Hrs binds directly to actinin-4 through distinct domains. (A) Recombinant hrs was incubated with immobilized actinin-4, washed, eluted, separated by SDS-PAGE, and bound hrs was quantitated using ¹²⁵I-secondary antibody and phosphorimaging. The EC₅₀ is ∼0.4 μM, and the stoichiometry is 0.1–0.5–1.0, depending on which protein is immobilized. A Ponceau S stain of actinin-4 is shown for a loading control. (B) Actinin-4 coimmunoprecipitates with hrs. Rat brain postnuclear supernatant (lane 4) was incubated with purified mouse IgG (lane 1) or a monoclonal anti-hrs (lanes 2 and 3). Hrs and actinin-4 also were coimmunoprecipitated from HeLa cell lysate (lane 3). (C) Seven GST-hrs fusion proteins (designated 1–7) were immobilized on glutathione-agarose and assayed for binding to actinin-4_(359–913) as in A. These results identify a specific domain in hrs essential for its interaction with actinin-4.

RNA interference specifically depletes actinin-4 and inhibits transferrin receptor recycling. (A) Extracts from HeLa cells transfected with either duplexed RNA oligonucleotides directed against the actinin-4 sequence (Act-4_RNAi) or a random RNA sequence (Cont_RNAi) were subjected to quantitative western-blots for actinin-4 (lanes 1 and 2). The same samples were probed with a pan actinin antibody that likely recognizes all actinin isoforms (lanes 3 and 4). Extracts also were probed with antibodies to actin, hrs, or eps15 (lanes 5–7). Ponceau S staining shows equal protein loading. HeLa cells were plated on coverslips (right), transfected with either Act-4_RNAi or Cont_RNAi, and probed with antibodies against EEA1, actin, and actinin-4. Bar, 4 μm. (B) Actinin-4 was depleted from HeLa cells as described above. Cells were incubated with ¹²⁵I-epidermal growth factor or ¹²⁵I-transferrin as described in Materials and Methods. At each time point, the amount of Tf or EGF internalized (Int.), recycled (Rec.), and degraded (Deg., EGF only) was determined as described (see Materials and Methods). The ratio of the amount of recycled or degraded to the amount internalized is plotted over time for Tf and EGF. Histograms of the mean slope (rate) of these ratios are shown for each condition (n = 3). The rate of recycling of transferrin was significantly inhibited by depletion of actinin-4 (n = 3, p = 0.0021; recycling rate in actinin-4–depleted cells, 0.0051 ± 0.0004; scrambled, 0.014 ± 0.001). The rate of recycling and degradation of EGF were not altered by depletion of actinin-4 (rate of recycling: scrambled, 0.001, and actinin-4, 0.001, n = 3; rate of degradation: scrambled, 0.0052, and actinin-4, 0.0057, n = 3). C. HeLa cells transfected with either Act-4_RNAi or Cont_RNAi were loaded with Tf-Alexa594 at 0°C for 60 min then shifted to 37°C for 10 min (0′). After 0- or 30-min chase period the location of Tf-Alexa594 was determined. Early endosomes (arrowheads alone) and perinuclear recycling endosomes (arrows) are indicated. The inset of the Act-4_RNAi panel at 30 min shows colocalization of the Tf receptor with Tf-Alexa 594. Bar, 16 μm.

The CART complex is formed in situ and in vitro, and BERP interactions are required for transferrin recycling. (A) BERP and actinin-4 are enriched with early endosomal membranes. HeLa cell lysate and endosomal membranes were isolated as described previously (Sun et al., 2003; Yan et al., 2004), separated on SDS-PAGE, and examined for the presence of BERP and actinin-4. The endosomal fractions do not contain detectable Na⁺/K⁺ ATPase, calnexin, LAMP1, or rab 7, but they are enriched for the early endosomal marker protein, EEA1. (B) Immunoprecipitation of hrs from HeLa cell lysate. HeLa cells were transfected with myc-hrs, FLAG-BERP, and HIS-myosin V and precipitated with either FLAG antisera (lane 1) or mouse IgG (lane 2). The precipitated material was separated by SDS-PAGE, and resulting blots were probed with antisera against hrs, FLAG, actinin-4, and HIS. (C) Hrs, actinin-4, BERP, and myosin V interact in the absence of other proteins and assemble in an ordered manner. Recombinant proteins were either immobilized on Sepharose or purified in a soluble form and incubated in various combinations (as indicated). All proteins bound when incubated with immobilized myosin V (lane 1), actinin-4 and hrs bound to immobilized BERP (lane 2), and hrs bound to immobilized actinin-4 (lane 3). Neither hrs nor actinin-4 bound to immobilized myosin V in the absence of BERP (lane 4). Hrs did not bind to immobilized myosin V in the presence of BERP, but not actinin-4 (lane 5). None of the recombinant proteins bound to immobilized GST (lane 6). (D) Expression of BERP (lane 3), but not N-BERP (lane 2), allowed coprecipitation of actinin-4 with myosin Vb. Although BERP is seen in the lysate (lane 3), N-BERP is not seen in this lane because of significantly smaller size. However, it was detected when separated on another gel, transferred, and probed with FLAG antibody (our unpublished data). (E) The domain structure of BERP and the N-BERP fragment. (F) The N-terminal domain of BERP inhibits the rate of transferrin recycling. Linear domain structure of BERP and N-BERP (top, see text for details). The N-terminal domain of BERP binds to actinin-4, but not to myosin V, which allows the formation of the hrs/actinin-4/N-terminal BERP complex but inhibits association with myosin V. In the presence of this BERP fragment, the transferrin recycling rate was significantly slowed. Tf recycling rates for the HeLa control were 0.0180 ± 0.0013 (n = 6) and for N-BERP, 0.0091 ± 0.0014 (n = 6). ^* denotes p ≤ 0.05.

