Molecular Basis for the Sorting of the SNARE VAMP7 into Endocytic Clathrin-Coated Vesicles by the ArfGAP Hrb

Summary SNAREs provide the specificity and energy for the fusion of vesicles with their target membrane, but how they are sorted into the appropriate vesicles on post-Golgi trafficking pathways is largely unknown. We demonstrate that the clathrin-mediated endocytosis of the SNARE VAMP7 is directly mediated by Hrb, a clathrin adaptor and ArfGAP. Hrb wraps 20 residues of its unstructured C-terminal tail around the folded VAMP7 longin domain, demonstrating that unstructured regions of clathrin adaptors can select cargo. Disrupting this interaction by mutation of the VAMP7 longin domain or depletion of Hrb causes VAMP7 to accumulate on the cell's surface. However, the SNARE helix of VAMP7 binds back onto its longin domain, outcompeting Hrb for binding to the same groove and suggesting that Hrb-mediated endocytosis of VAMP7 occurs only when VAMP7 is incorporated into a cis-SNARE complex. These results elucidate the mechanism of retrieval of a postfusion SNARE complex in clathrin-coated vesicles.

NRK cells were transfected using Fugene 6 transfection reagent (Roche Diagnostics Ltd.) at a ratio of 2µg DNA: 6 µl Fugene. NRK cells were transfected with ∆pMEP4-VAMP7 constructs and stable cell lines generated by growing cells in DMEM supplemented with 0.2 mg/ml hygromycin (Roche Diagnostics Ltd.).
∆pMEP4 contains a metallothionein promoter, hence expression of VAMP7-HA (stable cell lines) or CD8 constructs (transients) was induced by adding 10 µM CdCl 2 for 16 h to the tissue culture medium before experiments, NB treatment with cadmium results in some variation of expression levels in individual cells. In the case of CD8 constructs 21µM leupeptin was added with the CdCl 2 to inhibit lysosomal proteases.
For depletion of clathrin a single oligonucleotide chc-2 as described in (Motley et al., 2003) was used. There is a single mis-match between this oligonucleotide and the rat sequence, but after a single transfection of siRNA at 10nM there was a decrease in clathrin heavy chain levels (Supplementary Figure S4).
The corresponding anti-sense sequences were phosphorylated at their 5' end.
A single dsRNA oligonucleotide was also designed to human Hrb (used in EGF assays), the sense sequence was 5'-AGUCGUGGCAUCAGUUCAUdTdT-3'. The TSG101 and AMSH siRNA duplexes were as described (Garrus et al., 2001;McCullough et al., 2004). All dsRNA oligonucleotides were purchased from Dharmacon. RNAi experiments were performed using oligofectamine and dsRNA oligonucleotides at 100 nM.

Iodination of monoclonal anti-HA and binding studies.
20µg of purified monoclonal anti-HA (clone HA.11; Covance) was iodinated for 10 min using 125 I by the Iodogen method (Pierce) and free iodide removed by passage through an Econo-Pac 10DG desalting column (Bio-Rad). A typical iodination resulted in a specific activity of ~5,000 Ci/mmole. For binding studies, NRK cells cells stably expressing VAMP7-HA, and un-transfected NRK cells, were fixed with 4% paraformaldehyde for 20 min at 21˚C. For total binding, half of the cells for each condition were permeabilised with 0.1% (v/v) Triton X-100 for 10 min at 21˚C. All cells were then incubated in 0.2% (w/v) BSA in PBS for 1 h at 37˚C.
Cells were then incubated with the iodinated antibody (50 µCi/ml; 0.3ml/well of a 24 well dish) in PBS supplemented with 0.2% (w/v) BSA, for 1 h at 21˚C. Unbound antibody was removed by washing cells 8X with 1 ml of the PBS-BSA solution. Cells were solubilised with 200 µl of 0.2M NaOH, and the radioactivity counted. Total protein/well was also measured using the BCA protein assay (Pierce). Calculations of the % cell surface VAMP7-HA was determined by dividing the counts/µg protein for non-permeabilised cells by the counts/µg protein for permeabilised cells for each condition. Background counts were determined by using un-transfected NRK cells and background surface and total counts were taken into account in all calculations.

EGF uptake and degradation assays.
Epidermal growth factor (EGF) uptake and degradation was performed as previously described (Bowers et al., 2006). The graph in Supplementary Figure S4 represent counts in the TCA supernatants (i.e. degraded 125 I-EGF) per µg of protein in each sample as a percentage of the total counts bound at time zero (also per µg of protein).
The structure was determined by SIRAS using one native and one mercury derivative data as molecular replacement techniques failed. Seven mercury sites were identified in the derivative using autoSHARP (Vonrhein et al., 2006), and phasing was subsequently undertaken with SHARP (Bricogne et al., 2003). Although the derivative diffracted to 2.5 Å, little phase information was obtained beyond 5 Å resolution. Solvent flattening using SOLOMON (Abrahams and Leslie, 1996) and DM (Cowtan and Main, 1996) was carried out in SHARP. Despite the poor quality of the initial phases, the map after solvent flattening was interpretable and could be used for model building (see Supplementary Figure S2). SHARP optimized the solvent content at 54.6%, with three copies in the asymmetric unit; these molecules were named chains 1, 2, and 3.
Initial model building was carried out in O (Jones et al., 1991), while COOT (Emsley and Cowtan, 2004) was employed for rebuilding. After several rounds of refinement using REFMAC5, it became clear in the F O -F C map that the asymmetric unit contained four copies instead of three. Large patches of "missing" density were 5 observed, indicating density as yet unaccounted for in the model. These patches suggested the presence of β-sheet structure. PHASER (McCoy et al., 2007) was then employed successfully to locate the fourth copy, called chain 4. Molecule 1 was used as the search model, and the positions of molecules 1, 2, and 3 were listed as "known" in the input file. Early rounds of refinement were undertaken using Hendrickson-Lattman coefficients obtained from SHARP, while late rounds employed no prior phase information. The structure was refined to final R cryst and R free values of 22.8% and 28.6% respectively with no residues in disallowed regions (Davis et al., 2007). interacting residues conserved between the two proteins (see Supplementary Figure   9A). HrbL is able to localise to clathrin coated pits/vesicles as shown in Supplementary Figure 9B and Figure 5A). (1/20 th of total lysate) and Hrb isolated on the Sepharose beads after the pulldown.     Depletion of TSG101 and AMSH were used as controls because these respectively cause a reduction and increase in the rate of EGF degradation (Bowers et al., 2006).   Hrb simultaneously binds VAMP7 and AP2 appendage domains causing VAMP7 to be internalised as cargo by clathrin mediated endocytosis. Following delivery to the recycling endosome, VAMP7 can then be sorted by AP3 to its major intracellular location on late endocytic organelles. The SNARE complex in which VAMP7 was participating during its endocytosis must be disassembled and the other SNAREs to which it was bound recycled back to the cell's limiting membrane EH-domain binding NPF motifs are indicated by an asterisk. Identical residues between the two proteins are shaded black, conserved residues grey. The alignment was performed using ClustalW (http://www.ebi.ac.uk/clustalw/), using default parameters.