GAK x Tubulin

Giardia lamblia is a major cause of diarrhoeal disease worldwide. Since it has no known toxin, the ability of trophozoites to colonise the human small intestine is required for its pathogenesis. Mitosis in this protozoan parasite is a unique challenge because its two equivalent nuclei and complex cytoskeleton must be duplicated and segregated accurately. Giardial mitosis is a complex and rapid event that is poorly understood at the cellular and molecular levels. Higher eukaryotes have one to three members of the highly conserved Ser/Thr aurora kinase (AK) family that regulate key aspects of mitosis and cytokinesis. Giardia has a single AK orthologue (gAK) with 61% similarity to human AK A. In addition to the conserved active site residues, activation loop and destruction-box motifs characteristic of AKs, gAK contains a unique insert near the active site region. We epitope-tagged gAK at its C-terminus and expressed it under its own promoter. During interphase, gAK localises exclusively to the nuclei, but is not phosphorylated as shown by lack of staining with an antibody specific to phosphorylated AK A (pAK). In contrast, during mitosis pAK localises to the basal bodies/centrosomes and co-localises with tubulin to the spindle. During specific stages of mitosis, giardial pAK also localised dynamically to cytoskeletal structures unique to Giardia: the paraflagellar dense rods of the anterior flagella and the median body, whose functions are unknown, as well as to the parent attachment disc. Two AK inhibitors significantly decreased giardial growth and increased the numbers of cells arrested in cytokinesis. These inhibitors appeared to increase microtubule nucleation and cell-ploidy. Our data show that gAK is phosphorylated in mitosis and suggest that it plays an important role in the Giardia cell cycle. The pleiotropic localisation of AK suggests that it may co-ordinate the reorganisation and segregation of tubulin-containing structures in mitosis. We believe this is the first report of a signalling protein regulating cell division in Giardia.

Contrary to popular belief, there can be dry weather in the English Lake District and a week of glorious weather was enjoyed by the delegates at the 55th Harden Conference, staying at St Martins College, Ambleside. One might have expected, given the beauty of the landscape (Figure 1) and the rarity of dry weather in this part of the world, that a week of sunshine would have been serious competition for a meeting. However, thanks to the great organisation of the scientific programme by Viki Allan and Bob Burgoyne and the extremely high quality of talks and posters, the great outdoors got very little sight of the delegates (other than during the coffee breaks). One of the notable features of the meeting was the blend of long talks with many selected abstracts, which produced a wide range of talks and also different viewpoints on the same subjects. However, given space constraints and to avoid producing a list of speakers and titles, this report is focused on the talks from invited speakers. Blue skies over Lake Windemere highlighted the beauty of the English Lake District, an inspiring setting for discussion on membrane traffic. Since both membrane dynamics and organelle distribution are critically dependent on the cytoskeleton and molecular motors, it was an appropriate topic for the opening two sessions of the meeting. In the first talk, Viki Allan (University of Manchester) reported her findings using mammalian cell lines stably expressing the Golgi marker NA-GFP to investigate the role of cytoplasmic dynein 1 in segregation of the Golgi during mitosis. In prophase, in addition to the spindle-pole-directed movement an accumulation of peripheral structures was also detected. Although the central Golgi clusters still move in early prometaphase, indicating dynein is still active, no movement could be detected by metaphase. Dynein is also responsible for the cell-cycle-dependent movement of ER membranes in Xenopus egg extracts. In metaphase ER movement is decreased by 10-fold, in keeping with a decrease in the amount of CD and dynactin on the membrane. This is closely tied to the phosphorylation of the light intermediate chains (LICs) by cdc2kinase. A conserved site in the N-terminal domain of the LIC had previously been thought to regulate this association; however, recent experiments indicate this is constitutively phosphorylated, and therefore an unlikely site for regulation. A site close to the C-terminal region may be a more likely candidate, as suggested by use of phosphothreonine antibody. Results from experiments assessing the phosphorylation state of LIC and IC during somatic vs. embryonic mitosis highlight the importance of this post-translational modification in modulating motor activity. The Xenopus melanophore provides a unique and versatile model system for investigations into the coordination of motors, as described by Vladimir Gelfand (University of Illinois, Urbana-Champaign). In this system, melatonin induces dynein-mediated aggregation of pigment granules, with the bulk of movement going into the cell centre. In contrast, application of melanocyte-stimulating hormone (MSH) working through cAMP, directs movement toward the cell periphery. In addition to kinesin II, myosin V is also a component of the motor machinery involved in dispersion. Why are two motors necessary? Organelle-tracking experiments indicate that myosin V shortens dynein-driven minus-end runs during dispersion and by pulling organelles off microtubules. Therefore, one attractive hypothesis is that microtubules are used for rapid long-distance travel, whereas actin tracks serve for local transport to locations not accessible to the microtubules. Is the regulation due solely to a decrease in the amount of bound motors, or due to regulation of the activity of the motor? The answer seems to be both. Melanosomes from aggregating cells not only have substantially less myosin V than melanosomes from dispersing cells as assessed by Western blotting, but there is also an increase in phosphorylation of myosin-V, suggesting again that phosphorylation may play a role in disassociation of motors from their target membranes. Although phosphorylation by CAMKII at S1650 is sufficient to completely dissociate myosin V from the melanosome surface, it is probably not the only kinase involved in regulation. As pointed out in the melanophore system, as well as microtubules and their motors, the actin cytoskeleton and unconventional myosins also perform critical and extremely diverse functions in membrane transport and organelle distribution. The capabilities of myosin VI in this regard were described by Folma Buss (Cambridge Institute for Medical Research, Cambridge). Myosin VI associates with the endocytic pathway, the Golgi complex and plasma membrane ruffles. It would be predicted therefore that associating proteins might mediate organelle-specific interactions and, indeed, in a yeast two-hybrid screen a number of myosin VI binding partners were identified, including one protein associated with endocytosis, Dab2, and two proteins associated with the Golgi complex. In addition, a new protein was isolated which has been predicted to play a role in cell migration. Investigations into all aspects of myosin VI function may be furthered by characterisation of Snell's waltzer mouse. A null mutation in myosin VI in these mice results in fusion of stereocilia and subsequent loss of inner-ear hair cells, leading to deafness. Stable fibroblast cell lines have been isolated, and effects on Golgi morphology and function have been assessed. They not only have a 40% decrease in the size of the Golgi but also show decreased secretion of alkaline phosphatase. Both of these defects can be rescued by transfection of a full-length myosin VI. Since an extensive study of defects other than deafness in Snell's waltzer mice has yet to be conducted, it will be extremely interesting to see whether there are aberrations in processes such as neuronal migration and gastrointestinal function as well, all due to a lack of a multitalented myosin VI. Trina Schroer (Johns Hopkins University, Baltimore) explored some novel roles for dynactin in nuclear receptor trafficking as well as localisation of late endosomes. Over-expression of dynamitin resulted in constitutive import of glucocoticoid receptors (GRs) into the nucleus bypassing the need for ligand binding. GRs are normally held in the cytoplasm on microtubules (MTs); therefore, on MT disruption or dynamitin disruption, GRs enter the nucleus constitutively. This suggests that ligand binding may dissociate the GRs from the MTs, perhaps by interfering with the binding to dynactin complexes. Expression of one of the dynactin subunits, p27, a component of the ‘pointed end complex’, resulted in decreased nuclear transport, further supporting the proposition that it serves as a tether for GRs. Switching from the nucleus to endocytic compartments, the need for motor pairs directing movement and localisation of organelles was highlighted. The effects of dynamitin overexpression on endosomal localisation indicate a requirement for dynein-driven inward movement. In contrast, in cells overexpressing mutant kinesin 2, LAMP-1 positive structures collapsed toward the middle of the cell without any effects on the localisation of early endosomes or the recycling compartment. The interaction of the microtubule and actin cytoskeleton in governing exit pathways from the TGN was explored by Enrique Rodriguez-Boulan (Weill Medical College, Cornell University, New York). Results were presented from experiments looking at synchronised post-Golgi transport events using live-cell imaging and microinjection of cDNA encoding an apically targeted p75-GFP and two markers for the basolateral pathway, LDLR-GFP and NCAM-GFP. Exit of apical and basolateral membrane proteins occurs with