Dynamic and Nuclear Expression of Pdgfra and IGF-1R in Alveolar Rhabdomyosarcoma

Published OnlineFirst August 8, 2013; DOI: 10.1158/1541-7786.MCR-12-0598

Molecular Cancer Cell Cycle and Senescence Research

Dynamic and Nuclear Expression of PDGFRa and IGF-1R in Alveolar Rhabdomyosarcoma

M. Imran Aslam1,2,12, Simone Hettmer7,8,9,10,12, Jinu Abraham2, Dorian LaTocha1, Anuradha Soundararajan5, Elaine T. Huang2, Martin W. Goros6, Joel E. Michalek10, Shuyu Wang11, Atiya Mansoor3, Brian J. Druker1,4, Amy J. Wagers7,8,9,12, Jeffrey W. Tyner1, and Charles Keller2

Abstract Since the advent of tyrosine kinase inhibitors as targeted therapies in cancer, several receptor tyrosine kinases (RTK) have been identified as operationally important for disease progression. Rhabdomyosarcoma (RMS) is a malignancy in need of new treatment options; therefore, better understanding of the heterogeneity of RTKs would advance this goal. Here, alveolar RMS (aRMS) tumor cells derived from a transgenic mouse model expressing two such RTKs, platelet-derived growth factor (PDGFR)a and insulin-like growth factor (IGF)-1R, were investigated by fluorescence-activated cell sorting (FACS). Sorted subpopulations that were positive or negative for PDGFRa and IGF-1R dynamically altered their cell surface RTK expression profiles as early as the first cell division. Interestingly, a difference in total PDGFRa expression and nuclear IGF-1R expression was conserved in populations. Nuclear IGF-1R expression was greater than cytoplasmic IGF-1R in cells with initially high cell surface IGF-1R, and cells with high nuclear IGF-1R established tumors more efficiently in vivo. RNA interference– mediated silencing of IGF-1R in the subpopulation of cells initially harboring higher cell surface and total IGF-1R resulted in significantly reduced anchorage-independent colony formation as compared with cells with initially lower cell surface and total IGF-1R expression. Finally, in accordance with the findings observed in murine aRMS, human aRMS also had robust expression of nuclear IGF-1R.

Implications: RTK expression status and subcellular localization dynamics are important considerations for personalized medicine. Mol Cancer Res; 11(11); 1303–13. 2013 AACR.

Introduction muscle cancer, patients with alveolar rhabdomyosarcoma have the most dismal outcome despite maximal therapeutic inter- Rhabdomyosarcoma is the most frequent soft tissue sarcoma – of childhood. Of the common subtypes for this aggressive vention (1 3). We have developed and validated a genetically engineered, conditional mouse model of alveolar rhabdomyosarcoma (4, 5) and therein identified the receptor tyrosine Authors' Affiliations: 1Division of Hematology and Medical Oncology, kinase (RTK) targets platelet-derived growth factor receptor-a Knight Cancer Institute; 2Department of Pediatrics, Pediatric Cancer Biology a Program, Pape Family Pediatric Research Institute; 3Department of Pathol- (PDGFR ) and insulin-like growth factor-1 receptor (IGF- ogy, Oregon Health & Science University; 4Howard Hughes Medical Institute, 1R) as potential therapeutic targets (6, 7). In agreement with Portland, Oregon; 5Greehey Children's Cancer Research Institute; and our mouse model studies, careful work by other groups has 6Department of Epidemiology & Biostatistics, University of Texas Health fi Science Center, San Antonio, Texas; 7Department of Stem Cell and Regen- ascribed biologic and clinical signi cance to the expression of erative Biology, Harvard University; 8Harvard Stem Cell Institute; 9Joslin these and other RTKs for alveolar rhabdomyosarcoma (8, 9). 10 Diabetes Center, Cambridge; Division of Pediatric Hematology/Oncology, In the current era of biomarker-stratified and personalized Department of Pediatric Oncology, Dana-Farber Cancer Institute, Children's Hospital; and 11Harvard University, Cambridge, MA; 12Howard Hughes cancer therapy clinical trials, great importance is being Medical Institute ascribed to biomarker expression from the solitary biopsy Note: Supplementary data for this article are available at Molecular sample taken from each patient at a single time point (10). Cancer Research Online (http://mcr.aacrjournals.org/). However, intra and intertumor heterogeneity for individual

M.I. Aslam and S. Hettmer contributed equally to this work. patients is increasingly recognized (11). Several models developed to interpret the heterogeneity of individual cells that Current address for M.I. Aslam: Howard Hughes Medical Institute Medical compose a tumor have been proposed, with strong evidence Research Fellows Program, 4000 Jones Bridge Rd., Chevy Chase, MD 20815 that solid tumors consist of subsets of cells distinct at the Corresponding Authors: Charles Keller, Oregon Health & Science Uni- genetic and phenotypic level (11–15). However, the concept versity, 3181 SW Sam Jackson Park Rd., MC-L321, Portland, OR 97239. Phone: 503-494-1210; Fax: 503-418-5044; E-mail: [email protected]; of isolating unique populations of cells by expression of Jeffrey W. Tyner, E-mail: [email protected]; and Amy J. Wagers, E-mail: surface membrane proteins lends itself toward controversy [email protected] because cells reverting to a state reﬂecting the heterogeneity of doi: 10.1158/1541-7786.MCR-12-0598 the original tumor have been described as well (16). Thus, the 2013 American Association for Cancer Research. ability to proﬁle cell populations using cell surface markers,

www.aacrjournals.org 1303

Aslam et al.

