Role of Rab Proteins in Intracellular Traffic
Colloquium organized by David Sabatini et Bruno Goud from 15 – 20 april 2000
William E. BALCI (Scripps Research Institute, La Jolla, USA), François DARCHEN (Institut de biologie physico-chimique, Paris), Susan FERRO-NOVICK (Yale University School of Medicine, New Haven), Dieter GALLWITZ (Max-Planck-Institute, Gёttingen), Marcos GONZALEZ-GAITAN (Max-Planck-Institute, Gёttingen), Bruno GOUD (Institut Curie, Paris), Jean E. GRUENBERG (Université de Genève), Peter J. NOVICK (Yale University School of Medicine, New Haven), Suzanne R. PFEFFER (Stanford University School of Medicine, Stanford), David D. SAВATINI (New York University School of Medicine, New York), Miguel C. SEAВRA (Imperial College School of Medicine, Londres), Nava SEGEV (University of Chicago), Philip D. STAHL (Washington University School of Medicine, St-Louis), Harald STENMARK (The Norwegian Radium Hospital, Oslo), Thomas SUDHOF (University of Texas, Dallas), Yoshimi TAKAI (University Medical Sсhool, Osaka), Peter Van der SLUIJS (University Utrecht School of Medicine, Pays-Bas), William T. WICKNER, Dartmouth Medical School, Hanover, USA), Ahmed ZAHRAOUI (Institut Curie, Paris).
The cytoplasm of the eukaryotic cell is characterized by the presence of multiple, membrane-bounded, subcellular compartments, or organelles, and the interchange of proteins between them is essential to sustain die activities of the cell. Interorganellar protein transport between the ER, Golgi apparatus, secretory vesicles and granules, endosomes and lysosomes, as well as the plasma membrane, takes place by means of vesicular or tubular carriers, that emerge from a donor compartment, traverse the cytoplasm and deliver their specific cargo molecules to an acceptor compartment, in a process which involves membrane fusion. These transport steps take place in both anterograde and retrograde directions during the complex processes of exocytosis and endocytosis which lead, respectively, to the discharge of proteins at the cell surface or to the intake into the cell surface of nutrient or signaling molecules from the surrounding medium.
Interorganellar vesicular traffic requires the coordinate action of several classes of protein molecules: that in the donor membrane function in cargo selection and vesicle formation, then mediate vectorial targeting of the vesicles through the cell; and at the acceptor
membrane, effect its tethering at a receiving site, followed by docking, which involves a SNARE engagement, and finally membrane fusion.
The meeting that took place at les Treilles from the 15th to the 20th of April, 2000 focused on the role of the rab proteins: a large family of small GTP-binding proteins that in mammalian cells comprises over fifty highly conserved members. Rab proteins are molecular switches that can exist in GDP-bound (inactive) or GTPbound (active) forms. Exchange factors (GEFs) and GTPase-activating proteins (GAPs) function to convert one form into the other. In their membrane bound forms, individual rab proteins have characteristic subcellular distributions that reflect their role in regulating specific steps in transport. Rab proteins exist in the cytosol in their GDP form bound to GDI, a guanine nucleotide dissociation The involvement of rabs in protein transport, was first discovered in the yeast Saccharomyes cerevisiae, which has a total of eleven rab proteins (some of which are also called Ypts in yeast), only four of which (Ypt1p, Ypt31p, Ypt32p, and Seс4p) are involved in the exocytic pathway. The meeting began with presentations by several of the pioneers in this field, who continue to work with yeast.
The Rabs of Yeast
Susan Ferro-Novick (Yale University) has long been interested in ER to Golgi transport and, in particular, in the role of the rab protein Ypt1 and its functional relationship to the SNARE proteins. The latter are membrane components of the vesicular carriers and the receiving compartments (v-SNARES and t-SNARES), whose direct interaction with each other constitutes the docking step that precedes membrane fusion. She has discovered a multiprotein complex named TRAPP (for transport protein particle) that contains ten subunits (including BЕТ3, ВЕТ5 and eight Trs subunits) most of which are conserved between yeast and human cells. Her findings suggest that TRAPP is a receptor associated with Golgi membranes for the COPII vesicles that emerge from the ER. The TRAPP complex would play a tethering role for the incoming vesicle on the acceptor membrane. Activation of the Ypt1 rab protein on the surface of the vesicle, which appears to be сatalyzed by TRAPP, would follow the initial tethering and lead to the recruitment of other tethering factors, such as Uso1p and the Sec34p-Sec35p complex, that are thought to act before docking.
Peter Novick, also from Yale University, first discovered the role of the rab protein Seс4p in the delivery of post Golgi secretory vesicles to the yeast plasma membrane. Sec4p is activated by the exchange factor Sec2p and is bound to the surface of secretory vesicles that are normally targeted and delivered in a polarized fashion to the site of bud formation in the yeast cell plasma membrane.
This budding site is marked by the presence of the protein Sec3p, that ultimately becomes incorporated into “the exocyst“, a protein complex that serves as a landmark in the plasma membrane for sites of exocytosis, and whose localization is independent of ongoing secretion and the integrity of the cytoskelеton. The assеmbly of the exocyst begins with the recruitment of the protein Sec15p by an active Sec4p in the surface of the vesicle. This, in turn, through a chain of protein-protein interactions initiated by Sec15p, leads to the recruitment of other subunits including Sec3p, which completes the assembly of the tethering complex that links the vesicle to its ultimate site of fusion on the membrane. It is notеworthy that, although exocyst complexes are concentrated in the appropriate subdomains of the plasma membrane, the t-SNARES on which vesicle docking must take place are more widely distributed. This would be expected if the specificity of vesicle targeting is conferred by the distribution of the exocyst and not of the SNARES. Novick reported that the polarized delivery of secretory vesicles also depends on the protein actif and on муо2, a type V myosin. Disruption of these cytoskeletal elements, abolishes polarized secretion and cell surface growth. In yeast, the SNARE complexes that mediate vesicle fusion at the plasma membrane involve the vesicle integral membrane protein Snc, as well as the plasma membrane integral membrane protein Sso, and the peripheral membrane protein Sеc9. The v-SNARE — t-SNARE interaction may also be facilitated by the protein Sec1p, which, through the work of other investigators studying other docking events, has become generally regarded as a negative regulator of the docking step.
Dieter Gallwitz (Max Planck Institute, Göttingen) has previously identified an integral multipass Golgi membrane protein, Yip1p, that interacts with two rabs, Ypt1p and Ypt31p, and was proposed to serve as a receptor for those rabs. Hе described another structurally related protein, Yif1p, which forms a tight complex with Yip1p on Golgi membranes. The complex involves the C-terminal membrane-anchoring domain of the proteins, and it was suggested that the N-terminal portions play a role in recruiting components of the vesicle docking machinery. Hе also reported the identification of a family of rab-specific GAP proteins, designated Gyps, that accelerate more than 105 fold the rate of GTP hydrolysis in their cognate Ypts, and described point mutants of Gyp6p, the Ypt6p-specific GAP, that abolish its activity These findings, and a structural analysis, suggested that the Ypt1 GAP functions by a similar mechanism as the GAPs for Ras and Rho-GAP, which require an “arginine finger.”
Nava Segev (University of Chicago), who discovered Ypt1, has found that TRAPP, in addition to serving as a GEF for Ypt1, also manifests nucleotide exchange activity for Ypt31p and Ypt32p, and suggested that TRAPP might coordinate the function of Ypt1 with the function of these rabs in the process of entering and exiting from the Golgi apparatus. In addition, she determined that the capaсity of Ypt1p to hydrolyze GTP is not required for its function, since a GTPase-deficient mutant could replace the endogenous gene product and restore viability. She concluded that GTP hydrolysis is only required for the effective recycling of rabs between membranes. Her work with Ypt31p/Ypt32p indicates that rabs can participate in vesicle formation, as well as in vesicle targeting, since defects in those rabs prevent exit from the Golgi. Finally, the fact that the exchange factor for Ypt1p, TRAPP, is found in Golgi membranes, which serve as an acceptor for Ypt1p-containing vesicles destined to the Golgi, implies that, contrary to prevalent notions, rabs on transport vesicles may be activated at the tethering stage of transport.
