A Role for EHD4 in the Regulation of Early Endosomal Transport

Endogenous EHD4 localizes to early endosomes

Progress in elucidating the function of specific C-terminal EHD proteins has been hindered by their high level of homology and the difficulty in obtaining antibodies that exclusively recognize only a single EHD paralog. Accordingly, we identified a unique C-terminal EHD4 sequence and generated rabbit polyclonal peptide antibodies against it. We first tested the affinity-purified anti-EHD4 antibodies by immunoblotting analysis and found that it recognized a band of ~64 kDa from HeLa cell lysates, corresponding to the predicted molecular mass of endogenous EHD4 (Figure 1A, left lane). Incubation of the anti-EHD4 antibodies with the immunizing peptide abolished detection of this band, demonstrating specificity (Figure 1A, right lane). To further define the specificity of the EHD4 antibody, HeLa cells were transfected with green fluorescent protein (GFP)–EHD1, GFP–EHD2, GFP–EHD3 or GFP–EHD4. Lysates from these transfected cells were separated by SDS–PAGE and immunoblotted with the purified anti-EHD4 antibodies (Figure 1B). Expression of each of the four GFP–EHDs was confirmed using an anti-GFP antibody (Figure 1B, right panel). Note that the molecular masses of EHD2 and EHD4 are slightly higher than those of EHD1 and EHD3. When immunoblotted with anti-EHD4 antibodies (Figure 1B, left panel), an ~91-kDa GFP–EHD4 band was detected exclusively in cells transfected with GFP–EHD4. However, despite the high level of homology between EHD proteins, no bands were detected in lysates from cells transfected with the other EHD paralogs (Figure 1B, left panel). On the other hand, ~64 kDa endogenous EHD4 was detected in all the lysates. These experiments demonstrate the high specificity of this antibody for EHD4.

We next examined the subcellular localization of endogenous EHD4. As demonstrated, the anti-EHD4 antibody decorated a series of punctate structures, with occasional extended tubular membranes (Figure 2A). This pattern corresponded well with that observed upon overexpression of hemagglutinin-epitope tagged (HA)–EHD4 (Figure 2B) and GFP–EHD4 (data not shown), suggesting that when overexpressed, EHD4 displays a similar subcellular distribution to that of the endogenous protein.

It has been previously demonstrated that truncation of the EHD of EHD proteins (EHD ΔEH) results in the loss of tubular EHD membranes and the relocalization of EHD proteins exclusively to punctate structures (30,45,46). We analyzed the localization of EHD4 ΔEH and observed a similar phenotype, with EHD4 ΔEH no longer associated with tubular membranes and primarily localized to large punctate structures (Figure 2C). It has also been established that the conserved N-terminal EHD nucleotide-binding motif is required for the association of EHD1 and EHD3 with membranes (24,25). Mutations in the nucleotide-binding motif of EHD4 (G68R) render this protein cytosolic in a similar manner (Figure 2D).

Studies to date have implicated EHD proteins in the regulation of various steps within the endocytic pathways. Armed with this knowledge, and based on the subcellular distribution of the endogenous protein, we first investigated the subcellular compartment to which EHD4 localizes. Our immunofluorescence analysis revealed a significant colocalization of the endogenous early endosomal protein Rab5 with endogenous EHD4 or with HA–EHD4 in transfected cells (Figure 3A–F), suggesting that a considerable pool of EHD4 associates with early endosomes. Endogenous EHD4 also clearly displayed partial overlap with an additional marker for early endosomes, endogenous early embryonic antigen 1 (EEA1) (Figure 3G–I, see insets and arrows for colocalizations), as well as Rab4 (Figure S1G–I) and Sorting Nexins 1 and 2 (Figure S1M–R). Moreover, at the early time-points following internalization, transferrin could also be detected colocalizing with endogenous EHD4 (Figure 3J–L). Endogenous EHD4 also displayed partial colocalization with vesicles and tubules containing Arf6 (Figure 3M–O), suggesting that it may be involved in clathrin-independent endocytic events. On the other hand, little colocalization was observed with Rab11, a protein localized to the recycling compartment (Figure 3P–R), the lysosomal membrane marker, Lamp1 (Figure S1A–C), the cis Golgi membrane protein GM130 (Figure S1D–F) or the late endosomal protein, Rab7 (Figure S1J–L). Taken together, these data indicate that EHD4 localizes to early endosomes.

