CALM Depletion Abolishes Endocytosis of HA-Tagged VAMPs
(A) Effect of CALM depletion on surface expression of HA-tagged VAMPs. Control and siRNA-treated cells expressing HA-tagged VAMP2, VAMP3, or VAMP8 were mixed together, fixed without permeabilization, and labeled with anti-HA, then permeabilized and labeled with anti-CALM. Knocking down CALM increases the surface expression of all three VAMPs. The scale bar represents 20 μm.
(B) Endocytosis of anti-HA in cells expressing HA-tagged VAMPs. Antibody was bound to the cells at 4°C, then the cells were warmed to 37°C for 2–30 min and antibody remaining at the cell surface was quantified by flow cytometry. Each point is derived from at least three separate experiments; the error bars show the SEM. Knocking down CALM effectively abolishes the uptake of all three VAMPs. The specificity of the knockdown phenotype was demonstrated by stably transfecting VAMP8-HA-expressing cells with siRNA-resistant myc-tagged CALM (CALMres wt) and knocking down endogenous CALM. Expression of the CALM construct almost completely rescues the knockdown phenotype.
See also Figure S1.

CALM Binds Directly to VAMP8, VAMP3, and VAMP2
(A) GST Pull downs using His₆MycCALM_ANTH and the GST-fusion proteins indicated. Top: Coomassie blue stained gel. Lower: western blot probed with anti-myc. In this and all subsequent experiments, the lane adjacent to the Molecular weight markers (MWM) is loaded with His₆MycCALM_ANTH only. The ANTH domain of CALM binds directly to VAMP2, 3, and 8 in a concentration dependent manner.
(B) ITC quantitating the binding of CALM_ANTH to VAMP2, 3, and 8 (black squares). The Adaptor:SNARE interaction was tightest for VAMP8 (K_D17 ± 1μM) and weaker for both VAMP3 and VAMP2, (46 ± 7 μM and 48 ± 4 μM, respectively). Epsin_ENTH showed no VAMP binding (black triangles) indicating the CALM:SNARE interaction is specific. Data for epsin1 is offset by −0.3 μcal/sec for clarity.
See also Figure S2.

The N-Terminal Half of the Small R-SNARE VAMPs Binds to CALM_ANTH
(A) Sequence alignment of the SNARE motifs of VAMP2, VAMP3, Snc1p, and VAMP8. Conserved residues are boxed in gray. The position of mutated residues in VAMP8: L16 and V23 (open triangles), K24 and M27 (black circles), L44 and L51 (open circles) are indicated.
(B) Schematic representation of truncation mutants of VAMP8.
(C) GST pull-down experiments using His₆MycCALM_ANTH and the GST-fusion proteins indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. The minimal fragment of VAMP8 that is able to bind to CALM with a similar affinity as the full cytoplasmic portion of VAMP8 (residues 1–76) comprises residues 10–41.
(D) ITC quantitating the binding of CALM_ANTH to truncated VAMP8. VAMP8_10-41 (black triangles) exhibited an essentially identical binding affinity to CALM_ANTH as VAMP8_1-76 (black squares) (K_Ds 17 ± 1 μM and 20 ± 3 μM respectively). Data for VAMP8_(10-41) is offset by −1.3 μcal/s for clarity.
(E) Localization of anti-HA in cells expressing different wild-type or mutant HA-tagged VAMPs. The cells were allowed to endocytose the antibody for 40 min, then processed for immunofluorescence. The cells expressing wild-type VAMP2-HA, VAMP3-HA and VAMP8-HA have endocytosed the antibody, but the cells expressing the VAMP2-HAV43AM46A, VAMP3-HAV30AMetA, and VAMP8-HAK24AM27A mutants have retained the antibody on the plasma membrane. The scale bar represents 20 μm.
See also Figure S3.