Rab proteins contain a double cysteine motif near their C-termini, which is recognized by the geranyl-geranyl transferase that carries out the prenylation needed for rab-membrane association. The enzyme is a heterodimer, that consists a catalytic B component and a substrate recognition component called the rab escort protein, REP. In mammals there are two isoforms of REP: 1 and 2. REР2 is the major housekeeping isoform present in ail tissues, whereas REP1 has a limited distribution and a narrow substrate specificity, which, when mutated, leads to the retinal degenerative defect, choroiderremia (CHM). The REP proteins show sequence similarity (–30%) to GDI and regions of high indentity in these molecules appear to be responsible for their association with rab proteins. GDI, however, only recognizes prenylated rabs, whereas REPs can recognize rabs regardless of whether thеу are prenylated or not.
Bill Balch (Scripps Research Institute) and his associate, Christelle A1ory, took advantage of the sequence similarities, to design a mutagenic approach to study structure-function relationships of the yeast REP protein, Mrs6p, which is essential for yeast growth. They showed that certain point mutations within Mrs6p, in regions corresponding to those that in GDI x-ray crystallography showed are involved in rab binding, inhibit REP binding to rabs in vitro as well as the capacity of REP to promote prenylation.
Interestingly, theу found that certain single point mutations that markedly impair the ability of REPs to bind rabs and to support prenylation in vitro, nevertheless have only minimal effects on yeast growth. This indicated that in vivo REРactivity is normally in functional excess over that required to maintain a minimal pool of active rabs required for growth. Certain double mutations, however, could not support growth and, in fact, served as dominant negative inhibitors of growth in wild type yeast. The normal excess of Mrs6p in the cell was confirmed by the observation that, a short transient period of expression of a temperature sensitive variant of Mts6p, allowed for substantial subsequent growth. Mutations in other regions of the Mrs6p (domain II) had no effeсt on rab-binding, уеt inactivated the activity of the REP to promote in vitro prenylation and reduced its сapacity to support growth of a yeast strain full for the Mts6 gene. Various Mrs6p mutations had differential effects on transport between ER and Golgi or Golgi to the vacuole. It was suggested that this reflects the fact, that, the different residues in the rab Binding pocket of Mrs6p contribute in different ways to the strength of the interaction between the REP and different rab species
Wickner (Dartmouth University) has been studying the homotypic fusion of yeast vacuoles (the yeast equivalent of the lysosome), which is essential for proper vacuolar inheritance after mating and during bud formation, and is needed to maintain a 1ow Copy number of the organelle. Vacuole fusion is regulated by the rab Ypt7p. Wickner has designed an in vitro assay in which the priming, docking and fusion steps can be separated. The priming step, which occurs on separate vacuoles, consists of the disassembly of a large (65s) complex of chaperones, rab effectors and cis v-t SNARE complexes; i.e. v-SNAKE-t SNARE complexes in the same membrane. Priming is driven by ATP hydrolуsis in Sеc18p (the yeast N-etylmaleimide-sensitive fusion protein, NSF) which is promoted by Sec17p (the yeast SNAP or soluble NSF attachment protein). In addition to ATP, priming requires ergosterol and PI (4,5) Р2. Priming leads to release of the HOPS (for homotypic fusion and vacuole protein sorting) protein complex, which consists of at least six Vps (for vacuolar protein sorting)/vam (for vacuolar morphology) gene products, including a Sec1p homolog. HOPS and the SNARE Vam7 then catalyze the activation of the rab Ypt7p and remain bound to it as an effector complex on the surface of the vacuole. The activated Ypt7p:HOPS complex serves to mediate a reversible tethering event, followed by trans SNARE pairing (i.e. complex formation between v and t SNARES in opposite membranes). Trans SNARE pairing is followed by release from the vacuolar lumen of Cal+ ions, which bind to calmodulin, which regulates fusion. Wickner’s observations indicate that Rab effectors can interact with their rabs before rab activation has taken place.
This prompts one to wonder whether a similar situation occurs during other tethering-docking events?
The Rabs of Mammalian Cells
Rab3 and exocytosis in neuronal and neurosecretory cells
Among the rabs in mammalian cells, Rab3A has been of special interest to those studying exocytosis and, in particular, the release of neurotransmitters at synaptic endings of neurons; as well as the release of hormones from neuroendocrine cells. There are four members of the Rab3 subfamily: Rab3A and Rab3C are synaptic vesicle-associated, while Rab3D and RаЬ3В are predominantly expressed outside the brain.
Francois Darchen (Institut dе Biologie Physico-Chimique, Paris), presented his work on the role of Rab3A in the regulation of calcium-sensitive steps in exocytosis. Rab3A normally exerts а negative control on the secretory response. In chromaffin cells, depletion of Rab3A following injection of an antisense oligonucleotide led to а substantial change in the Са2+ dependence of ехоcytosis, measured through the increase in the capacitance of the plasma membrane in а system in which Са2+ levels were controlled by Са2+ dialysis through а patch pipette. It was thus found that when Rab3A levels were decreased, exocytosis was promoted at low levels of Са2+ that had no effect in controls. At high Са2+ concentration, the Rab3A antisense oligonucleotide did not affect the ехoсytоtiс response. In agreement with these findings, in a соmplementary experiment with Aplysia neurons from the buccal ganglion, intracellular chelation of сa2 ions with EGTA greatly potentiated the inhibitory effect on acetylcholine release of a microinjected GTPase-deficient Rab3A mutant protein (Rab3AQ80L). On the other hand, when intraterminal Са2+ was increased by repetitive stimulation, the inhibition bу the Rab3A active mutant protein was released. It was also shown that the intracellular injection of the active form of Rab3A, increased paired-pulse and train facilitation. These effects on short term synaptic facilitation (plasticity) mау reflect a Са2+-depеndent modulation of the function of GTP-bound Rab3A. Overall, these findings suggest that Rab3A mауcontrol a very late step in exocytosis, possibly after the trans SNARE interaction. Because injection of a GTPase-deficient Rab3A markedly delayеd the onset of the clostridial neurotoxin-induced inhibition of acetylcholine release in Aplysia neurons, it was suggested that the activated Rab3A stabilizes the SNARE complex and thus makes the targets of the toxin (the snares VAMP and SNAP-25) resistant to the proteolytic activity of the toxin.
The deactivation of Rab ЗА, which occurs upon stimulation of secretory activity, involves a Rabi-GAP – one of the two mammalian GAPs that have been identified. Darchen reported that the activity of the GAP is not stimulated directly bу Са2+ ions, and showed that mutations in the switch I region of Rab3A disrupt the interaction with its GAP. In addition, rabphilin3, a putative target of Rab3, has long been known to inhibit the activity of the GAP. Mutation of a single arginine residue, arg728, in the GAP abolishes its catalytic activity, but not its interaction with Rab3A, suggesting that the GAP acts by a mechanism similar to that of the GAPs for ras and rho, which also require a critical arginine residue.
Tom Sudhof (Texas Southwestern Medical Center) has long been interested in synaptic transmission, and the role of the Rab3 proteins and their putative effectors in the exocytotic discharge. Hе discussed the remarkable speed by which Са2+ influx triggers exocytosis at the active zone of the synapse. Only one or two synaptic vesicles are thought to be released after a single effective stimulation, from a total of 5-10 vesicles per active zone that constitute the readilуreleasable pool (RRP), i.e. the group of vesicles alrеady docked at the presynaptic membrane. Hypertonic solutions, such as 0.5M sucrose, can also trigger release of synaptic vesicles from the readily releasable pool, and they dо this by a Са2+-independent mechanism. Neurotransmitter release, however, whether triggered physiologically bу Са2+ or by hypertonic sucrose, is prevented by pretreatment with the botulinum or tetanus toxins, which cleave the SNARES involved in docking. These findings indicate that synaptic vesicles in the readily released pool are already docked by assembled SNARE core complexes, and Са2+ simply promotes the final events in fusion. Sudhof went on to describe the phenotypes of knockout mice for either Rab3A, or on or the other of its two putative effectors, rabphilin 3 aid RIM. Rabphilin, which is present in synaptic vesicles, was the first protein shown (by У. Yakai) to bind specifically to an active form of а rab. It contains an N-terminal Rab3-interacting domain, а central phosphorylation domain, and two C-terminal С2 domains that bind Са2+. Ru is localized at the presynaptic plasma membrane. It has an N-terminal Zn finger and а C-terminal PDZ domain, as well as two C-terminal С2 domains, as in rabphilin, but there domains lack the amimo acid residues involved in Са2+ binding.
Although Rab3A knockout mice appear normal, in them the readily released pool of synaptic vesicles was exhausted faster after stimulation, and there was а major increase in the release caused Ьу the Са2+ signal. Thus, synaptic exocytosis was enhanced, short term plasticity was normal, but long term potentiation (LTP) at mossy fiber synapses in the hyppocampal САЗ region, which depends on protein kinase А (PKA), was abolished — probably because Rab3A р1ауs а role in the recruitment of synaptic vesicles to the readily releasable pool for exocytosis.