particularly RTKs expressed by only a subpopulation of ly conducted and proteins visualized with chemiluminescent tumor cells, might be informative if RTK expression is stable. signal using Super Signal Western Pico or Dura Substrate Otherwise, understanding mechanisms that dynamically (Pierce Biotechnology). Mouse tumors were washed in PBS, modulate subcellular localization of RTKs is necessary and then lysed in radioimmunoprecipitation assay buffer (17–23). In this regard, the mechanism of intracellular supplemented with protease and phosphatase inhibitors. (nuclear) trafficking of RTKs has been described with most Lysates were homogenized and centrifuged at maximum details for the EGF receptor (EGFR; refs. 18, 20, 21). Other speed for 10 minutes at 4C and supernatant used for RTKs, such as FGFR, IGF-1R, and ErbB4 have also been Western blotting. Antibodies used for immunoblotting shown to localize to the cell nucleus in cancer (17, 19, 22). were PDGFRa (1:1000, #3164, Cell Signaling Technolo- Because PDGFRa and IGF-1R are reported to act in gies), IGF-1Rb (1:1000, sc-713, Santa Cruz Biotechnolo- concert for resistance to IGF-1R small-molecule inhibition gy), p-IGF-1Rb recognizing phospho-Tyr 1161 (1:1000, sc- in rhabdomyosarcoma (24), we sought to understand wheth- 101703, Santa Cruz Biotechnology), phospho and total er tumor cell subpopulations can be subdivided into distinct p42/44 MAP K from Cell Signaling Technology (used at RTK profiles and whether isolated cells behave as pheno- 1:1000), b-actin (1:10,000, Sigma), a-tubulin (1:10,000, typically static or dynamic populations. Furthermore, we Sigma), and Sp1 (1:1000, Millipore) and pan-Tyr antibody aimed to determine whether differential subcellular locali- (1:10,000, 4G-10, Upstate Biotechnology). Immunopre- zation of these RTKs had a biologically significant role in cipitation was conducted by standard protocol using Protein tumor establishment. A beads for rabbit immunoglobulin G (IgG; Invitrogen).

Cytosolic and nuclear fractionation Materials and Methods The nuclear fraction of cells was isolated by the following Cell culture protocol: cells were lifted from a 100-mm culture dish and Mouse tumor cell lines were established and cultured washed with PBS once. Cell pellet was resuspended in 600 previously described (7). Cells were grown in Dulbecco's mL buffer 1 (25 mmol/L HEPES, pH 7.9, 5 mmol/L KCl, fi modi ed Eagle medium (DMEM) supplemented with 10% 0.5 mmol/L MgCl2, 1 mmol/L dithiothreitol, 100 mmol/L FBS and penicillin (100 U/mL)/streptomycin (100 mg/mL; phenylmethylsulfonylfluoride in H O), then 600 mLof 2 Invitrogen) in 5% CO2 in air at 37 C. buffer 2 was added (buffer 1 þ 1% Nonidet-P40), and then centrifuged at 2,600 rpm for 5 minutes. The supernatant Flow cytometry (cytoplasmic fraction) was transferred to a new tube. Hun- Murine alveolar rhabdomyosarcoma cells were grown to dred microliter mL of buffer 3 (1:1 mix of buffer 1 and 2) was approximately 50% confluency then trypsinized, washed added to the cell pellet and gently mixed, centrifuged at with PBS, and stained with primary APC-conjugated anti- 2,600 rpm, and supernatant added to cytoplasmic fraction. body to PDGFRa (1:40, eBiosciences) in fluorescence-acti- Cell pellet was then washed two to three times with buffer 1 vated cell sorting (FACS) buffer (2% FBS in PBS) on ice for and supernatant discarded for purity of subsequent soluble 30 minutes, washed in FACS buffer, then stained with IGF- nuclear protein extraction. Five hundred microliter of buffer 1R (1:100, ab32823, Abcam) primary antibody for 30 4 (25 mmol/L HEPES, pH 7.9, 360 mmol/L NaCl, 10% minutes on ice. Cells were again washed in FACS buffer, dextrose, 0.05% Nonidet-P40) was added to cell pellet and then stained with anti-chicken secondary antibody conjugat- tube rotated for 1 hour at 4C. Tube was centrifuged at ed to Cy3.5 (1:80, ab97146, Abcam) for 30 minutes on ice. 13,200 rpm for 10 minutes and supernatant (soluble nuclear Cells were washed in FACS buffer and resuspended in buffer fraction) was transferred to a new tube. At this point, both and analyzed by a BD FACSAria III Cell Sorter (BD Bio- the cytoplasmic and nuclear fractions were clarified again by sciences). As a control, cells were stained with isotype centrifuged at 13,200 rpm and transferring supernatant to a antibody conjugated to APC at the same concentration as new tube. Protein was quantified using BCA Protein Assay PDGFRa antibody (#17-1401-81, eBiosciences) and subse- Reagent (Pierce) and immunoblotting was conducted as quently with secondary antibody conjugated to Cy3.5 only, at mentioned above. the aforementioned concentration. Stained cells were gated according to control and sorted by quadrant for PDGFRa RNA interference mediated silencing of IGF-1R and and IGF-1R expression. Once cells were sorted, reanalysis by anchorage-independent colony formation assay FACS was done to assess purity of sorted populations before For RNA interference (RNAi) studies, siRNA targeted to cells were recounted and allowed to recover in culture before IGF-1R (#SI01074017) and nonspecific siRNA (#1027280) further experiments were carried out. Analysis of .fcs files was were purchased from Qiagen. Experiments with siRNA were done with FlowJo (Tree Star, Inc., PC version 7.6). carried out at 100 nmol/L concentration out using Lipofec- tamine 2000 in Opti-MEM Reduced Serum Media (Invitro- Immunoblotting and immunoprecipitation gen). Cells were transfected with nonspecific siRNA or Cell lysates were prepared using Cell Lysis Buffer (#9803, siRNA-targeting IGF-1R and on day 2, trypsinized and in Cell Signaling Technology) supplemented with the appro- each well of a 6-well plate, 6,000 cells were suspended in 1.5 priate protease and phosphatase inhibitors (Sigma-Aldrich), mL of 0.7% agarose (at 37C) in DMEM with 10% fetal calf and standard immunoblotting procedures were subsequent- serum. The cells in agarose were plated atop a 1.4% agarose

