The serine͞threonine kinase cyclin G-associated kinase regulates epidermal growth factor receptor signaling

Lei Zhang, Ole Gjoerup, and Thomas M. Roberts*†

Department of Cancer Biology, Dana–Farber Cancer Institute, and Department of Pathology, Harvard Medical School, Boston, MA 02115

Communicated by R. John Collier, Harvard Medical School, Boston, MA, May 5, 2004 (received for review March 18, 2004) Cyclin G-associated kinase (GAK) is a serine͞threonine kinase that activated by autophosphorylation, internalized, at least in part features high homology outside its kinase domain with auxilin. through clathrin-mediated endocytosis, and sorted to early, and Like auxilin, GAK has been shown to be a cofactor for uncoating subsequently, late endosomes (13–16). Internalized receptors then clathrin vesicles in vitro. We investigated epidermal growth factor are delivered to the lysosome and degraded or recycled to the cell (EGF) receptor-mediated signaling in cells, in which GAK is down- surface. As the receptor transits this three-dimensional path regulated by small hairpin RNAs. Here, we report that down- through the cell’s interior, various EGF-mediated signaling mole- regulation of GAK by small hairpin RNA has two pronounced cules are activated including extracellular signal-regulated kinases effects on EGF receptor signaling: (i) the levels of receptor expres- 1͞2 (ERK1͞2) via the Ras͞raf pathway (11, 17–21), and Akt via the sion and tyrosine kinase activity go up by >50-fold; and (ii) the phosphatidylinositol 3-kinase pathway (22–24). Previous studies spectrum of downstream signaling is significantly changed. One have indicated that blocking receptor trafficking at a particular very obvious result is a large increase in the levels of activated point can have significant effects on receptor signaling. For in- extracellular signal-regulated kinase 5 and Akt. These two effects stance, studies by Vieira et al. (25) showed that expression of a of GAK down-regulation result from, at least in part, alterations in dominant negative allele of dynamin, a GTPase involved in clathrin- receptor trafficking, the most striking of which is the persistence of coated vesicle fission, inhibits endocytosis of EGFR, and con- EGF receptor in altered cellular compartment along with activated sequently, ERK1͞2 activation is attenuated. However, the roles extracellular signal-regulated kinase 5. The alterations resulting of trafficking events subsequent to formation of clathrin-coated from GAK down-regulation can have distinctive biological conse- vesicles in the regulation of EGFR signaling remain relatively quences: In CV1P cells, down-regulation of GAK results in out- unknown. growth of cells in soft agar, raising the possibility that loss of GAK In the present report, we examine the effects of down-regulating function may promote tumorigenesis. GAK expression upon EGFR signaling. By using small hairpin RNA to reduce GAK levels by Ͼ90%, we find that both the levels yclin G-associated kinase (GAK), also known as auxilin II (1, of EGFR in the cells and the signaling downstream from the C2), is a 160-kDa protein, which is highly homologous to receptor are strikingly altered. At least some of these dramatic auxilin I, with the notable exception of a kinase domain found changes are likely due to changes in receptor trafficking. At the at the N terminus of GAK, and totally absent in auxilin I. Unlike biological level, GAK down-regulation can have significant effects auxilin I, which has been reported to be expressed only in on cell growth. These results suggest that GAK plays a pivotal role neurons, GAK is ubiquitously expressed. In vitro GAK, like in regulating not only receptor trafficking but also receptor signal- auxilin I, is known to assist heat shock cognate 70 (Hsc70) in ing and function. uncoating clathrin-coated vesicles (3, 4). Thus, GAK is thought to function in trafficking just after the step regulated by dynamin. Materials and Methods GAK also represents a major portion of the kinase activity in DNA-Based RNA Interference (RNAi) Constructs and Cell Culture. The clathrin-coated vesicles (5). GAK has been reported to phos- retroviral pSuper vector used in the study was the same as that phorylate the ␮2 subunit of adaptor protein (AP)2, as has described in Brummelkamp et al. (26). Selected sequences were AAK1, the protein kinase most closely related to GAK in its submitted to BLAST searches against the human genome sequence catalytic domain (4–9). In the case of AAK1, phosphorylation of to ensure only GAK mRNA is targeted. For generation of stable ␮2 has been shown to have functional consequences in promot- GAK knockdown cells, HeLa (a human cervix epithelioid carci- ing the recruitment of AP2 to internalization motifs found on noma cell line, American Type Culture Collection) or CV1P (an membrane-bound receptors (9). Overexpression of either GAK African green monkey kidney cell line) cells were infected with the or AAK1 is sufficient to impair internalization of the transferrin retroviral pSuper vectors containing either one of two hairpin receptor (4, 7, 10). Thus, GAK may function in both the ultimate constructs capable of generating 19-nt duplex RNAi oligonucleo- disassembly of clathrin-coated vesicles through its auxilin ho- tides corresponding to human GAK sequences starting at the 145th mology domains and in their formation through its kinase base in the coding sequence (denoted 145) or the 525th base domain although there is no direct evidence that GAK can mimic (denoted 525). Cells were cultured in DMEM with 10% FCS, and ͞ AAK1 in regulating vesicle formation (6–9). Because so many 1% penicillin streptomycin in 10% CO2 humidified incubator. kinases play key regulatory roles, it occurred to us that GAK When necessary, 1.5 ␮g/ml puromycin was included for selection. might represent an important regulatory point in receptor trafficking. We were particularly interested to see whether GAK might regulate receptor tyrosine kinases, such as the EGF Abbreviations: GAK, cyclin G-associated kinase; EGF, epidermal growth factor; EGFR, EGF receptor; ERK, extracellular signal-regulated kinase; Hsc70, heat shock cognate 70; RNAi, receptor (EGFR), because this class of receptor is key to normal RNA interference; DAPI, 4Ј,6-diamidino-2-phenylindole; AP, adaptor protein. cell function and transformation, and because the signaling and *To whom correspondence should be addressed. E-mail: thomas࿝[email protected]. trafficking of receptor tyrosine kinases is well characterized. edu. EGFR-mediated signaling results in cell proliferation or differ- † In compliance with Harvard Medical School guidelines on possible conflict of interest, we entiation. Elevated levels or inappropriate activation of the EGFR disclose that T.M.R. has consulting relationships with Upstate Biotechnology and Novartis is associated with increased propensity for cell transformation and Pharmaceuticals. tumor formation (11, 12). Upon ligand stimulation, EGFRs are © 2004 by The National Academy of Sciences of the USA

