SGIP1α Is an Endocytic Protein That Directly Interacts with Phospholipids and Eps15*

SGIP1 has been shown to be an endophilin-interacting protein that regulates energy balance, but its function is not fully understood. Here, we identified its splicing variant of SGIP1 and named it SGIP1α. SGIP1α bound to phosphatidylserine and phosphoinositides and deformed the plasma membrane and liposomes into narrow tubules, suggesting the involvement in vesicle formation during endocytosis. SGIP1α furthermore bound to Eps15, an important adaptor protein of clathrin-mediated endocytic machinery. SGIP1α was colocalized with Eps15 and the AP-2 complex. Upon epidermal growth factor (EGF) stimulation, SGIP1α was colocalized with EGF at the plasma membrane, indicating the localization of SGIP1α at clathrin-coated pits/vesicles. SGIP1α overexpression reduced transferrin and EGF endocytosis. SGIP1α knockdown reduced transferrin endocytosis but not EGF endocytosis; this difference may be due to the presence of redundant pathways in EGF endocytosis. These results suggest that SGIP1α plays an essential role in clathrin-mediated endocytosis by interacting with phospholipids and Eps15.

elusive whether its tubulin binding is significant, but we have found that SGIP1␣ is an endocytic protein, which directly interacts with phosphoinositides and Eps15.

EXPERIMENTAL PROCEDURES
Tubulin Blot Overlay-Tubulin blot overlay was performed as described (15,16) with some modifications. Samples to be tested were subjected to SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked with PBS containing 5% fat-free powder milk overnight at 4°C and then equilibrated with PEM buffer (100 mM PIPES/NaOH, pH 6.8, 1 mM EGTA, 1 mM MgCl 2 , and 1 mM DTT) for 10 min at room temperature. The membrane was then incubated with 30 g/ml tubulin in PEM buffer containing 1% bovine serum albumin for 1 h at 4°C. After washing in PBS, the membrane was subjected to immunoblot analysis using anti-tubulin ␣ mAb (DM1A, Sigma-Aldrich).
Purification of p100 and Mass Spectrometry-All the purification procedures were carried out at 0 -4°C. Twenty rat brains were homogenized in 200 ml of buffer A (20 mM Tris/ HCl, pH 8.0, 2 mM EDTA, 5 mM EGTA, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, and 1 g/ml pepstatin A) followed by ultracentrifugation. The pellet was rehomogenized in 200 ml of buffer A followed by ultracentrifugation. The pellet was again homogenized in 100 ml of buffer A containing 1.5 M NaCl. The homogenate was gently stirred for 30 min followed by ultracentrifugation. The supernatant was then dialyzed against buffer B (20 mM HEPES/ NaOH, pH 7.5, 1 mM EDTA, and 1 mM DTT) overnight followed by ultracentrifugation. The pellet was resuspended in 65 ml of buffer C (buffer B containing 6 M urea) and gently stirred for 30 min. After the sample was centrifuged, the supernatant was applied to an SP-Sepharose Fast Flow column (2.6 ϫ 10 cm, GE Healthcare Bio-Science Corp.) equilibrated with buffer C. Elution was performed with 60 ml of buffer C containing 0.5 M NaCl. Fractions of 5 ml each were collected. The active fractions (fractions 6 -12) were collected and diluted with 210 ml of buffer D (20 mM Tris/HCl, pH 7.5, 1 mM EDTA, 1 mM DTT, and 4 M urea). The sample was applied to a HiPrep 16/10 Q Fast Flow column (1.6 ϫ 10 cm, GE Healthcare Bio-Science Corp.) equilibrated with buffer D. Elution was performed with a 200-ml linear gradient of NaCl (0 -0.5 M) in buffer D. Fractions of 4 ml each were collected. The active fractions (fractions [22][23][24][25][26][27] were collected. The sample was subjected to precipitation by the chloroform/methanol/water system (17) and dissolved in 15 ml of buffer E (10 mM potassium phosphate, pH 7.8, 1 mM DTT, 0.6% CHAPS, and 4 M urea). The sample was applied to a TSKgel (0.75 ϫ 7.5 cm, HA-1000, Tosoh) equilibrated with buffer E. Elution was performed with a 30-ml linear gradient of potassium phosphate (10 -500 mM) in buffer E. Fractions of 1 ml each were collected. The active fractions (fractions [11][12][13][14][15][16][17][18] were collected and subjected to precipitation by the chloroform/methanol/water system (17). The sample was dissolved in 5 ml of buffer F (20 mM bis-Tris/HCl, pH 5.5, 1 mM EDTA, 1 mM DTT, and 4 M urea) and applied to a Mono Q 5/50 GL column (GE Healthcare Bio-Science Corp.) equilibrated with buffer F. Elution was performed with a 20-ml linear gradient of NaCl (0 -500 mM) in buffer F. Fractions of 0.5 ml each were collected. The active fractions (fractions 16 -19) were subjected to SDS-PAGE followed by silver staining (18). After the protein band corresponding to p100 was excised and subjected to in-gel trypsin digestion, the resultant peptides were desalted by Zip tips C18 (Millipore) and subjected to infusion nanoESI QQTOF mass spectrometry (QSTAR Pulsar i Applied Biosystems/MDS SCIEX). Partial aa sequences were obtained with peptide ion fragmentation due to high collision energy. The sequences obtained were subjected to a search for sequence similarity against the current NCBI data base using the Mascot Search Program (Matrix Science Ltd.).