EHD4 controls traffic at the early endosome

Given the localization of EHD4 to early endosomes, we next sought to gain functional information on its potential role at this stage of the endocytic pathway by determining the effect of its depletion. Accordingly, by small interfering RNA (siRNA) knock-down, we were consistently able to reduce EHD4 expression by 60–75% (Figure 4A). We were unable to increase the efficacy beyond this level, possibly because of the long half-life of the EHD4 protein (data not shown). EHD4–siRNA knock-down was specific for EHD4 as it had no significant effect on the expression levels of EHD1, EHD2 and EHD3 (Figure 4A). EHD4–siRNA- and mock-treated cells were fixed after 72 h and subjected to immunostaining with Rab5 and EEA1 (Figure 4B,C,E,F, both markers of the early endosome) and Rab11 (Figure 4D,G, a marker for recycling endosomes). Depletion of EHD4 had a striking effect on the morphology of early endosomes, which appeared to be larger, clustered and in some cases even fused into much larger structures that were less prevalent in the periphery of the cell (Figure 4E,F). On the other hand, the subcellular distribution of Rab11 did not change upon loss of EHD4 expression (Figure 4D,G).

To provide additional evidence that EHD4 depletion is specifically responsible for the enlarged early endosome phenotype, we utilized an siRNA-resistant EHD4 construct engineered with a silent mutation in the region recognized by the oligonucleotide. To this end, as a control, we first expressed GFP in both mock- and EHD4–siRNA-treated cells (Figure 4H–K). As expected, upon immunostaining these cells with the early endosomal marker EEA1, the endosomes in the EHD4-depleted cells were greatly enlarged (compare 4J with 4H). Moreover, expression of the GFP protein in these cells (depicted by dashed yellow line) had no effect on the regulation of endosomal size. On the other hand, expression the siRNA-resistant silent HA–EHD4 ‘rescued’ the effect of the siRNA knock-down and caused the EEA1-positive endosomes to revert back to their normal size (Figure 4L–O, compare dashed yellow area in N with neighboring cells and cells in L). These data strongly suggest that EHD4 knock-down is indeed responsible for the enlargement of early endosomes.

To further confirm our data showing that siRNA knockdown of EHD4 caused the formation of enlarged early endosomal structures, we utilized an additional knockdown strategy based on short hairpin RNA (shRNA) (Figure 4P–R). As demonstrated by immunoblot analysis, transfection of the shRNA plasmids for EHD4 reduced EHD4 protein expression by >60% compared with either control shRNA plasmids or mock treatment with only the transfection reagent (Figure 4P). Indeed, upon transfection of the EHD4–shRNA followed by immunostaining for endogenous EHD4, ~50% of the cells had a marked decrease in the level of EHD4 expression (Figure 6Q, outlined cells). Similar to our results using siRNA knockdown (Figure 4A–G), in these cells, Rab5 showed a dramatically altered subcellular distribution by coalescing into very large perinuclear structures away from the cell periphery (Figure 4R, outlined cells). Collectively, these findings demonstrate that EHD4 controls early endosome morphology and suggest that EHD4 may be involved in the regulation of trafficking at the early endosome.