different kinetics and with different ratios of vesicular and tubular post-Golgi carriers. These ratios differ in polarised vs. nonpolarised cells. Inhibition of kinesin not only blocks exit of apically targeted p75 but also blocks formation of tubular transporters, as does inhibition of dynamin function. Using this system, it was also shown that mutant cdc42 not only inhibits exit of basolateral plasma membrane proteins from the Golgi but also stimulates exit of apical plasma membrane proteins. Regulation of Golgi exit pathways by cdc42 could be due to regulation of actin polymerisation or remodelling of the actin cytoskeleton, both of which might affect Golgi polarity, or regulation of interactions with adaptor and coat protein – or perhaps even a disruption in myosin VI function? The second part of the talk focused on sites of exocytosis of post-Golgi transporters in nonpolarised vs. polarised cells using spinning disc confocal and evanescent wave microscopy. It could clearly be seen that there was random fusion of both apical and basolateral post-Golgi transport intermediates (PGTIs) in nonpolarised cells, but only apical fusion of p75-GFP and lateral fusion of LDLR-GFP in polarised cells. The sites of fusion of apical and basolateral transporters correlated with the localisation of syntaxins. In nonpolarised cells, syntaxin 3 and 4 are random but upon polarisation, syntaxin 3 appears apically and syntaxin 4 and the exocyst laterally; both are absent from the basal membrane. Interestingly, the apical localisation of syntaxin 3 and apical fusion of p75-GFP were dependent on intact microtubules; this was not the case for syntaxin 4, exocyst or the lateral fusion of LDLR-GFP. Tubule formation out of the TGN continued to take centre stage with Alberto Luini's (Mario Negri Institute, Italy) talk focusing on the formation of post-Golgi carriers (PGCs) transporting VSV-G to the plasma membrane. Using an extraordinary approach correlating structures seen by live cell light microscopy with EM, he showed that the majority of transporters formed consist of tubular precursors which emanate along microtubules from the Golgi mass. The tubules detach due to tension (or in a manner facilitated by tension), releasing PGCs that are large and pleiomorphic. To test whether vesicles fuse to form these intermediates, cells were microinjected with antibodies to NSF. The results indicate that intermediates form faithfully and still look like pieces of the TGN; hence formation of these membranes is independent of individual fusion events. Formation of PGCs appears to be from TGN export domains that exclude resident markers and include machinery like SNARES. Intriguingly, they still contain TGN46, which is resident in the TGN but does recycle through the plasma membrane. The role of lipids in the formation of these carriers was then described. BARS-50 is an acyl transferase that specifically acts on lysophosphatidic acid (LPA) and induces fission of Golgi tubules. BARS overexpression inhibits VSV-G transport from TGN - PM by blocking formation of tubular carriers. BARS (is proposed to) work by interfering with PA metabolism and therefore, since PA and LPA have a different geometry, it was proposed that they may be instrumental in the formation of these carriers. The sessions on the second day opened with Rainer Pepperkok (EMBL, Heidelberg) describing work on the temporal and spatial organisation of membrane traffic from the ER to the Golgi. Using multicolour time-lapse imaging and FRAP, the movements of a number of GFP tagged proteins were studied. COP I and COP II coats were proposed to act sequentially and COP I association with vesicular-tubular clusters (VTCs) was required for their microtubule-dependent movement to the Golgi. However, COP I independent transport was also observed, suggesting distinct pathways exist between the ER and Golgi. Recent data suggest that these different pathways emerge from the same ER exit sites, but that different cargoes might be selected through the formation of cargo-specific COP II complexes. Two talks focused on Golgi structure and biogenesis. Francis Barr (Max-Planck-Institute of Biochemistry, Munich) described his work on the GRASP55 and GRASP 65 proteins which are involved in Golgi reassembly and cisternal stacking. Affinity chromatography with the C-terminal tail of GRASPs aimed to identify proteins involved in the association of GRASPs with the membrane. Among the clearest interactors were the p24 proteins, which themselves form hetero-oligomeric complexes that cycle between the ER and Golgi. While it is not clear whether this interaction is the means through which the GRASPs are recruited to the Golgi membrane, it does demonstrate that the Golgi matrix, of which the GRASPs are a part, is potentially able to ‘talk’ to the cargo inside the lumen via the p24 proteins. The second half of the talk focused on the interaction of Rab proteins with proteins of the Golgi. Firstly, it