1304 Mol Cancer Res; 11(11) November 2013 Molecular Cancer Research

Nuclear Receptor Tyrosine Kinases in Rhabdomyosarcoma

layer and were allowed to grow for 2 weeks before visualizing Results the colonies by light microscopy. Murine alveolar rhabdomyosarcomas express PDGFRa and IGF-1R by FACS and can be isolated by cell sorting Immunohistochemistry The murine alveolar rhabdomyosarcoma primary cell Human alveolar rhabdomyosarcoma tissue microarrays culture, U23674, was derived from a left upper extremity were obtained from the COG Biorepository (TMA #3000- primary tumor that developed in a 46-day-old mouse from 30-P6897). Histology was conducted using standard pro- our genetically engineered, conditional mouse model of tocol. Briefly, formalin-fixed paraffin-embedded sections alveolar rhabdomyosarcoma harboring the Pax3:Fkhr onco- (FFPE) were dewaxed and dehydrated in xylene and graded gene and homozygous p53 deletion (4, 5). This lesion was alcohol concentrations. Heat-induced epitope retrieval was histologically diagnosed as an alveolar rhabdomyosarcoma used with sodium citrate buffer (pH 6.0) for 20 minutes, tumor and used for these studies at low passage number (P < slides blocked with 2% normal goat serum (NGS) and for 10). Initially, U23674 cells were stained for PDGFRa (APC) endogenous biotin using an avidin/biotin blocking kit (Vec- and IGF-1R (Cy3.5) and analyzed by FACS, indicating that tor Labs). Primary antibody for IGF-1R (Santa Cruz) was approximately 35% of cells express both PDGFRa and IGF- used at 1:50 overnight in 2% NGS. VECTASTAIN Elite 1R, but approximately 15% of cells express either receptor ABC kit (Rabbit IgG) and ImmPACT DAB Peroxidase exclusively (Fig. 1A). However, tumor cells displayed a Substrate kit was used according to manufacturer's protocol continuum of expression of PDGFRa and IGF-1R rather to visualize staining (Vector Lab). Slides were counterstained than neatly distinct populations expressing either RTK with hematoxylin for 5 minutes, rinsed, dehydrated, and individually or together. Similar results were seen in two mounted with xylene-based mounting medium. other murine alveolar rhabdomyosarcoma cell lines analyzed by flow cytometry for cell surface expression of PDGFRa Immunocytochemistry and IGF-1R (Supplementary Fig. S1). Confocal microscopy was conducted as follows: cells were Next, after U23674 cells were stained for PDGFRa and allowed to recover postsort in culture for 2 to 3 days then IGF-1R, the cells were sorted by expression of each receptor replated in 8 chamber slides (BD Biosciences) at low con- into four populations. Gates were determined using control fluency and allowed to incubate for 2 to 3 days. Cells were cells stained with isotype control antibody in addition to then fixed in 4% paraformaldehyde in PBS at room tem- a Cy3.5 conjugated secondary antibody as a control. perature for 20 minutes. Slides were carefully washed with Cells positive or negative for PDGFRa or IGF-1R by the hi hi PBS three times and incubated at room temperature in 5% assigned gates were designated as PDGFRa /IGF-1R or lo lo NGS (Invitrogen) and 0.01% Triton-X in PBS for 1 hour to PDGFRa /IGF-1R , respectively (Fig. 1B). For instance, inhibit nonspecific binding of antibodies. Cells were washed cells sorted for high IGF-1R expression, but negative for lo hi again, and primary antibody diluted in 5% NGS in PBS was PDGFRa were denoted as PDGFRa IGF-1R . Immedi- added overnight at 4C. Alexafluor 488-conjugated or ately following sorting, cells were reanalyzed by flow cyto- Alexafluor 594-conjugated anti-rabbit IgG antibody (Invi- metry to assess purity of the sort (Fig. 1B). Figure 1C shows trogen) at 1:200 was added, and cells were incubated for 1 distinct morphology of cells grown in culture for 7 days hi hi hour at room temperature. Slides were mounted using the postsort. IGF-1R expression, but not PDGFRa expres- VECTASHIELD Mounting Medium with 40, 6-diamidino- sion, was associated with spindle morphology. 2-phenylindole (Vector Laboratories) and visualized with a Zeiss LSM 700 confocal microscope. PDGFRa antibody PDGFRa and IGF-1R cell surface expression is dynamic was used at 1:100 (#3164, Cell Signaling Technologies) and Once U23674 cells were sorted by PDGFRa and IGF- IGF-1Rb at 1:50 (sc-713, Santa Cruz Biotechnology). 1R expression, cells were allowed to recover postsort in Images were taken at 40 and 63 oil immersion for culture and reanalyzed by flow cytometry for surface further magnification. expression of both RTKs at serial time points. FACS analysis revealed that cell surface expression of both In vivo studies PDGFRa and IGF-1R varied dramatically as cells were All animals were treated humanely and the experiments grown in culture postsort. Two such time points are were carried out in accordance with the Institutional Animal shown in Fig. 2 at 2 days and 14 days postsort, compared Care and Use Committee approved protocols. SCID/hair- with day 0 (day 7 was similar to day 14; Supplementary scid hr less/outbred (Crl:SHO-Prkdc Hr ) mice purchased from Fig. S2). These results illustrated an initial dramatic Charles River at 8 weeks of age were injected with 1 105 change in expression of PDGFRa and IGF-1R postsort U23674 cells into the gastrocnemius muscle that had been within a 48-hour time period, tending toward a presort preinjured 24 hours prior with 0.85 mg/mouse cardiotoxin profile. With the observation that cell surface expression (from Naja mossambica, Sigma). Tumor volumes (cm3) were profile shifted considerably within a short period of time, measured three dimensionally with electronic calipers and we sought to understand whether altered receptor expres- calculated from formula (p/6) length width height, sion was a function of many or few rounds of replication. assuming tumors to be spheroid. Once tumor volume Thus, after a purity check was conducted for sorted reached 2 cm3, mice were euthanized and the tumor was populations postsort, cells were recounted and plated at harvested at necropsy. a known cell number. Exactly 48 hours after the initial

www.aacrjournals.org Mol Cancer Res; 11(11) November 2013 1305

Aslam et al.