We next aimed to determine whether the enlarged early endosomal structures generated in the absence of EHD4 were accessible to endocytic cargo. HeLa cells on cover-slips were either mock treated (Figure 5A,B) or treated with EHD4–siRNA (Figure 5D,E) for 72 h and then subjected to brief ‘pulse–chase’ experiments with fluorochrome-labeled transferrin [transferrin–Alexa Fluor 568 (Tf-568)]. As demonstrated, after a short 3-min pulse with Tf-568, both mock-treated and EHD4–siRNA-treated cells exhibited a similar subcellular distribution pattern of the internalized transferrin (Figure 5A,D). For the most part, the transferrin was observed in very small vesicles, including pre-endosomal and early endosomal structures and at this time-point still dispersed quite uniformly through the cell (Figure 5A,D). Although some concentration in the perinuclear region was beginning to be evident in mock-treated cells (Figure 5A), this coalescence at the ERC was particularly evident only following a 5-min ‘chase’ (Figure 5B). On the other hand, in the EHD4–siRNA-treated cells, the Tf-568 was observed in enlarged structures (correlating with those seen earlier for Rab5 and EEA1), and importantly, it remained well dispersed throughout the periphery of the cells and did not concentrate at the ERC (Figure 5E). Indeed, upon EHD4 depletion, even following a 2 h continuous uptake of transferrin, this cargo did not accumulate at the ERC (Figure 5, compare F with C) but remained primarily in the periphery, often in enlarged structures. However, the rate of transferrin recycling to the plasma membrane upon EHD4 depletion remained similar to mock-treated cells (Figure 5G), suggesting that direct recycling to the plasma membrane from early endosomes may compensate for the impaired transport to the ERC. These data indicate that the enlarged early endosomes generated in the absence of EHD4 are indeed accessible to endocytic cargo and impede normal transport downstream through this pathway.

Because EHD4 has been implicated in the Arf6 clathrin-independent endocytic pathway (49), we attempted to determine whether EHD4 regulates the trafficking of clathrin-independent cargo through the early endosome. To address this question, mock- and siRNA-treated cells were allowed to uptake antibodies against MHCI proteins (30) for 20 min. As demonstrated, a significant portion of MHCI was detected in early endosomal structures containing EEA1 (Figure 6A–C, see arrows and insets). As expected, early endosomes were enlarged upon depletion of EHD4, and these structures were accessible to MHCI (Figure 6D–F, see arrows and insets). In mock-treated cells following an additional 20 min of ‘chase’, MHCI was predominantly in a perinuclear region corresponding to the ERC, with little MHCI remaining colocalized with EEA1 (Figure 6G–I). However, depletion of EHD4 caused retention of MHCI in the early endosomal compartment where it remained highly colocalized with EEA1 (Figure 6J–L, see arrows in insets). These results were quantified and consistently indicated that twofold to threefold more MHCI was retained in EEA1-positive early endosomes at this time-point (Figure 6M). Indeed, even after 40 min of chase, the EHD4-depleted cells displayed considerable retention of MHCI in the early endosome compared with mock-treated cells (unpublished data). These data indicate that trafficking of both clathrin-dependent and clathrin-independent cargo through the early endosome is regulated by EHD4.

Early endosomes are also a nexus for the sorting of proteins to the late endosomal degradatory pathway. We therefore proposed to determine whether loss of EHD4 also impairs the trafficking of cargo that traverses this route. To this aim, mock- and EHD4–siRNA-treated cells were pulsed for 15 min with DiI-low-density lipoprotein (LDL), chased for 40 min and analyzed to determine the extent to which LDL was localized to EEA1-containing early endosomes. At this time-point in mock-treated cells, the vast majority of LDL molecules had already exited the early endosomes and no longer displayed colocalization with EEA1 (Figure 7A–C). On the other hand, in EHD4–siRNA-treated cells, a significant portion of LDL could still be observed colocalizing with EEA1 (Figure 7D–F, quantified in G). These data imply that EHD4 regulates exit from the early endosome toward both the recycling and the degradative pathways.

