RILP is required for the proper morphology and function of late endosomes

Ligand-mediated endocytosis of cell-surface receptors for growth factors, hormones and cytokines represents a major mechanism for attenuating cell signalling (Ceresa and Schmid, 2000; Sorkin and Von Zastrow, 2002; Gonzalez-Gaitan and Stenmark, 2003). The translocation of a receptor from the plasma membrane to an endosomal membrane is not by itself sufficient to terminate its signalling. On the contrary, certain signalling outputs are favoured when the receptor is located at an endosome. However, as soon as the receptor becomes sorted into ILVs that pinch off inwards from the limiting endosome membrane, signalling is turned off. When the resulting MVE fuses with a lysosome, the ILV and its content become degraded, which results in downregulation of the internalized receptor (Gruenberg and Stenmark, 2004).

Because of the physiological importance of this pathway of receptor silencing and downregulation, it has been studied intensively, and some of the molecular mechanisms are now understood to some extent. For instance, the mechanisms by which endocytosed receptors become sorted into MVEs have been elucidated through recent research, which has revealed how endocytosed nutrient receptors, such as transferrin receptors, are recycled to the plasma membrane, whereas most receptors for growth factors, hormones and cytokines are sorted into MVEs (Raiborg et al., 2003). A key sorting determinant appears to be the covalent attachment of one or more ubiquitin molecules to the latter receptors. Ubiquitylation functions as a signal for MVE sorting (and in some cases, also for endocytosis), and protein complexes that recognize ubiquitin have consequently been sought as possible sorting devices (Hicke and Dunn, 2003).

The current evidence suggests that a protein complex consisting of the hepatocyte-growth-factor-regulated tyrosine kinase substrate (Hrs) and signal-transducing adaptor molecule (STAM) initiates the sorting process by binding to ubiquitylated cargo in the endosome membrane (Raiborg et al., 2003). The Hrs-STAM complex is recruited to endosomal membranes by binding of the Hrs FYVE domain to the endosomal lipid phosphatidylinositol 3-phosphate [PtdIns(3)P] (Raiborg et al., 2001b). This complex then recruits the endosomal sorting complex required for transport I (ESCRT-I) through a direct interaction between Hrs and the ESCRT-I subunit Tsg101 (Bache et al., 2003a; Katzmann et al., 2003; Lu et al., 2003). Tsg101 contains a ubiquitin-binding UEV domain that possibly could acquire the ubiquitylated cargo originally captured by Hrs (Sundquist et al., 2004). Together with two functionally related complexes, ESCRT-II and ESCRT-III, ESCRT-I not only mediates cargo sorting, but also the formation of endosomal ILVs, by a mechanism that is still not understood (Hurley and Emr, 2006; Slagsvold et al., 2006; Williams and Urbe, 2007). Interestingly, ESCRT-II also contains a ubiquitin-binding subunit, Vps36, whose GLUE domain binds both to ubiquitin and endosomal lipids (Slagsvold et al., 2005).

In a typical cell, early endosomes have a rather peripheral distribution, whereas late endosomes and lysosomes are located in a more perinuclear location (Gruenberg and Stenmark, 2004). Endocytic trafficking is known to occur along microtubules, and the PtdIns(3)P-binding motor KIF-16B mediates the trafficking of PtdIns(3)P-containing endosomes to the cell periphery, towards the plus-end of microtubules (Hoepfner et al., 2005). Conversely, dynein is known to mediate minus-end-directed transport of endosomes towards the juxtanuclear microtubule-organizing centre (Aniento et al., 1994). Although the formation of MVEs and late endosomes appears to be coordinated with their minus-end-directed trafficking along microtubules to the perinuclear region (Gruenberg and Stenmark, 2004), there are currently no known mechanistic links between endosome motility and MVE biogenesis. However, a potential candidate for such a function has emerged recently: the Rab7-interacting lysosomal protein, RILP. This protein was identified as an effector of the small GTPase Rab7, which controls trafficking of late endosomes (Cantalupo et al., 2001). RILP was subsequently found to interact with the dynein-dynactin motor complex (Jordens et al., 2001), and recent data have revealed that RILP interacts with the C-terminus of the dynactin subunit p150^GLUED and suggested that motor recruitment to late endocytic compartments requires Rab7-GTP, RILP, βIII spectrin and a second Rab7-binding protein, ORP1L (Johansson et al., 2007). The potential link to MVE sorting and biogenesis comes from the observation that RILP also interacts with the ESCRT-II subunits Vps22 and Vps36 (Progida et al., 2006; Wang and Hong, 2006). So far, however, the function of RILP in MVE sorting remains unclear.