Figure 1. Murine alveolar rhabdomyosarcoma isolated by PDGFRa and IGF-1R expression. A, U23674 (murine alveolar rhabdomyosarcoma) cell expression proﬁle by FACS of PDGFRa (APC) and IGF-1R (Cy3.5) gated with respect to isotype and secondary only control. B, purity of isolated cells by FACS shown. Isolated populations according to cells positive and negative for each receptor designated as hi and lo, respectively. C, phase contrast microscopy of sorted populations showing differential morphology of cells grown in culture 7 days postsort. Scale bar, 200 mm.

sort, cells were recounted and FACS analysis for PDGFRa Total expression of PDGFRa and IGF-1R by and IGF-1R expression was done. The number of cells immunoblot analysis in sorted cells differs from cell comprising each of the four populations was calculated surface expression profile by FACS from the FACS profile of the initial sort and 48-hour To determine whether total expression of either PDGFRa postsort (Fig. 2). Doubling times were on the order of or IGF-1R in the four sorted cell populations reflected the 30 to 42 hours (Supplementary Table S1); therefore, dy- surface expression profile of each RTK by FACS, we con- namic alterations of PDGFRa and/or IGF-1R cell surface ducted immunoblot analysis on cells grown in culture for 7 expression seem to start after 0 to 1 cell divisions. days postsort. Surprisingly, total cellular PDGFRa expression

1306 Mol Cancer Res; 11(11) November 2013 Molecular Cancer Research

Nuclear Receptor Tyrosine Kinases in Rhabdomyosarcoma

Figure 2. Murine alveolar rhabdomyosarcoma show dynamic expression of PDGFRa and IGF-1R by FACS in a temporal manner. A, FACS proﬁle of PDGFRa and IGF-1R expression in U23674 cells at day of sort (Day 0), and 2 and 14 days postsort. Day 7 was similar to Day 14.

seemed to vary dramatically among the sorted populations and IGF-1R such as Akt and extracellular signal-regulated (Fig. 3A). Higher PDGFRa expression at day 7 correlated kinase (Erk)-1/2, respectively. Although we did not see a with the populations, which were sorted for high PDGFRa difference in Akt phosphorylation levels between the sorted expression by FACS initially. Conversely, total IGF-1R cells (data not shown), Erk1/2 phosphorylation levels were hi hi expression by immunoblot analysis did not show a dramatic remarkably elevated in the PDGFRa IGF-1R population lo lo difference in expression among sorted populations (Fig. 3A). and conversely decreased in the PDGFRa IGF-1R popu- In terms of activation, phosphorylation of PDGFRa was lation; p-Erk 1/2 levels seemed to correlate with the levels of hi lo hi elevated in the PDGFRa IGF-1R and PDGFRa IGF- phosphorylated PDGFRa and IGF-1R. hi 1R populations (Fig. 3A). Despite lack of differential Thus far, our results indicated that PDGFRa and IGF-1R expression for total IGF-1R, IGF-1R phosphorylation was in tumor cells are expressed in a dynamic manner with hi hi increased in the PDGFRa IGF-1R populations, as well as respect to surface expression proﬁle by FACS, as well as total lo hi (modestly) in the PDGFRa IGF-1R population. Because protein expression by immunoblot analysis. However, the we observed that activated receptor levels varied across sorted dramatic difference in total PDGFRa expression (Fig. 3A), populations, we looked at activation of common cell survival for example, did not reﬂect the percentage of cells positive for and proliferation pathways downstream of both PDGFRa PDGFRa by FACS analysis at the same time point (day

www.aacrjournals.org Mol Cancer Res; 11(11) November 2013 1307

Aslam et al.

Figure 3. Functional PDGFRa and IGF-1R expression postsort. A, immunoblot analysis of sorted populations grown in culture for 1 week postsort, showing phosphorylated, as well as total PDGFRa, IGF-1R, and Erk 1/2. B, representative data showing MFI of U23674 sorted cells 7 days postsort indicating surface expression of PDGFRa and IGF-1R.