Enlarged early endosomes have been observed previously when either wild-type Rab5 (50) or a GTP-locked Rab5 mutant was transfected into cells (51). We therefore hypothesized that the enlarged early endosomes in EHD4 knock-down cells may contain enhanced levels of GTP–Rab5 associated with their membranes. To test this hypothesis, we utilized three different methods. First, we took advantage of a plasmid containing a yellow fluorescent protein (YFP)/Rab5-binding domain of Rabaptin-5 fusion protein (YFP–R5BD) that binds specifically to GTP–Rab5 (52). Mock- and EHD4–siRNA-treated cells were transfected with YFP–R5BD, fixed and examined by confocal microscopy (Figure 8A–F). In mock-treated cells, endogenous Rab5 showed a typical early endosomal distribution pattern and was localized throughout the cell to small- and medium-sized vesicles (Figure 8A). The YFP–R5BD in these cells was mostly cytosolic as previously discussed (52), suggesting that relatively little GTP–Rab5 was localized to endosomal membranes (Figure 8A–C). In EHD4–siRNA-treated cells, as we have shown, the Rab5 localized to much larger structures (Figure 8D, see inset). YFP–R5BD was recruited from the cytosol to these structures, indicating that they were enriched in GTP–Rab5 (Figure 8D–F, see insets). Second, endogenous full-length Rabaptin-5 was also recruited to these enlarged endosomal membranes in EHD4-depleted cells (Figure 8G, H). Third, to obtain additional evidence supporting these findings that depletion of EHD4 leads to activation of Rab5, HeLa cells were mock treated or treated with both EHD4–siRNA and shRNA (to maximize the level of knock-down), followed by homogenization and separation to membrane and cytosolic fractions and then examined by immunoblotting with anti-Rab5 antibodies. To verify that membrane proteins are exclusively in the pellet fraction, immunoblotting was first performed with anti-transferrin receptor (TfR) (Figure 8I, right panel). As demonstrated in a representative experiment, cells depleted for EHD4 displayed more Rab5 in the membrane pellet, indicating an increase in the membrane-bound active Rab5–GTP (Figure 8I). In addition, a modest decrease in cytosolic Rab5 was also observed in EHD4-depleted cells (Figure 8I). To quantitatively evaluate the level of Rab5 activation, the membrane-to-cytosol ratios were calculated for mock- and EHD4–siRNA-treated cells from three independent experiments (Figure 8J). When mock ratios were normalized (to a value of 1 to permit fair comparison between separate experiments), the increase in membrane-to-cytosolic Rab5 in EHD4-depleted cells was consistently twofold to fivefold higher (Figure 8J, depicts a mean of fourfold increase in the ratio). Collectively, these data favor a model by which loss of EHD4 induces activation of Rab5.

Endogenous EHD4 interacts with EHD1

Several studies have recently demonstrated that EHD1 and its closest paralog, EHD3 (86% identical), are capable of oligomerizing through their central coiled-coil domains (25,42) and that loss of oligomerization leads to the protein becoming cytosolic. Because EHD4 also contains a predicted central coiled coil, we next determined whether this protein oligomerizes with the other EHD paralogs. Using a selective two-hybrid binding assay, we showed that EHD4 robustly binds to EHD1 and EHD3 (Figure 9A). However, while binding of EHD4 to EHD2 and EHD4 homo-oligomerization was also observed, it appears to be significantly weaker than binding of EHD4 to either EHD1 or EHD3 (Figure 9A). To confirm these observations in mammalian cells by coimmunoprecipitations, we transfected HeLa cells with HA–EHD4 and with GFP–Myc–EHD1, GFP–Myc–EHD2 or GFP-Myc-EHD3 (Figure 9B). As shown, expression levels of GFP–Myc–EHD1, GFP–Myc–EHD2 and GFP–Myc–EHD3 were relatively similar, and HA–EHD4 was coexpressed in each of these cells (Figure 9B, cell lysate panel). Despite similar expression levels, significantly more EHD1 was coimmunoprecipitated with EHD4 compared with EHD2 (Figure 9B, upper left panel). EHD3 was also immunoprecipitated at higher levels than EHD2 (Figure 9B). Although EHD4 could also homo-oligomerize, EHD4/EHD4 interactions identified by coimmunoprecipitation appeared to be weaker than those between EHD4 and either EHD1 or EHD3 (Figure 9C). This association of EHD4 with all three paralogs and its homo-oligomerization, as shown in Figure 9B,C, were quantified and are presented in Figure 9E (for calculation of values, see legend). To ascertain whether EHD4 hetero-oligomerizes with EHD1 through coiled-coil interactions, we cotransfected HA–EHD4 together with Myc–EHD1 V203P, a point mutant that we have previously shown to have a disrupted coiled coil (25). As demonstrated, this mutation significantly reduced EHD4/EHD1 interactions to <5% of the level of interactions between the two wild-type proteins (Figure 9D). These data show that EHD4 interacts with its EHD paralogs in a hierarchical manner through coiled-coil interactions.