10296–10301 ͉ PNAS ͉ July 13, 2004 ͉ vol. 101 ͉ no. 28 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0403175101 Downloaded by guest on September 27, 2021 A431 cells (a human epidermoid carcinoma cell line, American chemiluminescence (Amersham Biosciences). The signals were Type Culture Collection) are also used to compare the expression visualized by exposure to Kodak X-Omat blue XB-1 film. level of EGFR. EGF Internalization Assay and Indirect Immunofluorescence. Cells Abs. The following Abs were used: mAb clone 1C2 (anti-GAK, stably expressing a DNA-based RNAi construct targeting GAK Medical & Biological Laboratories); mAb clone MA1 065 (anti- were seeded on eight-well Labtek chamber slides. Twenty-four clathrin heavy chain, Affinity BioReagents, Golden, CO); mAb hours later, cells were serum-starved in DMEM without serum clone14 (anti-EEA1, BD Transduction Laboratories); 4G10 (an- overnight. The next day, cells were incubated with 0.5 ␮g/ml tiphosphotyrosine, Upstate Biotechnology, Lake Placid, NY); mAb tetramethylrhodamine-conjugated EGF (Molecular Probes) for 1 h EGFR (528) Ab (Santa Cruz Biotechnology); polyclonal EGFR on ice and then for 40 min at 37°C. Cells were fixed in 4% (1005) Ab (Santa Cruz Biotechnology); polyclonal phospho-p42͞44 paraformaldehyde, washed with PBS, permeabilized with 0.5% mitogen-activated protein kinase (Thr-202͞Tyr-204) Ab (Cell Sig- Triton X-100 in PBS for 10 min, and blocked with 10% goat serum naling Technology, Beverly, MA); polyclonal p42͞44 Ab (Cell for 1 h. Primary Abs were applied for 1 h and cells were then washed Signaling Technology); polyclonal phospho-ERK5 (Thr-218͞Tyr- with PBS. Cy2-conjugated secondary Abs (The Jackson Labora- 220) Ab (Cell Signaling Technology); polyclonal ERK5 Ab tory) were applied and incubated for 1 h. The same cells were then (Sigma); polyclonal phospho-Akt (Ser-473) or (Thr-308) Abs (Cell counterstained with 4Ј,6-diamidino-2-phenylindole (DAPI; 1 ␮g/ Signaling Technology); and polyclonal Akt Ab (Cell Signaling ml, Sigma) and mounted with Fisher Scientific mounting media. Technology). Cells were visualized with a Zeiss confocal microscope LSM510META͞NLO at ϫ63 magnification and images were cap- Western Blot Analysis and Immunoprecipitation. Cells were seeded in tured by Zeiss confocal microscope software Version 3.2. Ϸ 100-mm dishes and grown to 75% confluence in a 10% CO2 humidified incubator. The next day, cells were serum-starved with BrdUrd Incorporation Assay. Cells grown on chamber slides were DMEM without serum for 24 h wherever necessary. Cells were serum-starved overnight and labeled with BrdUrd for 15 h. Cells incubated with 50 ng/ml EGF (Upstate Biotechnology) for indi- were fixed and permeabilized as described above. Cells were then cated times. Cells were then harvested and lysed with Nonidet P-40 incubated with a mAb (Amersham Biosciences) against BrdUrd for lysis buffer (50 mM Tris⅐HCl,pH7.5͞150 mM NaCl͞1% Nonidet 1 h. Cy2-conjugated secondary Abs (The Jackson Laboratory) were P-40͞5 ␮g of leupeptin per ml͞5 ␮g of pepstatin per ml͞0.5 mM applied and were incubated for 1 h. Cells were washed and phenylmethylsulfonyl fluoride͞1 mM sodium fluoride͞100 ␮M counterstained with DAPI (1 ␮g/ml, Sigma) and mounted with sodium vanadate͞10 mM ␤-glycerol phosphate) for 10 min on ice. Fisher Scientific mounting media. BrdUrd-positive cells were Extracts were cleared of cell debris by centrifugation at 10,000 ϫ g counted among DAPI-stained cells in at least 10 randomly selected for 10 min in a microcentrifuge at 4°C. Equal amounts of protein fields on each slide under a Nikon Eclipse TE300 fluorescence lysate from each dish of cell lines was immunoprecipitated by mAb microscope. EGFR (528) Ab (Santa Cruz Biotechnology) for 2 h and then 45 ␮l of protein A Sepharose beads for 45 min. The beads were washed Soft Agar Assay. A total of 5 ϫ 104 CV1P cells (vector control three times with Nonidet P-40 lysis buffer. For Western blotting, the or 525) as single-cell suspensions in 2 ml of 0.27% Bacto agar immunoprecipitates or protein lysates were boiled with one third were overlayed on top of 0.6% agar in 60-mm culture dish. In volume of 3ϫ SDS buffer (150 mM Tris⅐HCl,pH6.8͞300 mM each experiment, at least two or more duplicate plates were DTT͞6% SDS͞0.3% Bromophenol blue͞30% glycerol) and sepa- prepared for each cell line. DMEM supplemented with 10% rated on a 7.5% or 10% SDS polyacrylamide gel wherever appli- FCS and 1% streptomycin͞penicillin was used as media for cable, followed by transfer to a poly(vinylidene difluoride) mem- agar. Colonies were counted after incubation at 37°Cina10% brane. The blots were then blocked with 5% dry milk or 2% gelatin CO2-humidified incubator for 21 days under a Nikon SMZ (for 4G10) and incubated with various Abs in appropriate dilutions microscope (magnification: ϫ7). as follows: EGFR Ab, phopho-ERK5, ERK5, phopho-p42͞44, and p42͞44; phospho-Akt, total Akt, and antiphosphotyrosine Ab Results (4G10) overnight or 45 min (4G10), washed with Tris-buffered Down-Regulation of GAK by Small Hairpin RNA Dramatically Enhances saline-Tween (150 mM NaCl͞10 mM Tris⅐HCl͞0.2% Tween 20), EGFR Expression. We used small hairpin RNA techniques to down- then incubated with horseradish peroxidase-conjugated anti-rabbit regulate endogenous GAK in mammalian cells. To compare our or anti-mouse secondary Ab (Amersham Biosciences) for 30 min, findings with earlier work using dominant-negative dynamin con- washed again, and developed by enhanced chemiluminescence structs (25), we focused our studies on EGFR signaling in HeLa (Amersham Biosciences). The signals were visualized by exposure cells. We also down-regulated GAK in CV1P cells to assess the to Kodak X-Omat blue XB-1 film. biological consequences of GAK down-regulation. Here, we report data obtained from HeLa cells in which GAK expression was stably Cell Surface-Associated EGFR Assay. Cells were seeded in 60-mm down-regulated through either of two different RNAi constructs. Ϸ dishes and grown to 75% confluence in a 10% CO2 humidified We have obtained similar signaling results by using oligonucleotide- incubator. The next day, cells were serum-starved with DMEM based methodologies to transiently down-regulate GAK in HeLa, without serum for 24 h. Cells were stimulated with 50 ng/ml EGF COS7, and CVIP cells (data not shown). For generation of stable (Upstate Biotechnology) for the indicated times. Biotinylation of GAK knockdown in HeLa and CV1P cells, cells were infected with intact cells were carried out as described by Levy-Toledano et al. retroviral pSuper vectors containing either one of two hairpin

