Ubiquitin-independent binding of Hrs mediates endosomal sorting of the interleukin-2 receptor β-chain

Endocytic membrane trafficking plays important roles in both the degradation of cell surface receptors and attenuation of receptor-mediated signal transduction. In ligand-dependent or ligand-independent manners, cell surface receptors are internalized and pinched off from the plasma membrane into vesicles, from which they are delivered to sorting endosomes and either recycled back to the plasma membrane or degraded via delivery to the lumen of the lysosome. Before receptor degradation in the lysosome, the limiting membrane loaded with receptors invaginates and buds into the lumen of the late endosome to form a multivesicular body (MVB). Mature MVBs fuse with the lysosome, thereby releasing the receptor-containing vesicles into its hydrolytic lumen for degradation. Recent studies have revealed that the endosomal-sorting complex required for transport (ESCRT) protein complexes ESCRT-I, ESCRT-II and ESCRT-III mediate endosomal sorting, thereby sorting endosomal cargoes into MVBs.

Ubiquitylation of endocytic and biosynthetic cargoes serves as a sorting signal for ESCRT-dependent delivery of these cargo proteins to MVBs (Bonifacino and Traub, 2003; Di Fiore et al., 2003; Hicke et al., 2005). ESCRT-I, which consists of the class E vacuolar protein sorting (Vps) proteins Vps23, Vps28 and Vps37, recognizes ubiquitylated cargo proteins via the ubiquitin-binding activity of one subunit, Vps23 (Katzmann et al., 2001). Recent crystal structure studies have indicated that Vps28 of the ESCRT-I complex tightly interacts with Vps36 of the ESCRT-II complex, and that Vps36 also binds ubiquitin through one of the two Np14 zinc finger (NZF) domains (Kostelansky et al., 2006). The ESCRT-II complex recruited by ESCRT-I stimulates assembly of the ESCRT-III complex on the endosomal membrane through binding of the ESCRT-III subunit Vps20 to Vps25 of ESCRT-II (Babst et al., 2002), resulting in sorting of the cargo proteins into the forming MVBs.

Hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs) was identified as a phosphotyrosine protein after stimulation with hepatocyte growth factor (Komada and Kitamura, 1995). Hrs and Vps27, a yeast orthologue of Hrs, bind to ubiquitin (Lloyd et al., 2002; Polo et al., 2002; Raiborg et al., 2002). Mutation of the Hrs UIM domain abrogates its ability to bind to ubiquitylated proteins (Bishop et al., 2002) and consequently prevents ubiquitylated cargo proteins from being sorted into the vacuole lumen (Shih et al., 2002). Hrs/Vps27 binds to Vps23/TSG101 (the mammalian orthologue of Vps23) via the P(S/T)AP motif (Bache et al., 2003; Bilodeau et al., 2003; Lu et al., 2003; Pornillos et al., 2003). Hrs depletion causes a reduction in membrane-associated ESCRT-I subunits (Bache et al., 2003), whereas mutations within the P(S/T)AP motif of Vps27 impair recruitment of Vps23 to endosomes (Katzmann et al., 2003), indicating that Hrs functions upstream of ESCRT-I. The interaction between ubiquitylated cargo proteins and Hrs therefore appears to be the first step of endosomal sorting to the MVB pathway.

Interleukin-2 (IL-2) is a T-cell cytokine that induces the growth, differentiation and activation of various types of hematopoietic cells through its interaction with IL-2 receptor (IL-2R) complexes. Functional IL-2R complexes are composed of α-, β- and γc-chains or β- and γc-chains, indicating that the β- and γc-chains are indispensable for the formation of functional IL-2R complexes (Sugamura et al., 1996). We previously found that Hrs is immediately tyrosine-phosphorylated after IL-2 stimulation (Asao et al., 1997). Although endocytosis of many cell surface receptors is mediated by the clathrin-dependent molecular machinery, which cooperates closely with ubiquitin-binding molecules such as Hrs and Epsin (Raiborg et al., 2001; Shih et al., 2002; Sigismund et al., 2005), IL-2R complexes are internalized in a clathrin-independent manner (Lamaze et al., 2001). After IL-2R complexes are internalized, IL-2Rβ and IL-2Rγc are sorted to late endosomes and degraded, whereas IL-2Rα is recycled to the plasma membrane (Hemar et al., 1995). On the other hand, a chimeric receptor of IL-2Rα containing a partial peptide of the IL-2Rβ cytoplasmic region was found to be ubiquitylated and delivered to the lysosome (Rocca et al., 2001). However, it remains unclear whether IL-2Rβ interacts with ubiquitin-binding molecules and whether this interaction plays a functional role in the endosomal sorting of IL-2Rβ. Thus, we investigated the interaction of IL-2Rβ with Hrs and the possible role of Hrs in its endosomal sorting in the present study.

Since sorting of membrane proteins to the vacuole lumen is inhibited in Hrs-deficient cells and Hrs UIM-deleted mutant cells (Bilodeau et al., 2002; Katzmann et al., 2003; Shih et al., 2002), the membrane proteins are considered to associate with Hrs in the endosomal pathway. To examine the association of Hrs with IL-2R, we initially performed coimmunoprecipitation assays. Functional IL-2R complexes are composed of three subunits: IL-2Rα contains only 13 amino acid residues in its cytoplasmic region, whereas IL-2Rβ and IL-2Rγc contain 286 and 86 cytoplasmic amino acid residues, respectively. The cytoplasmic regions of IL-2Rβ and IL-2Rγc, but not IL-2Rα, are critical for IL-2 signal transduction (Asao et al., 1993; Hatakeyama et al., 1989). Accordingly, HEK293T cells were transiently transfected with Flag-tagged IL-2Rβ or IL-2Rγc and a HA-tagged Hrs. Lysates of the HEK293T cells (2×10⁶ cells) were immunoprecipitated with an anti-Flag antibody and immunoblotted with an anti-Hrs antibody. Hrs was clearly coimmunoprecipitated with IL-2Rβ, but its association with IL-2Rγc was much weaker than that with IL-2Rβ (Fig. 1A). Thus, we focused our attention on the interaction between Hrs and IL-2Rβ. To further examine whether endogenous Hrs is able to bind to IL-2Rβ, we used HEK293Tβ4 cells, which stably expressed human IL-2Rβ.Lysates of HEK293Tβ4 and control HEK293T cells (5×10⁷ cells) were immunoprecipitated with an anti-IL-2Rβ antibody and immunoblotted with an anti-Hrs antibody. Hrs was coimmunoprecipitated with IL-2Rβ in the HEK293Tβ4 cells (Fig. 1B). This experiment reveals an interaction between endogenous Hrs and IL-2Rβ in HEK293Tβ4 cells.

