Cyclin G–Associated Kinase Is Necessary for Osteosarcoma Cell Proliferation and Receptor Trafficking

Published OnlineFirst September 23, 2010; DOI: 10.1158/1535-7163.MCT-10-0637

Molecular Cancer Preclinical Development Therapeutics

Michiro Susa1,2, Edwin Choy2, Xianzhe Liu1,2, Joseph Schwab1,2, Francis J. Hornicek1,2, Henry Mankin2, and Zhenfeng Duan1,2

Abstract Osteosarcoma is the most frequent primary malignant bone tumor among the children. The advent of neoadjuvant chemotherapy significantly improved the prognosis of patients with osteosarcoma in the 1980s, but it has since plateaued in the past decades. Recently, one of the most researched areas in sarcoma treatment is tyrosine kinases. Here, we describe research on a serine/threonine kinase, cyclin G–associated kinase (GAK), which has not been reported in osteosarcoma previously. In this study, a lentiviral based human shRNA library was utilized to screen for kinases in KHOS and U-2OS osteosarcoma cells. The expression of GAK was examined in osteosarcoma and the effect on cell proliferation was analyzed by GAK siRNA knockdown. The level of GAK expression and its correlation to prognosis was analyzed in osteosarcoma tissue microarray. The effect of GAK depletion on insulin-like growth factor and epidermal growth factor receptor– mediated signal transduction was analyzed by Western blot. We observed that GAK was overexpressed in both osteosarcoma cell lines and tissue samples when compared with human osteoblasts. GAK knockdown by siRNA decreased cell proliferation in both drug-sensitive and multidrug-resistant osteosarcoma cell lines. Immunohistochemistry of osteosarcoma tissue microarray revealed that overexpression of GAK was associated with poor prognosis. Finally, knockdown of GAK resulted in alterations of receptor trafficking and several downstream proteins. In conclusion, our results suggest that osteosarcoma cell proliferation and survival are dependent on GAK. These findings may lead to the development of new therapeutic options for osteosarcoma. Mol Cancer Ther; 9(12); 3342–50. 2010 AACR.

Introduction osteosarcoma has yet to be elucidated. Nevertheless, targeting kinases has tremendous potential as shown in Osteosarcoma has the highest prevalence (approxi- the treatment of gastrointestinal stromal tumors treated mately 60%) among patients with primary malignant bone with imatinib mesylate (Gleevec; Novartis Pharmaceu- tumors. Each year, there are more than 800 new patients, ticalsCorp;ref.2). younger than 20 years, with a diagnosis of osteosarcoma in Cyclin G–associated kinase (GAK), also known as the United States (1). Major progress has been made in the auxillin 2, is a 160-kDa serine/threonin protein kinase treatment of patients with osteosarcoma because of the that was cloned from rats and humans in 1997 (3, 4). It is use of chemotherapy, leading to an improved overall composed from N-terminal kinase domain that phos- survival rate of 65%. However, 30% to 40% of the patients phorylates the m-subunits of adaptor protein 1 and 2, with osteosarcoma still experience recurrence or metas- clathrin-binding domain, and a C-terminal J-domain that tasis despite the improved multimodality therapy. Many interacts with heat shock cognate 70 (Hsc 70). Although it will develop resistance to multiple types of chemotherapy shares 43% homology to auxillin, which is expressed only after prolonged periods of treatments. in the neuronal cells (5), GAK is expressed ubiquitously The importance of kinases has been established in various organs. GAK is known to assemble clathrin in many cancers, but their diversity of functions in into baskets during endocytosis and support Hsc 70 in uncoating clathrin-coated vesicles (CCV; refs. 6, 7). In yeast and Caenorhabditis elegans, auxillin depletion has Authors' Affiliations: 1Department of Orthopaedic Surgery, and 2Sarcoma been reported to cause accumulation of CCVs, impair Biology Laboratory, Center for Sarcoma and Connective Tissue Oncology, cargo delivery to the vacuoles, and slow down cell Massachusetts General Hospital, Boston, MA growth (8–10). GAK was identified in our lentiviral Corresponding Author: Zhenfeng Duan, Center for Sarcoma and Con- nective Tissue Oncology, Massachusetts General Hospital, 55 Fruit St. shRNA screen as being critical for osteosarcoma cell Jackson 1115, Boston, MA 02114. Phone: (617) 724-3144; Fax: (617) 724- proliferation. Because GAK has important functions as 7396. E-mail: [email protected] a regulator of receptor tyrosine kinase trafficking, which doi: 10.1158/1535-7163.MCT-10-0637 is situated at the upstream of many kinases that 2010 American Association for Cancer Research. have been reported to play key regulatory roles in the