Previous functional studies of RILP have almost exclusively relied on overexpression of full-length or truncated versions of the protein. An N-terminal deletion mutant of RILP has been found to inhibit degradation of receptors for both EGF and low-density lipoprotein (LDL) (Cantalupo et al., 2001). Moreover, overexpression of full-length RILP causes enlargement and clustering of lyso-bisphosphatidic acid (LBPA)-containing early endosomes and delays degradation of internalized EGF (Wang and Hong, 2006). Given the ability of RILP to interact with Rab7 and dynactin, as well as with Vps22 and Vps36, the functional consequences of overexpressing RILP could, in principle, be caused by sequestration of these molecules. In order to understand the function of RILP, it is therefore pertinent to ask what happens with endosomal morphology and receptor trafficking when RILP is depleted. In this report, we have used small interfering RNA (siRNA) to deplete RILP in HeLa cells and have assayed the consequences on endosome morphology as well as on trafficking of transferrin and EGF receptors.

In order to study the function of endogenous RILP, we designed two RILP-specific siRNAs and tested their ability to deplete RILP in HeLa cells. Analysis of lysates from siRNA-treated cells by SDS-PAGE and western blotting (Fig. 1A) revealed that the expression level of RILP was strongly reduced in cells incubated with either of the two siRNAs but not in cells treated with a control RNA duplex. Of the two siRNAs, the one named #2 yielded the strongest knockdown of RILP, but siRNA #1 did also cause depletion of RILP. Similar data were obtained with real-time PCR with RILP-specific primers (data not shown). These results indicate that the two siRNAs indeed reduce the levels of RILP mRNA and thereby cause depletion of RILP protein.

Because RILP interacts with ESCRT-II, we next asked whether RILP is required for the stability of the ESCRT-II subunit Vps22. For this purpose, we studied the levels of RILP and Vps22 when the respective binding partner had been depleted by siRNA. These experiments showed that depletion of RILP was accompanied by a strong reduction in the total levels of Vps22, and vice versa (Fig. 1B). This supports the idea that these proteins exist in a complex in vivo.

ESCRT-II is required for efficient ligand-mediated downregulation of EGFRs (Langelier et al., 2006; Malerød et al., 2007), and previous studies have shown that RILP interacts with the ESCRT-II subunits Vps22 and Vps36 (Progida et al., 2006; Wang and Hong, 2006). The finding that overexpression of RILP retards the degradation of EGFRs is indicative of RILP being a negative regulator of EGFR trafficking (Wang and Hong, 2006), possibly by sequestration of Vps22. If so, then depletion of RILP might facilitate ligand-mediated downregulation of EGFRs. To address this possibility, we treated cells with siRNAs against RILP or Vps22 and measured their effect on ligand-mediated degradation of EGFRs. For this purpose, we incubated control or siRNA-treated cells for 1 hour with 10 μg/ml cycloheximide in order to prevent new EGFR synthesis and then with 50 ng/ml EGF for up to three hours in order to allow lysosomal trafficking and degradation of EGFRs. Cell lysates were analyzed by SDS-PAGE and western blotting with antibodies against EGFR to assess the effect of the depletion of RILP. As a positive control, we used cells treated with siRNA against Vps22. As shown in Fig. 2A and quantified in Fig. 2B, the levels of the EGFR were strongly reduced in control RNA-treated cells after 2 hours incubation. By contrast, most of the EGFRs remained intact in Vps22-depleted cells, as reported previously (Langelier et al., 2006; Malerød et al., 2007). Interestingly, depletion of RILP with siRNA #2 had an effect similar to that of depletion of Vps22 (Fig. 2A,B), and this was also the case when RILP siRNA #1 was used (supplementary material Fig. S1). This indicates that RILP, like Vps22, is required for efficient ligand-mediated downregulation of EGFRs.