Previous reports have indicated that the UIM-domain of Hrs is necessary for ubiquitin binding to sort ubiquitylated cargo proteins into the MVB pathway (Bishop et al., 2002; Polo et al., 2002; Shih et al., 2002). However, our experiments showed that a UIM-deleted mutant could bind to IL-2Rβ (Fig. 1C). This finding prompted us to investigate whether other regions of Hrs that do not contain the UIM-domain may also contribute to the interaction with IL-2Rβ. To examine this possibility, we constructed Hrs mutants with C-terminal truncations at residues 340, 428, 451, 512, 557, 570, 611 and 641 (Fig. 2A), and performed coimmunoprecipitation experiments. Surprisingly, despite the fact that all the mutants contained the intact UIM domain, the truncation mutants comprising residues 1-340 and 1-428 were unable to bind to IL-2Rβ (Fig. 2B). Further analysis of the truncation mutants revealed that association of the Hrs mutants comprising residues 1-451 (dC2), 1-512, 1-557 and 1-570 (dC1) with IL-2Rβ was markedly weaker than that of full-length Hrs, whereas the mutants comprising residues 1-611 and 1-641 bound to IL-2Rβ as efficiently as full-length Hrs (Fig. 2B). These results indicate that residues 429-610 of Hrs are required for IL-2Rβ binding. Thus, we constructed Hrs deletion mutants lacking residues 428-443, 428-451, 428-466, 428-557, 428-602, 470-512 (dCC1), 468-555 (dCC2) and 432-573 (dM), and examined the detailed characteristics of the binding region to IL-2Rβ (Fig. 2C). As the size of the deleted region increased in 428-443, 428-451 and 428-466 deletion mutants, the association of the mutants with IL-2Rβ was reduced in that order (Fig. 2D). However, IL-2Rβ binding was hardly detected in lysates of HEK293T cells transfected with the mutants lacking residues 428-557, 428-602 and 432-573 (dM), each one of which lacks the coiled-coil region (Fig. 2D). By contrast, the mutants with deleted residues within the coiled-coil region, dCC1 and dCC2, bound to IL-2Rβ, suggesting that the coiled-coil region of Hrs is dispensable for IL-2Rβ binding. Therefore, we speculated that residues 428-466 would comprise one of the IL-2Rβ-binding regions in Hrs. Together with the data in Fig. 2B, residues 558-611 appear to have an effect on an IL-2Rβ-binding region. However, the mutant lacking residues 428-557 was unable to bind to IL-2Rβ (Fig. 2D), despite the fact that it contained residues 558-611. In addition, the Hrs mutant lacking residues 558-611, similarly to wild-type Hrs, bound to IL-2Rβ. The reduced binding property relative to the wild type of the Hrs mutant lacking both residues 428-466 and 558-611 with IL-2Rβ was similar to that of the Hrs mutant lacking only residues 428-466 (supplementary material Fig. S1). Therefore, residues 428-466 of Hrs constitute the essential domain required for IL-2Rβ binding. Since it is known that Hrs specifically bind to signal transducing adaptor molecule (STAM), and the Hrs/STAM complex recognizes some ubiquitylated cargo proteins, we examined whether STAM may affect the interaction between Hrs and IL-2Rβ. Coimmunoprecipitation experiments revealed that, similarly to wild-type Hrs, the Hrs mutant lacking residues 428-466 could bind to STAM (data not shown), whereas the mutant dCC2 lacking the STAM association could bind to IL-2Rβ (Fig. 2D). These results suggest that STAM is not involved in the Hrs–IL-2Rβ interaction.

To define the region of IL-2Rβ required for its interaction with Hrs, we constructed IL-2Rβ mutants truncated at residues 268, 348, 357, 380, 394 and 412 (Fig. 3A), and investigated their binding abilities toward Hrs by coimmunoprecipitation analyses (Fig. 3B). Hrs binding was significantly detected in lysates of HEK293T cells transfected with IL-2Rβ mutants comprising residues 1-357, 1-380 and 1-394, whereas Hrs bound to the mutant comprising residues 1-412 as efficiently as wild-type IL-2Rβ. However, Hrs was hardly detectable, if present at all, after coimmunoprecipitation with IL-2Rβ mutants comprising residues 1-268 and 1-348. These experiments suggested that residues 349-412 are indispensable for Hrs binding. Further experiments showed that deletion mutants lacking residues 268-410 and 349-410 of IL-2Rβ (Fig. 3A) were completely unable to bind to Hrs (Fig. 3B). These results suggest the possibility that the interaction between Hrs and IL-2Rβ may be independent of ubiquitylation of the cytoplasmic region of IL-2Rβ, because all the lysine residues modified by ubiquitin attachment in the cytoplasmic region of IL-2Rβ are located within residues 268-348 and intact in the deletion mutant lacking residues 349-410.