Cell Culture, Transfection, and Immunofluorescence Microscopy-Cells were cultured at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transfection was performed using Lipofectamine 2000 reagent (Invitrogen). Cells were fixed with 1% formaldehyde in PBS for 15 min, permeabilized with 0.2% Triton X-100 in PBS for 10 min, and immunostained with various Abs. DiIC 16 (3) (Invitrogen) staining was performed as described (20). Images were taken with a fluorescence microscope (BX51, Olympus) or a confocal microscopy system (BX50 and Fluoview FV300; Olympus) at room temperature. The fluorochromes used included Alexa Fluor 488 and 594, Texas-red conjugated EGF, and DiIC 16 (3) (Invitrogen). A ϫ60 oil immersion objective, NA 1.40 (Olympus) was used. Images were assembled with PhotoShop (Adobe). In each plate, photographs were cropped, and each fluorochrome was adjusted for identical brightness and contrast to represent the observed images.
The quantification of colocalization was performed using Meta-Morph imaging system software (21). Briefly, background was subtracted from unprocessed images, and the percentage of SGIP1␣ pixels overlapping Eps15, ␣-adaptin, or EGF pixels was measured. For the colocalization in the plasma membrane region, images were acquired to display the entire cell surface adhering to culture dishes (22). The colocalization percentage was determined in a restricted region of the images extending 15 pixels from the cell edge toward the cytoplasm. For the colocalization in the intracellular region, images were acquired at upper planes from the ventral surface. The colocalization percentage was determined in a region of the images excluding 15 pixels from the cell edge. Data were shown as the means ϩ S.E. of three independent experiments.
Liposome Binding and Liposome Tubulation Assays-Liposome binding assay was performed as described (12). PE/PC liposomes consisted of 70% PE, 20% PC, and 10% PI or various phosphoinositides. Where indicated, liposomes consisted of 80% PE, 20% PC, and various percentages of PS (with a corresponding reduction in PE). MBP-SGIP1␣-4 (50 g/ml) was incubated with 1 mg/ml liposomes for 15 min at room temperature followed by centrifugation. To calculate the K d value, various doses of GST-SGIP1␣-4 were incubated with 0.2 mg/ml liposomes (60% PE, 20% PC, and 20% PI(4,5)P 2 ) for 30 min at 4°C followed by centrifugation. Comparable amounts of the supernatant and pellet fractions were subjected to SDS-PAGE followed by CBB staining. The protein amount was quantified by scanning using NIH image (version 1.61). Liposome tubulation assay was carried out as described (12).
In Vitro Binding Assays and Immunoprecipitation-To examine the in vitro binding of SGIP1␣ to Eps15, MBP-Eps15 or MBP alone (200 pmol each) was incubated with GST-SGIP1␣-3 or GST alone (200 pmol each) immobilized on beads in PBS containing 0.1% Triton X-100. After washing, bound proteins were subjected to SDS-PAGE followed by CBB staining.