7; Fig. 2). Therefore, to determine whether the amount of rhabdomyosarcoma cells established tumors with the same lo lo PDGFRa or IGF-1R expressed on the surface membrane of latency as PDGFRa IGF-1R -sorted cells (Fig. 4D). hi the sorted cells was reflective of total expression of each To confirm the results that IGF-1R -sorted cells establish lo receptor, we looked at mean fluorescence intensity (MFI) of tumors in vivo earlier than IGF-1R cells, we sorted the cells stained for PDGFRa and IGF-1R at day 7. Figure 3B independent murine alveolar rhabdomyosarcoma primary hi lo shows MFI of both RTKs in each of the four sorted popula- cell culture U21459 into IGF-1R and IGF-1R cells and tions. The MFI values indicated that cells sorted initially for implanted the cells immediately postsort in recipient non- PDGFRa or IGF-1R displayed a higher cell surface expres- obese diabetic/severe combined immunodeficient (NOD/ lo hi sion of each receptor 1 week postsort. Taken together, these SCID) host mice. Compared with IGF-1R cells, IGF-1R results showed that surface expression did not completely cells sorted from U21459 engrafted with decreased latency correlate with total expression level of either receptor, indi- and led to shorter survival (Supplementary Fig. S3A–S3C). cating, among other possibilities, differential subcellular localization of these RTKs. PDGFRa and IGF-1R show nuclear localization in hi hi PDGFRa and IGF-1R -sorted cells and nuclear Total PDGFRa and IGF-1R expression in vivo reflects in IGF-1R is associated with increased anchorage- hi vitro behavior of tumor cells and IGF-1R -sorted cells independent colony formation ability lo establish tumors in vivo earlier than IGF-1R cells On the basis of the observation that cells sorted for either To determine whether tumor cells retained an expression PDGFRa or IGF-1R exhibited an apparent discrepancy profile of PDGFRa and IGF-1R in vivo reflective of in vitro between cell surface and total expression of each receptor, we behavior(Fig.3A),weusedanorthotopicallograftin vivo hypothesized that differential cellular localization of PDGFRa model of alveolar rhabdomyosarcoma and implanted cells and IGF-1R was the etiology for these disparate results. We immediately postsort into the right gastrocnemius muscle of first conducted confocal microscopy on sorted cells looking at immunocompromised mice. Immunoblot analysis of tumors PDGFRa and IGF-1R localization. Interestingly, we observed hi hi hi lo formed in mice showed that PDGFRa and IGF-1R expression that in PDGFRa IGF-1R and PDGFRa IGF-1R cells, of subpopulations (Fig. 4A) mirrored the expression of both PDGFRa was predominantly localized to the nucleus, while lo hi receptors in vitro (Fig. 3A). Leg measurements taken daily to chiefly cytosolic or membranous in PDGFRa IGF-1R and lo hi lo lo monitor tumor growth revealed that the PDGFRa IGF-1R PDGFRa IGF-1R cells (Fig. 5 and Supplementary Fig. S3). hi hi lo hi hi hi and PDGFRa IGF-1R -sorted cells more quickly established Similarly, in PDGFRa IGF-1R and PDGFRa IGF-1R hi lo tumors in mice compared with PDGFRa IGF-1R and cells, IGF-1R seemed to be principally localized to the nucleus, lo lo PDGFRa IGF-1R cells (Fig. 4B and C). The control group whereas more abundant in the cytosolic compartment/cell hi lo lo lo of mice implanted with unsorted U23674 murine alveolar membrane of PDGFRa IGF-1R and PDGFRa IGF-1R

1308 Mol Cancer Res; 11(11) November 2013 Molecular Cancer Research

Nuclear Receptor Tyrosine Kinases in Rhabdomyosarcoma

Figure 4. Cells sorted for IGF-1R establish more aggressive tumors in vivo. A, tumors established from murine aRMS cells sorted for PDGFRa and IGF-1R show a relative dynamic and static expression of each receptor, respectively, reflective of cells in hi lo vitro. B, PDGFRa IGF-1R -sorted cells show a significantly slower growth rate in an in vivo orthotopic allograft mouse model of alveolar rhabdomyosarcoma compared lo hi with PDGFRa IGF-1R -sorted hi hi cells. C, PDGFRa IGF-1R -sorted cells grow at a significantly faster growth rate in an in vivo orthotopic allograft mouse model of alveolar rhabdomyosarcoma compared lo lo with PDGFRa IGF-1R -sorted cells. D, tumor establishment or growth rate of unsorted U23674 cells in vivo do not differ from lo lo PDGFRa IGF-1R -sorted cells.

cells (Fig. 5 and Supplementary Fig. S4). In the biologically assessed the effect of IGF-1R knockdown on these cells' ability independent U21459 cell culture, immunocytochemistry also to form colonies in an anchorage-independent colony forma- revealed that IGF-1R was commonly localized to the nucleus tion assay (Fig. 6B). Knockdown efficiencies of IGF-1R in both hi lo in IGF-1R cells and IGF-1R cells, respectively (Supplemen- sorted cells are shown (Fig. 6B). Colony formation ability of hi hi tary Fig. S3D). These results subsequently led us to question PDGFRa IGF-1R was decreased approximately 45% com- hi whether the ability of IGF-1R cells to establish tumors earlier pared with control (nonspecificsiRNA)whenIGF-1R was lo lo lo in vivo compared with IGF-1R cells was due to higher nuclear silenced by RNAi, whereas PDGFRa IGF-1R cells show expression of IGF-1R in these cells. Focusing our efforts on only a approximately 27% reduction in colony formation ability IGF-1R because of the orthotopic allograft results, we next with IGF-1R knockdown. Similarly, studies with the nuclear confirmed whether IGF-1R was indeed expressed at higher IGF-1R-expressing murine alveolar rhabdomyosarcoma prima- hi levels in the nucleus in IGF-1R cells, compared with IGF- ry cell culture U21459 showed that three independent siRNA lo 1R cells, by separating cytosolic and nuclear proteins of all for IGF-1R suppressed colony formation in soft agar (Supple- four sorted populations 7 days following isolation of cells by mentary Fig. S3E). Taken together, these results indicate that FACS. We subsequently immunoblotted for IGF-1R, using nuclear expression of IGF-1R confers an increased tumorigenic Sp1 and a-tubulin as confirmation of enrichment of nuclear phenotype to murine alveolar rhabdomyosarcoma cells. and cytosolic cell fractions, respectively (Fig. 6A). These results hi confirmed a higher nuclear expression of IGF-1R in IGF-1R Normal human tissue expresses cytosolic IGF-1R, lo cells compared with IGF-1R cells. whereas human alveolar rhabdomyosarcoma has high To confirm our hypothesis that nuclear expression of IGF- frequency of nuclear IGF-1R by immunohistochemistry 1R contributed to the ability of tumor cells to establish more To determine whether nuclear IGF-1R occurs in aggressive tumors in vivo,weusedRNAitosilenceIGF-1R in human alveolar rhabdomyosarcoma, we conducted immu- hi hi lo lo PDGFRa IGF-1R and PDGFRa IGF-1R cells. We then nohistochemistry for IGF-1R on a human alveolar

www.aacrjournals.org Mol Cancer Res; 11(11) November 2013 1309

Aslam et al.