(27). Briefly, cells were rinsed with cold PBS twice and incubated constructs capable of generating 19-nt duplex RNAi oligonucleo- CELL BIOLOGY on ice for 2 min. Cells were then incubated with 3 ml of biotinylation tides corresponding to human GAK sequences starting at the 145th buffer [0.4 mg/ml NHS-LC-biotin (Pierce) in 50 mM sodium base in the coding sequence (denoted 145) or the 525th base ͞ ͞ phosphate, pH 8.5 110 mM NaCl 0.1% NaN3]onicefor20min (denoted 525, ref. 26). As shown in Fig. 1A, GAK protein levels in with occasional mixing. Cells were then washed three times with the resulting HeLa knockdown cell lines were reduced at least 90% ice-cold PBS buffer plus 0.1% NaN3 (pH 7.4). Cells lysates were compared with GAK levels in vector control cells (The 525 con- immunoprecipitated with mAb EGFR Ab and subsequently, West- struct was slightly better than the 145 construct). In CV1P cells, ern blotting was performed as described above. The blot was then however, only the 525 cell line, but not the 145 cell line, shows a incubated with streptavidin-horseradish peroxidase (Amersham significant reduction in GAK expression (Fig. 1A and data not Biosciences) for 30 min, and washed and developed by enhanced shown). We tried to test the possibility of reversing the phenotypes