To examine whether ubiquitylation of IL-2Rβ is required for the association with Hrs, we prepared a purified GST fusion protein of full-length Hrs and examined lysates of HEK293T cells transfected with IL-2Rβ mutant genes lacking residues 268-348, in which all the lysine residues are located, or residues 349-410, which constitute the Hrs-binding region in the cytoplasmic region. The fusion protein was immobilized on glutathione-Sepharose beads and incubated with the lysates. The beads were washed and bound material was immunoblotted with an anti-IL-2Rβ antibody. Consistent with the data shown in Fig. 3B, no interaction was observed between GST-Hrs and the IL-2Rβ mutant lacking residues 349-410. By contrast, GST-Hrs associated with the mutant lacking residues 268-348, in which all the lysine residues are located, as efficiently as wild-type IL-2Rβ (Fig. 4A). In addition, we examined whether these cytoplasmic regions were themselves ubiquitylated in HEK293T cells transfected with HA-tagged ubiquitin. The lysates were immunoprecipitated with an anti-IL-2Rβ antibody, and the precipitates were immunoblotted with an anti-HA antibody. Receptors modified by ubiquitin attachment were detected in lysates expressing wild-type IL-2Rβ or IL-2Rβ_d349-410, but not in that expressing IL-2Rβ_d268-348 (Fig. 4B). These experiments indicated that the interaction between Hrs and IL-2Rβ was independent of ubiquitylation of lysine residues in the cytoplasmic region of IL-2Rβ. Furthermore, GST-Hrs was incubated with an extract of Escherichia coli expressing His-tagged IL-2Rβ_269-551 comprising the cytoplasmic tail of IL-2Rβ and material bound to the glutathione-Sepharose was immunoblotted with an anti-IL-2Rβ antibody. The results revealed that His-tagged IL-2Rβ_269-551 associated with GST-Hrs, but not with GST alone (Fig. 4C), and therefore confirmed that Hrs interacts directly with IL-2Rβ in a ubiquitylation-independent manner.

To investigate the endocytic intracellular localization of IL-2Rβ, HEK293Tβ4 cells were analyzed by confocal microscopy using an anti-IL-2Rβ antibody together with antibodies against the early endosome marker EEA1, late endosome or lysosome marker LAMP1 or Hrs. The antibodies against the endosome markers revealed that IL-2Rβ colocalized with LAMP1-positive compartments but was rarely found in EEA1-positive and Hrs-positive compartments (Fig. 5A). IL-2Rβ was detected in dispersed intracellular vesicles in the cytoplasm rather than on the cell surface (Fig. 5A), suggesting that IL-2Rβ is constitutively internalized in the absence of the ligand as previously reported (Hemar et al., 1994; Morelon and Dautry-Varsat, 1998). To examine this possibility, HEK293Tβ4 cells were incubated with a ¹²⁵I-labeled anti-IL-2Rβ antibody (¹²⁵I-TU11). To evaluate the receptor internalization, cells were collected at the indicated times (Fig. 5B). Treatment of the cells with citric acid buffer (pH 2.5) was used to distinguish cell surface-bound ¹²⁵I-TU11 from intracellular ¹²⁵I-TU11. The radioactivity of surface-bound acid-removable ¹²⁵I-TU11 decreased rapidly (Fig. 5Ba) accompanied by a rapid increase in the radioactivity of intracellular acid-unremovable ¹²⁵I-TU11 (Fig. 5Bb), indicating that IL-2Rβ is constitutively internalized in the absence of the ligand. In addition, we generated HEK293Tαβγ9 cells, which stably expressed IL-2R α-, β-and γc-chains, and examined IL-2Rβ internalization in HEK293Tαβγ9 cells in the presence or absence of IL-2. An anti-IL-2Rβ antibody, TU11, does not block the receptor assembly among IL-2R α-, β- and γc-chains (Takeshita et al., 1992). Furthermore, TU11 neither stimulates nor inhibits the receptor-mediated cell growth (Ohbo et al., 1991). Interestingly, the kinetics of IL-2Rβ internalization in HEK293Tαβγ9 cells, with or without IL-2, were the same as those observed in HEK293Tβ4 cells (Fig. 5B). Thus, stimulation with IL-2 has less effect on IL-2Rβ internalization. The data in Fig. 1B indicated that endogenous Hrs bound to IL-2Rβ, whereas the data in Fig. 5A indicated that IL-2Rβ colocalized with LAMP1-positive compartments but not Hrs-positive compartments under steady-state conditions. To resolve this apparent conflict, we examined the kinetics of the receptor trafficking after internalization using confocal microscopy. HEK293Tβ4 and parental HEK293T cells were incubated at 0°C to suppress the receptor internalization and then treated with TU11. The receptors were covalently linked with TU11 by incubation with disuccinimidyl suberate (DSS), a chemical crosslinker. Next, the cells were incubated at 37°C for the indicated times and analyzed by confocal microscopy. IL-2Rβ was observed on HEK293Tβ4 cell surface but not HEK293T cell surface at 0 minutes (Fig. 5C), but then became internalized and sorted to Hrs-positive compartments at 10 minutes. At 60 minutes, the receptors were delivered to LAMP1-positive compartments and no longer present in Hrs-positive compartments (Fig. 5C). These results indicate that IL-2Rβ is constitutively internalized from the cell surface in the absence of the ligand, and then sorted to LAMP1-positive compartments through Hrs-positive compartments.