In order to establish this conclusion more firmly, we also studied the importance of RILP in EGFR trafficking with an alternative assay based on confocal microscopy (Fig. 3) (Bache et al., 2003b). As in the western blot assay, EGFR staining strongly decreased with time after addition of EGF in cells treated with control RNA. When cells had been treated with siRNA against Vps22 or RILP (siRNA #2), significant EGFR staining was detected even after 3 hours of incubation. A similar result was obtained with RILP siRNA #1 (supplementary material Fig. S2). These results thus confirm those from the western blot assay. Although it might seem surprising that depletion of RILP causes the same net effect as overexpression of RILP, this finding is not unprecedented in the EGFR trafficking pathway. Overexpression and depletion of Hrs also cause similar effects (Bache et al., 2003b; Raiborg et al., 2001a), indicating that some of the key components of degradative EGFR trafficking only function within certain critical thresholds of concentration.

The finding that RILP is required for efficient degradation of the EGFR raised the question of whether RILP is required for trafficking of the EGFR between early and late endosomes or between late endosomes and lysosomes. To study this question, we studied the localization of EGFRs in RILP-depleted cells. When cells were stimulated with EGF for 2 hours, EGFRs were hardly detected in cells treated with control RNA, as expected from the results in Figs 2 and 3 (Fig. 4). However, in cells treated with siRNA against RILP, internalized EGFRs were detected and found to colocalize strongly with the early-endosome marker EEA1. Conversely, there was only a limited colocalization between EGFR and the late endosome marker LAMP-1 at this time point. This indicates that RILP, like Vps22 (Langelier et al., 2006; Malerød et al., 2007), is required for receptor trafficking from early to late endosomes.

The cation-independent mannose-6-phosphate receptor (CI-M6PR) shuttles between the trans-Golgi network (TGN) and endosomes, in addition to being transported from the TGN to the plasma membrane (Ghosh et al., 2003). In order to determine whether RILP is required for normal endosome-to-TGN transport, we investigated the distribution of CI-M6PR in cells transfected either with control RNA or with siRNA against RILP. As seen in Fig. 5A, the CI-M6PR colocalized to the same extent with TGN46 in control and RILP-depleted cells, indicating that normal shuttling occurred between the TGN and endosomes in RILP-knockdown cells. Likewise, the CI-M6PR localized to EEA1-positive early endosomes in cells treated with siRNA against RILP, just as in control cells, implying normal endocytosis of cell-surface receptors. However, one striking difference between control cells and RILP-depleted cells was that CI-M6PR staining was markedly stronger in the latter. This was verified by image quantification of multiple cells (Fig. 5B). By contrast, TGN46 and EEA1 levels remained unchanged in the RILP-depleted cells. The increased level of the CI-M6PR in RILP-depleted cells was verified by western blotting (Fig. 5C), and the quantification of the western blot showed an increase similar to that detected by confocal microscopy (Fig. 5D). These results suggest that RILP depletion does not affect the distribution of CI-M6PR between endosomes and the TGN, whereas it inhibits its degradation in lysosomes.

The accumulation of EGFRs in early endosomes in Vps22- and RILP-depleted cells raised the possibility that depletions of Vps22 and RILP cause a general block of transport out of early endosomes. To address this, we studied the trafficking of transferrin, whose receptor is known to become constitutively endocytosed from clathrin-coated pits and to recycle efficiently from early endosomes to the plasma membrane after delivering the transferrin-bound Fe³⁺ in the lumen of the endosome (Yamashiro and Maxfield, 1984). Transferrin labelled with ¹²⁵I was first prebound to control or siRNA-treated HeLa cells at 4°C, a temperature at which endocytosis does not occur. The cells were then warmed to 37°C and incubated for different time periods before their surface-associated transferrin was stripped off by treatment with pronase. The cell-associated radioactivity, representing endosomal transferrin, was then detected by γ-counting. This analysis (Fig. 6) showed that transferrin was rapidly internalized in control cells, and internal transferrin peaked at 5 minutes, before gradually decreasing as a result of recycling. This is entirely consistent with previously published results for the trafficking of transferrin (van der Sluijs et al., 1992; Bucci et al., 1992). Interestingly, the internalization and recycling curves for ¹²⁵I-labelled transferrin were essentially identical in Vps22- and RILP-depleted cells, as in control cells. This indicates that Vps22 and RILP, despite their importance for trafficking of the EGFR out of early endosomes, are not required for efficient recycling of transferrin receptors.