Zhang et al. PNAS ͉ July 13, 2004 ͉ vol. 101 ͉ no. 28 ͉ 10297 Downloaded by guest on September 27, 2021 Fig. 2. In GAK knockdown cells, EGFR remains surface-associated for ex- tended periods after EGF stimulation and displays enhanced kinase activity. (A) The level of surface-associated EGFR in GAK knockdown cell remains high for a prolonged time. Cells were serum-starved overnight and incubated with Fig. 1. Down-regulation of GAK dramatically increases EGFR expression in 50 ng/ml EGF for indicated times. Cells were surface-labeled with biotin at the HeLa cells. (A) GAK expression was stably down-regulated in HeLa and CV1P indicated times, then the biotinylated receptors were isolated by means of cells. Cells were infected with retroviral pSuper vectors encoding either one of streptavidin affinity and visualized by Western blotting with an EGFR Ab. (B) two hairpin constructs capable of generating 19-nt duplex RNAi oligonucle- EGFR in GAK knockdown cells displays high tyrosine kinase activity in response otides corresponding to human GAK sequences, starting at the 145th base in to EGF stimulation. Cells were serum-starved overnight and incubated with 50 the coding sequence (denoted 145) or the 525th base (denoted 525). Western ng/ml EGF for indicated times. Total EGFR in cells was immunoprecipitated blotting of total cell lysates was carried out by using a mAb against GAK. As with EGFR Ab and blotted with an Ab against phosphorylated tyrosine (4G10). a positive control, a lysate from cells transiently overexpressing GAK was included. Levels of Hsc70 were used as loading controls. (B) EGFR expression was dramatically enhanced in GAK knockdown cells. Cells were serum-starved overnight and stimulated with 50 ng/ml EGF for the indicated times. EGFR was isolated by means of streptavidin affinity and visualized by Western detected by Western blotting using an Ab against it. (C) The level of EGFR blotting with an anti-EGFR antiserum. Fig. 2A shows that levels of expression in GAK knockdown cells was Ϸ75% of that of A431 cells. Cells were surface accessible EGFR in knockdown cells are approximately serum-starved overnight and EGFR expression was detected as described in B. Ͼ50-fold higher than the levels seen in controls before EGF Levels of Hsc70 were used as loading controls. stimulation. This finding suggests that most of the ‘‘excess’’ recep- tors in the knockdown cells are located normally, a fact which was later born out by immunofluorescence studies (see below). Surface- observed upon GAK knockdown through GAK reexpression; associated receptors displayed a significant decrease by 30 min after however, expression of the wild-type human GAK by using avail- EGF stimulation in knockdown cells. By 60 min after EGF stim- able promoters resulted in GAK accumulation in apparent vesicular ulation, the number of surface-associated EGFRs in knockdown aggregates (4). cells had decreased dramatically, indicating that most, but not all of We first characterized the gross levels of EGFR expression in the the receptors, had been internalized. HeLa knockdown cells. To our surprise, in GAK knockdown cells To examine the activation state of the EGFR in GAK knock- under serum starvation, EGFR levels were significantly increased (Ͼ50-fold) compared with the vector control cells (Fig. 1B). To down cells, the level of tyrosine phosphorylation of immunopre- better characterize the level of EGFR overexpression in GAK cipitated EGFR from knockdown cells was examined after EGF knockdown cell lines, we compared the EGFR numbers in knock- stimulation. Fig. 2B shows that the EGFR in knockdown cells was down cells with those in the tumor cell line A431, known to express highly phosphorylated on tyrosines upon EGF stimulation. There high levels of EGFR. Fig. 1C shows that the level of EGFR seemed to be at least 50-fold higher levels of activated receptors expression in knockdown cells is about three fourths of that seen in present compared with control cells, which indicated that most of A431 cells [reported to be 2 million EGFR molecules per cell (28)]. the additional receptors were functional. Receptor activation was We also measured the levels of EGFR expression as a function of sustained for at least 60 min. Notably, there is a substantial level of time after EGF stimulation. Normally, after cells are treated with activated EGFR in the knockdown cells even after serum starva- EGF, the EGFR is internalized and degraded rapidly in lysosomes tion, indicating that a subset of the EGFR is constitutively active. (29). However, the elevated level of EGFR in knockdown cells Thus, down-regulation of GAK elicits dramatic effects on the level persisted for hours after EGF stimulation (Fig. 1B and data not and activation of the EGFR. shown). This lack of down-regulation seen upon GAK knockdown may explain the enhanced levels of EGFR expression. A pulse– Down-Regulation of GAK Differentially Alters Signaling Downstream chase experiment showed that the half-life of the EGF stimulated of the EGFR. To investigate whether EGFR signaling is also altered EGFR in knockdown cells is longer than 4 h (versus Ͻ1 h in control in GAK knockdown cells, we analyzed several known EGFR- cells; data not shown). mediated signaling pathways. Previous work by Vieira et al. (25) had indicated that overexpression of a dominant-negative dynamin Down-Regulation of GAK Results in Elevated Levels of EGFR on the Cell allele suppresses EGFR-mediated ERK1͞2 signaling, so we exam- Surface and Enhanced EGFR Activation. We next wished to determine ined the activation of mitogen-activated protein kinases in the whether the EGFR in GAK knockdown cells is localized at the cell knockdown cells. Under conditions of serum starvation, there was surface or at some abnormal intracellular location. To examine constitutive activation of ERK1͞2 in knockdown cells but not in cell-surface exposure, receptors on the surface of living cells were control cells (Fig. 3A). This finding is consistent with the presence labeled with biotin. Subsequently, the biotinylated receptors were of constitutively active EGFR in the knockdown cells. Ten minutes

