Phosphatidylinositol-4-phosphate 5-Kinase-1 (cid:1) Is Essential for Epidermal Growth Factor Receptor-mediated Endocytosis*

Phosphatidylinositol-4,5-bisphosphate (PIP 2 ) is known to play an important role in signal transduction and membrane trafficking. We show that one enzyme responsible for PIP 2 production, phosphatidylinositol-4-phosphate 5-kinase type 1 (cid:1) (PIPK (cid:1) ), is essential for epidermal growth factor receptor (EGFR)-mediated endocytosis. Expression of murine PIPK (cid:1) in NR6 cells expressing EGFR strikingly increased receptor internalization. Moreover, the kinase was shown to form an immunoprecipitable complex with EGFR. Expression of either a truncated kinase or a kinase dead mutant inhibited EGFR endocytosis and also blocked the membrane recruitment of PIPK (cid:1) and both clathrin light chain and dynamin. Our results delineate a novel mechanism by which PIPK (cid:1) reg-ulates receptor-mediated endocytosis and receptor tyrosine kinase membrane traffic.

Phosphotidylinositols (PIs) 1 are known to play an essential role in membrane trafficking and signal transduction (1). PIs serve multiple functions via the recruitment of cytosolic proteins with PI phosphate (PIP) binding domains (2), the modification of the physical properties of the membranes in which they reside or by serving as a branch point in phosphoinositide metabolism as substrates for phosphoinositide-specific lipases (3). Recent work with the PI phosphatases indicates a central role for PIP 2 and PIP 3 in receptor-mediated endocytosis, although the precise nature of the PI kinases involved and their role have not been delineated (4,5).
Signaling by ligand-activated receptor tyrosine kinases (RTKs) triggers the rapid internalization and degradation of receptor-ligand complexes (6,7), which is the major mechanism for long term signal attenuation (8). Moreover, the intracellular trafficking of receptors may allow for differential signaling according to subcellular locale (9). Thus, receptor internalization impacts the biological responses to growth factor receptors and is a potential target for signal modulation. EGF receptor autophosphorylation is thought to result in a conformational change in the receptor revealing internalization motifs (10) that allow the receptor to interact, directly or indirectly, with endocytic proteins, such as AP-2, clathrin, and dynamin (10,11). Interestingly, recent work with colony-stimulating factor-1 receptor (CSF-1R) identified a possible connection between the endocytosis of activated RTKs and the type I phosphatidylinositol-4-phosphate 5-kinase (PIPK) (12). Indeed, PIP 2 , the product of PIPKs, has been shown to be involved in multiple steps of endocytosis, including the assembly of the clathrin coat, the regulation of adaptor proteins, and the production of endocytic vesicles via the regulation of dynamin (2).
Here, we demonstrate that overexpression of murine phosphatidylinositol-4-phosphate 5-kinase type I␤ (PIPK␤) or a truncated construct, PIPK␤-truncated, but not the mouse type 1␣ isoform (PIPK␣), causes dramatic alterations in the endocytic trafficking of the EGFR, including an effect on recruitment of dynamin and clathrin light chain. Our data indicate that PIPK␤ plays a key role in mediating the recruitment of the protein machinery required for clathrin-mediated endocytosis.
Transfection and Sindbis Virus Infections-NR6 monolayers were either transiently transfected with pcDNA3 vector using Fugene 6 (Roche) as described (12) and/or mock-infected or infected with the vector or recombinant viruses as described previously (13).
Receptor Internalization Studies-Mouse EGF (Life Technologies, Inc.) was iodinated with 125 I (PerkinElmer Life Sciences) using IODO-BEADS (Pierce), according to the manufacturer's protocol. The specific activities of labeled ligands were typically 150,000 cpm/ng (600 Ci/ mmol). Quiescent NR6-EGFR cells expressing different PIPK constructs were washed in binding buffer, incubated at 4°C for 3 h with 100 pM 125 I-EGF, and warmed up for different periods of time as indicated in each figure. Surface-bound and internalized ligands were discriminated essentially as described previously (15,16). Briefly, unbound ligand was removed by washing the monolayer three times with ice-cold buffer (20 mM Hepes, 130 mM NaCl, 5 mM KCl, 0.5 mM MgCl 2 , 1 mM CaCl 2 , 1 mg/ml polyvinylpyrrolidone, pH 7.4). Surface-bound ligand was then collected in ice-cold acid strip buffer (50 mM glycine HCl, 100 mM NaCl, 1 mg/ml polyvinylpyrrolidone, pH 3.0) for 2 min, and internalized ligand was released in 1 N NaOH overnight at room temperature. Nonspecific binding (Ͻ2%) was assessed in the presence of 200 nM unlabeled human EGF (Sigma) and subtracted from the total. To quantitate receptor down-regulation, NR6-EGFR cells transfected as described above were incubated for various times with EGF (100 nM) at 37°C and then rinsed with cold medium. Surface EGF was then removed by mild acid/salt treatment as described above. Remaining cell surface binding sites were then quantified by incubating the cells with 100 pM 125 I-EGF at 4°C for 3 h (15).