Despite increasing evidence that EHD proteins oligomerize and that oligomerization is important for biological function, to date, no studies have addressed the interaction between these endogenous proteins. Our coimmunoprecipitation studies with overexpressed EHD proteins showed that EHD4 interacted preferably with EHD1 and EHD3; however, for the purposes of endogenous coimmunoprecipitations, our anti-EHD3 antibodies are unsuitable. Accordingly, we aimed to determine whether endogenous EHD4 and endogenous EHD1 interact in HeLa cells (Figure 10). To this aim, untransfected HeLa cells were lysed and the proteins were separated by SDS–PAGE. The non-reducing conditions utilized here allowed us to distance the non-reduced immunoglobulin G (~155 kDa) from the EHD proteins (EHDs, ~60 kDa). Immunoblotting with either anti-EHD1 (Figure 10A, left panel) or anti-EHD4 (Figure 10A, right panel) showed the position of these EHD proteins on the gel, with EHD4 migrating at a slightly higher apparent molecular weight. Anti-EHD1 antibodies successfully immunoprecipitated EHD1 (Figure 10A, third lane), whereas pre-immune serum did not precipitate protein that was recognized by anti-EHD1 antibodies (Figure 10A, second lane), demonstrating the specificity of the anti-EHD1 antibodies for EHD1 immunoprecipitations. Moreover, when control immunoprecipitations were carried out with either EHD1 pre-immune serum or protein A–Sepharose beads, no EHD4 was detected by immunoblotting (Figure 10A, second panel from right). However, when lysates were immunoprecipitated with anti-EHD1 antibodies and immunoblotted with anti-EHD4, a band of ~64 kDa was detected (Figure 10A, fourth lane). A band of identical apparent molecular weight was observed when immunoprecipitations were carried out directly with anti-EHD4 and immunoblotted with the same antibody (Figure 10A, fifth lane). While the anti-EHD4 was generated against an entirely unique C-terminal amino acid sequence present only in EHD4, it was not possible to obtain useful antibodies against EHD1 that did not partially overlap with the sequence of EHD4. Although immunoblotting clearly showed a high preference of the EHD1 peptide antibodies for the 60-kDa EHD1 band (as opposed to the ~64-kDa EHD4 band), it was nevertheless necessary to rule out the possibility that EHD4 might be immunoprecipitated directly by the anti-EHD1 antibody. To this aim, we immunoprecipitated EHD1 from mock-treated cells as well as cells depleted for EHD1 with EHD1–siRNA (Figure 10B). As shown, EHD1–siRNA-treated cells showed a significant reduction (>90%) in the levels of EHD1 when assessed by immunoprecipitation and immunoblotting analysis (Figure 10B, top panel). Moreover, when EHD1 was depleted from the cells with siRNA, there was almost a complete loss of the coimmunoprecipitated EHD4 (which is often but not always observed as a doublet band), indicating that EHD4 precipitates as a result of its association with EHD1 and not because of direct binding with the anti-EHD1 antibody (Figure 10B, bottom panel). These data demonstrate that endogenous EHD4 and EHD1 interact in cells.