10298 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0403175101 Zhang et al. Downloaded by guest on September 27, 2021 Fig. 3. Down-regulation of GAK differentially enhances EGFR-dependent signaling. Cells were serum-starved overnight and stimulated with 50 ng/ml EGF for indicated times. Western blot analysis was carried out by probing with an Ab against phosphorylated ERK1͞2 or total ERK1͞2(A); and an Ab against phosphorylated ERK5 or total ERK5 (B); or an Ab against phosphorylated Akt or total Akt (C).

after EGF addition, the level of activated ERK1͞2 is higher in knockdown cells than in controls. However, the difference between control and knockdown cells is diminished by 20 min after EGF stimulation, and control cells actually display higher levels of activated ERK1͞2 at later times. This result is consistent with the idea that ERK1͞2 activation occurs more rapidly in the knockdown cells than in control cells. On the other hand, activation of ERK5, a kinase involved in cell proliferation and survival, is greatly up-regulated in the knockdown cells (Fig. 3B and refs. 30–32). As ͞ Fig. 4. Down-regulation of GAK alters intracellular trafficking of EGFR. Cells observed in the case of ERK1 2 activation, there is a substantial grown on chamber slides were serum-starved overnight. Cells were then level of active ERK5 under serum starvation. However, in contrast incubated with 0.5 ␮g/ml rhodamine-conjugated EGF on ice for 1 h. (A–D) Cells ͞ Ϸ to ERK1 2 activation, activation of ERK5 is 100-fold higher in were either fixed immediately (A and B) or incubated at 37°C in a 10% CO2 the knockdown cells than in controls, and this activation is sustained humidified incubator for 40 min (C and D), then fixed and counterstained with for Ͼ60 min (Fig. 3B and data not shown). We also examined the DAPI (blue) to localize the nucleus of the cell being studied. (E–L) Alternatively, activation of several other signaling proteins. We found that ty- cells were incubated with (E–H, K, and L) or without (I and J) rhodamine- rosine phosphorylation of p85, a regulatory subunit of phosphati- conjugated EGF (red signal) and fixed as described above, then incubated with primary Ab as indicated below. In each case, the Ab signal is green. Overlap of dylinositol 3-kinase, was greatly elevated in knockdown cells (data ͞ the EGF signal in red and Ab signal in green is yellow. Cells were visualized with not shown). Consequently, the serine threonine kinase Akt, a a Zeiss LSM 510 META͞NLO confocal microscope at ϫ63 magnification and downstream target of phosphatidylinositol 3-kinase, was highly images were captured by Zeiss confocal microscope software Version 3.2. (A activated in knockdown cells (Fig. 3C). On the other hand, tyrosine and B) Enhanced levels of EGFR visualized on the surface of GAK knockdown phosphorylation of SHC and phospholipase C ␥ was either sup- cells. Fluorescently labeled EGF is used to visualize EGFR on the plasma pressed (SHC) or not affected (PLC-␥) (data not shown). Whereas membrane without internalization. (C and D) EGFR is mislocalized after EGF it is not surprising that down-regulation of GAK affects signaling, stimulation. Fluorescently labeled EGF is used to visualize EGFR at 40 min after ͞ stimulation. Images captured by using differential interference contrast (DIC) the differential regulation of ERK1 2 versus ERK5 suggests that and fluorescence microscopy combined. (E and F) EGFR partially colocalizes

GAK plays a more complex role in cell signaling than expected. with clathrin. Costaining with an Ab against clathrin in addition to fluores- CELL BIOLOGY cently labeled EGF is used to determine the localization of clathrin in the Down-Regulation of GAK Alters EGFR Trafficking. In an attempt to compartment containing the EGFR. (G and H) EGFR colocalizes with EEA1 in understand the mechanism by which down-regulation of GAK GAK knockdown cells. Costaining with an Ab against EEA1, an early endosome might bring about such complex effects, we examined intracellular marker, in addition to fluorescently labeled EGF is used to determine the receptor trafficking in the knockdown cells by using confocal degree of colocalization between EEA1 and EGFR. (I and J) EEA1 is mislocalized even in the absence of receptor stimulation. An Ab against EEA1 is used to microscopy. We first used a fluorescently labeled EGF probe to show the status of the early endosome in the absence of EGF stimulation. (K track the receptor itself. As seen in Fig. 4, if cells are left on ice after and L) Activated ERK5 colocalizes with EGFR. Costaining with an Ab against application of ligand, the EGF probe is primarily seen on the plasma phosphorylated ERK5 in addition to fluorescently labeled EGF is used to show membrane in both knockdown (Fig. 4B) and control cells (Fig. 4A), the relationship between EGFR and activated ERK5. (Scale bar, 20 ␮M.)

Zhang et al. PNAS ͉ July 13, 2004 ͉ vol. 101 ͉ no. 28 ͉ 10299 Downloaded by guest on September 27, 2021 with a much stronger signal evident on the knockdown cells. After cells were left at 37°C for 40 min to allow EGF internalization, the probe relocalizes to the cytoplasm in the knockdown cells but is clearly mislocalized compared with controls (Fig. 4D versus C). By immunostaining with multiple markers, we found that in GAK knockdown cells, internalized EGFR was not colocalized with the clathrin AP2 (data not shown) and was partially colocalized with clathrin (Fig. 4F, compare with E). Interestingly, EGFR was well colocalized with EEA1, an early endosome marker (Fig. 4H, compared with G). We also examined whether any of these markers were relocalized in the knockdown cells in the absence of EGF stimulation. We observed that clathrin and AP2 localization was not affected (data not shown). However, EEA1 was mislocalized in knockdown cells even in the absence of EGF (Fig. 4J, compared with I). We also examined the localization of downstream signaling molecules. Whereas the relocalization of ERK1͞2 to the nucleus was not significantly altered compared with EGF-treated control cells (data not shown), we observed that ERK5 colocalized with receptors in the altered endosomal compartment (Fig. 4 L com- pared with K).