3342 Mol Cancer Ther; 9(12) December 2010

Cyclin G–Associated Kinase and Osteosarcoma

development of cancer, it was our aim to further examine in accordance with the policies of the institutional review its function in osteosarcoma cell lines. board of the hospital. All diagnoses were confirmed by In osteosarcoma, receptor tyrosine kinases such as light microscopy and immunohistochemistry. insulin-like growth factor 1 receptor (IGF-IR; refs. 11–15) and epidermal growth factor receptor (EGFR; refs. Lentiviral human kinase shRNA library screen 16–19) have been implicated in malignant growth and using MISSION LentiExpress human kinases invasion. When ligands bind to each respective receptor, shRNA library they are activated by autophosphorylation, internalized, A total of 673 human kinases were analyzed for at least in part through clathrin-mediated endocytosis, their effects on osteosarcoma cell growth, using MISSION and sorted to early and late endosomes. Activation of LentiExpress Human Kinases shRNA Library (Sigma) as signal transduction pathways and consequent sub- described previously (20). cellular reactions are mediated through the subcellular compartmentalization of each receptor tyrosine kinases. Synthetic GAK siRNA and transfection It is important to note that osteosarcoma is characterized Two different GAK siRNAs and 1 nonspecific by heterogeneity that has numerous genetic changes. siRNA that has no significant homology to any known Although there are several clinical trials (www. genesequencesfrommouse,rat,orhumanwere clinicaltrials.gov) currently undergoing that utilize var- purchased from Ambion. The siRNA sequences target- ious drugs to block the IGF-IR and EGFR signaling ing GAK (GenBank accession No. NM_005255.2) corre- cascade, it may be logical to do a sequential or combi- sponded to coding regions (1: sense 50-GUCCGUCGC- natorial application of several kinase inhibitors for the UAAUUAUGCAtt-30,antisense50-UGCAUAAUUAG- treatment of osteosarcoma. Therefore, it may be useful CGACGGACtg-30;2:sense50-CACCAGAAAUCAUA- to identify new kinases associated with osteosarcoma GACUUtt-30,antisense50-AAGUCUAUGAUUUCU- to develop appropriate treatment options for diverse GGUGtt-30)oftheGAK gene. The siRNA oligonucleo- patients with osteosarcoma. tides were dissolved in nuclease-free water at a con- In this study, we analyzed the expression of GAK and centration of 100 mmol/L and kept at 20Cuntilthe its function in osteosarcoma cell lines. In addition, we following transfection experiments. Transfections were hypothesized that GAK might have a role as a prognostic performed with Lipofectamine RNAiMAX according to indicator for osteosarcoma. These data may contribute to the manufacturer’s instruction (Invitrogen). Mixtures the growing information of kinases for clinical utility in (100 nmol/L) of siRNA, transfection agent, and medium the treatment of osteosarcoma. were subsequently diluted with the same medium and applied at the designated concentration to each well. Materials and Methods The medium was replaced with RPMI 1640 supplemen- ted with 10% FBS 24 hours after each transfection. Cell culture and tissues The human osteosarcoma cell line U-2OS was obtained Immunofluorescence microscopy from American Type Culture Collection. Dr Efstathios Osteosarcoma cells were grown with the addition of S. Gonos (National Hellenic Research Foundation) kindly either 100 nmol/L of GAK siRNA1 or 100 nmol/L of provided the human osteosarcoma cell line KHOS and nonspecific siRNA on 8-well Lab-Tek chamber slides the multidrug-resistant cell lines KHOSR2 and U-2OSR2. (Thermo Fisher Scientific, Rochester, NY) for 24 hours All cell lines were cultured in RPMI 1640 medium sup- and fixed in 3.7% buffered paraformaldehyde. Immunos- plemented with 10% FBS, 100 U/mL of penicillin, and tainings were done using antibodies against GAK (1:50 100 mg/mL of streptomycin (all obtained from Invitro- dilution; sc-7864; Santa Cruz Biotechnology) and actin gen). Cells were incubated at 37 Cin5%CO2, 95% air (1:200 dilution; Sigma-Aldrich) overnight. After washing, atmosphere, and passaged when near-confluent mono- the cells were incubated with Alexa Fluor secondary anti- layers were achieved using trypsin-EDTA solution. Resis- bodies (Invitrogen). To counterstain the nuclei, the cells tant cell lines were continuously cultured in 0.1 mmol/L were incubated with PBS containing 1 mg/mL of Hoechst of doxorubicin. Doxorubicin was obtained as unused 33342 (Invitrogen) for 15 minutes. Stained cells were then residual clinical material from the Massachusetts General visualized on a Nikon Eclipse Ti-U fluorescence micro- Hospital. Human osteoblast cells HOB-c were obtained scope (Nikon Instruments Inc.) equipped with a SPOT RT from PromoCell GmbH. Osteoblast cells were cultured in digital camera (Diagnostic Instruments Inc.). osteoblast growth medium (PromoCell GmbH) with 10% FBS. Cells were free of mycoplasma contamination Effect of GAK depletion on osteosarcoma as tested by MycoAlert Mycoplasma Detection Kit from cell proliferation Cambrex. Authentication of the cell line was not done. Proliferation of the cells was assessed using the Cell- Osteosarcoma tissue samples were obtained from Mas- Titer 96 Aqueous One Solution Cell Assay (Promega) and sachusetts General Hospital sarcoma tissue bank. Surgi- by direct enumeration. Briefly, 1 103 cells were plated cally treated patients with a diagnosis of osteosarcoma per well in a 96-well plate for the CellTiter 96 Aqueous were identified and utilized for the study and were used One Solution Cell Assay and 2 103 cells were seeded in

www.aacrjournals.org Mol Cancer Ther; 9(12) December 2010 3343

Susa et al.