Functional relationship between EHD4 and EHD1

To probe the relationship between EHD4 and EHD1, HeLa cells were either mock treated or treated with EHD4–siRNA, and the effect of EHD4 depletion on the subcellular distribution of EHD1 was examined. As expected, in mock-treated cells, endogenous EHD1 localized to an array of punctae and tubular membranes that displayed minimal overlap with endogenous Rab5 (Figure 11A–C). As described earlier, EHD4 depletion led to Rab5 activation and localization to enlarged early endosomes (Figure 11D). However, the pattern of EHD1 was greatly altered, the tubular EHD1 structures were no longer observed and EHD1 now appeared to colocalize significantly with Rab5 on the enlarged early endosomal membranes (Figure 11D–F). In a reciprocal experiment, the subcellular distribution of EHD4 was examined in mock-treated and EHD1–siRNA-treated cells (Figure 11G–L). In mock-treated cells, Rab5 localized to perinuclear and peripheral punctate endosomes, whereas EHD4 localized peripherally to punctae and tubular endosomes, with some colocalization in evidence (Figure 11G–I). However, in EHD1-depleted cells, both Rab5 and EHD4 localize primarily to the perinuclear ERC region, are largely absent from the periphery and show a strong colocalization with each other (Figure 11J–L). These data imply that EHD4/EHD1 interactions are important to maintain proper subcellular distribution of EHD and other endocytic regulatory proteins.

Recombinant DNA constructs

Human EHD1, EHD3 and EHD1 V203P constructs have been described previously (25,28,30). A full-length human EHD4 complementary DNA clone was purchased from GeneCopoeia and subcloned into a pCDNA 3.1(+) expression vector (Invitrogen) using a 5′ primer containing the nucleotide sequence coding for the HA epitope with BamHI and 3′ XhoI restriction sites. HA-tagged EHD4 ΔEH and HA-tagged EHD4 G68R mutants were generated utilizing the QuickChange site-directed mutagenesis kit (Stratagene) by introducing a stop codon at amino acids 429 and changing G68 to R68, respectively. Human EHD4 was also subcloned into EGFPC3 vector (Clontech) using 5′ XhoI and 3′ BamHI restriction sites with GFP at the N-terminus of the fusion protein. Two-hybrid control vectors (GAL4ad–SV40 large T-antigen and GAL4bd–p53) were obtained from BD Biosciences Clontech. pGADT7 and pGBKT7 two-hybrid vectors containing EHD1, EHD2 and EHD3 have been described previously (28), and EHD4 was subcloned into the multicloning sites of these vectors by engineering 5′ NdeI and 3′ BamHI restrictions sites. All constructs were sequence verified and are available upon request. The YFP–R5BD (Rab5-binding domain of Rabaptin-5) was a kind gift from Dr A. Sorkin (University of Colorado Health Sciences Center), the HA–Arf6 was from Dr Julie Donaldson (National Institutes of Health) and the GFP–Rab constructs were a gift from Dr Robert Lodge (Universite Laval).