was shown that the matrix complexes GRASP65-GM130 and GRASP55-Golgin45 interact with different rab proteins and function at distinct levels within the Golgi (GRASP65 complex –cis; GRASP 55 complex – medial). Work was also described in which a Rab6 effector was identified as a dynactin (p150 glued complex). In particular, Rab6 was demonstrated to bind the BICD2 (bicaudal) subunit. It was postulated that the strong binding of dynactin to microtubule ends may eclipse its Golgi localisation in immunofluorescence microscopy studies, and interestingly cold treatment of cells (which disrupts microtubules) allowed better visualisation of p150 at the Golgi apparatus. The Harden Lecture was given by Graham Warren (Yale University, New Haven) and provided interesting insights into Golgi partitioning and biogenesis. Two models of Golgi partitioning at cell division have been proposed – one in which the Golgi fragments and is then segregated between daughter cells (Golgi autonomous mechanism), and the second in which Golgi components merge with the ER and these components are then segregated as part of the ER. The Golgi reforms from components transported from the ER following cell division (ER dependent). Studies in the Warren lab on the behaviour of both matrix proteins (GM130) and Golgi enzymes (Mannosidase II) point strongly to the Golgi autonomous model. In particular, while many enzymes are seen to move to the ER prior to cell division, GM130 does not recycle and always remains distinct from the ER, demonstrating that the Golgi as an entity exists even when vesiculated and does not simply rebuild de novo every cell cycle. The observed autonomy of the Golgi led onto the question of Golgi biogenesis. How does a Golgi grow, to ensure it can be partitioned to give sufficient material to each daughter cell? To address this question two systems were studied, both of them parasites. The need to move away from mammalian cells was clear because observing growth when there are up to 50 interlinked Golgi compartments is too complex to get unambiguous answers. Both parasites described have a single Golgi apparatus, enabling the growth of this organelle to be studied in detail. In order to observe the behaviour of the Golgi in Toxoplasma gondii, two markers were introduced: GRASP-YFP and NAGF-YFP. Both of these proteins behave in a similar way to their behaviour in mammalian cells (e.g. in response to BFA). These studies gave rise to the idea that the Golgi of T. gondii may be viewed as a paired organelle which then grows, duplicates and divides (giving four Golgi), before then distributing two to each daughter, which then coalesce to give a single organelle. There was no involvement of the ER, and the inheritance was ordered (rather than random). Studies in Trypanosomes were aided by the knowledge of the genome sequence which has led to cloning of the single T. brucei GRASP protein. Again, there appeared to be a growth of a new Golgi near to a pre-existing one, distribution of the multiple Golgi and coalescence to give single organelles following cell division. An interesting comparison was drawn here with the mechanism seen for duplication of centrioles. Questions remaining are centred on the initiation process of a new Golgi, i.e. is there a ‘seed’ Golgi that acts as a template, or is there some kind of instructed assembly from individual components? Insights into protein trafficking from studies in Saccharomyces cerevisiae were given by both Hugh Pelham (Golgi to endosome and back) and Peter Novick (Golgi to plasma membrane). Both talks beautifully illustrated the power of a relatively simple model organism to identify and characterise the proteins involved in these largely conserved pathways. Hugh Pelham (Laboratory for Molecular Biology, Cambridge) discussed how proteins are able to reach and localise to distinct compartments within the secretory pathway. An important factor in localisation of membrane proteins is transmembrane domain length, with proteins at the plasma membrane having longer domains (> 20 aa) than those destined for the vacuole. Also of importance is the need for recycling of components both from the plasma membrane and from endosomal compartments back to the late Golgi. Work on t-SNAREs that define distinct compartments of the trafficking pathway has identified several key players in the sorting of these proteins, including Ypt6, Ric1 and Grd19. Another difference between endosomal and Golgi membranes is that the former have a higher PI(3)P content. This phospholipid is recognised by FYVE domains and PX domains. In yeast, 10 of the 15 proteins that contain PX domains are sorting nexins. Three of these are required for trafficking of the t-SNARE Snc1p back to the TGN, rather than to the vacuole, after its re-uptake into cells. Studies on the diffusion of membrane proteins was described. FRAP of a yeast plasma membrane t-SNARE Sso1p was compared in yeast and COS cells and was shown to be 