a 24-well plate for enumeration. After 24 hours of incuba- alone, medium was replaced and the cells were allowed tion with increasing concentrations of GAK siRNA, 100 to grow for another 48 hours. Protein lysates from cells nmol/L of nonspecific siRNA or medium alone, medium were generated through lysis with RIPA Lysis Buffer was replaced and the cells were allowed to grow for 6 to 7 (Upstate Biotechnology). The concentration of the protein days. GAK siRNA1 was utilized for CellTiter 96 Aqueous was determined by Protein Assay reagents (Bio-Rad) and a One Solution Cell Assay and GAK siRNA2 was applied spectrophotometer (Beckman DU-640, Beckman Instru- for direct enumeration. The absorbance for CellTiter 96 ments, Inc.). Western blot analysis was performed as Aqueous One Solution Cell Assay was read on a SPEC- previously reported (21). The mouse monoclonal antibody TRA max Microplate Spectrophotometer (Molecular to GAK was purchased from MBL International Corp. Devices) at a wavelength of 490 nm. The cells were Mouse monoclonal antibodies to EGFR and pSTAT3 were counted manually for enumeration by light microscopy, purchased from Cell Signaling Technologies. Rabbit poly- using Trypan blue exclusion. Each experiment was per- clonal antibodies to IGF-IR, pAKT, pmTOR, and pEGFR formed in triplicate. were purchased from Cell Signaling Technologies as well. P-gp1 monoclonal antibody C219 was purchased Tissue microarray slide and immunohistochemistry from Signet. The rabbit polyclonal antibody to human Osteosarcoma tissue microarrays were obtained from pIGF-IR (1158/1162/1163) was purchased from Invitro- the IMGENEX. The microarray was composed of 58 gen. Goat anti-mouse HRP and goat anti-rabbit HRP were osteosarcoma tissues and included duplicate core biop- purchased from Bio-Rad. SuperSignal West Pico Chemi- sies (2 mm in diameter) from fixed, paraffin-embedded luminescent Substrate was purchased from PIERCE. tumors. Five samples were lost during the staining pro- Bands were semiquantified by reverse image scanning cess, and 2 patients died because of other causes and densitometry with PhotoShop 7.0 (Adobe). therefore were excluded from the analysis. Slide was baked at 62C for 1 hour, deparaffinized in xylene for Statistical analysis 15 minutes, transferred through 100% ethanol for 5 min- The Student t test was used to compare the differences utes, and then rehydrated with graded ethanol. Endo- between groups (GraphPad PRISM 4 software; GraphPad genous peroxidase activity was quenched by 5-minute Software). The Kaplan-Meier method was used to deter- incubation in 3% hydrogen peroxide (H2O2) in methanol. mine the relationship between GAK staining levels and Rest of the staining was preformed with HRP-DAB Cell patient survival, and data were analyzed with the log and Tissue Staining Kit (R&D Systems) according to the rank test (GraphPad PRISM 4 software). Results are given manufacturer’s instruction. Primary antibody was as mean SD and results with P < 0.05 were considered applied at 4C overnight (1:50 dilution; Santa Cruz Bio- statistically significant. technology) in 1% bovine serum albumin. Slide was counterstained with hematoxylin QS (Vector Labora- Results tories) and mounted with VectaMount AQ (Vector Laboratories) for long-term preservation. Identification of GAK as a new regulator of osteosarcoma cell survival Evaluation of immunohistochemical staining We performed a screen that targeted 673 kinase genes The percentage of cells showing positive cytoplasmic by lentiviral shRNA and determined its effects on osteo- staining for GAK was calculated by reviewing the entire sarcoma cell survival. There were 9 kinases, when spot. On the basis of the percentage of cells with positive knocked down, which displayed inhibitory growth cytoplasmic staining, the staining patterns were classified effects on KHOS. We further validated these shRNA according to 4-grade scale: 0, absence of cell staining; 1þ, clones targeting the kinase hits in a secondary osteosar- weak staining; 2þ, moderate staining; and 3þ, intense coma cell line, U-2OS. Validated kinases included GAK, staining (20). For statistical analysis, patients with osteo- PLK1 (20), Mirk (22), and ROCK1. Although most of the sarcoma were subgrouped as either GAK low-staining kinases tested showed only limited effects, 3 of 5 shRNA group (scale grade: 0, 1) or GAK high-staining group (scale target sites of GAK inhibited osteosarcoma cell growth grade: 2, 3). Categorizing of immunostaining intensity (Fig. 1). Molecular signaling has become increasingly was completed by 2 independent observers. Discrepant important for the treatment of various cancers, and scores between the 2 observers were rescored to arrive at a because GAK has been implicated as an important kinase single final score. Light microscopic images were obtained in the trafficking of receptors that is at the very upstream using a Nikon Eclipse Ti-U fluorescence microscope of signal transduction, it was chosen to further investigate (Nikon Corp.) with a SPOT RT digital camera (Diagnostic its functional role in osteosarcoma cell lines. Instruments Inc.). GAK is overexpressed in osteosarcoma Western blot analysis cell lines and tissues Cells were seeded onto 6-well plates 4 105 cells/well). GAK protein expression was analyzed both in several After 24 hours of incubation with increasing concentra- osteosarcoma cell lines and in osteosarcoma tissues by tions of GAK siRNA, nonspecific siRNA, or medium Western blot. The results show that both drug-sensitive

3344 Mol Cancer Ther; 9(12) December 2010 Molecular Cancer Therapeutics

Cyclin G–Associated Kinase and Osteosarcoma

1: GAK lentiviral shRNA particle (x-1244slcl) 5: GAK lentiviral shRNA particle (x-4100slcl) 2: GAK lentiviral shRNA particle (x-3438slcl) 6: vector control (SHCOO1V) A 3: GAK lentiviral shRNA particle (x-1870slcl) 7: Nonspecific shRNA (SHCOO2V) 4: GAK lentiviral shRNA particle (x-2091slcl) 8: Medium control 1.2