Rab3A, Rab3C and rabphilin are colocalized in synaptic vesicles and dissociate from the vesicles after exocytosis. The finding that, in Rab3A knockout mice there is а 70% decrease in rabphilin – although none of 20 other synaptic ending proteins, including RI1, are affected – supports а model in which, the association of rabphilin with the synaptic vesicle, is dependent on its recruitment by Rab3A. In contrast to the Rab3A knockout, the rabphilin knockout mouse appears to be totally normal. In particular, it manifests normal LTP on mossy fibers, providing evidence that the Rab3A role in that process is not mediated by rabphilin. Мoreover, Rab3A is normally targeted to synaptic vesicles in these mice, indicating that it is not recruited there by rabphilin. Indeed, the Rab3A distribution within the brain was totally normal in the rabphilin knockout mouse. Sudhof also gave an initial description of the RINI knockout mouse, which shows a behavioral abnormality in not taking care of its latter.
Finally, he pointed out that synaptic vesicles also contain Rab5, a rab that plays an essential role in early endosome fusion. The presence of Rab5 – and the fact that it does not dissociate from synaptic vesicles after extensive stimulation, as does Rab3A – should be considered in light of a model in which, after neurotransmitter release, synaptic vesicles are regenerated from endosomes present at the synaptic endings. It is not known, however, if Rab5 is оnly present in the reserve pool of synaptic vesicles (which constitutes 90% of the vesicles) and not in the active recycling pool of vesicles, that includes the readily releasable pool of already docked vesicles.
Yoshimi Takai (Osaka University) has been a major contributor to our understanding of the cycle of function of rab proteins in mammalian cells, and in neurons in particular. His laboratory isolated, characterized and cloned the c-DNAs for Rab3A, B and C, rabGDI, rabphilin 3, Rabi GEF and the Rabi GAP. The rab GDIs bind to the GDP forms of all rabs and extract them from membranes, maintaining the rabs in the cytosol as a GDI-rab-GDP complex. GDI also deliver rabs to their putative membrane receptors, that may serve as GDI-dissociation factors (GDF). The latter would act before the exchange factors (GEFs’ originally called GDS for GDP dissociation stimulators) promote the release of GDP from their rabs and the Binding of GTP.
There are three forms of GDI, known as a, (3 and y; the first is specifically expressed in neuronal tissue, while the other two are ubiquitously expressed. As mentioned above, GDIs are promiscuous in their capacity to accept and act on the various rab proteins. The rab GDIa gene is located in the distal part of the chromosome Xy28, in a region where mutations have been associated with mental retardation and, in particular, with a form that presents epileptic seizures. Takai reported his findings with GDlα knockout mouse. In these animals, synaptic potentials in the CAl region of the hyppocampus display larger enhancements during repetitive stimulation. This is an effect opposite to that observed in Sudhof’s laboratory with Rab3A knockout mice. In addition, GDIα knock-out mice were hypersensitive to the seizure-inducing effects of the drug bicuculin, an antagonist of the inhibitory neurotransmitter GAGA. Takai interprets his findings as suggesting that rab GDIα plays a specialized role in Rab3A cycling and suppresses hyperexcitability by modulating prеsynaptic forms of plasticity.
Rabs in the endocytic pathway
Many different rabs participate in the endocytic pathway and are localized to its various subcompartments. Rab5 promotes endocytosis aid has been localized to the plasma membrane, clathrin coated vesicles, and еarly sorting endosomes and plays an essential role in early endosome fusion. Rab4 is also associated with early sorting and recycling endosomes and appears to regulate rapid recycling from the sorting endosome compartment to the cell surface. Rab7 participates in transport from early to late endosomes and Rab9 from late endosome to the TGN, in particular in the recycling to the Golgi apparatus of mannose-6-phosphate receptors that transport newly synthesized lysosomel enzymes to late endosomes or incipient lysosomes. Rabl1 is involved in the control of receptor recycling and is primarily localized in the perinuclear recycling compartment, and in polarized epithelial cells in an apical recycling compartment.
Phil Stahl (Washington University) has extensively studied receptor-mediated endocytosis and phagocytosis and, in recent years, his laboratory has focused on the role of Rab5 in endocytosis and endosome fusion and on the mechanisms that regulate Rab5 activity. Considerable progress has been made in elucidating the role of Rab5 in promoting endocytosis and endosome fusion. The latter role is strikingly demonstrated by the highly expanded early endosome compartment present in cells overexpressing wild type or constitutively forms of this rab.
Stahl demonstrated the promotion of endocytosis by Rab5, through measurements of the fluid phase uptake of horseradish peroxidase (НRP), which is enhanced by expression of the GTPase-deficient Rab5 mutant (Q79L) and depressed by the dominant negative one (S34N). This time lapse video fluorescence microscopy analysis of CIO cells expressing GFP-tagged versions of the wild type or Q79L mutant Rab5 showed that the large endosomes promoted by these rabs arise primarily by fusion of smaller vesicles. The fusion appears to occur by two different mechanisms: one leading to an explosive fusion, which takes place in seconds, and another more frequent and slower process, referred to as bridge fusion, which requires one to two minutes. In the latter case, the Rab5 protein can be seen to be concentrated at the site where the fusion occurs. The membrane of one endosome, the donor is then transferred to the acceptor endosome through a very narrow bridge between them. Fluorescence recovery, after photobleaching (FRAP) experiments in Stahl’s laboratory, showed that the Rab5 molecules located at the spots marking the fusion site are not freely diffusible, while other Rab5 molecules in the endosomal membrane are. Previous work from Stahl’s laboratory had shown that Rab5 is activated by upstream factors, including phosphatidylinositоl 3-kinase and Ras. It had been shown earlier by Bar Sagi and Feramisco that microinjection of activated Ha-Ras promotes active plasma membrane ruffling and endocytosis. The protein kinase PKB/akt, which is a component of signaling pathways initiated by various growth factor receptors, is also a regulator of Rab5. In fact, Stahl reported that the Ras-induced enhancement of endocytosis requires both PKB/akt and Rab5. Thus, the stimulation of endocytosis by wild type Rab5 is blocked by expression of a kinase-dead PKB/akt, which, however, has no effect if the cells express the constitutivеly active Rab5. This indicates that PKB/akt promotes endocytosis, by leading to the activation of Rab5. Moreover, the Ras effect on endocytosis in transfected cells is also abolished by coexpression of either a dominant negative Rab5, or a kinase-dead PKB/akt, which also has dominant negative properties. Wortmannin, a drug that inhibits the PI-3 kinase and normally suppresses endocytosis by preventing the activation of Rab5, has no effect on endocytosis in cells expressing constitutively active PKB/akt.
It is now clear that there are three isoforms of Rab5 (a,b,c) that differ primarily at their N and C termini, but are all able to activate endocytosis. It is well known that ligand binding to the epidermal growth factor receptor (EGFR), induces internalization of the complex, and that the internalized receptor continues to signal to its downstream effectors. It is less well known that EGF activation of its receptor also enhances overal1 cellular fluid phase, and receptor-mediated endocytosis. Stahl reported that the latter effect results primarily from the activation of Rab5a. Не also presented a scheme in which he integrated the results of his work with that of others (Honda et al., 1999, Cell 99:521-532) who showed that EGF stimulation (which leads to membrane ruffling) is accompanied by the activation of an Arf, рrobаblуArf6, the small GTP‑binding protein that promotes endocytosis at the plasma membrane, probаbly by stimulating AP2 adaptor recruitment. Ligand binding to the EGF receptor is known to also activate the GTP binding protein Rad, which also enhances membrane ruffling. EGF receptor activation could also contribute to the activation of Rab5, with subsequent enhancement of endocytosis. He also noted that the production of PI(4, 5)Р2, a phosphoinositide required for the activation of dynamin, a protein that controls clathrin coated vesicle-mediated endocytosis, also enhances actif phosphorylation and contributes to enhance membrane ruffling.