Structure of the CALM_ANTH Domain
(A) Ribbon representation of the structure of the CALM_ANTH domain colored from pink (residue 19) to purple (residue 288). Helices are numbered as in (Ford et al., 2001) with no α6. The insets show 2F_O–F_C electron density contoured at 1.2 σ for the well- and poorly-ordered α11 helix in the two CALM_ANTH molecules in the asymmetric unit (designated chains A and B).
(B) Surface representation of helices α1-α10 of the CALM_ANTH domain colored from high (dark green) to low (white) hydrophobicity oriented as in (A). The hydrophobic groove in which helix α11 and the ten preceeding residues sit can be clearly seen.
(C) GST pull-down expriments using GST, GSTVAMP8, and the His₆MycCALM_ANTH proteins indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. Residues 1–289 but not 1–264 of CALM_ANTH bound to GSTVAMP8.
(D) ITC quantitating the binding of residues 1–264 of CALM_ANTH to VAMP8. CALM_ANTH(1-264) (black triangles) showed no measurable binding to VAMP8 whereas WT CALM_ANTH(1-289) bound with a K_D of 20 ± 1μM (black squares). Data for CALM_ANTH(1-264) is offset by −0.2 μcal/sec for clarity.
(E) Structure of the CALM_ANTH(19-264) dimer. One monomer is colored pink/purple and the other blue. The view is rotated by 60° around the vertical axis relative to (A).
(F) Surface representation of the CALM_ANTH(1-264) dimer with one monomer colored by hydrophobicity as in (B) and the other monomer colored blue, oriented as in (A). Formation of the dimer obscures the surface buried by helix α11 in CALM_ANTH(1-289) which is proposed to be the VAMP8 binding site.
(G) Sequence of CALM_ANTH(1-289). Secondary structure is shown above the sequence representing helices α1 to α11; α6 has been omitted as per (Ford et al., 2001).
See also Figure S4.

Structure of the CALM_ANTH(1-264):VAMP8_(11-41) Chimera
(A) Overall structure of the complex with CALM_ANTH colored as in Figure 4 with VAMP8 colored from pale (residue 15) to dark (residue 37) green. The relative position of the membrane (gray bar) is inferred from the position of the PtdIns4,5P₂ (marked as PIP2) binding site on CALM_ANTH as in 1HF8. The dotted line is a schematic representation of how the remainder of the cytoplasmic portion of VAMP8 domain connects its CALM binding helix to its transmembrane helix.
(B) Orthogonal views of the CALM_ANTH:VAMP8 complex shown in molecular surface representation colored as in (A).
(C) Spatial complementarity of the CALM_ANTH:VAMP8 interface with key side chains involved in the binding of CALM_ANTH by VAMP8 shown. Molecular details of the interactions of the key residues (i) K24 and (ii) M27 are shown.
(D) Schematic representation of VAMP8 residues 15–38. CALM_ANTH residues that make hydrophobic interactions with VAMP8 are labeled in black, those that make salt bridge interactions with VAMP8 are labeled in blue and those that make hydrogen bonds with VAMP8 are labeled in turquoise.
(E) The final refined VAMP8 helix (green sticks) is shown in unbiased (F_O–F_C) electron density contoured at 3.5 σ calculated before the addition of the helix to the model (left panel). The VAMP8 helix lies on the same face as, but differs significantly from, the orientation of helix α11 (purple sticks) in the unliganded CALM_ANTH structure (right panel).
See also Figure S5.