1.0 0.8 B 0.6

0.4

OD proliferation 0.2

0.0 12 345678 Figure 2. Cyclin G–associated kinase (GAK) is expressed in osteo-

sarcoma. Osteosarcoma cell lines KHOS, KHOSR2, U-2OS, and Figure 1. Results of lentiviral shRNA knockdown in KHOS osteosarcoma U-2OSR2 overexpressed GAK when compared with a normal osteoblast cells. Each data represents a well from 96-well plate of MISSION Lenti- cell line HOB-c (A). All the osteosarcoma tissues examined expressed GAK Express Human Kinases shRNA library. The effects of cyclin G-associated (B). Lanes 1 to 6 represent 6 different patients’ tissue samples. Tissue kinase (GAK) knockdown in osteosarcoma cell lines were assessed using samples for lanes 1 and 2 were taken prior to treatment, and samples 3 to 6 GAK 5 different sequences of gene. Three of 5 sequences for GAK shRNA were taken after chemotherapies were done. (lanes 2, 3, and 5) effectively inhibited the growth of osteosarcoma cells compared with the negative controls (lanes 6 and 7). Medium only from patients were also analyzed. Samples 1 and 2 were environment, which contains untransduced cells and causes complete cell taken prior to chemotherapy, and samples 3 to 6 were death in 7 days, was used as a positive control (lane 8). taken after chemotherapies were performed. GAK was expressed in all the samples tested (Fig. 2B). and multidrug-resistant osteosarcoma cell lines overexpressed GAK when compared with a normal human Confirmation of GAK knockdown osteoblast cell line (Fig. 2A). To preclude the possibility using synthetic siRNA that GAK expression is not just an artifact from prolonged To assess the transfection efficacy of GAK siRNA, 2 cell culturing of the cell lines, 6 osteosarcoma tissue samples lines, KHOS and KHOSR2, were examined. Both GAK

A B

Figure 3. Effective knockdown of cyclin G–associated kinase (GAK) by siRNA. GAK protein expression was effectively inhibited with 2 sequences of GAK siRNA at a minimal concentration of C 10 nmol/L in both drug-sensitive KHOS (A) and drug-resistant

KHOSR2 (B) cell lines (M, medium; c-siRNA, 100 nmol/L of nonspecific siRNA). The results of the knockdown of GAK in both

KHOS (C) and KHOSR2 (D) were further confirmed by immunofluorescence microscopy. GAK (green) was localized using anti- D GAK antibodies and then stained with Alexa Fluor secondary antibody.

www.aacrjournals.org Mol Cancer Ther; 9(12) December 2010 3345

Susa et al.

D B

Figure 4. Effect of cyclin G–associated kinase (GAK) depletion on osteosarcoma cell lines. KHOS and KHOSR2 were transfected with respective siRNA or medium alone, and cell proliferation after transfection was determined by the CellTiter 96 Aqueous One Solution Reagent. Addition of varying concentrations of GAK siRNA1 showed decreased proliferation (*P < 0.05) as compared with medium alone and 100 nmol/L of nonspecific siRNA for both

KHOS (A) and KHOSR2 (B; mean SD). In addition, direct enumeration of the osteosarcoma cells with Trypan blue exclusion after inhibition of GAK with GAK siRNA2 showed decreased cell proliferation with minimum cell deaths for both KHOS (C) and KHOSR2 (D; mean SD). The experiment was repeated 3 times.

< siRNA were efficiently incorporated into drug-sensitive KHOSR2 (Fig. 4B) osteosarcoma cell lines (P 0.05). KHOS (Fig. 3A) and drug-resistant KHOSR2 (Fig. 3B), and Similarly, direct enumeration of the osteosarcoma cell expression of GAK was suppressed at a minimum con- lines by light microscopy after inhibition of GAK with centration of 10 nmol/L (Fig. 3A). Immunofluorescence GAK siRNA2 showed decreased cell proliferation with results further confirmed the knockdown of GAK in both minimum cell deaths for both KHOS (Fig. 4C) and drug-sensitive (Fig. 3C) and multidrug-resistant (Fig. 3D) KHOSR2 (Fig. 4D). osteosarcoma cell lines. GAK and its correlation to the survival GAK depletion inhibits proliferation of patients with osteosarcoma of osteosarcoma cell lines We further analyzed GAK expression by immunohis- To analyze the phenotype of GAK knockdown on tochemistry using tissue microarray (Fig. 5A–C). Patient osteosarcoma cell lines, CellTiter 96 Aqueous One prognosis was compared with expression levels of GAK. Solution Cell Assay and direct enumeration using Trypan Intensity of GAK staining was compared between sam- blue exclusion was performed. The absorbance for Cell- ples from survivors (survival at 60 months follow-up) Titer 96 Aqueous One Solution Cell Assay showed that and nonsurvivors (died during <60 months follow-up). In nonspecific siRNA did not affect the growth of both drug- 26 nonsurvivors, 18 of 26 (69.2%) patients expressed GAK sensitive and multidrug-resistant osteosarcoma cell lines, at high levels (>2þ), whereas in 25 survivors, 4 of 25 whereas 10 nmol/L of GAK siRNA1 was enough to (16.0%) patients expressed GAK in high levels (Table 1). significantly decrease the cell proliferation in both The average expression level of GAK staining for survi- drug-sensitive KHOS (Fig. 4A) and multidrug-resistant vors and nonsurvivors were 1.4 and 2.1, respectively

3346 Mol Cancer Ther; 9(12) December 2010 Molecular Cancer Therapeutics

Cyclin G–Associated Kinase and Osteosarcoma

A D 3 2.1 2 1.4 1 GAK staining 0 B P = 0.0024 Survivors Nonsurvivors E 100 80 Low-staining C 60 40 High-staining 20 Percent survival P = 0.001 0 0 50 100 150 200 Time after surgery (month)