Hrs and actinin-4 are localized on early endosomes. Ultrathin cryosections of HeLa cells were labeled with specific hrs and actinin-4 antisera and visualized with 15-nm gold particles conjugated to protein A. (A) Hrs is mostly found associated with early endosomal vacuoles and recycling tubules. Actinin-4 (B and C) is occasionally found associated with early endosomal vacuoles, but it is predominantly enriched to the proximity to the plasma membrane (D), at the actin-rich terminal web within 200 nm of the plasma membrane and the filopodial extensions. Arrowheads indicate the location of gold particles. ee, early endosome; er, endoplasmic reticulum; g, Golgi; pm, plasma membrane; n, nucleus. Bars, 200 nm. (E) Quantitation of hrs and actinin-4 labeling in eight subcellular compartments relevant for their distribution; including the plasma membrane (i.e., label on pm ± 20 nm), filopodia (i.e., actin rich cellular extensions from plasma membrane), terminal web under plasma membrane (actin-rich region <200 nm beneath the plasma membrane), early/recycling endosomes, late endosomes/lysosomes, Golgi/TGN, ER, and all other particles.

Inhibition of hrs/actinin-4 interaction reduces the rate of transferrin recycling. (A) Expression of hrs_(1–449) or actinin-4_(357–469) inhibits the interaction of hrs and actinin-4 in situ. Immunoprecipitation of endogenous hrs coprecipitates actinin-4 in wt control HeLa cells (lane 1), whereas expression of either actinin-4_(357–469) (lane 2) or hrs_(1–449) (lane 3), but not N-BERP (lane 4) inhibited the coprecipitation. Moreover, eps15 was coprecipitated with hrs under all conditions. Mouse IgG did not precipitate any proteins. (B) The hrs_(1–449) fragment was expressed in HeLa cells as described in Materials and Methods. The fate of both transferrin and EGF was then determined and represented as in Figure 3. (C) Schematic representation of the domain structure of actinin-4 with the region required for hrs binding (binds hrs), and a depiction of the region used to inhibit hrs/actinin-4 binding (actinin-4_357–469). Expression of hrs_(1–449) also significantly inhibited transferrin recycling (rate of control is 0.0096 ± 0.0012, hrs_(1–449) is 0.0031 ± 0.0003, n = 3, p = 0.001). Expression of hrs_(1–449) did not significantly affect EGF recycling or degradation (recycling rate in control is 0.0012 ± 5.77E-05, hrs_(1–449) is 0.0013 ± 8.82E-05, n = 3; degradation rate in control is 0.0047 ± 0.0003, hrs_(1–449) is 0.0049 ± 0.0002, n = 3). (D) The actinin-4_(357–469) fragment was expressed in HeLa cells as described in Materials and Methods. Expression of actinin-4_(357–469) did not significantly affect EGF recycling or degradation (recycling rate in control, 0.0025 ± 7.83E-05; actinin-4_(357–469), 0.0018 ± 1.09E-04, n = 12; degradation rate in control, 0.0028 ± 0.0001; actinin-4_(357–469), 0.0021 ± 0.0001, n = 12). The kinetics of transferrin recycling is shown (left) as is a bar graph representing the rate of transferrin recycling (right, recycling rate in control, 0.0252 ± 0.00036; actinin-4_(357–469), 0.0136 ± 0.000548). ^* denotes p ≤ 0.05.

Model of how Tf-containing recycling endosomes may link to actin filaments via the CART complex. (A) Recombinant protein binding experiments suggest that there is an ordered assembly for the complex: 1) hrs binds to actinin-4; 2) BERP binds to hrs/actinin-4; and 3) myosin V binds to hrs/actinin-4/BERP. BERP has been shown to bind to myosin V via its tail domain, whereas myosin V binds to actin filaments and moves processively via its head domain. Actinin-4 binds to BERP in its C-terminal region and to hrs in the region after the EF hands. The quaternary complex could bind to actin filaments or the tertiary hrs/actinin-4/BERP complex might bind to myosin V already attached to actin filaments. (B) Model for the role of the CART complex and the actin cytoskeleton in endosome recycling. Because much of hrs is cytosolic, it likely cycles on and off the early endosomal membrane (Komada et al., 1997; Tsujimoto et al., 1999), perhaps via binding to a protein receptor (SNAP-25, step a), before actinin-4 binding. This would occur after, or coincident with, endosomal sorting of Tf-R (step 1). Once the hrs/actinin-4 complex was formed, association with BERP and myosin V would facilitate rapid movement of Tf-R-enriched endosomes back to the plasma membrane on actin filaments. The majority of EGF receptor would bypass this recycling step (step 2) after ligand-induced internalization and would instead travel through the MVB en route to the lysosome for degradation. A minority of EGF receptor would recycle to the plasma membrane (from the MVB or recycling endosome) along with a fraction of the Tf receptor (through the recycling endosome), in CART complex-independent pathways.