The mouse pro-B cell line BAF-B03 is an IL-3-dependent cell line, and its transfectants with human IL-2Rβ genes have been used to analyze IL-2 signal transduction (Taniguchi and Minami, 1993). To analyze the internalization and degradation of IL-2Rβ_d349-410 lacking the Hrs-binding region in the cytoplasmic region, we used IL-2Rβ-deficient BAF-B03 cells. F7 cells are transfectants expressing wild-type human IL-2Rβ, whereas S25 cells stably express an IL-2Rβ mutant lacking residues 293-348 that binds to Hrs (data not shown) and BAFβ_d349-410 clone 3 and clone 4 cells stably express IL-2Rβ_d349-410. FACS and immunoprecipitation analyses indicated that there were no significant differences among the amounts of IL-2Rβ on the cell surfaces or in cell lysates of the transfectants (Fig. 6A,B). In addition, the counts of surface-bound ¹²⁵I-TU11 in F7, S25, BAFβ_d349-410 clone 3 and clone 4 cells were 15013, 14879, 14988 and 15004 cpm, respectively. These results also show that the cell surface expression levels of IL-2Rβ in different BAF transfectants were very similar. First, we examined the kinetics of receptor internalization in the transfectants. The above-described F7, S25, IL-2Rβ_d349-410 clone 3 and IL-2Rβ_d349-410 clone 4 cells were incubated with ¹²⁵I-TU11. To evaluate the receptor internalization, the cells were collected at the indicated times (Fig. 6C). The radioactivity of surface-bound acid-removable ¹²⁵I-TU11 decreased rapidly (Fig. 6Ca) accompanied by rapid increases in the radioactivity of intracellular acid-unremovable ¹²⁵I-TU11 (Fig. 6Cb), indicating that IL-2Rβ is constitutively internalized in the absence of the ligand, as shown in Fig. 5B. The kinetics of ¹²⁵I-TU11 internalization in BAFβ_d349-410 cells was the same as that in F7 cells (Fig. 6C). Subsequently, we evaluated the receptor degradation based on the amount of radioactivity released into the culture supernatants by the cells. The transfectants were incubated with ¹²⁵I-TU11 and the receptors were covalently linked with ¹²⁵I-TU11 using the chemical crosslinker DSS, before culture supernatants were collected at the indicated times (Fig. 6D). The radioactivity in the culture supernatants increased rapidly (Fig. 6Da) and the increase was quantitatively correlated with a decrease in the cell-bound radioactivity (Fig. 6Db). The radioactivity in the culture supernatants of BAFβ_d349-410 clones was significantly lower than that in the culture supernatants of F7 and S25 cells (Fig. 6Da), suggesting that the degradation rate of IL-2Rβ_d349-410 is lower than those of wild-type IL-2Rβ and IL-2Rβ_d293-348. Furthermore, we extracted the degraded short peptides from the culture supernatants by trichloroacetic acid (TCA) precipitation, and found lower amounts of degraded short peptides in culture supernatants from BAFβ_d349-410 cells compared with that of F7 cells (Fig. 6Dc). In addition, we also investigated the internalization and degradation in the presence of IL-2. The kinetics of the internalization and degradation in the presence of IL-2 (supplementary material Fig. S2) were the same as those in the absence of IL-2 (Fig. 6C,D), suggesting that IL-2 binding to IL-2Rβ has little effect on the internalization and degradation in BAF transfectants. Hence, the constitutive internalization and degradation of IL-2Rβ might act as a mechanism to attenuate IL-2 signals. These results suggest the possibility that the Hrs-binding region (residues 349-410) of IL-2Rβ is required for its precise transport to the lysosomes for degradation.

To investigate the details of IL-2Rβ endosomal sorting, we used a mouse embryonic fibroblast (MEF) cell line suitable for confocal microscopy analysis. cDNAs encoding the IL-2Rβ mutants lacking residues 268-348 and 349-410, as well as wild-type IL-2Rβ, were introduced into MEF cells to generate MEFβ_d268-348, MEFβ_d349-410-2 and MEFβ cells, respectively. FACS and immunoprecipitation analyses indicated that there were no significant differences among the amounts of IL-2Rβ on the cell surfaces or in cell lysates of the transfectants (Fig. 7A,B). Consistent with the results for HEK293Tβ4 cells, wild-type IL-2Rβ expressed in MEFβ cells was localized to LAMP1-positive compartments (Fig. 7C), whereas mutant IL-2Rβ lacking the Hrs-binding region expressed in MEFβ_d349-410-2 cells was hardly detectable in LAMP1-positive compartments but localized in punctate structures in the cytoplasm (Fig. 7C). These results suggest that the interaction between IL-2Rβ and Hrs is required for the endosomal sorting to LAMP1-positive compartments. We then examined whether Hrs deficiency affects the membrane trafficking of IL-2Rβ from early to late endosomes. Using HRSd cells, which exhibit enlarged vacuoles as consequence of Hrs-depletion as described in previous reports (Bache et al., 2003; Komada and Soriano, 1999; Raiborg et al., 2001), derived from Hrs-deficient mice (Kobayashi et al., 2005), we generated HRSdβ cells, which stably expressed wild-type IL-2Rβ. HRSdβ cells showed similar IL-2Rβ expression levels on the cell surface and in the cytoplasm to MEFβ cells, and exhibited enlarged endosomes as described above. Owing to the weak fluorescent EEA1 signals in HRSd cells, we used the GFP-SARA-FYVE domain as an early endosome marker (Itoh et al., 2002). IL-2Rβ expressed in HRSdβ cells was not localized to LAMP1-positive vacuoles, but to GFP-SARA-FYVE domain-positive vacuoles (Fig. 7C), indicating that IL-2Rβ is sorted to and accumulates in early endosomes under conditions of Hrs dysfunction. Accordingly, we examined whether IL-2Rβ_d349-410 is localized to early endosomes. However, IL-2Rβ_d349-410, as well as IL-2Rβ and IL-2Rβ_d268-348, was scarcely found in Hrs-positive early endosomes in the MEF transfectants (Fig. 7C). These findings raised the question of whether IL-2Rβ_d349-410 is delivered to early endosomes. Therefore, we investigated the kinetics of the receptor trafficking in MEFβ_d349-410-2 cells using confocal microscopy. MEFβ_d349-410-2 and MEFβ cells were treated with TU11, as described for Fig. 5C. IL-2Rβ_d349-410, as well as IL-2Rβ, was observed in Hrs-positive compartments at 10 minutes (Fig. 8A), had departed from these compartments at 30 minutes (data not shown) and was not detectable in Hrs-positive compartments at 120 minutes (Fig. 8A). A large part of IL-2Rβ was localized to LAMP1-positive compartments at 120 minutes (Fig. 8A). The sorting of IL-2Rβ_d349-410 to LAMP1-positive compartments was impaired (Fig. 8A, arrow) but not completely blocked (Fig. 8A, arrowhead). Therefore, it is considered that IL-2Rβ_d349-410 lacking the Hrs-binding region can at least be delivered to Hrs-positive early endosomes, similarly to wild-type IL-2Rβ, but its precise transport to LAMP1-positive compartments is impaired. However, mutant IL-2Rβ lacking all the lysine residues in the cytoplasmic region expressed in MEFβ_d268-348 cells was localized to LAMP1-positive compartments as efficiently as wild-type IL-2Rβ expressed in MEFβ cells (Fig. 7C). These results suggest that ubiquitylation of the cytoplasmic region is not necessary for the sorting to LAMP1-positive compartments, although residues 349-410 of the Hrs-binding region play an important role for endosomal sorting of IL-2Rβ from early to late endosomes.