Figure 5. Nuclear localization of PDGFRa and IGF-1R correlates with cells sorted by FACS for surface expression of the respective receptor. Representative confocal images of U23674 sorted populations showing localization of PDGFRa and IGF-1R (40) with inset box enlarging image of a single cell to appreciate localization. Scale bar, 100 mm. Larger magniﬁcations of insets are given in Supplementary Fig. S3.

rhabdomyosarcoma tumor microarray with 31 sections of understanding the complexity and potential pitfalls of 19 unique cases. We observed that 16 out of 19 (84%) assigning therapy based on the surface expression profiles specimens had nuclear expression of IGF-1R (>10% nuclei or other static biomarkers. Previously, several studies have of section positive). Tumor sections showed either cytosolic used cell surface markers (e.g., CD133, CD44, CD24, etc.) (Fig. 7A), or predominantly nuclear staining of IGF-1R as stable markers for a specific cell population (12, 15, 26). (Fig. 7B). Normal human tissue (pancreatic and skeletal RTKs, which are known to undergo ligand-dependent muscle) was used as positive controls for IGF-1R, and subcellular trafficking and recycling (18, 27, 28), may showed only cytosolic staining of IGF-1R; no nuclear however, be particularly undesirable as cell surface biomar- IGF-1R was seen in these sections (data not shown). Details kers. Nevertheless, our results then pointed us to differences of sample scores are given in Supplementary Table S2. in PDGFRa and IGF-1R subcellular localization that par- alleled membrane expression. IGF-1R nuclear localization, Discussion in particular, exhibited an important functional effect in In this report, we describe that the RTK PDGFRa and contributing toward a more aggressive tumor phenotype in IGF-1R exhibit a dynamic cell surface expression profile in vivo (summarized in Supplementary Fig. S5). murine alveolar rhabdomyosarcoma cells. Although we were Nuclear localization of RTKs or their substrates is an able to isolate relatively pure populations of cells expressing interesting phenomenon, which has been described in select high or low PDGFRa and/or IGF-1R, all sorted populations early reports (19–22, 29). IGF-1R had previously been rapidly and dynamically alter their cell surface expression shown to be a poor prognostic indicator in clear cell renal profile before the second cell division. Kept isolated in cell carcinoma (30), and a new study was able to further show culture, sorted populations eventually equilibrated their cell that among renal tumors expressing IGF-1R, intense or surface expression profile toward a presort profile. In the widespread nuclear expression by immunohistochemistry context of previous studies illustrating the phenotypically portends a decrease in survival probability (17). Interesting- dynamic state of cancer cells, including stem and non-stem ly, nuclear localization of the IGF-1R holoreceptor was cancer cells (16, 25), our study is a key next step into dependent upon tyrosine phosphorylation of putative

1310 Mol Cancer Res; 11(11) November 2013 Molecular Cancer Research

Nuclear Receptor Tyrosine Kinases in Rhabdomyosarcoma

Figure 6. IGF-1R is present in soluble nuclear fraction of sorted cells and is associated with anchorage-independent colony formation ability. A, cytosolic and nuclear fractions of sorted U23674 cells show that cells hi sorted by IGF-1R express higher levels of nuclear IGF-1R. B, cells with higher nuclear IGF-1R show a significantly further decrease in colony formation ability compared with control (nonspecific siRNA) when IGF-1R is silenced by siRNA. , P < 0.05 compared with nonspecific siRNA by Student t test.

residues on the IGF-1Rb subunit of the receptor (31). In our RTK is stimulated by its respective ligand is receptor hi sorted populations, the IGF-1R cells, which displayed a internalization and subsequent degradation (31). However, higher nuclear expression of IGF-1R, also had increased phosphorylation of IGF-1Rb is one of the two necessary phosphor-IGF-1Rb. A typical biologic response when an posttranslation modiﬁcations of the receptor required for

Figure 7. IGF-1R localization in human alveolar rhabdomyosarcoma (arms). A, human alveolar rhabdomyosarcoma sections with chieﬂy cytosolic IGF-1R expression. B, images of alveolar rhabdomyosarcoma sections showing both cytosolic (left) and nuclear (right) IGF-1R expression. Scale bar, 100 mmol/L.

www.aacrjournals.org Mol Cancer Res; 11(11) November 2013 1311

Aslam et al.