Down-Regulation of GAK Leads to Cell Proliferation and Partial Transformation. The large changes in receptor abundance, localiza- tion, and function occasioned by down-regulation of GAK might be expected to have biological consequences. We first examined HeLa cell entry into S phase under serum starvation by measuring the number of cells incorporating BrdUrd into DNA. Approximately 20% of GAK knockdown cells were BrdUrd-positive as compared with Ϸ3.4% of vector control cells (Fig. 5A). Thus, down-regulation of GAK exerts a significant biological effect by promoting cell Fig. 5. Down-regulation of GAK promotes cell proliferation and cell trans- formation. (A) Down-regulation of GAK promotes cell proliferation under proliferation in the absence of growth factors. To determine conditions of serum starvation. Cells grown on chamber slides were serum- whether knocking down GAK expression would cause cell trans- starved overnight and labeled with BrdUrd for 15 h. Cells were fixed and formation, we used RNAi to down-regulate GAK in CV1P cells incubated with an Ab against BrdUrd. BrdUrd-positive cells were counted because HeLa cells are already transformed. Fig. 5 B and C show among DAPI-stained cells in at least 10 randomly selected fields on each slide that knocking down GAK results in a marked increase in soft agar (n ϭ 3). (B) Down-regulation of GAK transforms CV1P cells. CV1P cells (vector colony formation by using CV1P cells. Notably, the colonies seen control and 525) were seeded in soft agar. Phase-contrast image of the are smaller than would be seen in fully transformed cells cultured colonies from a representative experiment is shown (magnification: ϫ7). (C) for the same period. Colony numbers (colony with Ͼ50 cells) from B were counted at 21 days after cells were seeded (n ϭ 3). Discussion The data presented here reveal an unexpected effect of GAK on whether it might simply result from the increased numbers of receptor signaling, suggesting that GAK is involved in consid- receptors but this possibility seems unlikely for two reasons. erably more than regulating the formation and uncoating of First, in our transient experiments with GAK siRNA, we some- clathrin-coated vesicles. If GAK were required for assembly of times see lower levels of receptor expression but the same effects coated vesicles, we would have expected to see receptor traf- on the early endosome. Second, we checked A431 cells and ficking block near the cell surface. Similarly, if GAK were simply required for uncoating, we might have expected to see EGFR found that, although they feature even higher numbers of EGFR, trafficking blocked in clathrin-coated vesicles. In the GAK they did not display the alterations of the EGFR signaling knockdown cells, receptors are found in structures containing reported here. It seems more likely that the altered signaling both clathrin and early endosomal markers. This finding suggests results from the fact that the receptors transit through trafficking that loss of GAK impairs the uncoating process and results in compartments such as the early endosome extremely slowly. We clathrin being present on vesicles from which it would normally note the similarity to data reported for the Trk family of receptors (33). When Trks are activated on the cell body they have been stripped. However, our data further suggest that GAK ͞ plays an active role in trafficking downstream of clathrin-coated signal through ERK1 2. When they are activated in a long axon vesicles in regulating the correct functioning of the endosome. and take part in a prolonged journey to the cell body, ERK5 This finding is implied by the fact that the receptors apparently signaling is seen. We hope that with further experimentation we are trapped not at the uncoating stage but in the endosome itself. may find that the signaling intermediates generated in GAK More importantly, the endosome appears to be altered in the knockdown cells will give us useful information on normal knockdown cells even in the absence of EGF stimulation, signaling. indicating that GAK plays an important role in the normal Our results differ significantly from those of Vieira et al. (25) who ontogeny of the endosome. observed that overexpression of dominant-negative dynamin inhib- How can we model the effects of down-regulating GAK on its EGFR-mediated ERK activation. Part of the differences be- signaling? It is reasonable to suppose that both the increased tween the effects of dominant-negative dynamin and GAK down- receptor numbers and alterations in receptor signaling result at regulation might lie in the technical details of the two studies. Vieira least in part from an underlying defect in receptor trafficking. et al. (25) measured EGFR signaling in HeLa cells treated with 3.5 The increase in numbers may simply reflect the fact that the nM EGF, whereas we are using a somewhat higher EGF concen- normal degradation of the receptor in the lysosome is blocked. tration (8 nM). At lower levels of EGF, internalization of EGFR has The signaling defect is perhaps more interesting. We wondered been reported to occur solely through clathrin-coated vesicles, while