Antibodies and reagents

The following antibodies were used in this study: affinity-purified rabbit polyclonal peptide antibodies directed against the C-terminus of human EHD1 (DLPPHLVPPSKRRHE), EHD2 (VERGPDEAMEDGEEGSDDEA) and EHD4 (SHRKSLPKAD) (AnaSpec), rabbit antibodies against Rab11 (US Biologicals) and EEA1 (Novus Biologicals, Inc.), mouse monoclonal antibodies directed against the Myc epitope (9E10) and HA epitope (Covance), against EEA1, Rab5, Rabaptin-5, Sorting Nexin 1 (SNX1) and Sorting Nexin 2 (SNX2) (Transduction Laboratories), against LAMP1 (Santa Cruz Biotechnology), against actin (Abcam), against anti-human HLA-A,B and C (Leinco Tech) and rat monoclonal antibody against Hsp70 (Stressgen Bioreagents). Secondary donkey anti-mouse Alexa 488 nm and goat anti-rabbit Alexa 568 nm antibodies were purchased from Invitrogen. Goat anti-mouse horseradish peroxidase (HRP) was obtained from Jackson Immuno Research Laboratories, Inc. Donkey anti-rabbit–HRP was obtained from Amersham Biosciences, and Goat anti-rat HRP was obtained from Jackson ImmunoResearch Laboratories, Inc. DiI-LDL was obtained from Biomedical Technologies. Fatty-acid-free bovine serum and protease inhibitor cocktail were obtained from Sigma.

Immunoprecipitations and immunoblotting

For immunoprecipitation experiments, cells were harvested and lysed for 15 min in buffer containing 25 mM Tris, pH 7.4, 150 mM NaCl, 0.5% Triton-X-100 (w/v), protease inhibitor cocktail (Roche Molecular Biochemicals) and 10 mM iodoacetamide. After removal of insoluble matter by centrifugation, the lysate supernatants were subjected to immunoprecipitations with goat anti-HA antibody-conjugated agarose beads (Bethyl laboratories). After 2-h incubation at 4°C, immunoprecipitates were washed four times in lysis buffer containing only 0.1% Triton-X-100 and boiled in 125 mM Tris (pH 6.8)/2% sodium dodecyl sulfate (SDS)/12% glycerol/0.02% bromophenol blue to elute the proteins. Proteins were separated by 10% SDS–PAGE, blocked in 5% non-fat milk in PBS and immunoblotted with either mouse anti-Myc (9E10) followed by mouse anti-rabbit light chain HRP (Jackson Immuno Research Laboratories, Inc.) or biotin-conjugated anti-GFP antibodies. Streptavidin–HRP was used to detect the presence of biotinylated anti-GFP antibodies by enhanced chemiluminescence. To immunoprecipitate endogenous EHD proteins, protein A–Sepharose beads prebound to pre-immune serum or to affinity-purified anti-EHD1 and anti-EHD4 antibodies were used. Immunoblotting was performed with affinity-purified anti-EHD1 and anti-EHD4 antibodies, and detection was with donkey anti-rabbit HRP. Densitometric analysis was utilized to calculate the intensity of protein bands on the immunoblots by multiplying the mean of luminosity by number of pixels as measured by ADOBE PHOTOSHOP software, subtracting the background and using actin expression levels for normalization of each lane.

Membrane fractionation

HeLa cells were harvested in 600 μL of ice-cold homogenization buffer (25 mM HEPES, 100 mM NaCl, 1 mM ethylenediaminetetraacetic acid, pH 7.4 and protease inhibitor cocktail) by 30 vertical strokes in a glass Dounce homogenizer (Thomas) at 4°C. After centrifugation at 800 × g for 10 min to remove unbroken cells and cellular debris, the supernatant was then ultracentrifuged at 108 000 ×g for 1 h at 4°C to yield a pellet of total cellular membranes and a supernatant representing the cytosolic fraction. The membrane pellet was dissolved in urea buffer (70 mM Tris–Hcl, pH 6.8, 8 M urea, 10 mM n-ethylmaleimide, 10 mM iodoacetamide, 2.5% SDS and 0.1 M DTT) by incubating at 37°C for 15 min. The entire membrane fraction and 3.5% of the cytosolic fraction were mixed with ×2 Laemmli buffer to a final concentration of ×1 and subjected to SDS–PAGE and immunoblot analysis.