Figure 5. Correlation of cyclin G–associated kinase (GAK) low-staining and high-staining on osteosarcoma patient survival. Osteosarcoma tissue microarray consisting of 51 samples was stained using antibodies against GAK. The staining of each tissue samples were classified according to 4-grade scale: 0, absence of cell staining; 1þ, weak staining; 2þ, moderate staining; 3þ, intense staining. A–C, representative pictures of immunohistochemical staining of tissue microarray. Left, H&E stainings. Right, weak-staining (1þ; A), moderate-staining (2þ; B), and high-staining (3þ; C) photographs of GAK expression in osteosarcoma tissue samples. D, distribution of GAK staining scores among the survivors (survival at 60 months follow-up) and nonsurvivors (died during <60 months follow-up) of osteosarcoma. There was significant difference in staining scores between the survivors and the nonsurvivors (P ¼ 0.0024). E, Kaplan-Meier survival curve of GAK low-staining and high-staining patient samples. For analysis, patients were grouped into either GAK low-staining group (scale grade 0, 1) or GAK high-staining group (scale grade 2, 3). The prognosis of GAK low-staining group was significantly better than the high-staining group (P ¼ 0.001).

(Fig. 5D). The level of GAK staining for samples from transduction, Western blot analysis was performed. GAK nonsurvivors was significantly higher than that for sur- siRNA did not suppress the expression of either IGF-IR or vivors (P ¼ 0.0024). Kaplan-Meier survival analysis of EGFR and their phosphorylated forms, but, interestingly, patients with osteosarcoma between the low-staining and the expression of both receptors gradually increased the high-staining groups showed that the prognosis for along with the increase in GAK siRNA concentration patients in the high-staining group was significantly (Fig. 6A; Supplementary Fig. S1). The same effect was worse than that in the GAK low-staining group observed in both drug-sensitive and drug-resistant osteo- (Fig. 5E). GAK expression level did not show any sig- sarcoma cell lines. P-glycoprotein (P-gp) has been impli- nificant differences between age, gender, or histological cated as one of the major causes of drug resistance in subtype (data not shown). various cancers (23) and in osteosarcoma (24). Effect of GAK knockdown on P-gp expression was also analyzed Effect of GAK knockdown on receptors to see whether GAK has any effect on the P-gp induction and downstream effectors of osteosarcoma in drug-resistant osteosarcoma cells. There was no GAK is a potential regulator of clathrin-mediated change in P-gp expression even with the addition of membrane trafficking of various receptors. To examine 100 nmol/L of GAK siRNA. To assess the effect of the effect of GAK depletion on IGF-IR and EGFR signal GAK knockdown on the downstream signaling cascade

Table 1. Distribution of GAK staining level and its correlation to the survival of patients with osteosarcoma

Expression level Prognosis Total %

Survivors, % Nonsurvivors, %

Low 0 2 (3.9) 0 (0) 2 (3.9) 29 (56.8) 1þ 19 (37.2) 8 (15.7) 27 (52.9) High 2þ 2 (3.9) 10 (19.6) 12 (23.5) 22 (43.1) 3þ 2 (2.0) 8 (17.6) 10 (19.6) Total 25 (51.0) 26 (49.0) 51 (100)

www.aacrjournals.org Mol Cancer Ther; 9(12) December 2010 3347

Susa et al.

A Figure 6. Western blot analysis was performed to assess the effect of cyclin G–associated kinase (GAK) depletion on osteosarcoma receptor expressions and downstream signaling. A, the expression of both pIGF-IR and pEGFR increased along with the increase in the concentration of GAK siRNA1. While the expressions of IGF-IR and EGFR were slightly increased or unchanged in

KHOSR2, there was significant increase in both receptors in KHOS after GAK inhibition with 100 nmol/L of GAK siRNA. B GAK depletion had no effect on the expression of P-gp. B, expressions of pAKT, pmTOR, and pSTAT3, which are downstream effectors of growth factor receptor signaling, decreased as the concentration of GAK siRNA increased. The analysis was repeated 3 times.

of IGF-IR and EGFR, expressions of pAKT, pmTOR, and similar effects on both drug-sensitive and drug-resistant pSTAT3 were analyzed. In contrast to the upregulation of osteosarcoma cell lines, which implicate that it exerts its pIGF-IR and EGFR by GAK depletion, expressions of effects independently of ATP-binding cassette transpor- pAKT, pmTOR, and pSTAT3 were suppressed in a dose- ters such as P-gp. This was further confirmed by Western dependent manner (Fig. 6B). blot analysis of P-gp, which did not show any effect on P- gp trafficking. Currently, extensive research is underway Discussion to develop novel agents to overcome the drug resistance In this study, we first performed kinase shRNA screen- of osteosarcoma, but the results have been unsuccessful ing that revealed that the loss of function of GAK, among to date, hampered by low efficiency and/or high toxicity others, resulted in marked growth inhibition of osteosar- (29–32). GAK has the potential to be a target for the coma cells. High-throughput screening of the kinome is a treatment of drug-naive osteosarcomas as well as multi- powerful tool with which one can identify multiple drug-resistant osteosarcomas. kinases related to the survival of cancer cells. We have GAK was originally identified as a protein that can be previously reported on other kinases in osteosarcoma— immunoprecipitated with cyclin G (3). Subsequent assays PLK1 (20) and Mirk (22)—which was discovered using showed that GAK does not regulate cell cycle, but it is the same lentiviral based shRNA library. Currently, PLK1 present in CCVs and that kinase domain of GAK was inhibitors are already in clinical trial for several cancers directed toward the m2-subunit of adaptor protein 2 in (25–28). Theoretically, the best gene for targeted therapy CCVs (33). Clathrin-mediated endocytosis is one of the would be ones that are highly expressed in tumor cells mechanisms in which various receptors are internalized while minimally expressed in the normal tissues. In as cargos. CCVs arise as a result of clathrin-coated pit contrast to the high expression of GAK in osteosarcomas, invagination and constriction. GAK and the molecular normal human osteoblasts expressed only low quantity chaperone Hsc 70 have been reported to be involved in of the protein. The result of kinase shRNA screening was the constriction of the vesicle neck and scission of the further confirmed by siRNA knockdown of GAK on CCVs from the membrane (34). GAK is also reported to several osteosarcoma cell lines. Although 100 nmol/L have a role in assembling clathrin triskelions into clathrin of nonspecific siRNA did not have any cytotoxic activity baskets and is known to interact with Hsc 70 in the on osteosarcoma cell lines, a concentration of as low as uncoating of clathrin from CCVs (6, 7, 35, 36). The J 10 nmol/L of GAK siRNA was enough to inhibit the domain of GAK binds to Hsc 70 and stimulates its proliferation of osteosarcoma cells. Importantly, it had ATPase activity, which destabilizes the clathrin–clathrin

