Identification of Bone-Derived Factors Conferring De Novo Therapeutic Resistance in Metastatic Prostate Cancer

Published OnlineFirst November 3, 2015; DOI: 10.1158/0008-5472.CAN-15-1215

Cancer Therapeutics, Targets, and Chemical Biology Research

Identiﬁcation of Bone-Derived Factors Conferring De Novo Therapeutic Resistance in Metastatic Prostate Cancer Yu-Chen Lee1, Song-Chang Lin1, Guoyu Yu1, Chien-Jui Cheng2,3, Bin Liu4, Hsuan-Chen Liu1, David H. Hawke1, Nila U. Parikh5, Andreas Varkaris5, Paul Corn5, Christopher Logothetis5, Robert L. Satcher6, Li-Yuan Yu-Lee7, Gary E. Gallick5, and Sue-Hwa Lin1,5

Abstract

Resistance to currently available targeted therapies significantly viable and expressed high levels of pFAK-Y397 and pTalin-S425, hampers the survival of patients with prostate cancer with bone mediators of integrin signaling. Accordingly, treatment of C4-2B4 metastasis. Here we demonstrate an important resistance mech- cells with integrin ligands resulted in increased pFAK-Y397 anism initiated from tumor-induced bone. Studies using an expression and cell survival, whereas targeting integrins with FAK osteogenic patient-derived xenograft, MDA-PCa-118b, revealed inhibitors PF-562271 or defactinib inhibited FAK phosphoryla- that tumor cells resistant to cabozantinib, a Met and VEGFR-2 tion and reduced the survival of PC3-mm2 cells. Moreover, inhibitor, reside in a "resistance niche" adjacent to prostate treatment of MDA-PCa-118b tumors with PF-562271 led to cancer-induced bone. We performed secretome analysis of the decreased tumor growth, irrespective of initial tumor size. Finally, conditioned medium from tumor-induced bone to identify pro- we show that upon treatment cessation, the combination of PF- teins (termed "osteocrines") found within this resistance niche. 562271 and cabozantinib delayed tumor recurrence in contrast to In accordance with previous reports demonstrating that activa- cabozantinib treatment alone. Our studies suggest that identify- tion of integrin signaling pathways confers therapeutic resistance, ing paracrine de novo resistance mechanisms may significantly 27 of the 90 osteocrines identified were integrin ligands. We contribute to the generation of a broader set of potent therapeutic found that following cabozantinib treatment, only tumor cells tools that act combinatorially to inhibit metastatic prostate positioned adjacent to the newly formed woven bone remained cancer. Cancer Res; 75(22); 4949–59. 2015 AACR.

Introduction of metastatic prostate cancer in bone is the induction of new bone formation, resulting in bone-forming lesions. This unique tumor Development of metastasis in bone is the lethal progression of microenvironment likely contributes to resistance to multiple prostate cancer. Although several targeted therapeutics have therapies. However, whether factors from prostate tumor-induced improved survival of patients with bone metastasis (1, 2), resis- bone are involved in therapy resistance is unknown. tance to targeted therapy invariably develops (3). A distinct feature Recently, cabozantinib, an oral multikinase inhibitor with potent activity against p-MET and p-VEGFR-2, demonstrated striking clinical and radiologic responses in patients with cas- 1 Department of Translational Molecular Pathology, The University of trate-resistant prostate cancer with bone metastasis (4, 5). How- Texas MD Anderson Cancer Center, Houston, Texas. 2Department of Pathology, School of Medicine, College of Medicine, Taipei Medical ever, although cabozantinib increased progression-free survival, it University, Taipei, Taiwan. 3Department of Pathology, Taipei Medical did not lead to improvement in overall survival in a phase 3 4 University Hospital, Taipei, Taiwan. Department of Genetics, Center clinical trial. Identifying the mechanism(s) of therapy resistance for Cancer Genetics and Genomics,The University of Texas MD Ander- son Cancer Center, Houston, Texas. 5Department of Genitourinary to cabozantinib and other current drugs for bone metastasis will Medical Oncology, The University of Texas MD Anderson Cancer lead to the development of strategies to treat this fatal disease. Center, Houston, Texas. 6Department of Orthopedic Oncology, The The current theory whereby patients acquire resistance is University of Texas MD Anderson Cancer Center, Houston, Texas. 7Department of Medicine, Baylor College of Medicine, Houston,Texas. adaptation of tumor cells in response to therapeutic agents. However, the possibility that prostate tumor-induced changes Note: Supplementary data for this article are available at Cancer Research in the bone microenvironment may confer "de novo"resistance, Online (http://cancerres.aacrjournals.org/). i.e., resistance that occurs prior to treatment with any therapy, Y.-C. Lee and S.-C. Lin contributed equally to this article. has not been examined previously. In this study, we used Corresponding Authors: Sue-Hwa Lin, The University of Texas MD Anderson cabozantinib as a study tool to examine the mechanism(s) by Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030. Phone: 713-794- which prostate tumor-induced bone formation confers therapy 1559; Fax: 713-794-4672; E-mail: [email protected]; and Gary E. Gallick, resistance. Our studies indicate that prostate cancer-induced [email protected]. bone provides a "resistance niche" that allows prostate cancer doi: 10.1158/0008-5472.CAN-15-1215 cells to survive under cabozantinib treatment and perhaps 2015 American Association for Cancer Research. other targeted therapies.

www.aacrjournals.org 4949

Lee et al.