The finding that RILP is needed for ligand-mediated EGFR degradation but not for transferrin recycling suggested that RILP is specifically required for degradative trafficking of receptors. As CI-M6PRs accumulated in RILP-depleted cells, RILP might be required for the degradative function of late endosomes, possibly by mediating their fusion with lysosomes. To investigate this further, we monitored the expression of two late-endosomal markers in control versus RILP-depleted cells. In cells treated with control RNA, endosomes containing lyso-bisphosphatidic acid (LBPA) and Lamp1 had a mainly perinuclear distribution and did not colocalize with EEA1, as expected (Fig. 7A, upper panels). While these late-endosomal markers failed to colocalize with EEA1 in RILP-depleted cells as well, these cells showed a striking increase in the detected levels of LBPA and Lamp1 (Fig. 7A, lower panels). Image quantification of multiple cells confirmed this (Fig. 7B). This is consistent with the finding that CI-M6PR levels were increased in RILP-depleted cells and indicates that RILP is required for the (slow) degradation of resident late-endosomal lipids and proteins.

It has been reported previously that depletion of RILP causes dispersal of CD63-positive late endosomes towards the cell periphery (Johansson et al., 2007). We were unable to detect any such redistribution of LBPA- and Lamp1-positive endosomes (Fig. 7A), and we also did not detect any peripheral translocation of CD63-positive endosomes in RILP-depleted cells (data not shown). Possible reasons for and consequences of these apparent differences are provided in the Discussion below.

To study the late endosomes of RILP-depleted cells at higher resolution, we employed cryo-immunoelectron microscopy, with CD63 as a marker for multivesicular late endosomes (Metzelaar et al., 1991). This analysis (Fig. 8) revealed striking differences in the morphology of late endosomes between control and RILP-depleted cells. While CD63 was found mainly in the ILVs of MVEs in control cells, RILP-depleted cells contained clustered endosomes with a very strong CD63 labelling on their limiting membrane. Most strikingly, the CD63-positive structures in RILP-depleted cells were almost devoid of intralumenal membranes. This indicates that RILP is not only required for the normal distribution of late endosomes/lysosomes but also for the formation of ILVs in late endosomes.

In this study, we have compared the importance of RILP and its interaction partner Vps22 for endocytic trafficking of two receptors that follow distinct pathways from early endosomes. Whereas recycling of the transferrin receptor remained unaffected by depletion of Vps22 or RILP, trafficking of EGFRs from early to late endosomes was strongly delayed when either Vps22 or RILP was depleted. This is consistent with the idea that both these proteins function specifically in degradative protein trafficking. While Vps22 and the rest of the ESCRT-II complex have been proposed to participate in the sorting of ubiquitylated cargo (such as the EGFR) into ILVs of MVEs, RILP is known to contact the dynein-dynactin microtubule motor complex (Johansson et al., 2007; Jordens et al., 2001). This has suggested that RILP might mediate the dynein-dependent motility of late endosomes towards the juxtanuclear microtubule-organizing centre. Importantly, we observed that depletion of RILP inhibited the biogenesis of ILVs in late endosomes and led to elevated levels of resident late-endosome lipid and protein molecules. Our results suggest that RILP serves to coordinate ILV biogenesis with dynein-dependent endosome motility.

Previous studies have shown that overexpression of RILP inhibits ligand-mediated degradation of EGFRs and causes juxtanuclear clustering of late endosomes (Cantalupo et al., 2001; Jordens et al., 2001; Wang and Hong, 2006). This phenotype is probably related to the coupling of RILP to the dynein-dynactin complex. RILP has been proposed to form a dimeric tripartite complex with Rab7-GTP and ORP1L, and to recruit p150^GLUED to this complex (Johansson et al., 2007). Depletion of RILP has been reported to cause a redistribution of CD63-positive late endosomes towards the cell periphery (Johansson et al., 2007), but the requirement of RILP for endosome morphology and function has not been addressed previously. Our findings that depletion of RILP strongly delayed degradation of the EGFR and led to an increase in the levels of CI-M6PR, CD63, Lamp1 and LBPA on late endosomes are consistent with the idea that RILP is crucial for the degradative functions of late endosomes. The function of RILP in this respect could be related to its involvement in the biogenesis of ILVs or the motility of endosomes leading to fusion with lysosomes, or both.