nuclear trafficking. SUMOylation of K1025, K1100, and of new (metastatic) tumors will be a key subject of further K1120 residues in the kinase domain of IGF-1Rb is also study and perhaps an operative step toward a temporal critical to nuclear import of the holoreceptor (32). Further biomarker-driven and personalized medicine approach in investigation has shown that nuclear IGF-1R originates from cancer therapy with regards to IGF-1R inhibition. surface IGF-1R only (33). Of note, we used an antibody to a sort cells based on surface IGF-1R expression (the subunit Disclosure of Potential Conflicts of Interest of IGF-1R) and a different antibody to detect IGF-1Rb at No potential conflicts of interest were disclosed. the nuclear fraction by immunoblot analysis, but given the covalent assembly of a and b subunits in the mature Authors' Contributions receptor, we do not expect this to have altered our results. Conception and design: M.I. Aslam, S. Hettmer, A. Soundararajan, B.J. Druker, A.J. Another intriguing aspect of our studies was that nuclear Wagers, J.W. Tyner, C. Keller PDGFRa did not seem to have an adverse functional role Development of methodology: M.I. Aslam, S. Hettmer, A. Soundararajan, J.W. in vivo Tyner, C. Keller with respect to tumor establishment . Acquisition of data (provided animals, acquired and managed patients, provided Although we were able to identify a difference in receptor facilities, etc.): M.I. Aslam, Simone Hettmer, J. Abraham, D. LaTocha, E.T. Huang, S. Wang, A. Mansoor, B.J. Druker, J.W. Tyner, C. Keller localization among the sorted populations of murine alveolar Analysis and interpretation of data (e.g., statistical analysis, biostatistics, compu- rhabdomyosarcoma, we did not observe a differential sen- tational analysis): M.I. Aslam, Simone Hettmer, D. LaTocha, M.W. Goros, J.E. sitivity of these cells to small-molecule inhibitors which Michalek, A. Mansoor, J.W. Tyner, C. Keller a Writing, review, and/or revision of the manuscript: M.I. Aslam, S. Hettmer, target PDGFR or IGF-1R (data not shown). An interesting A. Soundararajan, M.W. Goros, J.E. Michalek, A. Mansoor, B.J. Druker, A.J. Wagers, question these results raise is whether small-molecule inhi- J.W. Tyner, C. Keller bitors designed to inhibit kinase activity of RTKs are able to Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.I. Aslam, B.J. Druker, J.W. Tyner, C. Keller inhibit the function of these RTKs when localized to the Study supervision: M.I. Aslam, B.J. Druker, J.W. Tyner, C. Keller nucleus. Understanding whether kinase activity contributes to the role these receptors play in the nucleus would be one Grant Support step toward answering this question. Presumably, small- This work was supported by 5R01CA133229-05 and -06 to C. Keller. The studies molecule inhibitors would have an advantage over neutral- were also supported, in part, from a gift from the Ethan Jostad Foundation for izing antibodies in this respect. Prior studies have already Childhood Cancer. Pilot studies leading to this work were supported by an innovation award from Alex's Lemonade Stand. J.W. Tyner is supported by grants from the shown that certain RTKs, such as EGFR, interact with William Lawrence and Blanche Hughes Foundation, Leukemia & Lymphoma DNA-binding proteins to directly regulate the transcription Society, and the National Cancer Institute. M.I. Aslam is a Howard Hughes Medical Institute Medical Research Fellow. B.J. Druker is an investigator of the Howard of putative gene targets (34). The exact role of IGF-1R in the Hughes Medical Institute. nucleus of alveolar rhabdomyosarcoma will be an active area The costs of publication of this article were defrayed in part by the payment of page of further investigation, particularly given the potential charges. This article must therefore be hereby marked advertisement in accordance with translation from mouse to human and the expression of 18 U.S.C. Section 1734 solely to indicate this fact. nuclear IGF-1R in clinical FFPE samples. The mechanism Received October 17, 2012; revised June 25, 2013; accepted July 15, 2013; by which nuclear IGF-1R contributes to the establishment published OnlineFirst August 8, 2013.

References 1. Arndt CA, Crist WM. Common musculoskeletal tumors of childhood in rhabdomyosarcoma: correlation with overall survival in 105 patients. and adolescence. N Engl J Med 1999;341:342–52. Cancer 2007;110:2293–303. 2. Breneman JC, Lyden E, Pappo AS, Link MP, Anderson JR, Parham 9. Cen L, Arnoczky KJ, Hsieh FC, Lin HJ, Qualman SJ, Yu S, et al. DM, et al. Prognostic factors and clinical outcomes in children Phosphorylation proﬁles of protein kinases in alveolar and embryonal and adolescents with metastatic rhabdomyosarcoma–areportfrom rhabdomyosarcoma. Mod Pathol 2007;20:936–46. the Intergroup Rhabdomyosarcoma Study IV. J Clin Oncol 2003;21: 10. Kim ES, Herbst RS, Wistuba II, Lee JJ, Blumenschein GR Jr, Tsao A, 78–84. et al. The BATTLE trial: personalizing therapy for lung cancer. Cancer 3. Pappo AS, Shapiro DN, Crist WM. Rhabdomyosarcoma. Biology and Discov 2011;1:44–53. treatment. Pediatr Clin North Am 1997;44:953–72. 11. Gerlinger M, Rowan AJ, Horswell S, Larkin J, Endesfelder D, 4. Keller C, Arenkiel BR, Cofﬁn CM, El-Bardeesy N, DePinho RA, Capec- Gronroos E, et al. Intratumor heterogeneity and branched evolution chi MR. Alveolar rhabdomyosarcomas in conditional Pax3:Fkhr mice: revealed by multiregion sequencing. N Engl J Med 2012;366: cooperativity of Ink4a/ARF and Trp53 loss of function. Genes Dev 883–92. 2004;18:2614–26. 12. Bertolini G, Roz L, Perego P, Tortoreto M, Fontanella E, Gatti L, et al. 5. Nishijo K, Chen QR, Zhang L, McCleish AT, Rodriguez A, Cho MJ, et al. Highly tumorigenic lung cancer CD133þ cells display stem-like fea- Credentialing a preclinical mouse model of alveolar rhabdomyosar- tures and are spared by cisplatin treatment. Proc Natl Acad Sci U S A coma. Cancer Res 2009;69:2902–11. 2009;106:16281–6. 6. Abraham J, Prajapati SI, Nishijo K, Schaffer BS, Taniguchi E, Kilcoyne 13. Patrawala L, Calhoun-Davis T, Schneider-Broussard R, Tang DG. A, et al. Evasion mechanisms to IGF-1R inhibition in rhabdomyosar- Hierarchical organization of prostate cancer cells in xenograft tumors: coma. Mol Cancer Ther 2011;10:697–707. the CD44þalpha2beta1 þcell population is enriched in tumor-initiating 7. Taniguchi E, Nishijo K, McCleish AT, Michalek JE, Grayson MH, Infante cells. Cancer Res 2007;67:6796–805. AJ, et al. PDGFR-A is a therapeutic target in alveolar rhabdomyosar- 14. Shmelkov SV, Butler JM, Hooper AT, Hormigo A, Kushner J, Milde T, coma. Oncogene 2008;27:6550–60. et al. CD133 expression is not restricted to stem cells, and both 8. Armistead PM, Salganick J, Roh JS, Steinert DM, Patel S, Munsell M, CD133þ and CD133- metastatic colon cancer cells initiate tumors. et al. Expression of receptor tyrosine kinases and apoptotic molecules J Clin Invest 2008;118:2111–20.