Materials and Methods for the effect of PF562271 on PC3-mm2 cells, various concentration of PF562271 were added to the top layer of agarose Materials containing 0.1 million cells per well in RPMI containing 0.5% PC3-mm2 and C4-2B4, human prostate cancer cells, were FBS. PF562271 was replenished daily, 5 days per week. The cells maintained in RPMI 1640. HEK-293T cells were maintained in were grown for 18 days and the number of colonies counted. DMEM. All cell lines were verified by polymorphic short tandem repeat loci profiling and tested negative for mycoplasma Treatment of PCa-118b tumor with PF-562271 in vivo infection. The patient-derived xenograft (PDX), MDA-PCa-118b Mice with PCa-118b tumors grown for 4 to 6 weeks were (PCa-118b), was derived from a bone lesion of a patient with grouped according to their tumor sizes as small (30 mm3), castrate-resistant prostate cancer (6). Cabozantinib (XL184) was medium (80 mm3), and large (190 mm3). In each group of provided by Exelixis. Antibodies were from commercial sources: four mice, one was treated with vehicle, one with PF-562271 pTalin-S425 (ECM), Talin (Millipore), pFAK-Y397 (Invitrogen), (33 mg/kg), one with cabozantinib (20 mg/kg), and one with FAK (Cell Signaling), EpCAM (R&D), AE1/AE3 (Dako), penta-His both PF-562271 and cabozantinib, all through oral gavage twice a (Qiagen), mouse osteocalcin, and SPARC (Santa Cruz). day at 8- and 16-hour intervals for 5 days per week. Tumor sizes were measured weekly. After 2 weeks, treatments were stopped. Treatment of PCa-118b tumors with cabozantinib in vivo Mice in the "large" tumor group were sacrificed. Mice in the PCa-118b PDX was maintained by serial passage in SCID "medium" and "small" tumor groups were monitored for tumor mice as described (7). SCID mice bearing PCa-118b tumors regrowth for either 1 week or 3 weeks. In another study, the were treated with cabozantinib (20 mg/kg) through oral gavage treatments were initiated after the tumors reached palpable sizes. 2/day, 6 days/week. Control mice were administered with vehicle. Tumor sizes and bone formation were measured by Statistical analysis caliper and X-ray, respectively. Data are expressed as the mean SD unless otherwise stated. Student t test (two-tailed, paired) was used for statistical analyses. Histology and IHC PCa-118b tumors were fixed in formalin, decalcified with formic acid, and embedded in paraffin. IHC staining was per- Results formed as described (8, 9). Cabozantinib inhibits the growth of established osteogenic PCa-118b tumors Mass spectrometry analysis of conditioned medium from To study resistance specifically due to osteoblastic metastases, tumor-induced bone we use a bone-forming PDX, PCa-118b (6), as a model. PCa-118b PCa-118b tumor was digested with Accumax enzyme mixture has been shown to induce bone formation even when grown (eBioscience) and cells filtered through a cell strainer. The remain- subcutaneously (6, 7) and PCa-118b tumor has all the cellular ing bone (tumor-induced bone) was cultured in BGJb medium elements present in human bone metastases, including osteo- (Thermo Fisher Scientific) for 48 hours. The conditioned media blasts (6). from tumor-induced bone (Bone-CM) was analyzed by mass We first determined whether cabozantinib is effective in inhi- spectrometry (MS) as described (10). biting the growth of established PCa-118b tumors. Mice bearing PCa-118b tumors of various sizes were treated with cabozantinib. qPCR analysis Tumor size measurements showed that regression occurred in all Total RNA was prepared using TRIzol and RNeasy Mini Kit four tumors at one week after treatment (Fig. 1A, left). Larger (Qiagen) and analyzed by qRT-PCR. Human and mouse integrin tumors exhibited more pronounced decreases in size (Fig. 1A, primers, HINT-1 and MINT-1, respectively, were purchased from left). When calculated by percentage of tumor shrinkage, all four RealTimePrimers. GAPDH was used as a control. tumors showed decreased tumor volume within one week after cabozantinib treatment (Fig. 1A, right) and these decreases per- Expression and purification of osteocrines sisted throughout the remaining 3 weeks of treatment (Fig. 1A). RNA from MC3T3-E1 cells was used to generate cDNAs for The decrease in tumor sizes may be an underestimate of tumor osteocrines using primer sequences listed in Supplementary Table shrinkage as there was tumor-induced bone within the tumors S1. Recombinant osteocrines with His8-tag were expressed from (see Fig. 1B). HEK293 cells using pcDNA3.1 vectors and purified from conditioned media using Ni-NTA agarose (Qiagen; ref. 11). cDNAs in Tumor cells adjacent to tumor-induced bone are viable after pBMN-I-GFP retroviral vector (12) were used to generate C4-2B4 treatment cells stably expressed osteocrines. To further examine the effect of cabozantinib on PCa-118b- induced bone, X-rays were performed at the onset of treatment Soft agar colony assay and 4 weeks after treatment (Fig. 1B). No significant difference Soft agar colony assays were performed as described with was observed in the shape of the bone present in the PCa-118b modifications (13). In brief, soft agar colony assay was performed tumors (Fig. 1B); however, the bone density appeared somewhat in a 6-well plate using low melting point agarose. The base layer intensified in the cabozantinib-treated group (Fig. 1B, right). In was 0.8% agarose and the top layer was 0.48% agarose containing addition, decreases in tumor volume were observed in the X-ray 75,000 C4-2B4-LT cells per well and 10 mg/mL of lumican, images (Fig. 1B), consistent with the tumor size measurements osteopontin, SPARC, BIG-H3, or vitronectin in RPMI medium (Fig. 1A). Histologic analysis showed that cabozantinib treatment without serum. The cells were grown in soft agar for 14 days and induced cell death in large areas of the tumors (Fig. 1C). However, the number of colonies counted. Similarly, in the soft agar assay small clusters of viable tumor cells were found, most frequently

4950 Cancer Res; 75(22) November 15, 2015 Cancer Research

Tumor Microenvironment Confers De Novo Therapy Resistance

Figure 1. Cabozantinib inhibits the growth of established osteogenic PCa-118b tumors. A, mice with PCa-118b tumors of various sizes were treated with cabozantinib for 4 weeks. Tumor size (left) and percent tumor size changes (right) after cabozantinib treatment. B, X-ray of PCa-118b tumors before and after cabozantinib treatments. Right, higher magniﬁcation. C, left, histologic analysis of cabozantinib-treated PCa- 118b tumors. Cabozantinib treatment led to tumor cell death (asterisk). Viable tumor cells (green arrows) and stromal cells (blue arrowheads) adjacent to the tumor-induced bone were observed. Right, immunostaining of AE1/AE3 antigen and osteocalcin with DAPI counterstain. B, bone. 200.

adjacent to the tumor-induced bone. Further, osteoblasts rim- PCa-118 tumor is shown in Supplementary Fig. S1. These results ming the tumor-induced bone were found to be viable (Fig. 1C), suggest that osteoblasts are resistant to cabozantinib and may suggesting that PCa-118b-induced osteoblasts are resistant to protect tumor cells adjacent to the tumor-induced bone from killing by cabozantinib treatment. IHC staining for osteocalcin cabozantinib therapy. (an osteoblast marker) and AE1/AE3 (an epithelial cell marker) showed that the viable tumor cells are located adjacent to osteo- Tumor recurrence after discontinuation of cabozantinib blasts (Fig. 1C). Interestingly, PCa-118b tumor cells also express treatment osteocalcin (Fig. 1C), consistent with the reported osteomimicry To examine whether tumor recurrence may occur after discon- of prostate cancer cells (14, 15). IHC staining of control untreated tinuing cabozantinib treatment, PCa-118b tumors were allowed

www.aacrjournals.org Cancer Res; 75(22) November 15, 2015 4951

Lee et al.