Confocal Microscopy-NR6-EGFR cells, grown on glass coverslips, were examined by confocal microscopy in the absence or presence of 100 nM EGF in (␣-minimal essential medium containing 0.1% bovine serum albumin and 25 mM Hepes, pH 7.2) as described previously (15). Confocal microscopy was carried out on a Bio-Rad MRC1024 confocal microscope using a 63 ϫ, 1.4 numerical aperture bright-field objective and fluorescein filter sets.

RESULTS AND DISCUSSION
Type 1 PIPKs consist of three isoforms, ␣, ␤, and ␥ (17,18). Earlier work has shown that expression of PIPK-truncated, an enzymatically inactive fragment, alters the endocytosis of the CSF-1R (12). To more fully delineate the role of PIPK␤ in receptor-mediated endocytosis, we examined the effects of expressing PIPK␤ and PIPK␤-truncated on the internalization of 125 I-EGF. PIPK constructs (␤ and ␣), cloned into a Sindbis virus expression vector (13), were transiently expressed in NR6-EGFR cells (8,14), and EGFR endocytosis was examined by following the uptake of 125 I-EGF as described in the legend to Fig. 1. In cells expressing PIPK␤, endocytosis of 125 I-EGF was significantly enhanced, whereas in cells overexpressing PIPK␤-truncated, endocytosis of 125 I-EGF was potently inhibited. Although intracellular accumulation of EGF in cells expressing either PIPK␤ or PIPK␤-truncated approached a steady state at 15-20 min, cells expressing the wild type kinase accumulated about 5-fold more EGF than cells expressing the truncated construct. In addition, overexpression of a "kinase dead" mutant (PIPK␤-dead) also significantly inhibited 125 I-EGF uptake (Fig. 1).
Addition of EGF to cells results in the internalization and degradation of EGF (6 -8), and therefore we examined the effect of PIPK␤ and the two inhibitory mutants on down-regulation of EGFR following exposure to ligand. NR6-EGFR cells were infected with Sindbis virus, as described above, encoding PIPK␤ or the inhibitory constructs, and the rate of EGFR down-regulation was examined, as described under "Experimental Procedures." In control cells, 64% of surface EGFR was lost following exposure to EGF within 3 h. In cells overexpressing PIPK␤, 82% of surface EGFR was lost, whereas the majority of surface EGFR remained at the cell surface in cells expressing either PIPK␤-truncated (24% lost) or PIPK␤-dead (11% lost). Failure to internalize should lead to higher levels of 125 I-EGF binding in cells expressing either PIPK␤-truncated or PIPK␤-dead. Indeed, the expression of the two mutants, but not the wild type kinase, substantially increased EGF cell surface binding in transfected cells (data not shown).
We next set out to determine whether PIPK␤-truncated blocked EGFR-mediated endocytosis by preventing the recruitment of the cytosolic wild type kinase to the membrane (1). A cell fractionation study (15), using NR6-EGFR cells expressing the PIPK␤ or PIPK␤-truncated, revealed that PIPK␤ is localized to both membranes and cytosol, whereas PIPK␤-truncated is localized almost exclusively to the cytosolic fraction. Addition of EGF prior to fractionation resulted in a shift in the distribution of the cytosolic wild type kinase to the membrane, whereas PIPK␤-truncated cytosolic distribution was unaffected by EGF (data not shown).
The fact that PIPK␤-truncated is mostly cytosolic and enzymatically inactive (12) raises the possibility that PIPK␤-truncated may interact with the wild type enzyme such that targeting to the plasma membrane is prevented.