Immunoprecipitation was performed as follows. Cells were lysed in lysis buffer (50 mM Tris/HCl, pH 7.5, 5 mM EGTA, 5 mM EDTA, 150 mM NaCl, 15 mM NaF, 1.5 mM Na 3 VO 4 , 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 1 g/ml pepstatin A) followed by centrifugation. The supernatant (1 mg of protein) was incubated for 3 h at 4°C with anti-SGIP1␣ pAb or control rabbit IgG. Protein G-Sepharose beads were added to the sample, which was further incubated for 3 h at 4°C. After the beads were thoroughly washed with the same buffer, bound proteins were subjected to SDS-PAGE followed by immunoblot analysis.
RNAi-Stealth double-stranded RNAs were purchased from Invitrogen. The sequence of siRNA specific to SGIP1␣ was 5Ј-UUU ACC CAG AAU UCC UUG GUA UUG G-3Ј (corresponding nucleotide 1997-2021 relative to the start codon). A double-stranded RNA targeting luciferase (5Ј-CGU ACG CGG AAU ACU UCG AAA UGU C-3Ј) was used as a control. N1E115 cells were transfected with 20 nM siRNA using Lipofectamine 2000 reagent (Invitrogen). After 24 h, a second transfection was performed, and the cells were cultured for 72 h and subjected to various experiments.  Endocytosis Assays-Endocytosis in COS7 cells was assayed using fluorescent ligands as described (12,20). Endocytosis assays in N1E115 cells using fluorescent ligands were performed as described (12,20) with slight modifications. Briefly, cells were starved with serum-free Dulbecco's modified Eagle's medium for 2 h and incubated with 25 g/ml Alexa Fluor 594conjugated Tfn (Invitrogen) or 20 ng/ml Texas red-conjugated, biotinylated EGF (Invitrogen) for 10 min at 37°C. After acid stripping, cells were fixed with formaldehyde. The average intensity of internalized ligands per cell area was determined using MetaMorph imaging system software (23). Endocytosis assays in N1E115 cells using radioactive ligands were performed as described (24,25) with slight modifications. Briefly, cells were starved for 2 h and incubated with 10 nM 125 I-labeled Tfn (PerkinElmer) or 1.5 ng/ml 125 I-labeled EGF (PerkinElmer) for the indicated periods of time at 37°C in Dulbecco's modified Eagle's medium containing 1% bovine serum albumin. Internalized and surface radioactivity was quantified by a gamma counter. Nonspecific binding was measured for each time point in the presence of 100-fold molar excess of the same unlabeled ligand.

RESULTS
Identification of p100 as a Tubulin-binding Protein-To detect tubulin-and/or MT-binding proteins, we performed tubulin blot overlay. When various rat tissue homogenates were subjected to blot overlay, several bands with various molecular masses were detected (Fig. 1A). Of these protein bands, a thick band of about 100 kDa (p100) was detected only in the brain. By successive column chromatographies, p100 was copurified with another band of about 110 kDa (p110) (supplemental Fig.  1). These protein bands were separately excised out from the gels, digested with trypsin, and subjected to mass spectrometry. From the p100 protein band, four peptide sequences were obtained (supplemental Fig. 2A). A data base search revealed that the peptide sequences of p100 were identical to those of a mouse hypothetical protein (GenBank accession number BC017596). p110 was found to be CLIP-115, a brain-specific MT-binding protein (32). We performed PCR to amplify cDNA of the hypothetical protein using mouse brain cDNA as a template. Our cloned cDNA encoded a protein that consisted of 854 aa and showed a calculated molecular weight of 91,719 (DDBJ/EMBL/GenBank accession number AB262964) (supplemental Fig. 2A). This protein contained a proline-rich domain in the middle region, whereas it did not possess any conserved domains in the N-terminal or C-terminal region (see Fig. 2A).
During the study, the mouse hypothetical protein was found to be a recently reported neuronal protein, named SGIP1 (14). When the aa sequences of our clone and SGIP1 were aligned, they showed 94% identity (supplemental Fig. 2A). Our clone had two inserted peptide sequences at its N-terminal and middle regions. It is likely that our clone is a longer splicing variant of SGIP1; therefore, the protein encoded by our clone was named SGIP1␣. To confirm whether SGIP1␣ is detected by tubulin blot overlay, we expressed a Myc-tagged protein in COS7 cells. Myctagged SGIP1␣ was detected by blot overlay (Fig. 1B).