3348 Mol Cancer Ther; 9(12) December 2010 Molecular Cancer Therapeutics

Cyclin G–Associated Kinase and Osteosarcoma

interactions in CCVs (37). Although the functional to cathpesin D maturation, suggesting a defect in the mechanism of GAK has been extensively studied, its trans-Golgi network to endosome trafficking. GAK implication on cancer has not been determined. It has knockdown by siRNA in HeLa cells resulted in increased been suggested that the effect of GAK downregulation activation of ERK1/2, ERK5, and AKT (38), whereas varies between different receptors and between different introduction of dominant-negative dynamin, another cell lines. Downregulation of GAK has been reported to protein involved in receptor trafficking, inhibited MAPK transform CV1P (African green monkey kidney cell line) activation (41). GAK depletion has been linked to cells into highly proliferating cells (38). Similarly, GAK decrease in expression of Her4, a Notch target gene, in knockdown in HeLa cells (human cervix epithelioid zebrafish models (42). Two receptors (IGF-IR and EGFR) carcinoma cell line) resulted in the promotion of cell that have been repeatedly described in osteosarcoma proliferation (38). In contrast, GAK knockdown leads were analyzed to assess the effect of GAK depletion on to cell death in myeloma tumor cells as a part of results their expression. In our study, both receptors were upre- of high-throughput kinome siRNA screening (39). GAK gulated after GAK knockdown by siRNA. Although there expression has been implicated as a prognostic biomarker are multitudes of different kinases involved in the signal in prostate cancers also, in which GAK expression transduction of IGF-IR and EGFR, downstream effectors increases with prostate cancer progression to androgen such as AKT, mTOR, and STAT3 were significantly independence (40). Our results of osteosarcoma are com- decreased in activities as shown by a decrease in phos- patible with myeloma and prostate cancer for which GAK phorylation for each protein after GAK suppression. The inhibition leads to decreased cell proliferation. difference from the previous studies may be due to the There is scant data regarding pro-oncogenic function of difference in signal transduction of our receptor tyrosine GAK in tumors, but immunohistochemical staining of kinases in osteosarcoma compared with other cell types. osteosarcoma tissues suggest that there is an association The results imply that GAK knockdown leads to defect of between GAK expression levels and sarcoma progres- receptor trafficking, which, in part, may be due to the sion. Although we could not find any correlation between blockage of degradation of receptors during endocytosis. age, gender, primary site, or histological subtype based Our results suggest that GAK has profound effect on on our limited number of samples, further studies using receptor trafficking and signaling in osteosarcoma and greater number of patient samples are warranted. may be a potential target in therapy. Although the exact function of GAK overexpression in In summary, GAK is overexpressed in osteosarcoma advanced disease is still not clear, it has been suggested cells and required for osteosarcoma cell proliferation. that GAK overexpression correlates with increased GAK exerts its effects by perturbing the tyrosine receptor responsiveness of receptors in prostate cancers as they trafficking and causes alterations in signal transductions. progress in malignancy (40). Our results also suggest that These findings may lead to the development of new GAK expression correlates with worse outcome in therapeutic options for osteosarcoma. patients with osteosarcoma and can be a promising pre- dictor of osteosarcoma prognosis. Disclosure of Potential Conflicts of Interest GAK depletion has been linked to alterations in various receptor tyrosine kinase trafficking and signaling. No potential conflicts of interest were disclosed. According to Zhang et al., GAK knockdown caused EGFR upregulation in HeLa cells but did not cause Grant Support any change in either HER2 or IGF receptor (38). The same This project was supported by a grant from the Gaetagno and Wechsler group did not observe any change in PDGFR expression funds. Support has also been provided by the Kenneth Stanton Fund. Z. Duan is upon inhibition of GAK in NIH3T3 cells. Depletion of supported, in part, by a grant from the National Cancer Institute (Nanotechnol- GAK on various downstream signaling has been reported ogy Platform Partnership), R01-CA119617, and Sarcoma Foundation of America. E. Choy is supported by the Jennifer Hunter Yates Sarcoma Foundation. as well, with conflicting results. Zhang et al. (37) reported that transient depletion of GAK by siRNA in HeLa cells Received 07/07/2010; revised 09/08/2010; accepted 09/15/2010; caused defects in transferrin uptake and procathepsin D published 12/14/2010.