to grow until the tumors reached approximately 500 mm3. One Secretome analysis of conditioned medium from tumor- group of mice was left untreated (control) and a second group was associated bone treated with cabozantinib. Cabozantinib not only halted tumor Because small clusters of viable tumor cells were found adjacent growth but also led to reductions in tumor sizes (Supplementary to the bone after cabozantinib treatment (Fig. 1C), we hypothesized Fig. S2A). Histologic analyses showed that small clusters of viable that prostate tumor-induced osteoblasts secrete factors that induce tumor cells were present along with considerable cell death in signaling pathways in tumor cells, leading to therapy resistance. We cabozantinib-treated tumors at the time cabozantinib treatment termed these bone-secreted factors that modulate cellular function was discontinued (Supplementary Fig. S2A). However, tumors in a paracrine fashion as "osteocrines" ("osteo" ¼ bone; "crine" ¼ resumed growth after treatments were discontinued. In tumors paracrines). To identify osteocrines, PCa-118b tumor was digested collected 1 or 2 weeks after cabozantinib treatments, robust tumor with Accumax enzyme mixture to dissociate tumor cells. Tumor regrowth was observed that completely replaced the area of cell cells were removed and the remaining bone was cultured in BGJb death from prior cabozantinib treatment (Supplementary Fig. medium (Fig. 2A). The conditioned media of tumor-associated S2B and S2C). IHC analysis showed that cells in the recurrent bone (Bone-CM) from two independent PCa-118b tumors were tumors are positive for AE1/AE3 epithelial cell markers (Supple- resolved by SDS-PAGE (Fig. 2B, left). SPARC (also known as mentary Fig. S2B), indicating regrowth of cabozantinib-resistant osteonectin), a major bone secreted factor, was detected in Bone- tumor cells. CM (Fig. 2B, right). MS analyses of Bone-CM identiﬁed 65 and

Figure 2. Secretome analysis of Bone-CM. A, procedure for the preparation of Bone- CM for MS analysis. B, left, SDS-PAGE analysis of Bone-CM from two PCa-118b tumors. Proteins were transferred onto nitrocellulose and stained with Ponceau S. Right, Western blot detected SPARC in two Bone-CM. C, Venn diagram of the number of secretory proteins in two Bone-CM as determined by MS. D, expression of integrin ligands in PCa-118b tumor by qRT-PCR using mouse-speciﬁc primers. E, qRT-PCR for the expression of SPARC, SPP1, and lumican in PCa-118b tumor cells or whole tumor using human- or mouse- speciﬁc primers.

4952 Cancer Res; 75(22) November 15, 2015 Cancer Research

Tumor Microenvironment Confers De Novo Therapy Resistance

Figure 3. Integrins and integrin signaling in PCa-118b tumors. A, integrin expression in PCa-118b tumor as determined by qRT-PCR using human-speciﬁc primers. B, qRT-PCR of integrin av in isolated PCa-118b tumor cells, whole tumor, or prostate cancer cell lines using human- or mouse-speciﬁc primers. C, diagram of integrin signaling through pTalin-S425 and pFAK-Y397. D, IHC for the expression of pFAK-Y397 and pTalin-S425 in PCa-118b tumors after cabozantinib treatment. , dead cells.

75 secretory proteins in each sample. Fifty proteins were found in cells. qRT-PCR analysis using human-specific primers for 16a and both samples (Fig. 2C). Together, 90 secretory proteins were iden- 8b integrin subunits showed that PCa-118b tumor cells mainly tified (Supplementary Table S2). The most frequently identified express a6, av, and b1 integrins (Fig. 3A). Low levels of other group of proteins is ligands for integrins (27/90; Supplementary integrins, including a1, a2, a9, b3, b4, b5, and b8 were also Table S2). We (16) and others (17–20) have shown that integrin detected. To confirm that av is expressed in PCa-118b tumor, we activation plays a critical role in prostate cancer cell survival. performed RT-PCR using human- or mouse-specific integrin av Additionally, integrin activation has been associated with resistance primers on RNAs from isolated PCa-118b cells or PCa-118b to chemotherapy, radiotherapy, and targeted agents (21). Thus, the tumors. Human-specific primers detected av in both isolated secreted proteins that are integrin ligands are candidate osteocrines PCa-118b cells and total tumor (Fig. 3B). Mouse-specific av that may confer therapy resistance to prostate cancer cells. primers also detected av in total tumor, suggesting that mouse We employed mouse-specific primers in qRT-PCR analysis to cells that were recruited into PCa-118b tumor also expressed av. confirm the expression of several of these integrin ligands, includ- Interestingly, av is much more abundantly expressed in PCa-118b ing fibronectin, laminins, SPARC, SPP1 (osteopontin), BIG-H3 cells relative to PC3-mm2, DU145, and C4-2B cells (Fig. 3B, (TGFb induced, TGFbI), tenascin C (TNC), lumican (LUM), and right). vitronectin (VTN) in PCa-118b tumors. High levels of messages The ligands for avb1 are fibronectin, SPP1, and vitronectin, for mouse SPARC, SPP1, and lumican and medium levels for among others (23, 24). The ligands for a6b1 are laminins (23) mouse fibronectin, laminins (LAMA3, B1, and C1), and BIG-H3 and TNC (25), and others. Fibronectin, TNC, laminins (a1, a4, were observed (Fig. 2D). Because PCa-118b tumors contain both b1, and g1), SPP1, lumican, and vitronectin are also detected by human PCa-118b cells (22) and mesenchymal cells from mouse MS (Supplementary Table S2) and by qRT-PCR (Fig. 2D). These including osteoblasts, endothelial cells, and fibroblasts, we results demonstrate that PCa-118b cells express receptors for the employed human- and mouse-specific primers to determine the integrin ligands identified from the tumor-induced bone. origin of the osteocrines. High levels of mouse SPARC and SPP1 messages were found in PCa-118b tumors but not in purified PCa- 118b epithelial cells (Fig. 2E), consistent with previous reports FAK and Talin phosphorylation are associated with integrin that osteoblasts in PCa-118-induced bone are of mouse origin (6, activation 7), and the presence of SPARC in the Bone-CM (Fig. 2B). These To examine whether the integrin ligands secreted by the tumor- observations demonstrate that SPARC, SPP1, and lumican are induced bone activate integrin signaling in PCa-118b tumor, we secreted from tumor-associated osteoblasts. examined the phosphorylation of two integrin signaling mediators, focal adhesion kinase (pFAK-Y397) and pTalin-S425 Integrins and integrin ligands in PCa-118b tumors (Fig. 3C; ref. 9), in cabozantinib-treated tumors. Intense staining Because integrin ligands interact with different integrins, we of both pFAK-Y397 and pTalin-S425 was observed in viable next determined which integrins are expressed in PCa-118b tumor tumor cells that are adjacent to the bone (Fig. 3D). These results

www.aacrjournals.org Cancer Res; 75(22) November 15, 2015 4953

Lee et al.