Membrane Tubulation by SGIP1␣ Overexpression-To examine the binding of SGIP1␣ to tubulin and/or MTs in intact cells, we examined the subcellular distribution of SGIP1␣. When transiently expressed at a high level in COS7 cells, EGFP-SGIP1␣ (full-length) showed short tubular structures ( Fig. 2A). EGFP-SGIP1␣-1 (aa 1-280) and -4 (aa 1-97) also showed tubular structures, whereas other fragments, including EGFP-SGIP1␣-2 (aa 261-580), -3 (aa 561-854), and -5 (aa 98 -280), did not. These results indicate that the N-terminal region (aa 1-97) of SGIP1␣ is responsible for the formation of tubular structures. These structures appeared to be different from those of MTs but were reminiscent of those recently reported to be formed by phospholipid-binding BAR and EFC domains (11,12,20). To examine the possibility that the N-terminal region (aa 1-97) binds to the plasma membrane and deforms it into tubules, COS7 cells overexpressing EGFP-SGIP1␣-4 (aa 1-97) were stained with DiIC 16 (3). DiIC 16 (3) is a lipophilic fluorescent probe used for plasma membrane staining (33). The tubular structures of EGFP-SGIP1␣-4 were completely overlapped with DiIC 16 (3) staining (Fig. 2B). This result indicates that tubular structures originate from the plasma membrane. It remains elusive whether SGIP1␣ binds to tubulin and/or MTs in intact cells. Based on the observation that the N-terminal region (aa 1-97) of SGIP1␣ binds to membrane phospholipids as described below, we named this region the MP (membrane phospholipid-binding) domain, which shows no significant homology to any proteins in the current protein data base.
Direct Binding of SGIP1␣ to Phospholipids-To examine whether the MP domain directly interacts with membrane phospholipids in vitro, we performed liposome cosedimentation assay. Although MBP-SGIP1␣-4 (MP domain) did not bind to synthetic liposomes composed of PE and PC, it strongly bound to liposomes containing PS, PI 3-phosphate, PI 4phosphate, PI 3,4-bisphosphate, PI 3,5-bisphosphate, or PI(4,5)P 2 (Fig.  3A). The MP domain faintly bound to liposomes containing PI, PI 5-phosphate, or PI 3,4,5-trisphosphate. The percentage of PS in liposomes for maximal binding was about 10%. The MP domain bound to brain lipids (Folch fraction) rich in PS (about 50% of total lipids). This binding specificity of the MP domain is similar to that of the FBP17 EFC domain (11,12). The K d value of the MP domain for PI(4,5)P 2 was calculated to be about 5 ϫ 10 Ϫ7 M (Fig. 3B). This value is comparable with those of the PLC␦1 pleckstrin homology domain, the epsin ENTH domain, and the FBP17 EFC domain, which are known to bind PI(4,5)P 2 with high affinity (8,9,12). We next examined using synthetic liposomes containing rhodamine-conjugated PE whether the MP domain deforms liposomes into tubules. MBP-SGIP1␣-4 induced tubulation, whereas MBP alone did not (Fig. 3C). Taken together, these results indicate that SGIP1␣ binds to membrane phospholipids and deforms membranes into tubules through the MP domain.
Interaction of SGIP1␣ with Eps15 and the AP-2 Complex-We then attempted to identify an SGIP1␣-interacting protein(s) using the yeast two-hybrid method. We screened 2.5 ϫ 10 6 clones of a prey cDNA library from rat brain with a bait construct, pBTM116HA-SGIP1␣-7 (aa 428 -854), and 14 independent clones were obtained. Of these clones, pPrey clones 1-6 encoded C-terminal regions of Eps15 (Fig. 4A). Eps15 is an essential adaptor protein of clathrin-mediated endocytic machinery (34 -37). It directly interacts with the AP-2 complex and is involved in the formation of clathrin-coated pits. Eps15 consists of three Eps15 homology (EH) domains, a coiled-coil region, an Asp-Pro-Phe (DPF) repeat region, and two ubiquitin-interacting motifs (UIMs) (34). To examine the in vitro direct binding of SGIP1␣ and Eps15, we prepared MBP-Eps15 (aa 593-834). It bound to GST-SGIP1␣-3 (aa 561-854) immobilized on glutathione-Sepharose beads, whereas MBP alone did not (Fig. 4B). Neither MBP alone nor MBP-Eps15 bound to GST immobilized on beads. These results indicate that the C-terminal region of SGIP1␣ directly binds to the DPF repeat region of Eps15 in vitro.