Next, to investigate the receptor degradation in MEFβ, MEFβ_d268-348 and HRSdβ cells and two independent MEFβ_d349-410 clones, clone 2 and clone 22, the transfectants were incubated with ¹²⁵I-TU11 and the receptors were covalently linked with ¹²⁵I-TU11 using the chemical crosslinker DSS, as described for Fig. 6D. The counts of surface-bound ¹²⁵I-TU11 in MEFβ, MEFβ_d268-348, MEFβ_d349-410 clone 2, MEFβ_d349-410 clone 22 and HRSdβ cells were 6579, 5997, 6126, 6431 and 6628 cpm, respectively, indicating that the levels of IL-2Rβ expression on the cell surface were similar. The radioactivity in the culture supernatants increased rapidly (Fig. 8Ba) and this increase was quantitatively correlated with a decrease in the cell-bound radioactivity (Fig. 8Bb). The radioactivity in the culture supernatants of MEFβ_d349-410 clones was significantly lower than that in the culture supernatants of MEFβ and MEFβ_d268-348 cells (Fig. 8Ba). Treatment of the culture supernatants with TCA revealed that the degradation rate of IL-2Rβ_d349-410 was lower than that of wild-type IL-2Rβ and IL-2Rβ_d268-348 (Fig. 8Bc). Similarly to the results for IL-2Rβ_d349-410 degradation in BAF-B03 cells, the production of degraded short peptides in the culture supernatant of MEFβ_d349-410 cells was not completely blocked, because part of the IL-2Rβ_d349-410 expressed in MEFβ_d349-410-2 cells was delivered to LAMP1-positive compartments (Fig. 8A). The kinetics of the degradation in MEFβ_d268-348 cells expressing IL-2Rβ_d268-348 (which lacks Lys residues in the cytoplasmic region) were the same as those in MEFβ cells (Fig. 8B), indicating that the lysine residues in the cytoplasmic region are not required for the degradation of IL-2Rβ.

The confocal microscopy and degradation analyses in the MEF transfectants revealed that IL-2Rβ_d349-410 did not accumulate in early endosomes and exhibited impaired endosomal sorting to late endosomes or lysosomes. To verify the intracellular fate of IL-2Rβ_d349-410 in MEFβ_d349-410 cells, we compared the kinetics of the receptor trafficking with those of transferrin receptor trafficking. We treated MEFβ and MEFβ_d349-410 cells with TU11 and an anti-transferrin receptor antibody, as described for Fig. 5C. Wild-type IL-2Rβ and IL-2Rβ_d349-410 were localized to compartments positive for transferrin receptor at 10 minutes, indicating that both receptors localized to early endosomes. Notably, IL-2Rβ_d349-410 was still localized to transferrin-receptor-positive compartments at 40 minutes, whereas wild-type IL-2Rβ had departed from these compartments at 40 minutes (Fig. 9). To exclude the possibility that antibody crosslinking may cause mistargeting of target receptors, we performed the experiments without the crosslinker. We demonstrated that trafficking of IL-2Rβ and transferrin receptor was not changed, with or without the crosslinker (supplementary material Fig. S3), suggesting that the antibody crosslinking has no effect on an appropriate evaluation of endosomal sorting of IL-2Rβ and transferrin receptor. These findings suggest that loss of the direct interaction between IL-2Rβ and Hrs may cause IL-2Rβ to become mis-sorted to recycling endosomes or MVBs, in which the transferrin receptor is located, as previously reported (Harding et al., 1983). Taken together, these results suggest that Hrs is involved in the precise sorting of IL-2Rβ from early to late endosomes and in the pathway for IL-2Rβ degradation in a ubiquitin-independent manner.