10300 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0403175101 Zhang et al. Downloaded by guest on September 27, 2021 at higher concentrations, other pathways have been reported to be not shown). It seems likely that GAK is affecting the stability of operative (34). However, we have repeated our work at 4 nM EGF c-Cbl, either through direct phosphorylation or through some more and observed similar effects on Erk1͞2 activation as displayed in indirect route. GAK is also known to phosphorylate ␮ subunits of Fig. 3A (data not shown). Notably, other workers have reported that AP1 and AP2 in vitro (4, 9). If AP1 and AP2 are also targets of Ϫ even at an EGF concentration of 10 12 M, dominant-negative GAK in vivo, this phosphorylation might also contribute to the dynamin does not block ERK activation (35). We think that it is effects reported here. Potential effects of GAK on AP1 in particular likely that the major cause of differences between our data and might play a role in the effects of depleting GAK on trafficking earlier reports with dominant-negative dynamin stem from intrinsic steps distal from the membrane. Obviously other key targets for differences between the roles of GAK and dynamin in EGFR GAK may play a role in EGFR signaling. Further work will be internalization. required to delineate GAK targets in vivo. Most of the data shown here were obtained from HeLa cells in GAK down-regulation might be expected to have effects on the which GAK had been stably depleted by means of retroviral internalization and signaling of other receptors as well. We have expression of small hairpin RNA to GAK. It is fair to ask whether found that the effects of GAK down-regulation vary from receptor some of the effects observed might arise secondarily in the course to receptor and also vary with the cell type being studied. To date, of long term culture. Certainly, such indirect effects of GAK only the EGFR has been dramatically up-regulated by GAK depletion can occur. For instance, we know that GAK depletion by depletion in multiple cell types. Many receptors are unaffected. For transient transfection of GAK siRNA blocks transferrin uptake, instance, in HeLa cells, neither HER2 nor IGF receptor up- whereas cells in which GAK is stably depleted show only minor regulation is observed upon GAK knockdown. Similarly, in NIH inhibition of transferrin uptake (data not shown). This effect 3T3 cells, we have seen no effects on PDGF receptor after GAK presumably is selected for because transferrin is required for knockdown. The function of receptors outside the class of receptor long-term survival of the cells. We have also observed that, whereas tyrosine kinases has even been observed to be impaired (data not stable GAK depletion certainly causes an increase in EGFR shown). Even the effects that we see on EGFR in HeLa, CV1P, and stability as expected, it also results in an approximately 10-fold COS-7 cells are not seen in all cell types. It is well known that signal increase in the levels of mRNA for EGFR (data not shown). Thus, transduction of various receptors varies with cell type. Therefore, it it is possible that some of the quantitative effects we see in EGFR is perhaps not surprising to see different results in receptor traf- levels arise secondarily. However, we would like to stress that all of ficking in different cell types. the qualitative changes reported here (increased levels of EGFR Finally, the fact that GAK down-regulation can result in en- protein and signaling, increased activation of ERK5, and changes in hanced soft agar growth suggests GAK may function as a tumor localization of EEA1) all occur in cells where GAK has been acutely suppressor. Whereas the level of soft agar growth that we see in depleted via transfection of siRNA oligos. CV1P cells depleted of GAK is relatively modest, the effect of GAK Various molecular mechanisms might underlie the observed loss could easily be imagined to synergize with other mutations to changes in EGFR trafficking and signaling. For example, the facilitate tumor growth. Interestingly, the GAK chromosomal locus depletion of GAK’s auxilin-like activity would be expected to play 4p16.3 has been found to show loss of heterozygosity in multiple a significant role in the apparent defects in clathrin uncoating. In tumor types, including breast and colon cell carcinomas (40, 41). A addition the Cbl-mediated EGFR degradation pathway might be more detailed analysis of the GAK gene in these and other tumors altered in GAK knockdown cells. Cbl consists of an N-terminal could be a target for future work. phosphotyrosine binding domain and a C-terminal proline-rich domain separated by a zinc-RING finger motif (36, 37). Cbl has We thank Dr. Scott Pomeroy and Matthew C. Salanga for expert been reported to bind to various signaling proteins, including the technical assistance in obtaining confocal microscope images; and Drs. EGFR through its phosphotyrosine binding domain, and subse- Charles D. Stiles, Pamela A. Silver, Bruce Spiegelman, Sanja Sever, quently bring about their ubiquitination by means of the zinc-Ring Helen McNamee, Joanne Chan, Danielle K. Lynch, and Grace Williams finger motif (36, 37). Overexpression of Cbl promotes down- for critical reading of the manuscript. This work was supported by regulation of EGFR through receptor ubiquitination (38, 39). In National Institutes of Health Grants CA50661 (to T.M.R.) and P30 GAK knockdown cells, c-Cbl is significantly down-regulated (data HD18655 (to Scott Pomeroy).