We next examined the in vivo binding of SGIP1␣ to Eps15 by immunoprecipitation. For this purpose, we used N1E115 neuroblastoma cells because Northern and Western blot analyses showed that SGIP1␣ was predominantly expressed in neural tissues (supplemental Fig. 2B). When endogenous SGIP1␣ was immunoprecipitated from N1E115 cells with anti-SGIP1␣ pAb, endogenous Eps15 was co-precipitated (Fig. 4C). ␣-Adaptin, a component of the AP-2 complex, was also co-precipitated with SGIP1␣ and Eps15; however, ␥-adaptin, a component of the AP-1 complex, was not. We further examined the localization of SGIP1␣, Eps15, and the AP-2 complex by immunofluorescence microscopy. When transiently expressed at a low level in N1E115 cells, EGFP-SGIP1␣ showed a punctuate staining pattern and was colocalized with endogenous Eps15 and ␣-adaptin (Fig. 5A). The percentage of SGIP1␣-Eps15 colocalization in the plasma membrane region was similar to that in the intracellular region (38.6 Ϯ 3.2 versus 41.1 Ϯ 0.9%, respectively), whereas the percentage of SGIP1␣-␣ adaptin colocalization in the plasma membrane region was higher than that in the intracellular region (27.7 Ϯ 2.5 versus 14.7 Ϯ 1.8%, respectively). When EGFP-SGIP1␣ overexpression induced membrane tubulation in COS7 cells, Eps15 was recruited to membrane tubules (Fig. 5B). These results indicate that SGIP1␣ interacts with Eps15 and the AP-2 complex in intact cells.
Localization of SGIP1␣ at Clathrin-coated Pits/Vesicles-We examined the subcellular localization of SGIP1␣ when cells were incubated with EGF. EGFP-SGIP1␣-expressing N1E115 cells were incubated with EGF for 1 h at 4°C and moved to a 37°C incubator. At 0 and 1 min, SGIP1␣ was colocalized with EGF at the plasma membrane (Fig. 6A). The percentages of SGIP1␣-EGF colocalization were as follows: 0 min, 19.4 Ϯ 1.9% (plasma membrane region) and 3.5 Ϯ 0.8% (intracellular region); and 1 min, 21.5 Ϯ 1.7% (plasma membrane region) and 6.7 Ϯ 0.5% (intracellular region). When EGFP-SGIP1␣expressing COS7 cells were incubated with EGF at 4°C, SGIP1␣ was colocalized with EGF in tubules at the plasma membrane (Fig. 6B). Taken together with the above observation that SGIP1␣ is colocalized with Eps15 and the AP-2 complex, these results indicate that SGIP1␣ is localized at clathrin-coated pits/vesicles. Involvement of SGIP1␣ in Clathrin-mediated Endocytosis-We examined the involvement of SGIP1␣ in endocytosis by overexpression. When EGFP-SGIP1␣ was transiently expressed in COS7 cells, the uptake of Alexa Fluor 594-conjugated Tfn and Texas red-conjugated EGF was remarkably decreased (Fig. 7A). Within control cells, Tfn and EGF were internalized and observed as dots. Only about 25 and 30% of SGIP1␣-overexpressing cells showed Tfn and EGF uptake, respectively, whereas about 90% of control cells did. We next examined by knockdown using siRNA whether SGIP1␣ is involved in endocytosis. Western blot analysis revealed that SGIP1␣ siRNA reduced the expression level of SGIP1␣ in N1E115 cells (Fig. 8A). In cells with reduced SGIP1␣ expression, the uptake of Alexa Fluor 594-conjugated Tfn was remarkably reduced when compared with that in cells treated with control siRNA by microscopic analysis (Fig. 8B). In contrast, the uptake of Texas red-conjugated EGF in cells treated with SGIP1␣ siRNA was hardly reduced when compared with that in cells treated with control siRNA. 6 Similar results were obtained with a quantitative assay that measures the internalization of 125 I-labeled Tfn and EGF (Fig. 8C). When EGFP-SGIP1␣ mutant, in which siRNA target sequence was silently mutated, was transiently expressed in SGIP1␣ knockdown cells, Tfn uptake was restored (Fig. 9, A and B). Taken together, these findings indicate that SGIP1␣ is involved in clathrin-mediated endocytosis and required for Tfn endocytosis but not for EGF endocytosis in NIE115 cells.