There are several established lines of evidence for ubiquitin-independent membrane protein trafficking. In this process, direct interaction between the membrane cargo proteins and the molecules involved in the endosomal sorting machinery has not yet been elucidated. Sna3, a yeast protein of unknown function, is sorted into the vacuolar membrane in a ubiquitin-independent manner following mutation of its cytoplasmic lysine residues to arginine residues (Reggiori and Pelham, 2001). Rsp5, a yeast orthologue of mammalian Nedd4, is an E3 ubiquitin ligase that catalyzes monoubiquitylation of cargo proteins in the MVB pathway. Recent reports have shown that Rsp5 is involved in two routes of Sna3 sorting to the MVB pathway. The first is a ubiquitin-dependent route in which Rsp5 carries out ubiquitylation of Sna3, whereas the second is a ubiquitin-independent route, in which binding of Rsp5 to Sna3 is required (McNatt et al., 2007; Oestreich et al., 2007). However, the mechanism of ubiquitin-independent sorting to the MVB pathway remains unclear. Moreover, among G-protein-coupled receptors, cytoplasmic-lysine-deleted δ-opioid and β₂-adrenergic receptors have no effect on receptor sorting into lysosomes and receptor recycling to the plasma membrane, respectively (Hanyaloglu et al., 2005; Hislop et al., 2004). In both of these cases, an involvement of Hrs has been suggested, according to the results of Hrs knockdown experiments using small interfering RNAs (siRNAs). However, no interactions between Hrs and these receptors have been detected. Thus, the authors speculated that an adaptor molecule associated with Hrs may link the receptors to Hrs. In this regard, one possibility raised is that ubiquitylated adaptor molecules are involved in the sorting of nonubiquitylated cargo proteins. For example, this possibility has arisen from the findings that ubiquitin-independent sorting of lysine-deficient Sna3 requires the enzyme activity of Rsp5, and the loss of this activity disrupts ubiquitin-independent sorting of lysine-deficient Sna3 (Oestreich et al., 2007). Thus, this type of sorting may indicate indirect ubiquitin-dependent sorting of nonubiquitylated cargo proteins. Recently, overexpression or siRNA knockdown of Hrs was found to inhibit the recycling of β₂-adrenergic receptors, and the sequence involved in this inhibition was identified in the cytoplasmic region of the β₂-adrenergic receptors. However, no direct interaction between this sequence in the cytoplasmic region and Hrs was found (Hanyaloglu and von Zastrow, 2007). Therefore, direct involvement of Hrs in ubiquitin-independent endosomal sorting of membrane cargo proteins to the MVB pathway remains to be clarified. The present results indicate that Hrs directly interacts with IL-2Rβ in a ubiquitin-independent manner, and that this interaction is required for IL-2Rβ trafficking from early to late endosomes.

Functional IL-2R complexes consist of three subunits (α-, β- and γc-chains) or two subunits (β- and γc-chains) (Sugamura et al., 1996). After internalization of the receptor complexes, IL-2Rα, which only contains 13 amino acid residues in its cytoplasmic portion, colocalizes with transferrin-positive recycling compartments, whereas IL-2Rβ and IL-2Rγc are found in Rab7-positive late endocytic compartments (Hemar et al., 1995). In the cytoplasmic region of IL-2Rγc, a sequence of five amino acid residues (ESLQP), which differs from well-established sorting signals such as tyrosine-based or dileucine-based motifs, has been proposed as the sorting signal to the degradation pathway (Morelon and Dautry-Varsat, 1998). However, endosomal sorting of IL-2Rβ has been reported to be dependent on ubiquitylation of the cytoplasmic region, as evaluated in experiments using a chimeric receptor (Rocca et al., 2001). The discrepancy between our results and the findings for the chimeric receptor may be ascribed to the cytoplasmic tail of the chimeric receptor being too short. During the construction of the chimeric receptor, the sequence containing the tyrosine-based internalization signal (EPLSYTRF) of the transferrin receptor is first inserted at the C-terminus of IL-2Rα, and then the putative sorting signal of IL-2Rβ (amino acid residues 283-292) is inserted at the C-terminal extremity of the internalization signal-inserted receptor. Accordingly, the chimeric receptor has two lysine residues in the cytoplasmic portion. The chimeric receptor is dominantly localized in LAMP1-positive late compartments, whereas a mutant of the chimeric receptor lacking the two lysine residues accumulates in early transferrin-receptor-positive compartments rather than late endocytic compartments. By contrast, we found that IL-2Rβ_d268-348 lacking all the lysine residues in the cytoplasmic region was localized to LAMP1-positive compartments and underwent degradation, similarly to wild-type IL-2Rβ. Furthermore, the absence of both lysine residues of the chimeric receptor is necessary to suppress its degradation (Rocca et al., 2001). Regarding these two lysine residues, one is located in the inserted fragment derived from IL-2Rβ, whereas the other is located in the cytoplasmic domain of IL-2Rα in the chimeric receptor. These results may not reflect the authentic role of the lysine residue derived from IL-2Rα, because IL-2Rα is not degraded in the lysosome, but recycled back to the plasma membrane. Taken together with the finding that the truncated chimera has two lysine residues, and were ubiquitylated, probably leading to ubiquitin-dependent sorting of the receptor, the chimeric receptor may not be a good model to study ubiquitin-independent sorting of IL-2R.

In addition to ubiquitylation as an endosomal sorting signal, two major endosomal sorting signals of membrane proteins are established: tyrosine-based or dileucine-based signals within the cytoplasmic domain of the proteins (Bonifacino and Traub, 2003). Yxxϕ (ϕ: residue with bulky hydrophobic side chains) and [D/E]xxxL[L/I] sequences are the consensus motifs of the tyrosine-based and dileucine-based signals, respectively. Both sequences are recognized by the adaptor protein (AP) complexes AP-1, AP-2, AP-3 and AP-4, whereas another type of dileucine-based signal (DxxLL) is recognized by another family of adaptors known as Golgi-localized γ-ear-containing Arf-binding proteins (GGAs). The VHS domain of the GGAs, but not Hrs, interacts with the DxxLL motif. The VHS domain of Hrs has no effect on the binding between Hrs and IL-2Rβ (data not shown) and residues 349-410 of IL-2Rβ do not contain a DxxLL motif. Our results indicate that residues 428-466 of Hrs are important for the binding activity toward IL-2Rβ. The association of the Hrs mutants comprising residues 1-512, 1-557 and 1-570 were markedly weaker than that of full-length Hrs. Therefore, it may be necessary to build up the Hrs-specific conformation with residues 428-466 to form the entire binding site for IL-2Rβ. However, the binding region (residues 349-410) of IL-2Rβ to Hrs contains the portion referred to as the acidic region of IL-2Rβ. Some clusters of acidic residues are known to act as sorting signals (Bonifacino and Traub, 2003). In addition, this Hrs-binding region contains four tyrosine residues (Y364, Y381, Y384 and Y387). Each tyrosine residue corresponds to a Yxxϕ motif. However, we found that the Hrs-binding activity of IL-2Rβ_1-394, which has intact acidic and tyrosine residues in the binding region, was weaker than that of IL-2Rβ_1-412. This finding indicates that even the portion of the binding region outside the cluster of acidic residues and tyrosine-based motifs has an effect on Hrs binding. Thus, further analyses, such as solving the crystallographic structure, will be required to elucidate the nature of the binding between Hrs and IL-2Rβ.