References 1. Ottaviani G, Jaffe N. The epidemiology of osteosarcoma. Cancer Treat 4. Kimura SH, Tsuruga H, Yabuta N, Endo Y, Nojima H. Structure, Res 2010;152:3–13. expression, and chromosomal localization of human GAK. Genomics 2. Demetri GD, von Mehren M, Blanke CD, Van Den Abbeele AD, 1997;44:179–87. Eisenberg B, Roberts PJ, et al. Efficacy and safety of imatinib mesylate 5. Ahle S, Ungewickell E. Auxilin, a newly identified clathrin-associated in advanced gastrointestinal stromal tumors. N Engl J Med protein in coated vesicles from bovine brain. J Cell Biol 1990;111: 2002;347:472–80. 19–29. 3. Kanaoka Y, Kimura SH, Okazaki I, Ikeda M, Nojima H. GAK: a cyclin 6. Greener T, Zhao X, Nojima H, Eisenberg E, Greene LE. Role of cyclin G associated kinase contains a tensin/auxilin-like domain. FEBS Lett G-associated kinase in uncoating clathrin-coated vesicles from non- 1997;402:73–80. neuronal cells. J Biol Chem 2000;275:1365–70.

www.aacrjournals.org Mol Cancer Ther; 9(12) December 2010 3349

Susa et al.