Harald Stenmark (The Norwegian Radium Hospital, Oslo) reviewed studies on the Early Endosome Antigen 1 (EEA1), a protein that mediates the action of Rab5 in endosome fusion, and is the most extensively studied rab-effector. EEA1 was originally identified as an autoantigen in some patients with subacute systemic lupus erythematosus, and found to reside together with Rab5 as a peripheral protein in the membrane of early sorting endosomes. EEA1 is a 162 Kda (1411 aa) protein that comprises extensive, coiled-coil regions and contains both N and C terminal sequence motifs, similar to Zn fingers. The N-terminal motif is of the canonical C2H2 type, present in many transcription factors; whereas the C-terminal one (aa 1345-1410) corresponds to a distinct sequence, the FYVE motif, that includes four pairs of cysteine residues and two histidines, and has been shown to bind two Zn2+ ions. FYVE domains also contain a basic patch [R(R/K 11CR] motif, several conserved hydrophobic residues, and an arginine near their C-termini. The name FYVE corresponds to the first letters of the four proteins (Fabl, YOTB, Vaclp and EEA1) that were initially found to contain the motif. Subsequently, it was shown that FYVE domains bind vert’ specifically the phosphoinositide Ptdlns-3P. At present, eleven different proteins have been found to have FYVE motifs. EEA1 strongly binds to liposomes containing Ptdlns-3P and mutations in its FYVE domain that abolish binding to Ptdlns-3P lead to release of EEA1 into the cytosol. Release of EEA1 from membranes is also observed in living cells when levels of the phosphoinositide are diminished by treatment of cells with the Р1-3 kinase inhibitor, Wortmanin. Stenmark studied the сарасity of FYVE domains to bind Ptdlns-3P by Surface Plasmon Resonance (Biocor), and found it to be Zn2+ dependent aid а thousand times stronger than for the phosphoinositide PIP2.
Because the active GTP-containing form of Rab5 had been previously shown to counteract the inhibitory effect of Wortmanin in endosome fusion (observed in Р. Stahl’s laboratory), Stenmark hypothesized, and subsequently demonstrated, that EEA1 binds specifically to Rab5-GTP. This binding was demonstrated in а two hybrid interaction assay, as well as in а direct biochemical assay, using tagged fusion proteins. There are two binding sites for Rab5 in EEA1, one at each end of ‘ the protein. The N-terminal site involves the C2H2 Zn2+ binding domain, whereas the C-terminal one is contained within residues 1277 to 1348, а region that is immediately upstream, but does not include, the FYVE finger.
The structures of two FYVE domains have been determined by к-ray crystallography. Inc group solved at high resolution the ligand-free structure of the FYVE domain of the yeast protein Vps27, and modeled the binding of Ptdlns-3P in а pocket created by the backbone of one ß sheat. The carbon chains of the Ptdlns-3Р were modeled as buried in the membrane, and it was suggested that the FYVE motif stands out from the membrane surface. А different model was proposed by а group that solved the structure of the FYVE domain of Hrs, the putative mammalian homologue of Vps27, which was crystallized with citrate as а substitute ligand. In Hrs, FYVE domains formed homodimers with two ligand binding pockets formed by the (31 strand of on domain and the (34 of the other. In this model the domains lie horizontally with respect to the membrane.
Endosomal targeting of EEA1 requires both binding to Ptdlns- ЗР through the FYVE domain and binding to RabS-GTP. Presumably, this dual requirement restricts the binding of EEA1 to early sorting endosomes. Previous studies by Stah1 on the fusion of endosomes containing GFP-labeled Rab5 showed a concentration of the rab protein at sites where membranes of both endosomal vesicles contacted before their fusion took place. Stenmark remarked that the in vitro fusion assay, immunodepletion of EEA1 from the required cytosolic protein fraction inhibits fusion, and that addition of recombinant EEA1 restores it. It was suggested that EEA1 tethers the membranes of both endosomes through its binding to Rab5 at both ends. Indeed, a C-terminal fragment of EEA1 has a dominant negative effect in the in vitro endosome fusion test system, as well as in transfected BIK cells that express it, together with the active Rab5Q79L and are permitted to endo- cytose horseradish peroxidase. In these cells, the EEA1 C-terminal region (with an intact FYVE domain) counteracted the activity of the Rab5 active mutant; instead of the large endosomal vesicles containing НRP that result from endosome fusion promoted by the active Rab5, clustering of unfused small endocytic structures was observed.
The binding specificity of FYVE domains, has made it possible to devise a probe to localize by immunofluorescence Ptdlns-3P- enriched membrane domains. The probe consists of an epitopetagged recombinant protein containing 2 FYVE domains in tandem. In transfected cells, the probe was found to localize selectively in early endosomes. Using the probe for immunoelectron microscopy, it was shown that Ptdlns-P3P is, indeed, highly concentrated in early endosomes, and that it is present in the internal vesicles of multivesicular endosomes. It was speculated that Ptdlns- ЗР promotes invaginations in the endosomal membrane needed to generate the vesicles, and that once internalized the Ptdlns-3Р is degraded or modified, since the phosphoinositide is not detected in lysosomes. During the invagination hydrolysis of GTP in Rab5 would lead to the release of this rab, as well as of the associated EEA1, from the membrane before formation of the vesicle is completed. Ptdlns-3P mау, indeed, be required for the invagination of the endodomal membrane since others have shown that Wortmanin prevents multivesicular endosome formation.
In the cell, ЕЕА1 mау exist as а parallel homodimer and the homodimer, with а higher affinity for Ptdlns-3P than the monomer, would effect the tethering function needed for endosome fusion. The following scenario was proposed: the assemblу of а tethering platform on the endosomal membrane appears to be initiated by the formation of clusters of Rab5-GTP in the membrane of an endosome. These clusters are proposed to form by virtue of the fact that the exchange factor for Rab5, rabex-5, is in а complex with rabapatin 5 – а Rab5-СТР effector that binds to aid stabilizes Rab5 in the GTP-state. Thus, the exchange factor, once recruited and maintained in the membrane by the Rab5-GTPrabaptin interaction, would act in an amplifying mode, since it would be able to activate adjacent Rab5 molecules, which are in the membrane still in the GDP form. This would lead to the generation of а patch of active Rab5, aid the PI-3 kinase molecules recruited to this patch by their direct interaction with Rab5 GTP, could create а local high concentration of Ptdlns-3P sites, that pro-motes the recruitment of EEA1 bу Rab5. Incoming endocytic vesicles containing Rab5-GTP, acquired at the plasma membrane, would then become tethered to the existing endosome because their rab-GTP would bind to the free N-terminal Rab5-binding domain of EEA1, and this would facilitate the subsequent fusion events. EEA1 would then be able to initiate docking by interacting directly with the SNAKE syntaxin 13 on the endosomal membrane.In fact, Zerial’s group has shown that endosomal membranes contain oligomeric high molecular weight complexes that include rabaptin 5, rabex, ЕЕА1, syntaxin 13 and NSF.
Finally, Stenmark raised the question of whether EEA1 also participates in transport between early endosomes and the TGN. Не also mentioned that rab22 – а rab of unknown function that is ubiquitously expressed and shows 53% identify to Rab5A – is present in the plasma membrane, and is capable of interacting with EEA1 through the latter’s N-terminal RabS-binding domain. When mус-tagged rab22 was expressed in transfected cells it was found to colocalize with EEA1. Stenmark’s comprehensive review of the work on EEA1, illustrates how much has been learned of the detailed molecular mechanism by which RabS regulates endosomal trafficking.
Peter van der Sluijs (Utrecht University) presented his work on the role of Rab4 in endocytic and recycling pathways. Two isoforms of Rab4, а and b, have been identified in mammalian cells, and Rab4a has been localized to early endosomes. Rab4, at difference with RabS, is not present in the plasma membrane. van der Sluijs reported that Rab4 overexpression reduces fluid phase endocytosis and enhances the fraction of transferrin receptors found in the plasma membrane, suggesting а role of Rab4 in promoting recycling from early endosomes to the cell surface. Overexpression of the dominant negative Rab4 mutant (Rab4 S22N) in neuroendocrine РС12 cells led to an expansion of the peripheral tubular component of early endosomes. Because the GTPase-deficient Rab4 mutant (Rab4 Q67L), overexpressed in transfected cells is localized to perinuclear juxta-Golgi vesicular structures, it seems likely that Rab4 participates in transport between sorting endosomes aid the pericentriolar recycling compartment.
Van der Sluijs has identified, using а two hybrid screen, а putative effector of Rab4a, named rabaptin 4, which is nearly identical to rabaptin 5, an effector of Rab5 extensively studied in Marino Zerial’s laboratory (EMBL, Heidelberg). Rabaptin 4, however, lacks а 33 amino acid segment in the C-terminal region of rabaptin 5, close to the previously identified C-terminal site that in rabaptin 5 binds Rab5, and still remains functional in rabaptin 4. van der Sluijs found that both rabaptins 5 and 4 have, in fact, two (N and C terminal) Binding sites for Rab5. Surprisingly, rabaptins 4 and 5 appear to be the products of different genes.