suggest that osteoblast-secreted integrin ligands likely activate vector control (Fig. 4D). SPP1 is a ligand for a5b1, avb1, avb3, integrin signaling in tumor cells. avb5, a4b1, and a8b1, and lumican has been shown to be a ligand for a5b1, avb1, and a2b1. Thus, the effects of these two Integrin expression in C4-2B4 and PC3-mm2 prostate cancer osteocrines on C4-2B4 cells are consistent with the expression cells of a5 and b1 integrins in C4-2B4 cells. In contrast, addition of To investigate whether integrin signaling confers therapy resis- BIG-H3 (a ligand for avb3, avb5, and a6b4 integrins) and tance, we examined the effects of integrin ligands on prostate vitronectin (a ligand for avb1 and avb5 integrins) did not have cancer cells. Because the PCa-118b cells isolated from PCa-118b a significant effect on C4-2B4 cell colony formation, likely due to tumors do not survive for more than 2 days in standard culture the low to undetectable levels of expression of av and b1 in C4- conditions (22), C4-2B4 and PC3-mm2 cells were used for this 2B4 cells (Fig. 4A). SPARC (osteonectin), which is known as a analysis. qRT-PCR using human-specific integrin primers showed counter-adhesive protein (26–29), also did not have a significant that PC3-mm2 cells mainly expressed a5, a6, and b1 integrins effect on C4-2B4 cell colony formation (Fig. 4D). Additionally, we (Fig. 4A). C4-2B4 cells expressed low message levels of integrins expressed SPARC, SPP1, lumican, BIG-H3, and vitronectin in C4- compared with PC3-mm2 and PCa-118b cells, with a5, a6, b1, 2B4 cells through retroviral transduction. Soft agar colony assays and b5 being the major integrins expressed (Fig. 4A). Therefore, showed that overexpression of SPP1 or lumican in C4-2B4 cells expression of integrins in prostate cancer cells is heterogeneous led to an increase in cell survival (Supplementary Fig. S3), similar and this may lead to varied responses to different integrin ligands to the results observed with exogenous addition of purified present in the tumor microenvironment. recombinant proteins.

Effect of integrin ligands on C4-2B4 cell survival in vitro Effect of FAK inhibitor PF-562271 on PC3-mm2 in vitro To examine the effects of integrin ligands on C4-2B4 cell A consequence of integrin activation is phosphorylation of survival, cDNAs encoding SPP1, SPARC, BIG-H3, lumican, FAK, which is implicated in cell survival (30). PF-562271 is a and vitronectin with His8-tags were generated and recombinant competitive inhibitor for ATP binding to FAK (31), and is under- proteins were expressed in HEK293 cells (Fig. 4B). Treatment of going clinical trial (32). We have shown previously that C4-2B4 cells with recombinant SPP1 and lumican increased the integrin signaling pathways are constitutively activated in phosphorylation of FAK-Y397, indicating that these osteocrines PC3-mm2 cells and that FAK is constitutively activated (16). activate integrin signaling (Fig. 4C). Addition of SPP1 or lumican Therefore, we examined the effect of FAK inhibition on cell increased the colony formation of C4-2B4 cells compared with survival. To characterize the effect of PF-562271 on prostate

Figure 4. Integrin ligands confer C4-2B4 cell survival in vitro. A, integrin expression in PCa-118b, PC3-mm2, and C4-2B4 cells using human-specific primers in qRT-PCR analysis. B, Western blot of five integrin ligands expressed and purified from HEK293 cells using anti- His antibody. C, effect of integrin ligands on FAK activation. D, effect of integrin ligands on anchorage-independent growth of C4-2B4 cells. Bottom, quantification of soft agar colony numbers. , P < 0.05.

4954 Cancer Res; 75(22) November 15, 2015 Cancer Research

Tumor Microenvironment Confers De Novo Therapy Resistance

cancer cells, PC3-mm2 cells were first treated with PF-562271. Decreased FAK-Y397 phosphorylation was observed as early as 3 hours and persisted for 24 hours (Fig. 5A), with no change in total FAK expression. A dose–response analysis showed that treatment for 24 hours with PF-562271 at greater than 0.1 mmol/L was sufficient to decrease pFAK-Y397 by 80% in PC3-mm2 cells (Fig. 5B). Immunofluorescence showed that a 3-hour treatment with 5 or 10 mmol/L PF-562271 reduced pFAK-Y397 in focal adhesion sites as well as in the cytosol in PC3-mm2 cells (Fig. 5C, inset). Although Talin is not a direct target of PF-562271, phosphorylation of Talin-S425 was similarly inhibited, albeit for a shorter duration and requiring a higher concentration of PF- 562271 (Fig. 5A and B). These results suggest that PF-562271 effectively inhibits the phosphorylation of FAK-Y397 and Talin- S425 in PC3-mm2 cells. We further examined the effect of PF- 562271 on the survival of PC3-mm2 cells. We observed a dose- dependent inhibition of anchorage-independent growth by PF- 562271 in soft agar colony assays (Fig. 5D). We next examined the effects of PF-562271 on C4-2B4 cells, in which the integrin/FAK pathway is not constitutively activated. C4-2B4 cells were treated with 10 mg/mL SPARC, SPP1, lumican, BIG-H3, or vitronectin with or without 5 mmol/L PF-562271 and their growth in soft agar examined. Incubation of C4-2B4 cells with integrin ligands increased C4-2B4 cell survival (Fig. 5E). Treatment with PF-562271 significantly inhibited the osteocrine- induced C4-2B4 cell survival (Fig. 5E). Recently, several second-generation FAK inhibitors, including defactinib (VS-6063 PF-04554878), are being tested in clinical trials. We thus examined the effects of defactinib on pFAK inhibition in PC3-mm2 cells. Treatment of PC3-mm2 cells with defactinib led to a time- and dose-dependent inhibition of FAK phosphorylation similar to that, which was observed with PF- 562271 (Supplementary Fig. S4A and S4B). Defactinib treatment also resulted in a dose-dependent inhibition of PC3-mm2 cell survival in soft agar colony formation (Supplementary Fig. S4C). Together, these results suggest that inhibition of FAK activity decreases prostate cancer cell survival in vitro.