There could be several reasons why we did not observe the redistribution of late endosomes towards the cell periphery reported previously in RILP-depleted HeLa cells (Johansson et al., 2007). As we used different siRNA oligonucleotides, the most obvious explanation might be related to differences in the efficiencies of knockdown. However, it is worth noting that the siRNA-mediated RILP depletions we obtained clearly were sufficient to cause strong effects on the degradation of the EGFR and the morphology of endosomes. This raises the possibility that these effects are not related to the proposed function of RILP as a tether between endosomes and the dynein-dynactin complex. More likely, the observed effects might be due to the functional relationship between RILP and ESCRT-II in the sorting of EGFRs and the biogenesis of MVEs. Our finding that depletion of RILP retards endocytosed EGFRs in early endosomes is reminiscent of previous studies with knockdown of ESCRT subunits (Bishop et al., 2002; Bache et al., 2003b; Bache et al., 2006; Malerød et al., 2007) and suggests that RILP participates together with ESCRT proteins in the maturation of early endosomes into late endosomes.

In normal HeLa cells, late endosomes typically contain ILVs and whorl-shaped membranes. Such intralumenal membranes were rarely observed in CD63-positive late endosomes in RILP-depleted cells. Instead, these cells contained clusters of CD63-positive endosomes that contained few intralumenal membranes. In light of the fact that overexpression of RILP causes clustering of endosomes (Cantalupo et al., 2001; Jordens et al., 2001), it is perhaps surprising that depletion of RILP has a similar effect. However, this finding agrees with the observation that expression of a truncated RILP construct lacking the N-terminal half causes clusters of endosomes with a peripheral distribution (Cantalupo et al., 2001) and might imply that the concentration of RILP must be kept within critical concentration thresholds for its correct function in endocytic trafficking. The abnormal morphology of late endosomes in RILP-depleted cells begs the question of how RILP controls the biogenesis of these organelles. The fact that RILP interacts with ESCRT-II is likely to be of relevance in this context, and, because depletion of RILP causes a co-depletion of Vps22, some of the effects of RILP depletion might be related to the accompanying loss of Vps22. However, this is unlikely to account for all the endosomal effects observed by RILP depletion as RILP-depleted endosomes are morphologically distinct from Vps22-depleted endosomes (Malerød et al., 2007). The former contain even fewer ILVs and are often found clustered together, consistent with the finding that expression of an N-terminal deletion mutant of RILP results in scattered clusters of endosomes (Cantalupo et al., 2001). This indicates that an additional function of RILP is to prevent endosomal clustering, and the interaction of RILP with the dynein-dynactin complex is likely to be of relevance for this. Because RILP controls the morphology of endosomes (by means of ESCRT-II) and their tubulin-dependent motility (through dynein-dynactin), it is tempting to speculate that RILP serves to coordinate the biogenesis of ILVs with minus-end-directed transport during the maturation of endosomes.

In conclusion, this is the first study to determine the requirement of RILP for endosomal functions and morphology. We found RILP to be specifically required for degradative endocytic trafficking and for the morphology of late endosomes. In further studies, it will be important to set up assays that can establish the biochemical functions of RILP and ESCRT-II in ILV biogenesis and microtubule-dependent endosome motility.

Affinity-purified rabbit antibodies against RILP have been described previously (Cantalupo et al., 2001). A mouse monoclonal antibody against α-tubulin was obtained from Sigma Aldrich (St Louis, MO). Antiserum raised against recombinant human Vps22 (as a fusion with maltose binding protein, MBP) was prepared by Eurogentec (Herstal, Belgium). The rabbit serum was affinity purified on beads containing recombinant Vps22, prepared using Affi-Gel (Bio-Rad). Sheep anti-EGFR came from Fitzgerald (Concord, MA). Human anti-EEA1 antiserum was a gift from Ban-Hock Toh (Monash University, Melbourne, Australia). Mouse anti-human CD63 was obtained from Developmental Studies Hybridoma Bank (University of Iowa, IA). Rabbit anti-human Lamp-2 and mouse-anti LBPA were kindly given by Gillian Griffiths (University of Oxford, UK) and Jean Gruenberg (University of Geneva, Switzerland), respectively. Rabbit anti-human-CI-M6PR was a gift from Bernard Hoflack (Dresden University of Technology, Germany). Rabbit-IgG-HRP, mouse-IgG-HRP, sheep-IgG-HRP and Cy2-, Cy3- and Cy5-labelled secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Oligofectamine was provided by Invitrogen (Carlsbad, CA), and cycloheximide and human EGF were from Sigma Aldrich (St Louis, MO). Precise Protein Gels, SuperSignal West Pico Chemiluminescent Substrate and SuperSignal West Dura Extended Chemiluminescent Substrate were purchased from Pierce (Rockford, IL). Protein-A-gold was from CMC (Utrecht, The Netherlands).