Rabaptin 4 is recruited to recycling endosomes by the active GTPase-deficient Rab4a, and van der Sluijs suggested that whereas Rab5 and rabaptin 5 promote fusion of early endosomes, after fusion the rabaptin 5 on tubules derived from sorting endosomes maу interact with rabaptin 4, recruited by active Rab4 to recycling endosomes. This interaction would Occur through coiled-coil domains present in both rabaptins, and could facilitate the docking of carriers derived from sorting endosomes on the perinuclear recycling compartment that contains Rab4a. Alternatively, rabaptin 4 could be first recruited to the Rab5-rabaptin 5 complex, and then the super complex be brought to the recycling compartment by RabS. These suggestions are partly based on the fact that, in cells doublу transfected with rabaptin 4 and the active Rab4 Q67L, there is a marked expansion of the recycling compartment, identified by its perinuclear localization, its enrichment in cellubrevin aid accumulation of internalized fluorescent transferrin. van der Sluijs also reported that rabaptin 4 binds to the ear domain (between aminoacids 502-592) of the a-subunit of the AP1 adaptor complex localized in early endosomes, but not to the other sub-units of that adaptor.
Finally, he also discussed the cell cyсlе regulation of the phosphorylation of Rab4a, which takes place at serine 196 and is catalуzed by the p34cdc2 kinase. Не pointed out that Rab4b lacks this phosphorylation site aid is, therefore, not modified during mitosis. In addition, a mutant Rab4a lacking the C-terminal domain, which cannot acquire prenyl groups for membrane association, is still a substrate for the phosphorylation. He proposed that the phosphorylation of Rab4 is likelу to be a mechanism to inhibit endocytic membrane transport during mitosis. Phosphorylated Rab4 molecules, in the GTP, as well as in the GDP, states, are released from membranes and accumulate in the cytosol. Whereas in interphase cells, only the GDP form of Rab4 is found in the cytosol, and these molecules are complexed with GDI; in mitotic cells phosphorylated Rab4 GTP is also found in the cytosol. van der Sluijs found that cytosolic phosphorylated Rab4 molecules are present in а complex with Pin1, а 20Кд peptidyl proline isomerase. Pin1 specifically recognizes а phosphorylated (Ser/Thr)-Pro motif in its target proteins and catalyzes the cis-trans isomerization of the proline bond. The resulting change in conformation of Rab4, caused bу Pin1, mау regulate its interaction with other proteins and, in this manner, interfere with the essential role of the rab protein in the recycling pathway.
David Sabatini (New York University) presented the studies on the role of rab i 1 in receptor recycling that he carried out with Mindong Ren and Milton Adesnik. Rabi11 was originally purified from rat liver subcellular fractions in Takai’s laboratory where its cDNA was also cloned. It was later found that there are two rab11 isoforms, а and b, that differ in а few amino acids near the C-terminus. The b isoform, whose function has not been extensively studied, was originally identified in the synaptic vesicles of the cholinergic synapse of the stingray. Rabi was originally reported to be present in post Golgi secretory vesicles of neuroendocrine cells, and also to be particularly abundant within the tubulovesicular system that in gastric parietal cells contains the proton ATPase responsible for acid secretion. In these cells, upon stimulation, the rab11-containing tubular system fuses with the apical plasma membrane canalicular system – а phenomenon akin to а regulated plasma membrane protein recycling.
Endocytosis and recycling of the transferring receptor play an important nutritional role in all cell types, as thеу provide the means for cells to obtain an adequate supply of iron. Holotransferrin, containing bound iron, binds to the transferrin receptor at the extracellular neutral pH. The complex is internalized bу clathrin-dependent endocytosis and delivered to sorting endosomes in а series of processes, promoted by rab5. At the 1ow pH of the sorting endosomes, the iron dissociates from the carrier protein, but the resulting apotransferrin remain bound to the receptor. The iron-free complex is then segregated to tubular regions of the sorting endosomes, whereas other interiorized proteins destined to be degraded, such as LDL, remain primarily in the vesicular portions of the sorting endosome, to be transferred to late endosomes in а process dependent on rab7.
Interiorized transferrin receptors can be returned to the cell surface bу on of two routes, either directly from sorting endosomes, or by passage through а recycling endosomal compartment, which is composed of tubular elements, that in manу ccli types are concentrated in pericentriolar regions of the cytoplasrn, by а mechanism that requires an intact microtubular network.
Previous studies demonstrated an essential role of the active form of rab11 in promoting the passage of the transferrin receptor from sorting to recycling endosomes, and in the control of the release of iron-free interiorized transferrin into the medium. Indeed, in nonpolarized cells, Marino Zerial’s laboratory (EMBL, Heidelberg) and Ren et а1., in the laboratories of Sabatini aid Adesnik (NYU, New York) had found most of the cellular corplement of rab11 is localized in the perinuclear recycling compartment. However, some of the rab11 has also been localized to the TGN and evidence has been presented by Wandinger-Ness (University of New Mexico) that it participates in the transport of some proteins (the VSV-Г, but not imfluenza HA) from the TGN to the cell surface, possibly through the endosome.
Sabatini presented experiments in which CHI cells, lacking an endogenous transferrin receptor, were cotransfected with plasmids encoding the human receptor and one of several гаb11 mutants. They then measured in these cells, the return to the medium of previously internalized labelled transferrin, as well as assessed by immunofluorescence the distribution of the transferrin, its receptor, and rab11. Their observations suggested that exit from the recycling compartment requires hydrolysis of GTP on rab11, since expression of the GTPase-defective rab 11 allowed effective accumulation of transferrin in the recycling compartment but, nevertheless, inhibited release of transferrin into the medium.
Sabatini went on to report the cloning and characterization of а 912 amino acids rab11-binding protein (Rab11BP), that is largely cytosolic, but also exits in а membrane-bound form concentrated in the recycling compartment. The protein, which specifically binds to the active form of rab 1 1 and requires an intact effector domain in rab11, was identified in а biotting assay using rab11-GTP to probe an immobilized denatured ccli extract. The гab11 binding site in Rabi 11BP is located between residues 349-379, and mutation of two EEI tripeptides within this region completely eliminated the binding. The C-terminal region of Rab11ВР, extending from residues 504 to 912, includes seven WD40 domains, which are likely to mediate interactions with other proteins. Intact undenatured Rab11ВР is unable to bind Rab11 in an in vitro binding assay. Within the сеll, however, Rab11ВР must undergo а conformational change in which the Rab11 binding site becomes exposed since, when coexpressed with гab11 in transfected cells, the two proteins form easily detectable complexes in association with membranes. Furthermore, although overexpression of гаЫ1ВР did not affect transferrin recycling, overexpression of а truncated form of гаЫ1ВР(1-504), that includes the гаЫ1 binding site but lacks the WD40 domains, inhibited recycling more strongly; as does а dominant negative rab11 mutant protein that does not bind GTP. Strikingly, the inhibition caused by the truncated гab11ВР was completely prevented when the cells also expressed а C-terminally-deleted, nonprenylatable, form of гab11 that, by itself, has no effect on recycling. Sabatini proposed that Rab11ВР is an effector for rab11, whose association with the GTP-binding protein is dependent on the action of another membrane-associated factor that promotes the unmasking of the rab11-binding site in Rab11ВР. Не also showed – using a permeabilized cell assay with CHO cells that permanently express the transferrin receptor and in which the pericentriolar recycling compartment was preloaded with transferrin – that a truncated form of Rab11ВР, containing the intact rab11 binding site, was a powerful inhibitor of recycling and that this inhibition did not occur when the binding site for rab11 was mutated. This experiment conclusively demonstrated that the active form of rab11 is required for the last step in recycling, i.e. from the pericentriolar compartment to the plasma membrane.
In a search for proteins that interact with Rab11BP, Sabatini’s group identified the sigma subunit of the API adaptor, which has been found to be associated with clathrin-coated vesicles forming on early endosomes. The interacting domain within Rab11BP corresponds to the region encompassing the first two WD40 domains. This suggests a model in which Rab11ВР is recruited to a budding vesicle by the AP1 adaptor. This would alter the conformation of Rab11ВР, exposing the site for binding rab11-GTP This would ensure that rabi11 is contained within vesicles that contain the proper cargo molecules, initially recruited the AP 1 adaptor. Rab11 would then function in a later stage to facilitate post budding steps such as targeting, tethering or docking.