Effect of PF-562271 on tumor growth Because the expression of pFAK-Y397 was increased in the resistant tumor cells (Fig. 3D), we reasoned that: (1) inhibition of FAK activation alone might lead to the decrease of tumor growth; and (2) a combination of a FAK inhibitor with cabozantinib might reduce the number of resistant cells resulting from integrin ligand-mediated FAK activation. We first examined the effect of PF-562271 on FAK phosphorylation in vivo. SCID mice bearing PCa-118b tumors were treated with or without PF- 562271 for 1, 2, or 5 days. Western blot and immunostaining showed that pFAK-Y397 was inhibited in the tumor as early as 1 day after treatment with PF-562271 (Fig. 6A and B). Next, we tested whether PF-562271 is able to inhibit tumor growth. Mice bearing PCa-118b tumors were grouped according to tumor sizes: one group with tumors approximately 30 mm3,a second group with tumors approximately 80 mm3, and a third Figure 5. 3 PF-562271 inhibited PC3-mm2 cell survival in vitro. A, left, time course of group with tumors approximately 190 mm . Each mouse in the PF-562271 on the phosphorylation of FAK-Y397 or Talin-S425. Right, group was treated with vehicle, PF-562271, cabozantinib, or quantification of Western blot. B, left, dose–response of PF-562271 on the cabozantinib in combination with PF-562271. Mouse body phosphorylation of FAK-Y397 or Talin-S425. Right, quantification of weights in all groups except the cabozantinib alone group Western blot. C, immunostaining of PC3-mm2 cells showed that remained similar throughout the course of treatment, suggesting phosphorylated FAK is localized at focal adhesions (inset). D, dose- dependence of PF-562271 on soft agar growth of PC3-mm2 cells. PF-562271 that there was limited treatment toxicity (Supplementary Fig. S5). was replenished daily for 5 days/week. E, effect of PF-562271 on integrin Tumor growth over 14 days was calculated as the fold increase ligand-mediated C4-2B4 colony formation. , P < 0.05.

www.aacrjournals.org Cancer Res; 75(22) November 15, 2015 4955

Lee et al.

Figure 6. PF-562271 inhibits PCa-118b tumor growth in vivo. A, Western blot of pFAK-Y397 and pTalin-S425 in PCa-118b tumors from mice treated with PF- 562271 for 1, 2, or 5 days. B, IHC analyses of pFAK-Y397 in AE1/AE3-positive tumor cells one day after treatment. C, tumor growth in mice treated with PF-562271, cabozantinib, or combination of cabozantinib plus PF-562271 for 2 weeks. D, histology (H&E) and IHC analyses of pFAK-Y397 and Ki67 expression in tumors collected 2 weeks after treatment. , dead cells. E, tumor recurrence after treatment cessation. Mice bearing PCa-118b tumors were treated with PF-562271, cabozantinib, or combination of cabozantinib plus PF-562271 for 2 weeks, and then treatment was discontinued (dashed line). Tumor regrowth was monitored for 1 to 3 weeks. F, diagram illustrating therapy- induced de novo resistance. Cabozantinib treatment eradicates most of the prostate cancer tumor cells except those in proximity to the bone. Combination with PF-562271 reduces some of the resistant cells. Resistant tumor cells regrew upon treatment cessation. G, working model of integrin ligand-induced therapy resistance. Prostate cancer cells secrete factors that increase new bone formation. Osteoblasts in the new bone produce factors (osteocrines). Some of the osteocrines activate integrin signaling in prostate cancer cells, resulting in phosphorylation of FAK-Y397 and Talin-S425, which leads to therapy resistance. Inhibition of FAK activity decreases therapy resistance.

over the initial tumor size (Fig. 6C). We found that PF-562271 cell death was observed in cabozantinib-treated tumors (Fig. 6D, treatment alone resulted in an approximately 40% decrease in H&E). Mice treated with both cabozantinib and PF-562271 tumor size when compared with control, regardless of initial had fewer viable tumor cells compared with those treated tumor sizes (Fig. 6C), suggesting that PF-562271 can inhibit with cabozantinib alone (Fig. 6D, H&E). Immunoﬂuorescence tumor growth. Treatment with cabozantinib or cabozantinib plus analysis showed that pFAK-Y397 levels in PCa-118b tumor are PF-562271 resulted in complete tumor growth inhibition regard- heterogeneous, with some tumor areas expressing a higher level of less of the initial tumor size (Fig. 6C). pFAK-Y397 than other areas (Fig. 6D, control). PF-562271-treated Histologic analyses of the tumors showed that a majority of tumors have decreased levels of pFAK-Y397 (Fig. 6D). Cabozan- tumor cells were still viable after treatment with PF-562271 [Fig. tinib-resistant tumor islets have increased level of pFAK compared 6D, hematoxylin and eosin (H&E)]. In contrast, signiﬁcant tumor with those in untreated tumors (Fig. 6D), supporting FAK