In conclusion, Hrs is one of the UIM-domain-containing molecules involved in the ubiquitin-dependent sorting machinery, and our results further suggest that it also functions as sorting machinery for nonubiquitylated cargo proteins. Future studies will reveal whether other ubiquitin-dependent sorting machineries involving the ESCRT complexes are involved in ubiquitin-independent endosomal sorting.

Expression vectors containing HA-tagged wild-type Hrs (pKU-HrsHA) and its derivative mutants HrsHAd451 (dC2), HrsHAd570 (dC1) and HrsHAd432-573 (dM) were used (Asao et al., 1997). Additional Hrs mutants were generated by PCR-based site-directed mutagenesis using pKU-HrsHA as a template (Makarova et al., 2000). HrsHAd340, HrsHAd428, HrsHAd512, HrsHAd557, HrsHAd611, HrsHAd641, HrsHAd257-277 (dUIM), HrsHAd428-443, HrsHAd428-451, HrsHAd428-466, HrsHAd428-557, HrsHAd428-602, HrsHAd470-512, HrsHAd468-555, HrsHAd558-611 and HrsHAd428-466+d558-611 were Hrs mutants with C-terminal truncations at positions 340, 428, 512, 557, 611 and 641 or deleted regions between positions 257-277, 428-443, 428-451, 428-466, 428-557, 428-602, 470-512, 468-555, 558-611 and 428-466+558-611, respectively. pSRB5 is a human IL-2Rβ expression vector (Takeshita et al., 1992). An expression vector for Flag-tagged IL-2Rβ (Flag-IL2Rβ) was generated by inserting the Flag epitope (DYKDHDIDYKDDDDK) between amino acid positions Tyr38 and Asn39 in the IL-2Rβ coding region of pSRB5 using PCR. A Flag-tagged IL-2Rγ expression vector was also constructed by inserting the above Flag epitope between amino acid positions Thr36 and Thr37 in the IL-2Rγc coding region of the human IL-2Rγ expression vector pSRG (Takeshita et al., 1992). A series of IL-2Rβ mutants were created by PCR using the above-described Flag-IL-2Rβ template. Flagβd268, Flagβd348, Flagβd357, Flagβd380, Flagβd394, Flagβd412, Flagβd268-410, Flagβd268-348 and Flagβd349-410 were IL-2Rβ mutants with C-terminal truncations at positions 268, 348, 357, 380, 394 and 412 or deleted regions between positions 268-410, 268-348 and 349-410, respectively. pMXsβ was constructed by insertion of human IL-2Rβ into a Moloney murine leukemia virus-derived vector, pMXs. pMXsβd268-348 and pMXsβd349-410 were created by PCR using the pMXsβ template. pcDNA3-HA-ubiquitin (HA-UB) was generously provided by K. Miyazono (University of Tokyo, Tokyo, Japan). EGFP-fused SARA-FYVE (GFP-SARA-FYVE) was constructed by ligating the FYVE domain (amino acid sequence from Leu587 to Met665) of hSARA into pEGFP-C3 (Clontech Laboratories). All constructs were sequenced with an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems) to verify the amino acid changes.

With the exception of BAF-B03 cells, all cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and antibiotics. Cell clones stably expressing human IL-2Rβ were derived from transfected human embryonic kidney-derived (HEK) 293T cells and designated HEK293Tβ4 cells. HEK293Tαβγ9 cells were HEK293T-transfected cell clone stably expressing human IL-2Rα, β and γc. The mouse IL-3-dependent pro-B cell line BAF-B03 was maintained in RPMI 1640 medium supplemented with 10% FBS, 10% conditioned medium derived from WEHI-3 cell line cultures (as a source of IL-3) and 50 μM 2-mercaptoethanol. F7 and S25 cells were BAF-B03-transfected cell clones stably expressing wild-type and mutant human IL-2Rβ lacking the serine-rich region (S region), respectively, and were generously provided by T. Taniguchi (University of Tokyo, Tokyo, Japan). BAF-B03β_d349-410-3 and BAF-B03β_d349-410-4 were stable clones expressing a Flag-tagged IL-2Rβ mutant lacking residues 349-410. HRSd cells were Hrs-deficient fibroblastoid cells derived from E9.5 hrs^floxP/floxP mouse embryos, whereas MEF cells were wild-type fibroblastoid cells derived from E9.5 mouse embryos (Kobayashi et al., 2005). HRSdβ0 and MEFβ11 cells were transfected HRSd and MEF cell clones, respectively, which stably expressed wild-type IL-2Rβ. MEFβ_d268-348 and two MEFβ_d349-410 clones, clone 2 and clone 22, were transfected MEF cell clones that stably expressed IL-2Rβ mutants lacking amino acids 268-348 and 349-410, respectively.

Cell surface marker expression was examined by flow cytometry. Briefly, cells (1×10⁶) were stained with an anti-IL-2Rβ monoclonal antibody (TU11) (Suzuki et al., 1989) for 30 minutes on ice, washed three times with 1% FBS in phosphate-buffered saline (PBS) and stained with a FITC-conjugated secondary antibody (MP Biomedicals) for 30 minutes on ice. After washing, the cells were fixed with 1% paraformaldehyde in PBS prior to analysis in a FACScalibur machine (Becton Dickinson).