In addition to his studies on Rab3 in neuronal cells, Takai also reported on the isolation and characterization of a protein that binds preferentially to the GTP form of rab11, which he named rabphilin 11. This is the same protein that had been identified and studied by Ren et al. and designated, Rab11ВР. Takai found that in MDCK cells, this protein localizes with the active form of rab11 in the perinuclear region, presumably the Golgi complex and recycling endosomes. In HeLa cells grown on fibronectin, Takai found that rabphilin 11 was also localized along microtubules that extend towards lamiellapodia. Expression of a fragment of rabphilin 11 (aa 607-730) in HeLa cells inhibited accumulation of transferrin in perinuclear regions and also inhibited migration. This is, presumаblу, because the latter process most likely depends on the recycling to the front of the cell of vesicles endocytosed at the trailing end. Не also reported that, in a yeast two hybrid assay, the mammalian Sec13p protein – which is homologous to a component of the COPA coat of the vesicles that in yeast mediate transport out of the ER – binds to rabphilin 11, specifically at a region (residues 504634) that contains the first two of six WD40 domains.
Finally, Takai also presented data showing, by immunofluores- cence, that overexpression of the segment of rabphilin 11 that binds to Sec13p, which was expected to disrupt the endogenous interaction between the two proteins, inhibits the transport of GFP-tagged VSV-G from the Golgi to the plasma membrane, as was also the case when a dominant negative (GDP form) of rab11 was expressed.
Jean Gruenberg (University of Geneva) discussed aspects of the regulation of endocytic trafic by Rab5. Like all rabs, the сарасity of Rab5 to cycle between the cytosolic and membrane bound forms, essential for its function, depends on GDI — the guanyl nucleotide dissociation inhibitor that following GTP hydrolysis extracts rab proteins from membranes. GDI serves as a vehicle that, presumably with the aid of a putative GDI-displacement factor, delivers rab proteins to membranes where thеу can be activated again. It had been known for some time that GDI is phosphorylated on serine residues, and it has been speculated that this phosphorylation controls its function. Gruenberg’s laboratory showed that the activity of GDI in extracting Rab5 from endosomal membranes is markedly stimulated by a cytosolic GDI-activating factor. Incubation of GDI with cytosol, in the presence of the nonhydrolyzable ATP analogues ATPyS or AMPPNP, which prevent its phosphorylation, inhibited the activation of GDI. The cytosolic factor was purified and shown to be a protein complex containing as one subunit the mitogen activated protein kinase (p38 MАPК).
The MAPК, is known to be activated in the stress response cause elicited in cells incubated with 1202 (50µM) or menadione (100µM). As expected, this treatment increased the phosphorylation of GDI and the amount of GDI-Rab5 complexes present in the cytosol. In addition, treatment of cytosolic protein fractions with p38 МАР kinase inhibitors, such as SB203580, reduced their capacity to activate GDI, while it did not affect the phosphorylation state of p38 itself. As expected from these observations, an antibody to phospho-p38 inhibited the cytosol mediated activation. The activation of GDI could also be observed in a reconstituted in vitro system, in which the GDI was incubated with the purified recombinant proteins GST-p38 and GST-МКК6(Е) (a сonstitutively active form of the kinase that activates p 38) and it was shown that Ser121 in GDI was required for the activation. Gruenberg went on to demonstrate that the p38-dependent stress response regulates Rab5 GTPase cycle and rab function in vivo. Thus, 1202 mediated stress decreased the level of Rab5 associated with endosomes and lead to a release of the Rab5 еffector, ЕЕA1, into the cytosol. Previous studies had shown that overexpression of wild type or the constitutively active (GTP) form of Rab5 markedly accelerates endocytosis. The enhancement of MAPK caused by cellular stress had a similar effect, as expected from an acceleration of the cyсle of rab function. On the other hand, the inhibition of MAPK activity (and GDI activation) reduced endocytosis, as has been observed in cells expressing the dominant negative (GDP) form of Rab5.
Gruenberg concluded his presentation with a discussion of the importance of the lipid lysobisphosphatidic acid (LBPA). This highly hydrophobic phospholipid was identified as the epitope recognized by a monoclonal antibody (6G4), which was prepared using endosomal membranes from BIK cells as antigens. Gruenberg aid his associates have shown that LBPA is also а specific antigel for human auto-antibodies, associated with the antiphospholipid syndrome. LBPA is very abundant in late endosomes, constituting up to 17% of the intraluminal membranes that characterize multivesicular bodies and late endosomes. It was hypothesized that LBPA mау be synthesized in situ within the acidic organelles of the endocytic pathway. It is а poor substrate for phospholipases and is, therefore, resistant to degradation in lysosomes. LBPA has the shape of an inverted cone, which might facilitate the invagination of the endosomal membrane and the generation of the intraluminal vesicles characteristic of multivesicular bodies.
One of the important functions of late endosomes is the sorting of the mannose-6-P-receptor (MPR), which after delivering the hydrolases must be returned to the TGN. When cells were allowed to endocytose the 6G4 antibody, abnormal late endosomes were observed that contained in their interior accumulated 6G4 antibodies, together with missorted МРК and 1gр120, а protein normally present only at the endosomal surface. Other authors have found that LBPA faci1itates the degradation of glycosphingolipids and Gruenberg proponed that LBPA also functions in the cholesterol еfflux from late endosomes. Thus, when antibodies to LBPA are interiorized bу endocytosis, as mау occur in patients with antiphospholipid syndrome (aPL), they would bind to LBPA in endosomes causing an endosomal accumulation of cholesterol released from LDL. In Nieman Pick type С (NPC) Syndrome, cholesterol accumulates in late endosomes and lysosomes. Gruenberg suggests that the LBPA-rich membranes of endosomes are affected in this disease, possibly because the protein mutated in NPC normally serves as а sensor of cholesterol levels, and functions in conjunction with LBPA.
Bruno Goud (Institut Curie) has long taken advantage of the intracellular trafficking of the Shiga toxin (ST) to illuminate mechanisms regulating retrograde transport at the interface of the endocytic and biosynthetic pathways. This toxin consists of a catalytic subunit, which has N-glyсosidase aсtivity, and a homopentameric B subunit (B5) that serves as the targeting moiety for the holotoxin. At the сеll surface, the B subunit binds to its globotriaosoyl ceramide receptor (Gb3) and this leads to uptake of the toxin by endocytosis. The toxin there reaches the Golgi apparatus, from where it is transferred to the endoplasmic reticulum, the location where the A subunit is thought to be translocated into the cytosol where it inhibits protein synthesis by removing a specific adenine residue from the 28S ribosomal RNA. Goud and his colleagues have followed the route taken by a recombinant B subunit, which, in some cases, had an added sulfation site that allowed them to monitor arrival of the protein to the TGN region of the Golgi apparatus, where a sulfotransferase adds a radioactive sulfate moiety.
Goud’s laboratory has been studying both the retrograde transfer of the toxin from endosomes to the TGN region of the Golgi apparatus, and its transfer from the Golgi to the ER. Hе presented evidence that the route used by the toxin directly links early recycling endosomes to the TGN, and, therefore, does not include the late endosomes, from where the MPR and furin are retrieved to the TGN. Studies by Goud and his colleagues with HeLa cells indicate that Rabl1 plays a role in Controlling the early step that transfer the toxin to the TGN. The work on the subsequent retrograde transfer from the Golgi apparatus to the ER has revealed that the Shiga toxin utilizes a novel pathway that is regulated by Rab6a. Unlike Pseudomonas exotoxin A (PE) or Choiera toxin – which possess KDEL, or KDEL-like carboxyterminal retrieval signals (absent from STB) and rely on the signal mediated COPI- independent pathway for ER arrival from the Golgi – the Shiga toxin, and the related Shiga-like verotoxins, appear to reach the ER indepеndently of the COPI-mediated mechanism. Goud reported that the Shiga toxin B arrives in the ER at the same rate, whether or not a KDEL sequence is added to it, and that microinjection of anti-COPI antibodies did not inhibit transport of Shiga toxin B subunit to the ER, as it does for the KDEL containing toxins.
The role of Rab6s was investigated in two different types of experiments. Firstly, video microscopy observations of living cells expressing a GFP-Rab6a chimeric protein, revealed its presence in highly dynamic transport carriers that emerge from the Golgi apparatus and rive along microtubules towards the сell periphery, where ER entry sites appear to be located. The carriers are discrete structures with globular or long, tubular shape and, in addition to Rab6a, contain the Shiga toxin, but are distinct from endosomes, lysosomes, Golgi to plasma membrane transport carriers, and the Rab8 containing membrane system. Secondly, a dominant negative mutant of Rab6a, T27N, was found to inhibit the toxicity of the Shiga holotoxin and to block transport of the Shiga toxin B sub-unit to the ER.