HeLa cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS), 2 mM glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin, in a 5% CO₂ incubator at 37°C. For RNAi, we used the following oligonucleotides: siRNA-Vps22, sense sequence 5′-CUUGCAGAGGCCAAGUAUATT-3′ and antisense sequence 5′-UAUACUUGGCCUCUGCAAGTT-3′; siRNA-RILP1, sense sequence 5′-GCAGCGGAAGAAGAUCAAGTT-3′ and antisense sequence 5′-CUUGAUCUUCUUCCGCUGCTT-3′; and siRNA-RILP2, sense sequence 5′-GAUCAAGGCCAAGAUGUUATT-3′ and antisense sequence 5′-UAACAUCUUGGCCUUGAUCTT-3′. As a negative control we used a scrambled sequence: sense scrambled control 5′-ACUUCGAGCGUGCAUGGCUTT-3′ and antisense scrambled control 5′-AGCCAUGCACGCUCGAAGUTT-3′. All chemically synthesized oligonucleotides were purchased from a commercial supplier (MWG-Biotech). Transfection of HeLa cells with siRNA was performed as described previously (Cabezas et al., 2005). Briefly, HeLa cells were plated 1 day before transfection in 5 cm dishes (∼4×10⁵ cells/dish). The cells were first transfected with siRNA using Oligofectamine (Invitrogen, Carlsbad, CA) for 72 hours, replated and left 48 hours before performing further experiments.

Cells grown on coverslips were permeabilized with 0.05% saponin, fixed in 3% paraformaldehyde and stained with antibodies, as previously described (Simonsen et al., 1998). The coverslips were examined with a Zeiss LSM 510 META confocal microscope, and the Zeiss LSM 510 software (version 3.2) was used to quantify intensities of the staining. Confocal pictures used for quantification were scanned at the same pinhole, offset gain and amplifier values below pixel saturation.

Degradation of the EGFR was determined in control RNA, RILP or Vps22 siRNA-transfected cells. The cells were pretreated with 10 μg/ml cycloheximide for 1 hour, which prevents synthesis of new EGFR during the stimulation by EGF, before they were continuously stimulated with 50 ng/ml EGF for 15, 60, 120 or 180 minutes. The levels of undegraded EGFR were determined by western blotting and quantitative immunofluorescence microscopy. The intensity of EGFR labelling at each time point was calculated relative to the intensity measured at 15 minutes (which was set to 100%).

HeLa cells, transfected with control RNA or RILP siRNA (see above), were fixed in 4% formaldehyde/0.1% glutaraldehyde in 0.1 M phosphate buffer at room temperature for 40 minutes, washed, scraped and pelleted in 12% gelatin at 10,000 g. Specimens were infiltrated with 2.3 M sucrose, mounted on silver pins and frozen in liquid nitrogen. Ultrathin cryosections were cut at –110°C (Leica EM FCS ultramicrotome) and collected with a 1:1 mixture of 2% methyl cellulose and 2.3 M sucrose. Sections were transferred to formvar/carbon-coated grids and labelled with a mouse CD63 primary antibody followed by a bridging secondary anti-mouse antibody and protein-A-conjugates, essentially as described previously (Slot et al., 1991). Sections were observed at 60 kV in a JEOL JEM-1230 electron microscope. Micrographs were recorded with a digital CCD camera (Morada) using iTEM (SIS) software and further processed with Adobe Photoshop (7.0) software.

HeLa cells treated with siRNA against RILP, Vps22 or a scrambled RNA duplex were incubated with 100 nM ¹²⁵I-labelled human transferrin (hTf) at 0°C for 60 minutes, washed three times with PBS-0.1% BSA and further incubated at 37°C for different times. Endocytosed ¹²⁵I-labelled hTf was estimated as the residual radioactivity after pronase treatment compared with total bound ¹²⁵I-labelled hTf. Quantification of ¹²⁵I-labelled hTf in relevant samples was performed by γ-radiation counting.

L.M., S.S. and A.B. are postdoctoral, predoctoral and senior research fellows, respectively, of the FUGE programme. This work was also supported by Telethon-Italy (Grant No. GGP05160), the Norwegian Cancer Society, the Research Council of Norway, the Novo Nordisk Foundation and the Hartmann Family Foundation.