1. Kanaoka, Y., Kimura, S. H., Okazaki, I., Ikeda, M. & Nojima, H. (1997) FEBS Lett. 402, 21. Hindley, A. & Kolch, W. (2002) J. Cell Sci. 115, 1575–1581. 73–80. 22. Fruman, D. A., Meyers, R. E. & Cantley, L. C. (1998) Annu. Rev. Biochem. 67, 481–507. 2. Kimura, S. H., Tsuruga, H., Yabuta, N., Endo, Y. & Nojima, H. (1997) Genomics 44, 23. Cantley, L. C. (2002) Science 296, 1655–1657. 179–187. 24. Klippel, A., Kavanaugh, W. M., Pot, D. & Williams, L. T. (1997) Mol. Cell. Biol. 17, 338–344. 3. Greener, T., Zhao, X., Nojima, H., Eisenberg, E. & Greene, L. E. (2000) J. Biol. Chem. 25. Vieira, A. V., Lamaze, C. & Schmid, S. L. (1996) Science 274, 2086–2089. 275, 1365–1370. 26. Brummelkamp, T. R., Bernards, R. & Agami, R. (2002) Cancer Cell 2, 243–247. 4. Umeda, A., Meyerholz, A. & Ungewickell, E. (2000) Eur. J. Cell Biol. 79, 336–342. 27. Levy-Toledano, R., Caro, L. H., Hindman, N. & Taylor, S. I. (1993) Endocrinology 133, 5. Korolchuk, V. I. & Banting, G. (2002) Traffic 3, 428–439. 1803–1808. 6. Conner, S. D. & Schmid, S. L. (2002) J. Cell Biol. 156, 921–929. 28. Lokeshwar, V. B., Huang, S. S. & Huang, J. S. (1989) J. Biol. Chem. 264, 19318–19326. 7. Conner, S. D. & Schmid, S. L. (2003) J. Cell Biol. 162, 773–779. 29. Wiley, H. S. (2003) Exp. Cell Res. 284, 78–88. 8. Conner, S. D., Schroter, T. & Schmid, S. L. (2003) Traffic 4, 885–890. 30. Abe, J., Kusuhara, M., Ulevitch, R. J., Berk, B. C. & Lee, J. D. (1996) J. Biol. Chem. 9. Ricotta, D., Conner, S. D., Schmid, S. L., von Figura, K. & Honing, S. (2002) J. Cell Biol. 271, 16586–16590. 156, 791–795. 31. Kato, Y., Tapping, R. I., Huang, S., Watson, M. H., Ulevitch, R. J. & Lee, J. D. (1998) 10. Zhao, X., Greener, T., Al-Hasani, H., Cushman, S. W., Eisenberg, E. & Greene, L. E. Nature 395, 713–716. (2001) J. Cell Sci. 114, 353–365. 32. English, J. M., Pearson, G., Hockenberry, T., Shivakumar, L., White, M. A. & Cobb, 11. Schlessinger, J. (2000) Cell 103, 211–225. M. H. (1999) J. Biol. Chem. 274, 31588–31592. 12. Pawson, T. & Hunter, T. (1994) Curr. Opin. Genet. Dev. 4, 1–4. 33. Watson, F. L., Heerssen, H. M., Bhattacharyya, A., Klesse, L., Lin, M. Z. & Segal, R. A.

13. Benveniste, M., Schlessinger, J. & Kam, Z. (1989) J. Cell Biol. 109, 2105–2115. (2001) Nat. Neurosci. 4, 981–988. CELL BIOLOGY 14. Schlessinger, J., Schreiber, A. B., Levi, A., Lax, I., Libermann, T. & Yarden, Y. (1983) 34. Sorkin, A., McClure, M., Huang, F. & Carter, R. (2000) Curr. Biol. 10, 1395–1398. CRC Crit. Rev. Biochem. 14, 93–111. 35. Johannessen, L. E., Ringerike, T., Molnes, J. & Madshus, I. H. (2000) Exp. Cell Res. 260, 15. Schreiber, A. B., Libermann, T. A., Lax, I., Yarden, Y. & Schlessinger, J. (1983) J. Biol. 136–145. Chem. 258, 846–853. 36. Thien, C. B. & Langdon, W. Y. (2001) Nat. Rev. Mol. Cell Biol. 2, 294–307. 16. Sorkin, A. (1998) Front. Biosci. 3, D729–D738. 37. Sanjay, A., Horne, W. C. & Baron, R. (2001) Sci. STKE 2001, PE40. 17. Robinson, M. J. & Cobb, M. H. (1997) Curr. Opin. Cell Biol. 9, 180–186. 38. Dikic, I. (2003) Biochem. Soc. Trans. 31, 1178–1181. 18. Morrison, D. K., Kaplan, D. R., Rapp, U. & Roberts, T. M. (1988) Proc. Natl. Acad. Sci. 39. Yokouchi, M., Kondo, T., Houghton, A., Bartkiewicz, M., Horne, W. C., Zhang, H., USA 85, 8855–8859. Yoshimura, A. & Baron, R. (1999) J. Biol. Chem. 274, 31707–31712. 19. Williams, N. G., Paradis, H., Agarwal, S., Charest, D. L., Pelech, S. L. & Roberts, T. M. 40. Shivapurkar, N., Maitra, A., Milchgrub, S. & Gazdar, A. F. (2001) Hum. Pathol. 32, (1993) Proc. Natl. Acad. Sci. USA 90, 5772–5776. 169–177. 20. Mason, C. S., Springer, C. J., Cooper, R. G., Superti-Furga, G., Marshall, C. J. & Marais, 41. Shivapurkar, N., Sood, S., Wistuba, I. I., Virmani, A. K., Maitra, A., Milchgrub, S., R. (1999) EMBO J. 18, 2137–2148. Minna, J. D. & Gazdar, A. F. (1999) Cancer Res. 59, 3576–3580.

Zhang et al. PNAS ͉ July 13, 2004 ͉ vol. 101 ͉ no. 28 ͉ 10301 Downloaded by guest on September 27, 2021