To delineate the role of PIPK␤-truncated in inhibiting endocytosis, we examined the effect of expressing PIPK␤-truncated on the targeting of PIPK␤ to the plasma membrane in living cells. A transfection/infection protocol, as described in the legend to Fig. 2A, was used to first express PIPK␤ and then either GST-PIPK␤-truncated or GFP-PIPK␣-truncated. Data in Fig.  2A, panel II, show that GST-PIPK␤-truncated is localized totally to the cytoplasmic compartment. In cells expressing GST-PIPK␤-truncated, PIPK␤ is also seen predominantly in the cytoplasm (panel I), suggesting that the truncated construct is blocking the targeting of PIPK␤ to the plasma membrane. In control cells (panels I and IV, arrows), PIPK␤ is localized to the plasma membrane. A merged image (panel III) shows the colocalization of PIPK␤ and GST-PIPK␤-truncated in the cytoplasmic compartment. As an additional control, we used an N-terminal truncation of PIPK␣, which is truncated from the same point in the conserved core kinase homology domain (17,18) as PIPK␤-truncated. The PIPK␣ isoform has been shown to play a role in membrane ruffling (19 -22) and has no effect on EGFR uptake. 2 Surprisingly, GFP-PIPK␣-truncated localized to the nucleus as shown in panel V, and the merged image (panel VI) shows the distinct localization of PIPK␤ and GFP-PIPK␣-truncated, indicating that the truncated PIPK␣ isoform does not block PIPK␤ plasma membrane targeting.
To determine whether PIPK␤ and PIPK␤-truncated are physically associated, co-immunoprecipitation experiments were carried out using NR6-EGFR cells expressing both GFP-PIPK (␣ or ␤) and GST-PIPK␤-truncated. A second set of control experiments were carried out with cells co-expressing HA-  (8,14) were infected with Sindbis virus alone (ࡗ), virus encoding PIPK␤ (f), PIPK␤-dead (q), or PIPK␤-truncated (OE) as described previously (15). The levels of expression were at least 10-fold over background as detected by Western blotting. At 6-h post-infection, cells were incubated with 100 pM 125 I-EGF at 37°C for the indicated times, and the amount of ligand internalized was quantified as described previously (15). Error bars indicate the S.E. of the mean of three independent assays.

Phosphatidylinositol-4-phosphate 5-Kinase-1␤ and Endocytosis
PIPK␤ and GFP-PIPK␣-truncated. Fig. 2B, panels I and II, show Western blots of total lysates indicating the approximate levels of cellular expression of the GFP, GST, and HA constructs. Cell lysates were prepared and immunoprecipitated with anti-GFP antibodies (Fig. 2B, panels III and IV) and following SDS-PAGE, Western blotting with anti-GST antibodies (panel III, upper) indicated that immunoprecipitation of GFP-PIPK␤ results in the co-immunoprecipitation of GST-PIPK␤-truncated. On the contrary, GFP-PIPK␣ failed to coimmunoprecipitate GST-PIPK␤-truncated. As a control (panel IV), immunoprecipitation of GFP-PIPK␣-truncated failed to co-immunoprecipitate HA-PIPK␤. We speculate that the formation of a complex between the wild type enzyme and truncated construct may interfere with the correct targeting of the kinase to the plasma membrane. This speculation is further supported by gel exclusion chromatography analysis, which suggests that the PIPK␤ exists as large multimeric complexes. 3 Previous studies indicate that one or more phosphoinositide kinase activities partially co-purify with EGFR (16) and that EGF stimulates phosphoinositide kinase activity in several cell types (23). To determine whether PIPK␤ and PIPK␤-truncated are physically associated with EGFR, cells expressing PIPK␤, PIPK␤-truncated, or both constructs were stimulated with EGF, and immunoprecipitation studies were carried out as described in the legend to Fig. 2C. Fig. 2C, panel I, shows EGFR detected with both anti-EGFR and anti-phosphotyrosine antibodies, and Fig. 2C, panel II, shows a co-immunoprecipitation of PIPK␤ with EGFR detected with a PIPK␤-specific antibody. The data also indicate that ligand activation of the receptor results in a significant increase in EGFR-PIPK␤ association. On the contrary, EGFR and PIPK␤-truncated failed to co-immunoprecipitate. Moreover, co-expression of PIPK␤-3 N. Davis, personal communication.