Mutation of Key Residues in the CALM_ANTH:VAMP8 Interface Abolishes Their Interaction In Vitro and the Endocytosis of VAMP8 In Vivo
(A) Structure of the CALM_ANTH:VAMP8 complex “opened out like a book” as indicated by the arrows. Residues whose mutation affect binding between CALM and VAMP8 are colored and labeled in red, while mutations that have no effect are colored and labeled in gray.
(B) Pull-down experiments using GSTVAMP8 and WT or mutant His₆MycCALM_ANTH as indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. The mutations L219S, F240A, M244K, I247D, and L251S in His₆MycCALM_ANTH abolished binding to VAMP8.
(C) ITC quantitating the binding of certain point mutant versions of CALM_ANTH to VAMP8. The wild-type CALM_ANTH binds with a K_D17 ± 1μM (black squares) whereas CALM_ANTH L219S (black circles) and M244K (open triangles) showed no measurable interaction with VAMP8. Data for CALM_ANTH L219S and M244K are translated by 0.3 μcal/s and −0.7 μcal/s respectively.
(D) GST pull-down experiments using His₆MycCALM_ANTH and WT or mutant GSTVAMP8 fusion proteins as indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. Wt GSTVAMP8 bound CALM_ANTH whereas the K24AM27A VAMP8 did not. The GSTVAMP8 K24A mutant interacted weakly with CALM_ANTH, however, the single M27A mutation of VAMP8 was sufficient to completely abolish the interaction with CALM_ANTH.
(E) ITC quantitating the binding of K24AM27A mutant version of VAMP8 to CALM_ANTH. Wt GSTVAMP8 bound wt CALM_ANTH with a K_D17 ± 1μM (black squares) whereas GSTVAMP8 K24AM27A (black circles) showed no measurable interaction. Data for GSTVAMP8 K24AM27A is translated by −0.3 μcal/s for clarity.
(F) Cells expressing either VAMP8-HA alone, or VAMP8-HA plus Myc-tagged siRNA-resistant CALM (CALMres: wt, L219S, or M244K) were mixed together and endogenous CALM was depleted by siRNA treatment. The cells were then fixed without permeabilization and labeled with anti-HA antibody, and then permeabilized and labeled with anti-CALM. The scale bar represents 20 μm.
(G) Endocytosis of anti-HA in cells coexpressing VAMP8-HA and siRNA-resistant wild-type, L219S or M244K mutant myc-tagged CALM. Antibody was bound to the cells at 4°C, then the cells were warmed to 37°C for 2–30 min and antibody remaining at the cell surface was quantified by flow cytometry. Each point is derived from at least 3 separate experiments; the error bars show the SEM. Expression of the CALM construct rescues the knockdown phenotype, but expression of the two CALM mutants does not.
(H) Localization of anti-HA in cells expressing wild-type or K24AM27A mutant VAMP8-HA. The cells were allowed to endocytose the antibody for 40 min, then processed for immunofluorescence. Unlike the cells expressing wild-type VAMP8, the cells expressing the mutant have retained the antibody on the plasma membrane. The scale bar represents 20 μm.
(I) Endocytosis of anti-HA in cells expressing wild-type or K24AM27A mutant VAMP8-HA, using the flow cytometry assay described in Figure S1B. There is negligible endocytosis of the mutant construct.