7. Umeda A, Meyerholz A, Ungewickell E. Identification of the universal 25. Schoffski P. Polo-like kinase (PLK) inhibitors in preclinical and early cofactor (auxilin 2) in clathrin coat dissociation. Eur J Cell Biol clinical development in oncology. Oncologist 2009;14:559–70. 2000;79:336–42. 26. Garland LL, Taylor C, Pilkington DL, Cohen JL, Von Hoff DD. A phase I 8. Gall WE, Higginbotham MA, Chen C, Ingram MF, Cyr DM, Graham TR. pharmacokinetic study of HMN-214, a novel oral stilbene derivative The auxilin-like phosphoprotein Swa2p is required for clathrin function with polo-like kinase-1-interacting properties, in patients with in yeast. Curr Biol 2000;10:1349–58. advanced solid tumors. Clin Cancer Res 2006;12:5182–9. 9. Pishvaee B, Costaguta G, Yeung BG, Ryazanstsein S, Greener T, 27. Jimeno A, Li J, Messersmith WA, Laheru D, Rudk MA, Maniar M, et al. Greene LE, et al. A yeast DNA J protein required for uncoating of Phase I study of ON 01910.Na, a novel modulator of the Polo-like clathrin-coated vesicles in vivo. Nat Cell Biol 2000;2:958–63. kinase 1 pathway, in adult patients with solid tumors. J Clin Oncol 10. Greener T, Grant B, Zhang Y, Wu X, Greene LE, Hirsh D, et al. 2008;26:5504–10. Caenorhabditis elegans auxilin: a J-domain protein essential for 28. Mross K, Frost A, Steinbild S, Hedborn S, Rentschler J, Kaiser R, et al. clathrin-mediated endocytosis in vivo. Nat Cell Biol 2001;3:215–9. Phase I dose escalation and pharmacokinetic study of BI 2536, a 11. Kappel CC, Velez-Yanguas MC, Hirschfeld S, Helman LJ. Human novel Polo-like kinase 1 inhibitor, in patients with advanced solid osteosarcoma cell lines are dependent on insulin-like growth factor I tumors. J Clin Oncol 2008;26:5511–7. for in vitro growth. Cancer Res 1994;54:2803–7. 29. Szakacs G, Paterson JK, Ludwig JA, Booth-Genthe C, Gottesman 12. MacEwen EG, Pastor J, Kutzke J, Tsan R, Kurzman ID, Thamm DH, MM. Targeting multidrug resistance in cancer. Nat Rev Drug Discov et al. IGF-1 receptor contributes to the malignant phenotype in human 2006;5:219–34. and canine osteosarcoma. J Cell Biochem 2004;92:77–91. 30. Kaye SB. Reversal of drug resistance in ovarian cancer: where do we 13. Duan Z, Choy E, Harmon D, Yang C, Ryu K, Schwab J, et al. Insulin- go from here?J Clin Oncol 2008;26:2616–8. like growth factor-I receptor tyrosine kinase inhibitor cyclolignan 31. Krishna R, Mayer LD. Multidrug resistance (MDR) in cancer. Mechan- picropodophyllin inhibits proliferation and induces apoptosis in multi- isms, reversal using modulators of MDR and the role of MDR mod- drug resistant osteosarcoma cell lines. Mol Cancer Ther 2009;8: ulators in influencing the pharmacokinetics of anticancer drugs. Eur J 2122–30. Pharm Sci 2000;11:265–83. 14. Pollak MN, Polychronakos C, Richard M. Insulinlike growth factor I: a 32. Sonneveld P, Suciu S, Weijermans P, Beksac M, Neuwirtova R, potent mitogen for human osteogenic sarcoma. J Natl Cancer Inst Solbu G, et al. Cyclosporin A combined with vincristine, doxorubicin 1990;82:301–5. and dexamethasone (VAD) comparedwithVADaloneinpatients 15. Burrow S, Andrulis IL, Pollak M, Bell RS. Expression of insulin-like with advanced refractory multiple myeloma: an EORTC-HOVON growth factor receptor, IGF-1, and IGF-2 in primary and metastatic randomized phase III study (06914). Br J Haematol 2001;115: osteosarcoma. J Surg Oncol 1998;69:21–7. 895–902. 16. Hughes DP, Thomas DG, Giordano TJ, Baker LH, McDonagh KT. Cell 33. Korolchuk VI, Banting G. CK2 and GAK/auxilin2 are major protein surface expression of epidermal growth factor receptor and Her-2 kinases in clathrin-coated vesicles. Traffic 2002;3:428–39. with nuclear expression of Her-4 in primary osteosarcoma. Cancer 34. Newmyer SL, Christensen A, Sever S. Auxilin-dynamin interactions Res 2004;64:2047–53. link the uncoating ATPase chaperone machinery with vesicle forma- 17. Wen YH, Koeppen H, Garcia R, Chiriboga L, Tarlow BD, Peters BA, tion. Dev Cell 2003;4:929–40. et al. Epidermal growth factor receptor in osteosarcoma: expression 35. Jiang R, Gao B, Prasad K, Greene LE, Eisenberg E. Hsc70 chaperones and mutational analysis. Hum Pathol 2007;38:1184–91. clathrin and primes it to interact with vesicle membranes. J Biol Chem 18. Oda Y, Wehrmann B, Radig K, Walter H, Rose I, Neumann W, et al. 2000;275:8439–47. Expression of growth factors and their receptors in human osteosar- 36. Yim YI, Scarselletta S, Zang F, Wu X, Lee DW, Kang YS, et al. comas. Immunohistochemical detection of epidermal growth factor, Exchange of clathrin, AP2 and epsin on clathrin-coated pits in per- platelet-derived growth factor and their receptors: its correlation with meabilized tissue culture cells. J Cell Sci 2005;118:2405–13. proliferating activities and p53 expression. Gen Diagn Pathol 37. Zhang CX, Engqvist-Goldstein AE, Carreno S, Owen DJ, Smythe E, 1995;141:97–103. Drubin DG. Multiple roles for cyclin G-associated kinase in clathrin- 19. Freeman SS, Allen SW, Ganti R, Wu J, Ma J, Su X, et al. Copy number mediated sorting events. Traffic 2005;6:1103–13. gains in EGFR and copy number losses in PTEN are common events 38. Zhang L, Gjoerup O, Roberts TM. The serine/threonine kinase cyclin in osteosarcoma tumors. Cancer 2008;113:1453–61. G-associated kinase regulates epidermal growth factor receptor sig- 20. Duan Z, Ji D, Weinstein EJ, Susa M, Choy E, Yang C, et al. Lentiviral naling. Proc Natl Acad Sci U S A 2004;101:10296–301. shRNA screen of human kinases identifies PLK1 as a potential 39. Tiedemann RE, Zhu YX, Schmidt J, Yin H, Shi CX, Que Q, et al. therapeutic target for osteosarcoma. Cancer Lett 2010;293:220–9. Kinome-wide RNAi studies in human multiple myeloma identify vul- 21. Susa M, Iyer AK, Ryu K, Choy E, Hornicek FJ, Mankin H, et al. nerable kinase targets, including a lymphoid-restricted kinase, GRK6. Inhibition of ABCB1 (MDR1) expression by an siRNA nanoparticulate Blood 2010;115:1594–604. delivery system to overcome drug resistance in osteosarcoma. PLoS 40. Ray MR, Wafa LA, Cheng H, Snoek R, Fazli L, Gleave M, et al. Cyclin One 2010;5(5):e10764. G-associated kinase: a novel androgen receptor-interacting transcrip- 22. Yang C, Ji D, Weinstein EJ, Hornicek FJ, Wood KB, Liu X, et al. The tional coactivator that is overexpressed in hormone refractory pros- kinase Mirk is a potential therapeutic target in osteosarcoma. Carci- tate cancer. Int J Cancer 2006;118:1108–19. nogenesis 2010;31:552–8. 41. Vieira AV, Lamaze C, Schmid SL. Control of EGF receptor signaling by 23. Yague E, Arance A, Kubitza L, O’Hare M, Jat P, Ogilvie CM, et al. clathrin-mediated endocytosis. Science 1996;274:2086–9. Ability to acquire drug resistance arises early during the tumorigenesis 42. Bai T, Seebald JL, Kim KE, Ding HM, Szeto DP, Chang HC. Dis- process. Cancer Res 2007;67:1130–7. ruption of zebrafish cyclin G-associated kinase (GAK) function 24. Susa M, Iyer AK, Ryu K, Hornicek FJ, Mankin H, Amiji MM, et al. impairs the expression of Notch-dependent genes during neuro- Doxorubicin loaded polymeric nanoparticulate delivery system to genesis and causes defects in neuronal development. BMC Dev overcome drug resistance in osteosarcoma. BMC Cancer 2009;9:399. Biol 2010;10:7.

3350 Mol Cancer Ther; 9(12) December 2010 Molecular Cancer Therapeutics

Cyclin G−Associated Kinase Is Necessary for Osteosarcoma Cell Proliferation and Receptor Trafficking

Michiro Susa, Edwin Choy, Xianzhe Liu, et al.

Mol Cancer Ther 2010;9:3342-3350. Published OnlineFirst September 23, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-10-0637

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2010/09/23/1535-7163.MCT-10-0637.DC1

Cited articles This article cites 42 articles, 16 of which you can access for free at: http://mct.aacrjournals.org/content/9/12/3342.full#ref-list-1

Citing articles This article has been cited by 9 HighWire-hosted articles. Access the articles at: http://mct.aacrjournals.org/content/9/12/3342.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/9/12/3342. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.