1312 Mol Cancer Res; 11(11) November 2013 Molecular Cancer Research

Nuclear Receptor Tyrosine Kinases in Rhabdomyosarcoma

15. Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, et al. BMS-754807, in a human rhabdomyosarcoma model. Cancer Res Identification of human brain tumour initiating cells. Nature 2004;432: 2010;70:7221–31. 396–401. 25. Chaffer CL, Brueckmann I, Scheel C, Kaestli AJ, Wiggins PA, Rodri- 16. Gupta PB, Fillmore CM, Jiang G, Shapira SD, Tao K, Kuperwasser C, gues LO, et al. Normal and neoplastic nonstem cells can spontane- et al. Stochastic state transitions give rise to phenotypic equilibrium in ously convert to a stem-like state. Proc Natl Acad Sci U S A populations of cancer cells. Cell 2011;146:633–44. 2011;108:7950–5. 17. Aleksic T, Chitnis MM, Perestenko OV, Gao S, Thomas PH, 26. Li C, Heidt DG, Dalerba P, Burant CF, Zhang L, Adsay V, et al. Turner GD, et al. Type 1 insulin-like growth factor receptor Identification of pancreatic cancer stem cells. Cancer Res 2007;67: translocates to the nucleus of human tumor cells. Cancer Res 1030–7. 2010;70:6412–9. 27. Le Roy C, Wrana JL. Clathrin- and non-clathrin-mediated endocytic 18. Ettenberg SA, Magnifico A, Cuello M, Nau MM, Rubinstein YR, Yarden regulation of cell signalling. Nat Rev Mol Cell Biol 2005;6:112–26. Y, et al. Cbl-b-dependent coordinated degradation of the epidermal 28. Wang Y, Pennock SD, Chen X, Kazlauskas A, Wang Z. Platelet-derived growth factor receptor signaling complex. J Biol Chem 2001;276: growth factor receptor-mediated signal transduction from endo- 27677–84. somes. J Biol Chem 2004;279:8038–46. 19. Johnston CL, Cox HC, Gomm JJ, Coombes RC. Fibroblast growth 29. Wu A, Chen J, Baserga R. Nuclear insulin receptor substrate-1 acti- factor receptors (FGFRs) localize in different cellular compartments. A vates promoters of cell cycle progression genes. Oncogene 2008;27: splice variant of FGFR-3 localizes to the nucleus. J Biol Chem 397–403. 1995;270:30643–50. 30. Parker A, Cheville JC, Lohse C, Cerhan JR, Blute ML. Expression of 20. Liao HJ, Carpenter G. Role of the Sec61 translocon in EGF receptor insulin-like growth factor I receptor and survival in patients with clear trafficking to the nucleus and gene expression. Mol Biol Cell 2007;18: cell renal cell carcinoma. J Urol 2003;170:420–4. 1064–72. 31. Mukherjee S, Tessema M, Wandinger-Ness A. Vesicular trafficking of 21. Lin SY, Makino K, Xia W, Matin A, Wen Y, Kwong KY, et al. Nuclear tyrosine kinase receptors and associated proteins in the regulation of localization of EGF receptor and its potential new role as a transcription signaling and vascular function. Circ Res 2006;98:743–56. factor. Nat Cell Biol 2001;3:802–8. 32. Sehat B, Tofigh A, Lin Y, Trocme E, Liljedahl U, Lagergren J, et al. 22. Sardi SP, Murtie J, Koirala S, Patten BA, Corfas G. Presenilin-depen- SUMOylation mediates the nuclear translocation and signaling of the dent ErbB4 nuclear signaling regulates the timing of astrogenesis in the IGF-1 receptor. Sci Signal 2010;3:ra10. developing brain. Cell 2006;127:185–97. 33. Deng H, Lin Y, Badin M, Vasilcanu D, Stromberg T, Jernberg-Wiklund 23. Duncan JS, Whittle MC, Nakamura K, Abell AN, Midland AA, Zawis- H, et al. Over-accumulation of nuclear IGF-1 receptor in tumor cells towski JS, et al. Dynamic reprogramming of the kinome in response to requires elevated expression of the receptor and the SUMO-conju- targeted MEK inhibition in triple-negative breast cancer. Cell gating enzyme Ubc9. Biochem Biophys Res Commun 2011;404: 2012;149:307–21. 667–71. 24. Huang F, Hurlburt W, Greer A, Reeves KA, Hillerman S, Chang H, et al. 34. Lo HW. Nuclear mode of the EGFR signaling network: biology, prog- Differential mechanisms of acquired resistance to insulin-like growth nostic value, and therapeutic implications. Discov Med 2010;10: factor-I receptor antibody therapy or to a small-molecule inhibitor, 44–51.

www.aacrjournals.org Mol Cancer Res; 11(11) November 2013 1313

Dynamic and Nuclear Expression of PDGFRα and IGF-1R in Alveolar Rhabdomyosarcoma

M. Imran Aslam, Simone Hettmer, Jinu Abraham, et al.

Mol Cancer Res 2013;11:1303-1313. Published OnlineFirst August 8, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-12-0598

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2021/03/13/1541-7786.MCR-12-0598.DC1

Cited articles This article cites 34 articles, 17 of which you can access for free at: http://mcr.aacrjournals.org/content/11/11/1303.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/11/11/1303.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://mcr.aacrjournals.org/content/11/11/1303. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.