4956 Cancer Res; 75(22) November 15, 2015 Cancer Research

Tumor Microenvironment Confers De Novo Therapy Resistance

activation in the survival of tumor cells. The combination of significantly reduce resistance to cabozantinib treatment in the cabozantinib with PF-562271 reduced the levels of pFAK-Y397 PCa-118b model (data not shown). These results may be due to a in the survived tumor cells compared with cabozantinib alone failure in blocking a6 integrin that is highly expressed in PCa- (Fig. 6D), although staining was heterogeneous. Importantly, we 118b cells. As most integrins signal through FAK, inhibiting pFAK found that the tumor cells resistant to PF-562271, cabozantinib, may be a strategy to overcome the heterogeneity in integrin or combination treatments were Ki67 positive (Fig. 6D), indicat- expression in tumor cells. In addition to PF-562271 (38), several ing that these tumor cells have the potential to resume their FAK inhibitors, e.g., VS-6063 (defactinib, PF-04554878) and VS- growth upon cessation of treatments. 4718, are currently being tested in clinical trials. Although our studies showed that PF-562271 is able to reduce PCa-118b growth Effects of PF-562271 on reducing therapy resistance in as a single agent and improve the therapy outcomes when PCa-118b tumor combined with cabozantinib, PF-562271 may not be suitable Next, we examined tumor regrowth after treatment cessation. for clinical application. PF-562271 exhibits potent inhibition of After 2 weeks of treatments, tumors in one group of mice were CYP3A, leading to dose- and time-dependent nonlinearity in PF- allowed to re-grow for 1 week and another group for 2 to 3 weeks 562271 pharmacokinetics (32). As a result, second-generation after stopping the treatments. Upon cessation of treatments, FAK inhibitors VS-6063 and VS-4718 may be better agents than cabozantinib-treated tumor re-grew rapidly (Fig. 6E), consistent PF-562271 in clinical application. with the presence of resistant tumor cells. Treatment with cabo- Although integrin signals through FAK, FAK is also known to be zantinib plus PF-562271 delayed tumor regrowth compared with activated by many growth factor receptors in various types of the tumor regrowth with cabozantinib alone (Fig. 6E), consistent human cancer (for review, see ref. 39). Thus, FAK activation in the with the reduction in resistant tumor cells by PF-562271 (Fig. 6D). tumor cells that are resistant to cabozantinib treatments may also Similar results were observed in another set of studies in which be due to integrin-independent signaling pathways other than by treatments were started as soon as the tumors were palpable integrin activation. As a result, FAK inhibitors may be used in (Supplementary Fig. S6). Together, these results support our therapy resistance arising from pathways other than integrin. interpretation that cabozantinib eradicates most of the tumor Besides integrin activation, osteocrines may also be involved in cells except those adjacent to the bone (Fig. 6F). Cabozantinib- other resistance mechanisms. Several of the identified osteocrines, resistant cells have higher levels of pFAK, which are sensitive to PF- e.g., periostin and tenascin C, have been shown to regulate 562271 treatment, leading to a delayed tumor recurrence after maintenance of cancer stem cell-like properties in other solid treatment cessation in combination therapy compared with those tumor types (40, 41). Interestingly, FAK is also a survival factor for with cabozantinib alone. cancer stem cells (42). FAK has been shown to play a role in self- renewal, tumorigenicity, and maintenance of mammary cancer stem cells (42). Shapiro and colleagues (43) showed that FAK Discussion inhibitor treatment preferentially eliminates cancer stem cells. Our results suggest that a novel form of "preexisting" or de Thus, the combination of a FAK inhibitor with chemotherapy may novo therapy resistance stems from the osteoblastic nature of also eliminate cancer stem cells that are resistant to chemotherapy. prostate cancer metastasis, in which osteocrines secreted from Other nonintegrin osteocrines may also confer resistance through tumor-induced bone contribute to therapy resistance. We pro- different mechanisms. These osteocrines together may contribute pose a working model of how the newly formed bone can to therapy resistance in human prostate cancer bone metastasis. induce "de novo"resistanceinprostatecancercellspriortothe Recently, it was reported that vascular heterogeneity may repre- initiation of therapy (Fig. 6G). Within the tumor-induced sent another source of primary therapy resistance (44). Systematic newly formed woven bone, osteoblasts secrete factors (osteo- dissection of tumor–microenvironment interactions may reveal crines), many of which are integrin ligands that activate integrin other important mechanisms underlying therapy resistance. signaling in prostate cancer cells, thus increasing cell survival Investigations into the various mechanisms of resistance from and "de novo" therapy resistance. Inhibition of FAK activity tumor microenvironment may lead to different therapy combi- reduces therapy resistance, suggesting that osteocrine-mediated nations that further overcome resistance. FAK activation through integrin signaling is one of the de novo In addition to targeting the osteocrine-mediated resistance resistance mechanisms. pathways in the tumor cells, another approach would be to target Increasing evidence demonstrates that integrin-mediated sur- the release of osteocrines from osteoblasts. Thus, it is possible that vival pathways play a major role in tumorigenesis and therapy the bone-targeting agents, e.g., radium-223 (Alpharadin), could resistance (17, 20, 33, 34). As a result, several integrin antibodies be considered for preventing therapy resistance from tumor- are being tested in clinical trials for various cancers (21, 35–37). induced bone. Radium-223 dichloride is a bone-seeking, a-emit- We found that C4-2B4, PC3-mm2, and PCa-118b cells express ting radionuclide that has been shown to result in increased different panels of integrins. These observations raise concern that overall survival, i.e., 3.6 months (45, 46), and has received FDA the heterogeneity of integrin expression in prostate cancer cells approval. As the mechanism of resistance we have identified is may limit the therapeutic efficacy of integrin antibodies. Our therapy-independent, we predict that these bone-targeting agents unpublished data support such a possibility. We examined the may also be effective in preventing therapy resistance from the anti-integrin b1 antibody mAb 33B6 (16) in combination with bone microenvironment by blocking the release of osteocrines DI17E6, an anti-integrin av antibody that has been tested in phase when used in combination with numerous therapeutic agents 1 clinical trials (35), on PCa-118b tumor growth. We found that, approved or in clinical trial for bone-metastatic prostate cancer. when tested in established PCa-118b tumors, the combination of In summary, we identify for the first time that factors secreted mAb 33B6 and DI17E6 did not show a significant inhibition of from tumor-induced bone constitutes one type of microenviron- tumor growth (data not shown). These antibodies also did not ment-mediated therapy resistance. Although cabozantinib was

www.aacrjournals.org Cancer Res; 75(22) November 15, 2015 4957

Lee et al.

used in this study, our studies identify a novel mechanism of de Acquisition of data (provided animals, acquired and managed patients, novo therapy resistance from tumor-induced bone that leads to provided facilities, etc.): Y.-C. Lee, S.-C. Lin, G. Yu, C.-J. Cheng, D.H. Hawke, resistance prior to any therapy application. Our observations A. Varkaris, P. Corn, G.E. Gallick Analysis and interpretation of data (e.g., statistical analysis, biostatistics, suggest that pathways mediating such resistance need to be computational analysis): Y.-C. Lee, C.-J. Cheng, B. Liu, C. Logothetis, L.-Y. Yu- targeted simultaneously with therapeutic regimens for bone Lee, G.E. Gallick, S.-H. Lin metastasis. As bone metastasis is the lethal progression of prostate Writing, review, and/or revision of the manuscript: Y.-C. Lee, S.-C. Lin, cancer, strategies that combine inhibitors of preexisting therapy P. Corn, C. Logothetis, R.L. Satcher, L.-Y. Yu-Lee, G.E. Gallick, S.-H. Lin resistance with currently approved drugs may improve therapy Administrative, technical, or material support (i.e., reporting or organizing outcomes and prolong patient survival. data, constructing databases): Y.-C. Lee, H.-C. Liu, N.U. Parikh, S.-H. Lin Study supervision: G.E. Gallick, S.-H. Lin Disclosure of Potential Conflicts of Interest C. Logothetis reports receiving a commercial research grant from Astellas, Grant Support BMS, Karyopharm, Sanofi, J&J, Excelixis, Pfizer, Novartis, Cougar Biotechnol- This work was supported by grants from the NIH P50 CA140388, CA174798, ogy, Medivation, Bayer, and Glaxo Smith Kline; has received speakers bureau CA16672, the Prostate Cancer Foundation, Cancer Prevention and Research honoraria from Bayer, J&J, AstraZeneca, Pfizer, Novartis, Helsinn HC, Astellas, Institute of Texas (CPRIT RP110327, RP150179, and RP150282) and funds and BMS; and is a consultant/advisory board member for AStellas, BMS, J&J, from the Sister Institute Network Fund, Institutional Research Grant Program, Pfizer, Novartis, Bayer, AstraZeneca, Helsinn, and Excelixis. No potential con- and Prostate Cancer Moonshot Program at the M.D. Anderson Cancer Center. flicts of interest were disclosed by the other authors. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in Authors' Contributions accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Conception and design: Y.-C. Lee, C. Logothetis, R.L. Satcher, G.E. Gallick, S.-H. Lin Received May 8, 2015; revised August 24, 2015; accepted September 7, 2015; Development of methodology: Y.-C. Lee, H.-C. Liu, D.H. Hawke published OnlineFirst November 3, 2015.