Expression of the Rabбa mutant also blocked the normal recycling of Golgi glycosylation enzymes through the ER. There observations are consistent with a previous report from Goud’s laboratory, that expression of the constitutively active Rab6a-GTP, increases the addition of Ga1Nac to the ER resident invariant chain Iip33 of the МНС class II complex, whereas the dominant negative (Т27N) decreases it.
The physiological role of the COPI-independent retrograde transport pathway, regulated by Rab6a, remains to be determined, but Goud speculated that one of its functions could be to effect the recycling to the ER of 1ipids necessary to maintain vesicular flow out of this organelle.
Suzanne Pfeffer (Stanford University) discussed her work on the role of Rab9 and its effector, тIP47, in the recycling of the mannose 6-phosphate receptor from late endosomes to the TGN a transport step for which she has developed an in vitro assay. There are two mannose phosphate receptors (MPR) in mammalian cells:a cation independent one (CI-MPR) of 163 Kda that also binds insulin-like growth factor II, and a cation dependent one (CDMPR) composed of 45 Kda subunits. In the TGN region of the Golgi apparatus, the receptors, which are transmembrane proteins, bind, through their luminal domains newly synthesized lysosomal hydrolases bearing the mannose 6-phosphate marker in their oligosaccharides. The receptor-enzyme complexes are then transported in clathrin-coated vesicles, assembled via adaptors, to late endosomes (LE) or incipient lysosomes, in which the hydrolases accumulate. Until recently, the adaptors were thought to be of the AP 1 type, but are now known to be members of the GGA family of proteins. A fraction of the MPR receptor population transits through the plasma membrane, where their cytoplasmic tails are recognized by AP2 adaptors that mediate their endocytosis in clathrin-coated vesicles. The bulk of the MPR receptor population is, however, recycled back from endosomes to the TGN in a process that requires the participation of Rab9.
A key question concerning the routing of MPRs concerns the nature of the signals in their cytoplasmic tails that, in the different compartments, are recognized by the adaptor-like proteins that mediate the assembly of the carrier vesicles, in which the receptors are incorporated as cargo. Previous work from S. Kornfeld’s laboratory (Washington University) and others have identified sequence motifs in the cytoplasmic tails of the receptors that interact with either AP1 or AP2 adaptors, and may be responsible for the selection of the receptors as cargoes at different sites. Pfeffer indicated that two mechanisms could, and do, operate to ensure selectivity of adaptor recruitment at specific sites. One, involves an organelle-specific coreceptor that specifically enhances, or stabilizes, the interaction of the signal in the receptor tail with the corresponding adaptor. Another, is an organelle, or site-specific modification of the receptor tail, that enhances its association with the adaptor.
Pfeffer has identified a 47 Kda protein (TIP47) that binds to the cytoplasmic domain of both types of MPR, but does lot bind to the LDL receptor, or to the cytoplasmic domains of furin or TGN38 – two protein that also recycle from the plasma membrane to the TGN through endosomes and bind to the clathrin adaptor AP2 at the cell surface. Endocytosis of the CI-MPR is directed by the sequence YKYSKV whereas the CD-MPR uses a phenylalaninе sequence, FPHLAF, as well as a YRGV sequence. These signals are likely to interact dirеctly with the AP2 clathrin adaptor. In the CDMPR, a C-terminal dileucine motif, LLHV and the tyrosine based endocytosis signal YKYSKV are important for lуsosomal enzyme delivery to lysosomes. The CD-MPR also contains the dileucine signal HLLPM, which is also important for enzyme sorting. These signals are thought to be binding sites for the API. adaptor.
Pfeffer showed that TIP47 is present in the cytosol as well as on endosomal membranes, and that it is required for trafficking of MPR from endosomes to the TGN in vivo and in the in vitro system, which she developed. A signal containing W and F residues in the tail of MRP, appears to be essential for retrieval of the receptor from late endosome to the TGN, sine Kornfeld and others have shown that mutation of that sequence causes mislocalization of the MPR to lysosomes, probably because the MPR fails to be retrieved from endosomes. She has found that late endosome (LE) to TGN in vitro transport is markedly inhibited by immundepletion of TIP47 from the cytosol, and that transport is restored when the recombinant protein is added. Moreover, she showed that in cells depleted of TIP47 by antisense treatment, MPRs undergo a rapid degradation, as expected if TIP47 serves as a cargo selection devic (adaptor-like or coat-like function) that mediates retrieval of the MPRs from endosomes. It was previously shown that a reversible palmitoylation of the tail of the 45 Kda receptor may Occur in endosomes that creates a loop in the polypeptide tail. The loop would serve to present the WF signal in the receptor tail to TIP47, thus promoting efficient return of the receptor to the TGN, preventing its accumulation in lysosomes. The discovery in her laboratory that Rab9, in its GTP state, binds to TIP47 and that the two form a ternary complex with the cytoplasmic domain of the MPR, suggest that active Rab9 located in endosomes plays a key role in ensuring the selectivity of TIP47 binding to MPR molecules in the endosomal membranes, rather than in other membranes. Thus, the dissociation constant (Kda) of TIP47 for the cytoplasmic domain of the CI-MPR is 1µМ and TIP47 is present in the cytosol at the 1ow concentration of 300nM, which is sufficient to bring only —10% of TIP47 molecules to membrane-associated MPR. However, the presence of Rab9-GTP decreased the Kda of Т1Р47 for MPR to 300nM, a value, which in the cell, would ensure that 50% of the TIP47 molecules, upon binding to Rab9, can bind to MPR. Thus, Pfeffer inferred that sine Rab9 enhances the association of TIP47 with endosomes and binding to its cargo (the MPR), this rab functions during vesicle formation. In fact, expression of a dominant negative mutant Rab9 (S21N) does not lead to the accumulation of transport vesicles containing MPR, implying that the active rab is required for exit from late endosomes. Pfeffer, therefore, concluded that Rab9 is a coreceptor in endosomal membranes for the interaction of TIP47 with MPR.
Pfeffer mentioned that phosphorylation of the cytoplasmic tail of the MPR takes place as it leaves the TGN, and that dephosphorylation, which occurs in the late endosomes, facilitates the association with TIP47 aid Rab9. She noted that E Maxfield’s laboratory (Cornell Medical College) laboratory has shown that TGN38 is returned from the plasma membrane to the TGN via recycling
endosomes, whereas furin, like the MPR is returned via late endosomes. Nevertheless, the cytoplasmic tail of furin does not appear to be recognized by TIP47 aid must be sorted by an alternative mechanism.
Marcos A. Gonzalez Gaitan (Max Planck, Gottingen) presented his studies on the role of endocytosis in the formation and maintenance of a long range concentration gradient of the Drosophila Melanogaster TGF-β growth factor homologue, Decapentaplegic (Dpp). This protein is a wing morphogen, i.e., it especifies cell fate along the anterior posterior (A/P) axis of the wing. Dpp is produced and secreted by a strip of cells at the wing anterior/posterior (А/P) compartment boundary. It subsequently forms a long-range activity gradient, necessary for the activation of different target genes in cells at different distances, to which it confers positional information from the source. Gonzalez Gaitan presented evidence that the morphogen travels through the tissue by an endocytic pathway, involving uptake via its receptors, Tkv and Punt. Intracellularly, DPP can effect its signaling function, which involves activation of the transcription factor MAD, and leads to expression of the genes 0MB and Spaltz. Expression of the latter, was used by Gonzalez Gaitan, as a measure of the extent and slope of the Dpp gradient. These properties of the gradient are determined by the fraction of interiorized Dpp, that either undergoes recycling to the extracellular space, or degradation through the late endosome/lysosome pathway.