20-fold slower in yeast. This was not the case when measuring diffusion rates of vacuolar membrane proteins. The rate of diffusion was increased in cells which have a lower sterol content at the plasma membrane. This result gave rise to the interesting possibility that the high level of sterols in the yeast plasma membrane may make it behave as a single ‘lipid raft’ and as such it would be a very stiff membrane. This stiffness may be required to maintain a low rate of plasma membrane diffusion, which is potentially important in maintaining cell polarity. The spatial regulation of secretion was described by Peter Novick (Yale University, New Haven). Secretion in wild-type yeast is highly polarised and this is shown to require proteins associated with the vesicles, the plasma membrane and the actin cytoskeleton. On vesicles the Rab protein Sec4p is activated by its exchange factor Sec2p, which in turn is recruited by another Rab protein Ypt32p. Secretory vesicles are transported along actin cables by the type V myosin, Myo2p, to the plasma membrane where they encounter the large tethering complex, the exocyst. This 8-subunit complex appears to assemble at the site of fusion, with Sec3p being able to localise to the bud tip in an actin-independent but Cdc42p-dependent manner. Another component Exo70p interacts with the Rho protein, Rho3p. Other components appear to assemble as the vesicle reaches the site. Other collaborative work from the lab has recently provided insight into ER inheritance in yeast. Studies on the role of the exocyst protein Sec3p have shown that if its gene is deleted, then, unlike the other components of the complex, cells are viable. They do, however, have disrupted ER inheritance. In these Δsec3 cells Golgi inheritance appears normal. However, overexpression of Sec3p leads to accumulation of ER at the bud tip, giving rise to the possibility that Sec3p is a polarity marker not just for secretion but also for organelle inheritance. Finally arriving at the plasma membrane and ready for exocytosis, we reached the last day of talks at the meeting. Weiping Han (Howard Hughes Medical Institute/University of Texas South-western Medical Center, Dallas, TX) opened the session with a description of the role of synaptotagmin-1 as a calcium sensor in the fast phase of neurotransmitter release. It is a type I membrane protein that is postulated to function immediately prior to opening of the fusion pore during exocytosis. Syt1–/– mice were generated and these die shortly after birth. Knock-in mutants were described in which the calcium binding loops of syt-1 were mutated. Specific mutations correlated with the efficacy with which calcium triggers neurotransmitter release. Ca2+-dependent phospholipids binding for double C2 domains of Syt-1 using resin-based GST pull-down and liquid-based liposome centrifugation were discussed. It was emphasised that such measurements using GST pull-down may be unreliable. It was also noted that other cell types, e.g. PC12 cells, do not contain Syt1 or 2, but these cells still undergo calcium-triggered release of exocytic vesicles. In PC12 cells, Syt7 appears to play an important role at the plasma membrane in regulating the response to calcium. This is in contrast to the vesicular localisation of Syt1. In order to probe the regions of the synaptotagmins that allowed these different cellular localisations, chimeras of Syt1 and Syt7 were made. Bob Burgoyne (The Physiological Laboratory, Liverpool) described the elegant technique of amperometry, in which single exocytic events from adrenal chromaffin cells are recorded. Using this approach, the role of individual proteins in facilitating stages of exocytosis was investigated. The kinetics of exocytic release observed as a spike were used to distinguish types of fusion events: in particular, whether there was full fusion in which membrane then needs to be retrieved through a dynamin-dependent step, or whether the fusion was of the ‘kiss-and-run’ variety, in which the fusion pore opens to release contents but the vesicle is then resealed and recycled without a requirement for dynamin. Cysteine string protein (CSP), a protein carried on exocytic vesicles, was over-expressed and this was shown to change the amperometric characteristics of release, resulting in a larger peak with a longer-lasting period of release. Mutation of the Protein kinase-A phosphorylation site of CSP caused the change in spike kinetics in Csp1 over-expression to be abolished. A different effect on spike characteristics was seen when the Munc18/nSec1 protein was mutated. In this case the spike rise was faster and shorter. The shorter release time in the munc18 mutant is consistent with a switch to kiss-and-run exocytosis. PKC activation gave similar spike characteristics to the Munc18 mutant, and Munc18 was subsequently shown to be phosphorylated at sites close to syntaxin binding which weakens the interaction. Interestingly, studies on the SNARE