References 1. Dayyani F, Gallick GE, Logothetis CJ, Corn PG. Novel therapies for 15. Josson S, Matsuoka Y, Chung LW, Zhau HE, Wang R. Tumor-stroma co- metastatic castrate-resistant prostate cancer. J Natl Cancer Inst 2011; evolution in prostate cancer progression and metastasis. Semin Cell Dev 103:1665–75. Biol 2010;21:26–32. 2. Thoreson GR, Gayed BA, Chung PH, Raj GV. Emerging therapies in 16. Lee YC, Jin JK, Cheng CJ, Huang CF, Song JH, Huang M, et al. Targeting castration resistant prostate cancer. Can J Urol 2014;21(2 Supp 1):98–105. constitutively activated beta1 integrins inhibits prostate cancer metastasis. 3. Ramos P, Bentires-Alj M. Mechanism-based cancer therapy: resistance to Mol Cancer Res 2013;11:405–17. therapy, therapy for resistance. Oncogene 2015;34:3617–26. 17. Fornaro M, Manes T, Languino LR. Integrins and prostate cancer metas- 4. Lee RJ, Saylor PJ, Michaelson MD, Rothenberg SM, Smas ME, Miyamoto tases. Cancer Metastasis Rev 2001;20:321–31. DT, et al. A dose-ranging study of cabozantinib in men with castration- 18. Cress AE, Rabinovitz I, Zhu W, Nagle RB. The alpha 6 beta 1 and alpha 6 resistant prostate cancer and bone metastases. Clin Cancer Res 2013;19: beta 4 integrins in human prostate cancer progression. Cancer Metastasis 3088–94. Rev 1995;14:219–28. 5. Smith DC, Smith MR, Sweeney C, Elfiky AA, Logothetis C, Corn PG, et al. 19. Sroka IC, Anderson TA, McDaniel KM, Nagle RB, Gretzer MB, Cress AE. The Cabozantinib in patients with advanced prostate cancer: results of a phase laminin binding integrin alpha6beta1 in prostate cancer perineural inva- II randomized discontinuation trial. J Clin Oncol 2013;31:412–9. sion. J Cell Physiol 2010;224:283–8. 6. Li Z, Mathew P, Yang J, Starbuck M-W, Zurita AJ, Liu J, et al. Androgen 20. Sakamoto S, McCann RO, Dhir R, Kyprianou N. Talin1 promotes tumor receptor–negative human prostate cancer cells induce osteogenesis invasion and metastasis via focal adhesion signaling and anoikis resistance. through FGF9-mediated mechanisms. J Clin Invest 2008;118:2697–710. Cancer Res 2010;70:1885–95. 7. Lee YC, Cheng CJ, Bilen MA, Lu JF, Satcher RL, Yu-Lee LY, et al. BMP4 21. Jahangiri A, Aghi MK, Carbonell WS. beta1 integrin: critical path to promotes prostate tumor growth in bone through osteogenesis. Cancer Res antiangiogenic therapy resistance and beyond. Cancer Res 2014;74:3–7. 2011;71:5194–203. 22. Lee YC, Gajdosik MS, Josic D, Clifton JG, Logothetis C, Yu-Lee LY, et al. 8. Chu K, Cheng CJ, Ye X, Lee YC, Zurita AJ, Chen DT, et al. Cadherin-11 Secretome analysis of an osteogenic prostate tumor identifies complex promotes the metastasis of prostate cancer cells to bone. Mol Cancer Res signaling networks mediating cross-talk of cancer and stromal cells within 2008;6:1259–67. the tumor microenvironment. Mol Cell Proteomics 2015;14:471–83. 9. Jin JK, Tien PC, Cheng C-J, Song JH, Huang C, Lin S-H, et al. Talin1 23. Humphries JD, Byron A, Humphries MJ. Integrin ligands at a glance. J Cell phosphorylation activates b1 integrins: a novel mechanism to promote Sci 2006;119(Pt 19):3901–3. prostate cancer bone metastasis. Oncogene 2015;34:1811–21. 24. Plow EF, Haas TA, Zhang L, Loftus J, Smith JW. Ligand binding to integrins. 10. Heinemann ML, Ilmer M, Silva LP, Hawke DH, Recio A, Vorontsova MA, J Biol Chem 2000;275:21785–8. et al. Benchtop isolation and characterization of functional exosomes by 25. Katoh D, Nagaharu K, Shimojo N, Hanamura N, Yamashita M, Kozuka Y, sequential filtration. J Chromatogr A 2014;1371C:125–35. et al. Binding of alphavbeta1 and alphavbeta6 integrins to tenascin-C 11. Liang AK, Liu J, Mao SA, Siu VS, Lee YC, Lin SH. Expression of recombinant induces epithelial-mesenchymal transition-like change of breast cancer MDA-BF-1 with a kinase recognition site and a 7-histidine tag for receptor cells. Oncogenesis 2013;2:e65. binding and purification. Protein Expr Purif 2005;44:58–64. 26. Sage H, Vernon RB, Funk SE, Everitt EA, Angello J. SPARC, a secreted protein 12. Lira CB, Chu K, Lee YC, Hu MC, Lin S-H. Expression of the extracellular associated with cellular proliferation, inhibits cell spreading in vitro and domain of OB-cadherin as an Fc fusion protein using bicistronic retroviral exhibits Caþ2-dependent binding to the extracellular matrix. J Cell Biol expression vector. Protein Expr Purif 2008;61:220–6. 1989;109:341–56. 13. Yu G, Lee YC, Cheng CJ, Wu CF, Song JH, Gallick GE, et al. RSK promotes 27. Young BA, Wang P, Goldblum SE. The counteradhesive protein SPARC prostate cancer progression in bone through ING3, CKAP2, and PTK6- regulates an endothelial paracellular pathway through protein tyrosine mediated cell survival. Mol Cancer Res 2015;13:348–57. phosphorylation. Biochem Biophys Res Commun 1998;251:320–7. 14. Koeneman KS, Yeung F, Chung LW. Osteomimetic properties of pros- 28. Murphy-Ullrich JE, Sage EH. Revisiting the matricellular concept. Matrix tate cancer cells: a hypothesis supporting the predilection of prostate Biol 2014;37:1–14. cancer metastasis and growth in the bone environment. Prostate 29. Murphy-Ullrich JE. The de-adhesive activity of matricellular proteins: is 1999;39:246–61. intermediate cell adhesion an adaptive state? J Clin Invest 2001;107:785–90.