DISCUSSION
In this study, we identified SGIP1␣ as a tubulin-binding protein by tubulin blot overlay. It remains elusive whether its tubulin binding is significant. However, the phosphoinositide-binding ENTH domain in epsin has been shown to bind to tubulin and MTs (38). Moreover, the FCH domain, which constitutes the phospholipid-binding EFC domain together with the coiled-coil domain (11,12), was originally identified as an MTbinding domain (39). It is generally thought that MTs are not required for endocytosis, but the tubulin-and/or MT binding property of these phospholipidbinding proteins/domains may play a role in endocytosis.
The MP domain binds to phospholipids but shows no significant homology to other phospholipidbinding domains. However, the MP domain exhibits similar properties to BAR and EFC domains (7, 10 -13, 20). First, these domains do not show high specificity for phosphoinositides; MP, BAR, and EFC domains show relatively nonspecific binding to negatively charged phospholipids. Second, the SGIP1␣ MP domain as well as BAR and EFC domains forms an oligomer. 4 The oligomerization of BAR and EFC domains is thought to be important for membrane tubulation (7,11,20). The BAR domain is proposed to sense and generate membrane curvature for vesicle formation during endocytosis (10,13). Therefore, the MP domain, via its membrane-deforming property, may also act in vesicle formation during endocytosis. We have shown, by overexpression, knockdown, and localization studies, that SGIP1␣ is involved in clathrin-mediated endocytosis. SGIP1␣ overexpression as well as knockdown inhibits endocytosis, and this effect may be due to sequestration of its binding protein(s) from functional clathrin-mediated endocytosis machinery. This idea is consistent with the observation that Eps15 is recruited to membrane tubules formed by SGIP1␣ overexpression. Eps15 is an important adaptor protein of clathrin-mediated endocytic machinery (34 -37). Several lines of evidence suggest that Eps15 regulates clathrin coat assembly by interacting with its binding partners including the AP-2 complex. This role of Eps15 in clathrin coat assembly is likely to be closely related to its unique localization at the growing edges of clathrin-coated pits (40). SGIP1␣ may contribute to the unique localization of Eps15 for clathrin coat assembly during endocytosis. It has been shown that Eps15 and the AP-2 complex are required for Tfn endocytosis but not for EGF endocytosis (35,37,41,42). Consistently, we have shown by knockdown that SGIP1␣ is required for Tfn endocytosis but not for EGF endocytosis. This different requirement of SGIP1␣ in clathrin-mediated endocytosis may be due to the presence of redundant pathways, such as the Cbl-CIN85-endophilin pathway, in EGF endocytosis (43). It is most likely that SGIP1␣ plays an essential role in clathrin-mediated endocytosis not only by deforming the plasma membrane but also by interacting with Eps15 and the AP-2 complex. Further studies are necessary for our better understanding of the role of SGIP1␣ in endocytosis. FIGURE 8. Different Effects of SGIP1␣ knockdown on Tfn and EGF endocytosis. A, reduced expression level of SGIP1␣ by RNAi. N1E115 cells were transfected with control or SGIP1␣ siRNA. Cell lysate (40 g of protein) was subjected to SDS-PAGE (10% polyacrylamide gel) followed by immunoblot (IB) analysis using anti-SGIP1␣ pAb and anti-tubulin ␣ mAb. B, inhibition of Tfn endocytosis by SGIP1␣ RNAi. N1E115 cells transfected with control or SGIP1␣ siRNA were incubated with Alexa Fluor 594-Tfn for 10 min at 37°C. Samples were examined with a fluorescence microscope. Scale bar, 20 m. C, ineffectiveness of SGIP1␣ RNAi on EGF endocytosis. N1E115 cells transfected with control or SGIP1␣ siRNA were incubated with 125 I-labeled Tfn or EGF for the indicated periods of time at 37°C. Internalization ratio was expressed as the ratio between intracellular (internalized) and surface radioactive ligands. E, control RNAi; F, SGIP1␣ RNAi. Data are the means ϩ S.E. of three independent experiments.