Binding of CALM_ANTH and SNARE Complex Formation by VAMP8 Are Mutually Exclusive
(A) Comparison of the mode of VAMP8 binding to CALM_ANTH and to Syntaxin7:Syntaxin8:Vti1b (PDB 1GL2) (Antonin et al., 2002) made by superimposing the VAMP8 residues 15–39 from the two complexes. VAMP8 adopts the same superhelical twist in both structures and helices α9 and α10 of CALM_ANTH correspond to the helices of Syntaxin 8 and Syntaxin 7, respectively.
(B) Formation of GSTVAMP8:SNAP23:Synatxin3(195–253) SNARE complexes (see Supplemental Information) and their binding to CALM_ANTH. Pull-down experiments using His₆MycCALM_ANTH and the GST fusion proteins indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. SNARE complexes formed with GSTVAMP8 wt and K24AM27A but not with GSTVAMP8 L16PV23P or L44PL51P. Neither the wt VAMP8 SNARE complex nor the K24AM27A SNARE complex bound CALM_ANTH indicating that SNARE complex formation and CALM_ANTH binding shown are mutually exclusive events.
(C) Avidity of CALM binding to PtdIns4,5P₂ and VAMP8. Shown are the sensorgrams for the concentration dependent binding of CALM to membranes with captured VAMP8 (top: calculated K_D0.9 ± 0.15 μM), membranes with PtdIns4,5P₂ (middle: calculated K_D1.9 ± 0.45 μM) and membranes with both, PtdIns4,5P₂ and captured VAMP8 (bottom: calculated K_D0.17 ± 0.03 μM). Given are the mean values and SD of four independent measurements The affinity of CALM increases by 8.5 fold over the average K_D for the two ligands when VAMP8 and PtdIns4,5P₂ are bound simultaneously.
(D) Pull-down experiments using His₆MycCALM_ANTH and wt and mutant GSTVAMP8 fusion proteins as indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. The GSTVAMP8 mutants L16PV23P and K24AM27A (see also Figure 6) do not bind CALM_ANTH. However, GSTVAMP8 L44PL51P bound CALM_ANTH with a similar strength to wt GSTVAMP8.
(E) ITC quantitating the binding of point mutated versions of VAMP8 to CALM_ANTH. The binding of wt CALM_ANTH and GSTVAMP8 L44PL51P (open circles) was comparable to that of wt CALM_ANTH and wt GSTVAMP8 (black squares) (K_Ds of 23 ± 2μM and 17 ± 1μM respectively). However, GSTVAMP8 L16PV23P (open triangles) and GSTVAMP8 K24AM27A (black circles) both showed no measurable interaction with CALM_ANTH. Data for GSTVAMP8 L16PL23P, L44PL51P and K24AM27A are translated by −0.3 μcal/s, −1.5 μcal/s and −1.9 μcal/s respectively for clarity.
(F) Localization of anti-HA in cells expressing different VAMP8-HA constructs. The cells were allowed to endocytose the antibody for 40 min, then processed for immunofluorescence. The cells expressing wt VAMP8 and the L44PL51P mutant have endocytosed the antibody, but the cells expressing the L16PV23P mutant have mainly retained the antibody on the plasma membrane similar to the K24AM27A mutant. The scale bar represents 20 μm.
(G) Schematic representation of the model of VAMP8 trafficking from the plasma membrane. The clathrin adaptor CALM binds simultaneously to the R-SNARE VAMP8 and PtdIns4,5P₂ (labeled PIP₂) at the plasma membrane. CALM is released from the surface of an endocytosed vesicle when PtdIns4,5P₂ hydrolyzed and the clathrin cage disassembled and the hydrophobic CALM-binding helix of VAMP8 now “lies” on the vesicle's surface (Ellena et al., 2009). Finally VAMP8 forms a trans-SNARE complex with its cognate SNAREs on an early endosome to drive vesicule fusion. Thus throughout the interaction of the hydrophobic VAMP8 SNARE motif with the aqueous environment is minimized.

Endocytosis Assay, Related to Figure 1
(A) Western blot of nontransfected HeLa cells and of cells stably expressing HA-tagged VAMP3 or VAMP8, probed with an antibody against either the relevant VAMP, the HA tag, or syntaxin4 as a loading control. In both cases the tagged VAMP is expressed at considerably lower levels than the endogenous VAMP.
(B) Schematic diagram of the assay for endocytosis of HA-tagged VAMPs. Anti-HA was bound to the cell surface at 4°C, then the cells were washed and either shifted to 37°C for 2–30 min or kept at 4°C. They were then incubated with a fluorescent secondary antibody at 4°C, and the surface fluorescence was quantified by flow cytometry. A decrease in fluorescence at 37°C indicates endocytosis of the anti-HA.
(C) Western blots showing the specificity of the CALM knockdown and rescue. Extracts of cells expressing HA-tagged VAMP8 alone or coexpressing HA-tagged VAMP8 and myc-tagged siRNA-resistant CALM (wt or mutant) were probed with either anti-CALM, anti-Myc, or an antibody against the AP-2 α subunit (as a loading control). The blot shows that the three constructs are expressed at a similar level to endogenous CALM, and that the knockdown depletes endogenous CALM but not the constructs.
(D) Endocytosis of ¹²⁵I-labeled transferrin and EGF in control and CALM-depleted cells. The ligand was bound to the cell surface at 4°C, then the cells were shifted to 37°C for 2–30 min. The medium was harvested, ligand remaining on the plasma membrane was removed with an acid wash, the cells were solubilized with NaOH, and the label in all three was quantified. The graphs show the cell-associated counts as a percentage of the total counts. Each point is derived from at least 3 separate experiments; the error bars show the SEM.