FIG. 2. PIPK␤-truncated, but not PIPK␣-truncated, inhibits
125 I-EGF uptake by blocking the targeting of PIPK␤ to the plasma membrane. A, confocal images of PIPK␤, PIPK␤-truncated and PIPK␣-truncated. NR6-EGFR cells were transiently transfected with PIPK␤ using a pcDNA3 expression vector for 18 h. After transfection, the cells were infected with Sindbis virus encoding GST-PIPK␤truncated or GFP-PIPK␣-truncated as described in the legend to Fig. 1. After infection, the cells were incubated with 100 nM EGF for 10 min at 37°C, washed, fixed, and processed for confocal microscopy as described previously (15). Proteins were detected with antibodies specific for PIPK␤ (12) (panels I and IV) or polyclonal GST (Sigma) (panel II), and they were visualized with the Alexa antibodies (Molecular Probes). GFP-PIPK␣-truncated was visualized via the intrinsic fluorescence of GFP (panel V). The arrows indicate uninfected cells. Individual confocal sections are shown. Shown is one experiment representative of three independently performed. Bar, 10 m. B, co-immunoprecipitation of GFP-PIPK␤ and GST-PIPK␤-truncated. NR6-EGFR cells were transiently transfected with GFP-PIPK␤ or GFP-PIPK␣ using the pcDNA3 expression system; transfected cells were then infected with Sindbis virus encoding GST-PIPK␤-truncated. In a second experiment, pcDNA3 was used to express HA-PIPK␤ followed by expression of GFP-PIPK␣truncated using the Sindbis expression vector. Transfected/infected cells were incubated with 100 nM EGF for 10 min at 37°C, and after incubation, the cells were washed, lysed, and immunoprecipitated as described previously (12). Equal aliquots of cell lysates were resolved by SDS-PAGE, transferred, and visualized by GFP-(Zymed Laboratories Inc.), GST-(Sigma), or HA-(CLONTECH) specific antibodies as shown in Panels I and II. Panel III, GFP-PIPK (␣ and ␤) were immunoprecipitated using a polyclonal anti-GFP antibody (CLONTECH), and following SDS-PAGE, GST-PIPK␤-truncated was detected by immunoblotting with monoclonal anti-GST antibody (panel III, upper). The immunoprecipitated GFP-PIPK␣ and -␤ were detected with the monoclonal anti-GFP antibody (Zymed Laboratories Inc.) (panel III, lower). Panel IV, HA-PIPK␤ and GFP-PIPK␣-truncated were detected by immunoblotting with polyclonal anti-HA (Sigma) and anti-GFP antibodies, respectively (panel IV). The immunoprecipitations were performed with polyclonal anti-GFP antibody followed by immunoblotting with anti-HA antibody and subsequent re-probing with monoclonal anti-GFP antibody. Shown is one experiment representative of three independently performed (12). C, PIPK␤ co-immunoprecipitates with the EGFR, which is blocked by overexpression of PIPK␤-truncated. NR6-EGFR cells were infected with virus encoding either PIPK␤, PIPK␤-truncated, or both the wild type and truncated constructs as described previously (15). After 1.5 h, the cells were starved for an additional 4 h and then incubated in the presence or absence of 17 nM EGF for 10 min at 37°C. The cells were lysed, and 150 g of total protein was used for immunoprecipitations with an antibody specific for the N terminus of EGFR (AB5, Oncogene). Panel I, total EGFR in the assay (upper) and tyrosinephosphorylated EGFR (lower) was resolved by SDS-PAGE, Western blotted, and probed with antibodies specific for EGFR (Santa Cruz) or phosphotyrosine (Transduction Laboratories), respectively. Panel II, PIPK␤ and PIPK␤-truncated associated with EGFR were resolved by SDS-PAGE and Western blotted with an antibody specific for PIPK␤ (Upstate Biotechnology). The figures represent relative units (RU) of PIPK␤ (upper) or the truncated mutant (lower) associated with the EGFR, with 1 relative unit being the amount of PIPK␤ associated with the receptor in the absence of EGF stimulation. A kinase assay (12) was carried out on equivalent immunoprecipitates, and the relative units for the kinase activities were 1, 1.3, 0.14, and 0.22 for the kinase in lanes  1, 2, 5, and 6, respectively. The gel shown is one experiment representative of three independently performed.
Phosphatidylinositol-4-phosphate 5-Kinase-1␤ and Endocytosis truncated with the wild type kinase blocked the co-immunoprecipitation of an EGFR⅐PIPK␤ complex. PIPK enzymatic activity, which is regulated by phosphorylation (24), was quantitated in the immunoprecipitates and, as indicated in the legend of Fig. 2C, corresponds with the quantitation of the Western blots (i.e. the lipid kinase which is recruited to EGFR is active). Another factor that may influence the recruitment of PIPK␤ is the lipid binding site described by Anderson and colleagues (1), who have suggested that lipid kinase recruitment may include both a lipid binding site and specific protein targeting sites. Our data suggest that the expression of the PIPK␤-truncated inhibits endocytosis of 125 I-EGF by blocking the targeting of the PIPK␤ to EGFR at the plasma membrane. Interestingly, Tolias et al. (19) have shown that the thrombin activation of PIPK␣ and ␤ is regulated by Rac, although only PIPK␣ is involved in Rac-dependent actin assembly. These new data now delineate different functions associated with different isoforms of type 1 PIPKs.