proteins suggest that the formation of the highly stable complex as seen in the crystal structure is not necessary to drive membrane fusion, and that fusion events might be mediated by an intermediary structure. This structure intermediate may be reversed in the case of kiss-and-run fusion, or could collapse into the stable structure which then requires dynamin for re-uptake. Guy Rutter (University of Bristol, Bristol) described the movement and exocytosis of insulin secretory vesicles. A number of GFP-based probes were used to study movement of the vesicles in pancreatic beta-cells using total internal reflection fluorescence (TIRF) and evanescent wave microscopy. Glucose was shown to stimulate release of insulin from secretory vesicles by activating recruitment of vesicles from a reserve pool located some distance from the plasma membrane. Movement of vesicles to the plasma membrane required microtubules and conventional kinesin. After expression of a dominant-negative kinesin (KHC T93N) mutant, no vesicle movement was observed and the sustained phase (but not the immediate phase) of insulin release was inhibited. The concept of ‘kiss-and-glide’ fusion was introduced with vesicles apparently remaining close to the membrane after release but moving laterally along it. This type of fusion was, however, only able to release certain cargoes depending on size. Finally, the role of intravesicular calcium was explored. Vesicles were shown to accumulate calcium to high total levels (100 mMol) via an ATP-dependent Ca2+ pump, most of which was tightly bound (free [Ca2+] ∼ 50 μm). Release of calcium from the vesicles was shown to be important for both movement and fusion of the vesicles. One of the first speakers to address the question of regulation of clathrin-coated vesicle formation was George Banting (University of Bristol, Bristol). Although it has been known for some time that constitutive phosphorylation of proteins is associated with clathrin coated vesicle (CCV) formation, the specific kinases involved remained elusive. Characterisation of kinase activity associated with a ‘stripped protein’ fraction from purified CCVs resulted in the identification of two kinases, CK2 which phosphorylates the majority of CCV-associated proteins and GAK/auxilin2 which specifically acts on the λ2 subunit of the AP-2 complex. Interestingly, CK2 only becomes active after uncoating and is not active when associated with CCVs; this inhibition appears to be associated with phosphoinositides present in the vesicle membrane, rather than due to protein components of the CCV. The lipids may act not only to help anchor the kinase to the vesicle but also to inhibit activity while CK2 is bound. An attractive model was put forward proposing that CK2 may phosphorylate peripheral membrane proteins once they have been released from the vesicle preventing their immediate re-association. Bruno Goud (Institut Curie, Paris) gave convincing evidence that two isoforms of Rab6, Rab6A and Rab6A′ which differ only in 3 amino acids can mediate very distinct pathways. Rab6A is implicated in Golgi to ER transport, whereas Rab6A′ is required for the retrograde transport from early or recycling endosomes to the TGN. This pathway also requires the TGN-localised t-SNARE syntaxin 6, syntaxin 16 and Vti1a in addition to the v-SNAREs VAMP3 and VAMP4. The characterisation of two new Rab6 interacting proteins, Rab6IP2A and IP2B, revealed their preferential association with the Rab6A′-associated pathway, as overexpression of the Rab6-binding domain of Rab6IP2 inhibited transport of the Shiga toxin B subunit between the plasma membrane and the Golgi with no effect on passage through the secretory pathway. An interesting question becomes whether these proteins mediate delivery from several retrograde pathways or only the Shiga toxin route. The finding that more than one pathway exists from the endocytic pathway to the TGN was supported by data from Kirsten Sandvig (Institute for Cancer Research, The Norwegian Radium Hospital, Oslo). Using a sulfatable form of ricin, she showed that increasing cytosolic calcium by either thapsigargin or A23187 increased delivery of ricin but not M6PR to the TGN, substantiating earlier data that ricin is transported independently of the M6PR and Rab9. Increases or decreases in cellular cholesterol levels induced by addition of methyl-β-cyclodextrin (mβCD) or mβCD/cholesterol inhibit delivery of ricin to the TGN. Increases in cholesterol also lead to fragmentation of the Golgi stack. Whether this is the sole basis for the inhibition in ricin delivery remains to be elucidated in future experiments. Kathryn Ayscough (University of Glasgow, Glasgow) described the importance of the actin cytoskeleton for facilitating endocytosis in yeast. Studies focusing on the Sla1 protein have revealed that it interacts with both the endocytic machinery and with activators of actin dynamics. Using FRET analysis the lab has shown that