Yeast two-hybrid analysis

The Saccharomyces cerevisiae strain AH109 (Clontech) was maintained on Yeast extract, Peptone, Dextrose (YPD) agar plates. Transformation was carried out by the lithium acetate procedure as described in the instructions for the MATCHMAKER two-hybrid kit (Clontech). For colony growth assays, AH109 cotransformants were streaked on plates lacking leucine and tryptophan and allowed to grow at 30°C usually for 3 days or until colonies were large enough for further assays. An average of three to four colonies was then chosen and suspended in water, equilibrated to the same optical density of 600 nm and replated on plates lacking leucine and tryptophan (+HIS) as well as plates also lacking histidine (−HIS).

Gene knock-down by RNAi and rescue using RNAi-resistant EHD4 expression

RNA interference (RNAi) duplexes (synthesized by Dharmacon) were transfected using Dharmafect (Dharmacon) as previously described (28). Calibration experiments showed that 72 h of treatment was sufficient to attain significantly decreased expression levels of both EHD1 and EHD4. The sequence used for EHD1 (base pairs 943–963) was gaaagagatgcccaatgtc, and the sequence for EHD4 (base pairs 222–241) was gaccaccttcatcagatac. Rescue of EHD4 RNAi was performed by expression of an EHD4 construct engineered with a silent mutation at base pairs 231, 234, 235 and 237, which maintain phenylalanine, isoleucine and arginine residues in these positions. This construct was transfected using FuGENE HD (Roche Applied Science) in mock-treated and EHD4–siRNA-treated cells 48 h after the start of siRNA treatment. For the shRNA-mediated knock down of EHD4, the same sequence of EHD4 as mentioned above was cloned into RNAi-Ready pSIREN-Shuttle vector (BD Biosciences) as per the manufacturer’s instructions. Negative control shRNA was provided by the manufacturer. Efficacy of EHD1 and EHD4–RNAi was confirmed by immunoblotting of transfected and control cell lysates with affinity-purified rabbit polyclonal antibodies specific for EHD1 and EHD4, respectively.

Immunofluorescence and transferrin, MHCI and LDL uptake assays

HeLa cells were grown on cover glasses, transfected with FuGENE-6 (Roche Diagnostics) and fixed with 4% (v/v) paraformaldehyde in PBS as described previously (55). Fixed cells were incubated with primary antibodies prepared in staining solution [0.2% saponin (w/v) and 0.5% (w/v) BSA in PBS] for 1 h at room temperature. After washes in PBS, the cells were incubated with the appropriate fluorochrome-conjugated secondary antibody mixture in staining solution for 30 min at room temperature. All images were acquired using a Zeiss LSM 5 Pascal confocal microscope (Carl Zeiss) by using a ×63 1.4 numerical aperture objective with appropriate filters. Transferrin uptake was studied by first starving the cells in DMEM-lacking serum (but containing 0.5% BSA) for 30 min and then applying a 3-min pulse of 16.7 μg/mL Tf-568 (Invitrogen) followed by chase in complete media for 5 min. Cells were fixed and mounted as described above for image analysis. For MHCI pulse–chase assays, cells were pulsed with anti-MHCI antibody for the indicated times, followed by a 90-second incubation in stripping buffer (0.5 M NaCl and 0.5% acetic acid, pH 3.0) to remove surface-bound MHCI antibody. Cells were either fixed directly or incubated in complete media for the indicated times to ‘chase out’ the antibody. Colocalization coefficients were calculated with LSM 5 PASCAL software. For LDL uptake, cells were first starved in media containing fatty-acid-free serum for 1 h, followed by a brief pulse with DiI-labeled LDL and ‘chased’ in complete media for 40 min prior to fixation. All images shown are representative images from experiments that have been repeated at least three times. Quantification of colocalization was performed with IMAGEJ software utilizing JACoP (http://rsb.info.nih.gov/ij/plugins/track/jacop.html) (56) to calculate Pearson’s coefficient (57). Each pairwise comparison was performed on eight sets of images acquired with the same optical settings. Bars represent the mean ± standard error calculated from eight sets of images.