AP180_ANTH Does Not Bind VAMP2, VAMP3, or VAMP8, Related to Figure 2
(A) GST Pull downs using His₆MycAP180_ANTH and the GST-fusion proteins indicated. Top panel: Coomassie blue stained gel. Lower panel: western blot probed with anti-myc. The lane adjacent to the molecular weight markers (MWM) is loaded with His₆MycAP180_ANTH only. The ANTH domain of AP180 does not bind to VAMP2, 3, and 8.
(B) Isothermal titration calorimetry of the binding of CALM_ANTH and AP180_ANTH to Ins(1,4,5)P₃ gave comparable binding affinities (K_D24 ± 5μM and K_D23 ± 3μM respectively) confirming both proteins were folded and functionally active. Data for CALM is translated by −1.2 μcal/s for clarity.
(C) Surface Plasmon Resonance of CALM_ANTH and AP180_ANTH also showed AP180_ANTH did not bind to GST VAMP8. The SNARE was coupled directly to a CM5 chip and the binding of the ANTH domain proteins was assayed by flowing 200 nM analyte (CALM_ANTH or AP180_ANTH) across the chip for 60 s at a flow rate of 30μl/min in 20 mM HEPES (pH 7.4), 150 mM NaCl, 4 mM DTT. Data were recorded on a Biacore 3000 (GE Healthcare) at 20°C and the binding of analyte to a mock channel (GST) has been subtracted.

Analogous Mutations in the SNARE Motifs of VAMP2, VAMP3, and VAMP8 Abolish Their Endocytosis, Related to Figure 3
Endocytosis of anti-HA in cells expressing HA-tagged VAMPs, using the flow cytometry assay described in Figure S1B. The wild-type constructs were efficiently internalized, but the VAMP2-HAV43AM46A mutant, the VAMP3-HAV30AM33A mutant, and the VAMP8-HAK24AM27A mutant remained on the plasma membrane.

Superposition of CALM_ANTH Domains and Dimerization of CALM_ANTH(1-264), Related to Figure 4
(A) The structure of CALM_ANTH(1-289) (purple) from this study overlays with the previously published structure of CALM_ANTH (1HF8; gray) (Ford et al., 2001) with Cα atom rmsds of 0.43 Å (chain A) and 0.47 Å (chain B). Insets highlight the poorly-ordered C-terminal section for the structures of the ANTH domains.
(B) The weight-averaged molar mass (thin lines) is shown across the elution profile (A_280 nm, thick lines) of CALM_ANTH constructs subjected to size-exclusion chromatography and MALS. The expected molar masses of CALM_ANTH(1-289) monomers (M_r = 33,486) and CALM_ANTH(1-264) dimers (M_r = 62,792) are shown for reference (dotted grey lines).

CALM_ANTH Does Not Undergo Major Structural Reorganization upon Binding VAMP8, Related to Figure 5
(A) Residues 19–264 of the ANTH domains of unbound CALM (pink) and VAMP8-bound CALM (purple) shown in ribbon representation.
(B) Superimposition of Cα traces of residues 19–264 of unbound CALM (pink) and VAMP8-bound (purple) CALM_ANTH. Inset shows reorganisation of side chains upon binding at the VAMP8 interaction interface colored as above.

Localization of Anti-HA in Cells Expressing Different VAMP8-HA Proline Mutants, Related to Figure 7
Control and siRNA-treated cells were mixed together and allowed to endocytose the antibody for 40 min, then processed for immunofluorescence. Knocking down CALM prevents antibody uptake in cells expressing the L44PL51P mutant, but in cells expressing the L16PL23P mutant the antibody stays on the cell surface whether or not CALM is knocked down. The scale bar represents 20 μm.