4958 Cancer Res; 75(22) November 15, 2015 Cancer Research

Tumor Microenvironment Confers De Novo Therapy Resistance

30. Halder J, Kamat AA, Landen CN Jr, Han LY, Lutgendorf SK, Lin YG, et al. 38. Crompton BD, Carlton AL, Thorner AR, Christie AL, Du J, Calicchio ML, et al. Focal adhesion kinase targeting using in vivo short interfering RNA delivery High-throughput tyrosine kinase activity profiling identifies FAK as a can- in neutral liposomes for ovarian carcinoma therapy. Clin Cancer Res 2006; didate therapeutic target in Ewing sarcoma. Cancer Res 2013;73:2873–83. 12:4916–24. 39. Tai YL, Chen LC, Shen TL. Emerging roles of focal adhesion kinase in cancer. 31. Roberts WG, Ung E, Whalen P, Cooper B, Hulford C, Autry C, et al. BioMed Res Int 2015;2015:690690. Antitumor activity and pharmacology of a selective focal adhesion kinase 40. Malanchi I, Santamaria-Martinez A, Susanto E, Peng H, Lehr HA, Delaloye inhibitor, PF-562,271. Cancer Res 2008;68:1935–44. JF, et al. Interactions between cancer stem cells and their niche govern 32. Infante JR, Camidge DR, Mileshkin LR, Chen EX, Hicks RJ, Rischin D, et al. metastatic colonization. Nature 2012;481:85–9. Safety, pharmacokinetic, and pharmacodynamic phase I dose-escalation 41. Oskarsson T, Acharyya S, Zhang XH, Vanharanta S, Tavazoie SF, Morris PG, trial of PF-00562271, an inhibitor of focal adhesion kinase, in advanced et al. Breast cancer cells produce tenascin C as a metastatic niche compo- solid tumors. J Clin Oncol 2012;30:1527–33. nent to colonize the lungs. Nat Med 2011;17:867–74. 33. Goel HL, Li J, Kogan S, Languino LR. Integrins in prostate cancer progres- 42. Luo M, Fan H, Nagy T, Wei H, Wang C, Liu S, et al. Mammary epithelial- sion. Endocr Relat Cancer 2008;15:657–64. specific ablation of the focal adhesion kinase suppresses mammary tumor- 34. Barthel SR, Hays DL, Yazawa EM, Opperman M, Walley KC, Nimrichter L, igenesis by affecting mammary cancer stem/progenitor cells. Cancer Res et al. Definition of molecular determinants of prostate cancer cell bone 2009;69:466–74. extravasation. Cancer Res 2013;73:942–52. 43. Shapiro IM, Kolev VN, Vidal CM, Kadariya Y, Ring JE, Wright Q, et al. Merlin 35. Wirth M, Heidenreich A, Gschwend JE, Gil T, Zastrow S, Laniado M, et al. A deficiency predicts FAK inhibitor sensitivity: a synthetic lethal relationship. multicenter phase 1 study of EMD 525797 (DI17E6), a novel humanized Sci Transl Med 2014;6:237ra68. monoclonal antibody targeting alphav integrins, in progressive castration- 44. Varkaris A, Corn PG, Parikh NU, Efstathiou E, Song JH, Lee YC, et al. resistant prostate cancer with bone metastases after chemotherapy. Eur Integrating murine and clinical trials with cabozantinib to understand roles Urol 2014;65:897–904. of MET and VEGFR-2 as targets for growth inhibition of prostate cancer. 36. Uhl W, Zuhlsdorf M, Koernicke T, Forssmann U, Kovar A. Safety, tolera- Clin Cancer Res 2015. bility, and pharmacokinetics of the novel alphav-integrin antibody EMD 45. Parker C, Nilsson S, Heinrich D, Helle SI, O'Sullivan JM, Fossa SD, et al. 525797 (DI17E6) in healthy subjects after ascending single intravenous Alpha emitter radium-223 and survival in metastatic prostate cancer. N doses. Invest New Drugs 2014;32:347–54. Engl J Med 2013;369:213–23. 37. Carbonell WS, DeLay M, Jahangiri A, Park CC, Aghi MK. beta1 46. Sartor O, Coleman R, Nilsson S, Heinrich D, Helle SI, O'Sullivan JM, et al. integrin targeting potentiates antiangiogenic therapy and inhibits the Effect of radium-223 dichloride on symptomatic skeletal events in patients growth of bevacizumab-resistant glioblastoma. Cancer Res 2013;73: with castration-resistant prostate cancer and bone metastases: results from 3145–54. a phase 3, double-blind, randomised trial. Lancet Oncol 2014;15:738–46.

www.aacrjournals.org Cancer Res; 75(22) November 15, 2015 4959

Identification of Bone-Derived Factors Conferring De Novo Therapeutic Resistance in Metastatic Prostate Cancer

Yu-Chen Lee, Song-Chang Lin, Guoyu Yu, et al.

Cancer Res 2015;75:4949-4959. Published OnlineFirst November 3, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-15-1215

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2015/12/10/0008-5472.CAN-15-1215.DC1

Cited articles This article cites 45 articles, 20 of which you can access for free at: http://cancerres.aacrjournals.org/content/75/22/4949.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/75/22/4949.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://cancerres.aacrjournals.org/content/75/22/4949. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.