cortical actin and components of the endocytic machinery are found in both distinct and overlapping complexes at the plasma membrane. This and other data were used to suggest a model in which actin was recruited to premarked sites of endocytosis. Changes in actin dynamics were then required to drive the internalisation process. Two-hybrid screens with different domains of Sla1p identified novel proteins with potential roles in regulating actin and in membrane trafficking. Ysc84p has homologues in mouse and human and associates with the actin cytoskeleton. It appears to be localised to the actin cytoskeleton via interactions with the Arp2/3 activator Abp1p. A second Sla1p interactor, Lsb5p has characteristics of the GGA proteins in that it has a VHS domain and a GAT domain. However, it does not have a gamma-ear-interacting domain but instead has an NPFXD motif at its extreme C-terminus. Localisation data indicate that Lsb5p resides largely at the plasma membrane, suggesting that it may act as an adaptor protein between the endosomal system and the cell surface. In the penultimate talk, Judith Klumperman (Department of Cell Biology. University Medical Centre Utrecht, the Netherlands) presented a further characterisation of the ‘bi-layered coat’, a coat until recently in search of an adaptor. Although it has clearly been shown to contain clathrin heavy chain, the ‘adaptor’ protein associated appears not to be one of the classic adaptor complexes, AP-1 nor AP-2, but rather Hrs. TfRs are excluded from these coats, but both EGFR and GHR are concentrated in these flat lattices. This process is dependent on ubiquitination and therefore one obvious question centres on the relationship of the bi-layered coat to the yeast ESCRT complexes that have recently been shown to be necessary for formation of internal vesicles. To address this question, wild-type or dominant negative human Vps4 (hVps4EQ) was expressed and the number and size of coats were determined as well as the effects on EGFR sorting. Overexpression of hVps4EQ resulted in impairment of the formation of internal vesicles, an increase in the size and area of the bi-layered coat and retention of EGFR at the limiting membrane. A controversial pathway, that is governed by AP-3, was the subject of the second half of the talk. Although AP-3 is known to play a role in sorting lysosomal membrane proteins to lysosomes, the exact pathway has yet to be elucidated. Using two new monoclonal antibodies, AP-3 was predominantly found on membrane buds evolving from tubular endosomes, a third of which were clathrin positive, whereas the majority of AP1 was clathrin coated and localised to the TGN. It was postulated that AP-3 may provide a second trap for lysosomal proteins internalised from the cell surface, but the actual mechanism of this capture is still unclear. Harvey McMahon (Laboratory for Molecular Biology, Cambridge) kept us engaged for the last talk of the meeting, providing insights into membrane curvature and how a clathrin-coated lattice invaginates to form a clathrin-coated vesicle. A number of proteins regulate the assembly of the clathrin lattice and subsequent vesicle budding, including CALM, its brain-specific homologue AP180, and epsin. These three proteins contain a conserved amino-terminal lipid-binding domain and a C-terminal clathrin/adaptor binding domain. As an aside, according to Harvey, ‘If you want a specific inhibitor of clathrin-mediated endocytosis, AP180-C-terminus is your best bet.’ The structure of the ANTH (AP180 N-terminal homology) domain bound to PtdIns(4,5)P2 revealed an unusual binding site containing 3 protruding lysines and a histidine residue. These residues lie on the surface rather than in a pocket, as had been suggested for the epsin-N-terminal homology (ENTH) domain binding site. In addition to the ANTH domain, new findings based on the crystallisation of the epsin ENTH domain complexed to Ins(1,4,5)P3 were presented. In contrast to the structure without ligand, IP3 binding results in the formation of a new N-terminal helix, termed ‘helix-0’, with an unusual hydrophobic outer surface. The proposal is that this ampipathic helix inserts into the surface of the lipid bilayer with the hydrophobic residues in the hydrophobic phase of the membrane, This insertion was proposed to be essential for membrane curvature. In order to test this hypothesis, two helix-disrupting mutants, L6Q and L6E were introduced. In contrast, to the PtdIns(4,5)P2-binding mutants, R63L and H73L, which were found in the cytosol and profoundly inhibited AP-2 localisation as well as TfR uptake, L6Q and L6E had no affect on lipid head-group binding; however, they did result in reduced curvature. Another member of the epsin family, epsin R colocalises not with AP-2 but with AP-1 and GGAs, and may well bind to a differentially phosphorylated head-group. We look forward to hearing more about the structure–function relationship of this epsin in future meetings.