Our data indicate a key role for PIPK␤ in the early steps of internalization. PIP 2 , the product of PIPK␤, has been shown to bind specific domains (25), found in many of the proteins required for endocytosis and signaling. Dynamin and AP2, for example, bind to PIP 2 and regulate EGF receptor-mediated endocytosis (26). To investigate whether the expression of PIPK␤, PIPK␤dead, or PIPK␤-truncated affects the localization of proteins controlling clathrin-coated vesicle formation, cells were transfected with Sindbis virus encoding PIPK␤ or the two inhibitory mutants, and the localization of clathrin light chain (CLC) or dynamin, following EGF stimulation, was analyzed by confocal microscopy. In Fig. 3A, we show that the expression of PIPK␤truncated or PIPK␤-dead blocks the targeting of dynamin to the plasma membrane. GFP-PIPK␤ (panel I) is found on the plasma membrane as is dynamin (panel II), where they co-localized (panel III). In addition, dynamin is also localized to intracellular structures. GFP-PIPK␤-truncated localization is cytoplasmic (panel IV),and cells expressing PIPK␤-truncated are unable to recruit dynamin to the plasma membrane (panel V); however, some dynamin in the Golgi region remained (panel V, arrowhead). GFP-PIPK␤-dead localizes to the plasma membrane (panel VII), but dynamin recruitment is still blocked (panel VIII), indicating the requirement of an active type 1 kinase for dynamin recruitment in response to EGF.
Recruitment of clathrin light chain is also an index of activated endocytosis, and we examined the effect PIPK␤, PIPK␤truncated, or PIPK␤-dead on the localization of CLC. Cells stably expressing GFP-clathrin light chain (GFP-CLC) were infected with Sindbis encoding the PIPK␤ constructs, stimulated with EGF, and examined by confocal microscopy as described above. In Fig. 3B, panel I, GFP-CLC is localized to punctate spots throughout the cytoplasm. Panel II shows the plasma membrane localization of PIPK␤, and the partial localization of the two proteins in the merged image is seen in panel III. In cells expressing PIPK␤-truncated (panel IV), punctate localization of GFP-CLC is substantially lost, compared with the neighboring untransfected cell (panel IV, arrows), and PIPK␤-truncated is seen localized in the cytoplasm (panel V). Panel VI shows the substantially reduced co-localization of GFP-CLC and PIPK␤-truncated. Last, we examined PIPK␤dead in cells expressing GFP-CLC. As with the truncated construct, the recruitment of CLC in response to EGF was lost Phosphatidylinositol-4-phosphate 5-Kinase-1␤ and Endocytosis (panel VII). PIPK␤-dead is localized to the plasma membrane (panel VIII), but the absence of co-localization is shown in the merged image (panel IX). These experiments show that recruitment of an active PIPK␤ to the plasma membrane in response to EGF is required for the subsequent recruitment of dynamin and clathrin light chain. These observations explain the effect of the truncated kinase on receptor-mediated endocytosis.
Our data indicate that PIPK␤ plays a central role in receptor-mediated endocytosis via its recruitment to the EGFR and via the production of its product, PIP 2 , which in turn affects the binding of specific factors, required for endocytosis. Several groups (19 -22) have reported that activation of membrane ruffling is tightly coupled to PIPK␣. Thus, it may well be that the various type 1 PIPK isoforms have both independent and possibly sequential functions in signaling leading to endocytosis and ruffling. Expression of PIPK␤-truncated blocks the recruitment of PIPK␤, but additionally, expression of the truncated kinase or the kinase dead mutant appears to block coated vesicle formation by sequestering or preventing recruitment and/or assembly of additional limiting components essential for coated vesicle budding. It is interesting to note that endocytic proteins such as epsin and AP180, which contain ENTH domains that bind PIP 2 , have also been implicated in the endocytosis of the EGFR (27,28). An additional factor in the regulation of PIPK activity is phospholipase D (PLD). PLD has a required PH domain and is reported to be recruited via the action of PIPK␣ (29). The product of PLD, phosphatidic acid, is known to be an activator of PIPK activity (30), creating a possible positive feedback scenario for the amplified activation of PIPK␤. Studies in which synaptojanin, a PI(4,5)P 2 5-phosphatase, has been deleted confirmed a central role for PIP 2 in endocytosis (31). Our studies now directly demonstrate a central and fundamental role for PIPK␤ in receptor-mediated endocytosis. However, many additional factors need to be identified, including those factors that are responsible for the recruitment, activation, and deactivation of the kinase.