Immunoprecipitation and immunoblotting were carried out as described previously (Asao et al., 1997). Briefly, HEK293T cells (1×10⁶) were transfected with 2 μg of each indicated vector using a calcium phosphate precipitation method, and lysed with NP-40 cell extraction buffer (1% NP-40, 20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM Na₃VO₄, 2.5 mM sodium pyrophosphate, 1 mM β-glycerol phosphate and 1 mM aprotinin). After preclearing, the lysates were immunoprecipitated with the indicated antibodies immobilized on Protein-A-Sepharose beads (GE Healthcare) at 4°C overnight. The immunoprecipitates were extensively washed with NP-40 cell extraction buffer without aprotinin, separated by 10% SDS-PAGE and transferred onto Immobilon-P membranes (Millipore). After blocking with 0.1% Tween 20 in Tris-buffered saline, the membranes were incubated with the indicated primary antibodies at 4°C overnight, washed and incubated with horseradish-peroxidase-conjugated secondary antibodies (GE Healthcare or Cell Signaling Technology) at room temperature for 1 hour. After thorough washing, signals were visualized by the ECL Western Blotting Detection System (GE Healthcare). The primary antibodies utilized were: anti-HA antibody and rabbit anti-IL-2Rβ antibody (C20) (Santa Cruz Biotechnology Inc.); anti-Flag monoclonal antibody (M2) (Sigma); rat anti-Hrs monoclonal antibody (Imos-1) (Asao et al., 1997); rabbit anti-Hrs antibody (Asao et al., 1997); and anti-IL-2Rβ monoclonal antibody (TU11).

Cells were fixed with 4% paraformaldehyde in PBS for 15 minutes on ice, permeabilized with 0.1% Triton X-100 in PBS for 10 minutes at room temperature and blocked with 10% FBS in PBS for 30 minutes. For immunostaining, the samples were incubated with the indicated primary antibodies at 4°C overnight, washed three times and incubated with appropriate secondary antibodies [anti-rabbit, anti-mouse and anti-rat IgG antibodies conjugated with Alexa Fluor 488, Alexa Fluor 594 and FITC, respectively (Molecular Probes)] at 37°C for 1 hour. Fluorescence images were captured using a confocal microscope (TCS SP2; Leica). The primary antibodies utilized were: rabbit anti-IL-2Rβ antibody (C20); rat anti-Hrs monoclonal antibody (Imos-1); rabbit anti-Hrs antibody; mouse anti-EEA1 monoclonal antibody (BD Biosciences); mouse anti-human LAMP1 monoclonal antibody (H5G11) and rat anti-mouse LAMP1 monoclonal antibody (1D4B) (Santa Cruz Biotechnology).

A GST-fused Hrs (GST-Hrs) expression vector was constructed by insertion of wild-type Hrs into pGEX-4T-1 (GE Healthcare). A His-tagged IL-2Rβ mutant (IL-2Rβ_269-551His) expression vector was generated by ligating the cytoplasmic tail of IL-2Rβ (amino acids 296-551) into pET23d (Novagen). For preparation of GST-Hrs and IL-2Rβ_269-551His recombinant proteins, E. coli strain BL21 cells were separately transformed with the two expression vectors, and grown in 3 ml LB medium containing 0.1 mM IPTG at 30°C for 24 hours. Cells were harvested and sonicated in 1 ml PBS. After centrifugation at 12,000 g for 30 minutes, each supernatant was subjected to a pull-down assay. To examine the interaction between Hrs and IL-2Rβ, glutathione-Sepharose 4B beads (GE Healthcare) containing immobilized GST or GST-Hrs were incubated with cell lysates of HEK293T cells (1×10⁶) transfected with wild-type IL-2Rβ, the indicated IL-2Rβ mutants or the IL-2Rβ_269-551His recombinant protein at 4°C overnight. The beads were washed twice with PBS, and bound proteins were analyzed by immunoblotting with an anti-IL-2Rβ antibody. Total cell lysates from transfected HEK293T cells were examined as controls. The antibodies utilized were: rabbit anti-IL-2Rβ antibody (C20); and anti-GST monoclonal antibody (26H1) (Cell Signaling Technology).

TU11 was radiolabeled with Na¹²⁵I (MP Biomedicals) by the chloramine-T method. Degradation and internalization assays with ¹²⁵I-TU11 were carried out as reported previously (Fujii et al., 1986). For degradation assays of ¹²⁵I-TU11, cells (2×10⁶) were incubated with 0.5% BSA-PBS medium containing ¹²⁵I-TU11 (9.97×10⁶ dpm/pmol) at 0°C for 60 minutes and then washed three times with PBS. Next, IL-2Rβ on the cell surface was crosslinked with ¹²⁵I-TU11 by incubation with 270 μM DSS in PBS (pH 8.3) containing 1 mM MgCl₂ for 20minutes on ice, and the reaction was terminated by the addition of PBS containing 50mM Tris-HCl (pH 7.4). The cells were then suspended in RPMI medium supplemented with 10% FBS and incubated in the presence or absence of 1nM IL-2 at 37°C for the indicated times. After centrifugation of the cell suspensions, the radioactivity of the culture supernatants and cell pellets was counted. The culture supernatants were further treated with 10% TCA. The radioactivity of the TCA-soluble fractions was counted. For internalization assays, cells (2×10⁶) were incubated with 0.5% BSA-PBS medium containing ¹²⁵I-TU11 at 0°C for 60 minutes, washed three times with 0.5% BSA-PBS, suspended in RPMI medium supplemented with 10% FBS and incubated in the presence or absence of 1nM IL-2 at 37°C for the indicated times. After centrifugation of the cell suspensions, the culture supernatants were harvested, and the cell pellets were treated with citric acid buffer (10 mM citric acid, pH 2.5, 150 mM NaCl). The radioactivity of the acid-removable citric acid buffer fractions and acid-unremovable cell precipitates was then counted.

We thank K. Miyazono (University of Tokyo, Tokyo, Japan) for providing the pcDNA-HA-ubiquitin, T. Taniguchi (University of Tokyo, Tokyo, Japan) for providing the F7 and S25 cells and T. Kitamura (University of Tokyo, Tokyo, Japan) for providing the pMXs. We also thank the Instrumental Analysis Research Center for Human and Environmental Sciences at Shinshu University for technical assistance in the DNA sequencing, flow cytometry analyses and confocal microscopy.