRIP domain protein, Imh1, which binds directly to yeast Arl1 [18,22]. Those metazoans examined mostly have homologues of the four GRIP domain proteins found in humans (Caenorhabditis elegans appears to have lost golgin-97), and a single GRIP domain protein is found in all other species with an Arl1, including plants, Dictyostelium and Trypanosomatid protozoa [29–33]. The crystal structure of a complex of human Arl1-GTP bound to the GRIP domain of golgin-245 revealed that the GRIP domain forms a homodimer, consistent with it being part of a coiled-coil homodimeric protein [8,9]. One Arl1-GTP was bound to each GRIP domain (Figure 2), and this bivalency will increase the avidity of the interaction and the selectivity for Arl1 over other Arf family members.
Y2H screens with a GTP-locked form of mammalian Arl1 also revealed interactions with Arfaptin 2/POR1 (a Golgi-localized BAR domain protein of unknown function), MKLP1 (a mitotic kinesin), SCOCO (a short coiled-coil protein of unknown function), pericentrin (a centrosomal coiled-coil protein) and the δ-subunit of phosphodiesterase (PDEδ) [17,25]. For SCOCO and Arfaptin 2, the interaction was confirmed biochemically and shown to be GTP-specific . However, Arfaptin 2 and MKLP1 are also effectors for Arf1, and PDEδ is an effector for Arl2 and Arl3, and so the significance of these observations is as yet unclear as Arf family GTPases can show a greater promiscuity in vitro compared with in vivo [25,34]. Moreover, of these proteins, only SCOCO has a detectable yeast homologue, and this does not appear to bind to yeast Arl1 . Finally, affinity chromatography with yeast Arl1 revealed a GTP-dependent interaction with GARP (Golgi-associated retrograde protein)/VFT (Vps fifty-three), a putative vesicle-tethering complex that acts in endosome to Golgi traffic, and which also binds the yeast Rab6 homologue Ypt6 .
In vivo studies of Arl1 function
Arl1 function has been addressed by ablating the protein by both genetic methods in whole organisms, and by RNAi (RNA interference) or dominant-negative constructs in cultured cells. Mutation of Arl1 in Drosophila results in zygotic lethality, although the basis of this has not been investigated in detail . In contrast, yeast lacking Arl1 grow normally, indicating that the protein is not required for exocytosis . However, such mutants show a partial defect in sorting of proteins to the vacuole and defects in K+ uptake, suggesting mis-sorting of receptors and transporters in the endocytic system [35–37]. In addition, when Arl1 deletion is combined with mutations in genes involved in endosome to Golgi transport, such as Ypt6, there is a much stronger phenotype than with either mutation alone, often resulting in comprised or lost viability . This ‘synthetic lethality’ suggests that Arl1 acts at least in one of the two overlapping or partially redundant pathways for transport from endosomes to the Golgi. At the molecular level, deletion of yeast Arl1 causes delocalization of the GRIP domain protein Imh1 [18,22]. Imh1 deletion shows similar phenotypes and genetic interactions to loss of Arl1, suggesting that Arl1's primary function in yeast may be to recruit Imh1 [37,39]. In contrast, the Golgi localization of the GARP/VFT complex does not appear to be perturbed when Arl1 is absent , and indeed Arl1 is synthetically lethal with GARP/VFT subunits, implying that at least some function of this complex is not dependent on Arl1 .
Studies in mammalian cells have also indicated that Arl1 acts on the Golgi to facilitate the arrival of membrane traffic from the endosomal system. Overexpression of mutant forms of Arl1 locked in the GDP-bound state or GTP-bound state perturb Golgi structure [17,25]. Moreover, removal of Arl1 by RNAi was found to cause golgin-97 and golgin-245 to fall off the Golgi , and endocytosed Shiga toxin to accumulate in endocytic compartments rather than recycle back the trans-Golgi . In addition, overexpression of a GRIP domain-containing fragment has been found to displace endogenous GRIP domain proteins from the Golgi, presumably because it can titrate-out the endogenous Arl1-GTP. In such cells, Shiga toxin recycling was defective, and proteins such as TGN46 which recycle between early endosomes and the Golgi were mislocalized [40,41]. In contrast, proteins that recycle from late endosomes appeared unaffected . At present, it is not clear if these effects are due to a failure to correctly target GRIP domain proteins, or reflect displacement of other effectors. Removal of golgin-97 by RNAi does not appear to affect Shiga toxin recycling , but since the precise role of the GRIP proteins is unknown, it is not clear how much redundancy there is between golgin-97 and the others.
Finally, RNAi studies of Arl1 have also been reported for C. elegans and Trypanosoma brucei. Injection of dsRNA (double-stranded RNA) to remove C. elegans Arl1 had no discernible effect, but the efficacy of Arl1 depletion was not determined . In contrast, dsRNA-mediated removal of T. brucei Arl1 demonstrated that it is required for growth of the bloodstream form of the parasite, although the protein is not even expressed in the insect-borne procyclic form . In the bloodstream form, removal of Arl1 caused only a slight delay in secretion, but the Golgi became swollen and the trans face vesiculated, with vesicles accumulating in the cytoplasm.
Over the last few years, Arl1 has emerged as a protein with a key role in Golgi function in diverse eukaryotes. Although its precise role is probably not fully elucidated, studies from several laboratories have revealed a consistent picture. The protein is present on the membranes of the trans side of the Golgi, and it recruits by direct binding GRIP-domain coiled-coil proteins, and possibly recruits or activates other effectors. These effectors appear to be required for the Golgi to act as an acceptor compartment for traffic from a subset of the endosomal compartments. Alteration of Arl1 activity affects both Golgi structure and recycling from endosomes, and at present it is not clear if one of these phenotypes is a secondary consequence of the other, or if Arl1 and its effectors are directly involved in both structural maintenance and recycling. The identification of the GEFs and GAPs, as well as more Arl1 effectors, and finding the functions of those effectors already known are the next challenges, and the answers should clarify Arl1's true in vivo role.
I am grateful to R. Behnia, K. Röper and R. Sinka for comments on the manuscript and to A. Ridley and M. Seabra for the invitation to the meeting.
Localization and Activation of Ras-like GTPases: Focused Meeting held at the Royal Agricultural College, Cirencester, U.K., 21–23 March 2005. Organized and Edited by A. Ridley (Ludwig Institute of Cancer Research, London, U.K.) and M. Seabra (Imperial College London, U.K.).
Abbreviations: Arf, ADP-ribosylation factor; Arl, Arf-like; dsRNA, double-stranded RNA; GAP, GTPase-activating protein; GARP, Golgi-associated retrograde protein; GEF, guanine nucleotide-exchange factor; RNAi, RNA interference; VFT, Vps fifty-three; Y2H screen, yeast two-hybrid screen
- © 2005 The Biochemical Society