Gonzalez Gaitan used transgenic and mosaic flies expressing Green fluorescent protein (GFP)-tagged Dpp, which could be secreted at the A/P boundary in a controlled manner by a change in temperature, which affects the level of expression of the transcription factor Ga14. For these studies he also exploited the availability of mutant Hies with temperature sensitive mutations in α-daptin, clathrin heavy char, and dynamin, which at the nonpermissive temperature, are defective in endocytosis. By establishing blocks in endocytosis, he obtained phenotypes that resemble those of flies defective in Dpp or its receptors. Hе also was able to test aspects of the hypothesis that the shape and extent of the morphogen gradient are controlled by a balance between its recycling and degradation, using transgenic flies with mutations in various rab proteins that promote, or inhibit, traffic in the endocytic pathway. Не also expressed the same rabs as fluorescently tagged chimeras that could be localized intracellularly, at various stages of development. Expression of а dominant negative Rab5 restricted the signaling range of Dpp. Conversely, overexpression of an active Rab5 expanded this range. As expected from his model, overexpression of Rab7, а rab that promotes transport from early sorting endosomes to late endosomes and, hence, enhances degradation of the interiorized ligand, reduced the expression of the reporter gene to а narrow region in the vicinity of the А/Р boundary. These studies unveil а novel and important role of endocytosis and recycling in Controlling the establishment of the morphogenetic gradients responsible for pattern formation in development.
Ahmed Zahraoui (Institut Curie) has long been interested in the regulation of the assembly of the tight junctions that separate the luminal and basolateral domains of polarized epithelial cells. А number of rab proteins have been implicated in the development and maintenance of epithelial polarity. Thus, it has been reported that rab8 controls the targeting of proteins to the basolateral surface of epithelial cells; that Rab3B regulates traffic at regions near the apical junctional complex; and that rab13 controls the assemblу of the tight junctions themselves. Не noted that гab13 has а high degree of homology to rabs 8 aid 10, and to the yeast Sec4p, which is involved in polarized secretion during bud formation. In fibroblasts, rab13 is associated with vesicles dispersed throughout the cytoplasm, but in polarized epithelial monolayers, such as those of Сасо-2 cells, rab13 colocalizes in the plasma membrane with ZO-1, а peripheral membrane protein that is specific component of tight junctions. Zahraoui and his associates have also shown that, both in epithelial and in endothelial cells, the characteristic distribution of гаb13 depends on the integrity of the tight junctions.
The precise function of rab13 remain unclear, although it was hypothesized that in epithelial cells this rab may play а role in regulating the targeting of at least some membrane proteins to the two plasma membrane domains, or in modulating the barrier function of the tight junctions. Using а yeast two hybrid screen, Zahraoui’s group has identified as а гаЫ3 interacting protein (i.p). the protein that in the retinal rods serves as the delta subunit of the phosphodiesterase (PDEδ) involved in signal transduction. That protein interacted with гаb13 in а biochemical assay, but did not interact with rab8 or rab6.
The δPDE subunit was first identified as а polypeptide that can be coprecipitated with the soluble а and (3 subunits of the PDE in retinal rods, but the 5 subunit does not exert а regulatory action on the activity of PDE and, in contrast to the other subunits of the retinal PDE, is expressed in many tissues. This suggests а different function for δPDE, which must be conserved during evolution, since proteins with high degree of identity (75%) to 5 are present in C-elegans and in human cells.
Other investigators had shown that the bovine retinal rod δPDE promotes the solubilization of membrane-bound PDE а and subunits. Zahraoui, and his associates, showed that purified human δPDE also has the property of removing, in а stoichiometric manner, а large fraction (50-60%) of the гаb13 molecules present in membranes preincubated with GDP – without affecting the distribution of rab6 or Rab4. Strikingly, neither the а or isoforms of GDI were able to remove гаb13, although theу could effectively extract rab6 and Rab4 from the same membranes. It was noted that the C-terminal region of δPDE contains sequence motifs (SRV and FYV), that have been described as necessary and sufficient to mediate an interaction with the PDZ domains that characterize manу proteins found at synaptic contacts, or in septate or tight junctions. PDZ domains (so named for the first letters of the proteins PS095, D1g and ZO-1, in which they first identified) are thought to mediate the assembly of submembranous protein networks at specific plasma membrane microdomains. The localization by immunofluorescencе of a significant fraction of an epitopetagged version of the δPDE, expressed in transfected cells to vesicular structures near the plasma membrane, led to the suggestion that docking of those vesicles with their membrane targets mаy provide a means to remove rab13 from those membranes. Zahraoui also reported that rab13 is found in the cytosol in a complex with δPDE and nit with GDI. He, therefore, proposed that δPDE and not GDI, normally controls the dynamics of the association of this rab with cellular membranes.
Finally, he also reported that the distribution of apical, basolateral or junctional markers was not altered in permanent transformants of MDCK cells that express constitutively the dominant negative (GDP) or the GTPase-deficient and constitutively active (GTP) rab13 mutant proteins. However, treatment of cells expressing the constitutively active rab13 mutant with the phorbol ester, TPA, caused a dramatic change in the morphology of the cells. The PKC activator led to the appearance of prominent cellular processes, and the concomitant dissociation of the intercellular junctions. Hеconcluded that an active rabl3 plays a role in the dissociation of tight junctions.
The meeting concluded with a presentation by Miguel Seabra (Imperial College School of Medicine, London), whose laboratory has carried out a comprehensive analysis of mammalian rab protein sequences, searching for characteristic motifs, or rab specific features, that maу serve to define the rab protein family and to distinguish a putative rab unambiguously from other small GTPases. It is widely agreed that the Ras superfamily includes five different families (Ras, Rho/Rac, Rab, Arf, Ran), of which the rab family is the largest. Rab proteins are prenylated by the enzyme Rab geranylgeranyl transferase, and the presence of the double cysteine prenylation motif – which may exist in several configurations (XXXCЕ, XXCCX, XCCXX, CCXXX, or XXCXC) – at the C-terminus of a protein, allows its recognition as a rab. There are, however, bona fide rabs, recognized by other criteria, such as Rabs 8 and 13, that have only a single C-terminal cysteine residue (CXXX).
Like all GTP-binding proteins, rab proteins contain sequences involved in binding the guanine (G1-G3) or the phosphate (PM1 — PM3) moieties, but these regions are conserved in all members of the Ras superfamily and cannot be used to recognize a candidate protein as a rab. Using computer algorithms and profile hidden larkov models, Pereira-Leal aid Seabra were able to recognize the existence of ive highly conserved short stretches of residues, designated F1 to F5, that define the rab family. Although these sequences are rab specific, some residues in them are also conserved in other small GTPases. Seabra showed the location of the F motifs within the solution structure of Rab3A-GTP, which has been solved by NMR. The F1 feature (prototypical sequence: GVDF) is located within the, so-сalled, effector domain (loop 2 —β2). The other prototypical sequences and their relation to the Rab3A structure are: F2: KLQIW (β 3); F3: RFsiT (loop 4); F4: YVRGA (α 2 — loop 5); F5: LVYDIT ((β 4 — loop 6). F4 is almost adjacent to F3 and is followed closely by F5. The F1-F6 features allowed the identification as rabs of 52 sequences in Genebank. Seabra showed among these 52, several newly identified rabs and proposed their renaming. A plylogenetic analysis of the complete rab family carried out using a neighbor joining algorithm, revealed clusters of related rabs. Within these clusters, rabs with unusually high homolоgy, are designated isoforms (e.g. Rab1a and Rab1b), while others are defined as subfamilies, and it is sometimes problematic to differentiate between “related rabs,” “rab isoforms” and “subfamilies”.
Within each rab family sequence, subfamilies (SF1-4) can be recognized whose members show substantially higher identity between themselves, than to the general sequence that defines the subfamily. Most notably, among these is the rab F4 subfamily motif, which is located within the so-called hypervariable C-terminal domain of the rabs. Among the whole rab family the average identify in this region is 1ow (–14.4%) but within the F4 subfamily the average identity is 58%.
The 3D structure of Rab3A in complex with its effector, rabphilin ЗА, is now also available. It shows that certain regions of the polypeptide, designated rab complementary regions (rab CDRI, II and III) form a pocket in the rab for binding of the effector, and these regions are found to correspond to subfamily specific sequences. The finding that two subfamily specific surfaces are present in Rab3A, suggest that this rab could interact with more than one effector, and that, in general, different effectors could bind to different combinations of rab subfamily (SF) regions. Switch regions, i.e. regions thought to be affected by the nucleotide state of the rab, are part of these surfaces, which could explain the nucleotide sensitivity of effector binding.
In summary, the recent completion of the human genome sequence has revealed the existence of 60 rab protein genes, but Seabra showed that taking into account the five rab family specific motifs (F1 to F5), in conjunction with the double cysteine box and the PM/G landmarks, the unequivocal definition of a GTPase as a rab can be made. The challenge is now to relate the structural sequence, and structural features, of specific rabs to their regulators (e.g. GEFs, GDI, REPs) and effectors to then understand the, apparently, multiple and diverse roles of rab proteins within the cell.