Augmentation of Therapeutic Responses in Melanoma by Inhibition of IRAK-1,-4

Published OnlineFirst October 4, 2012; DOI: 10.1158/0008-5472.CAN-12-0337

Cancer Therapeutics, Targets, and Chemical Biology Research

Ratika Srivastava1, Degui Geng1, Yingjia Liu1, Liqin Zheng3, Zhaoyang Li1, Mary Ann Joseph1, Colleen McKenna1, Navneeta Bansal1, Augusto Ochoa3, and Eduardo Davila1,2

Abstract Toll-like receptors (TLR) are expressed by a variety of cancers, including melanoma, but their functional contributions in cancer cells are uncertain. To approach this question, we evaluated the effects of stimulating or inhibiting the TLR/IL-1 receptor-associated kinases IRAK-1 and IRAK-4 in melanoma cells where their functions are largely unexplored. TLRs and TLR-related proteins were variably expressed in melanoma cell lines, with 42% expressing activated phospho-IRAK-1 constitutively and 85% expressing high levels of phospho-IRAK-4 in the absence of TLR stimulation. Immunohistochemical evaluation of melanoma tumor biopsies (n ¼ 242) revealed two distinct patient populations, one that expressed p-IRAK-4 levels similar to normal skin (55%) and one with signiﬁcantly higher levels than normal skin (45%). Levels of p-IRAK-4 levels did not correlate with clinical stage, gender, or age, but attenuated IRAK-1,-4 signaling with pharmacologic inhibitors or siRNA-enhanced cell death in vitro in combination with vinblastine. Moreover, in a xenograft mouse model of melanoma, the combined pharmacologic treatment delayed tumor growth and prolonged survival compared with subjects receiving single agent therapy. We propose p-IRAK-4 as a novel inﬂammation and prosurvival marker in melanoma with the potential to serve as a therapeutic target to enhance chemotherapeutic responses. Cancer Res; 72(23); 1–8. 2012 AACR.

Introduction leukin (IL)-1 receptor-associated kinase (IRAK) signaling – The incidence of melanoma has been on the rise in the (8 10). IL-1, IL-18, and IL-33 can also activate IRAK signal- United States and worldwide over the last 30 years and has the ing. IRAK-4 kinase activity is regulated by autophosphoryla- – fastest rising cancer incidence in the United States (1–3). tion (Ser346, Thr342, and Thr34; refs. 8 11), which in turn Melanoma is the 5th/6th most common cancer in men and can activate IRAK-1. IRAK-4,-1 activation results in the women, respectively (1–3). The median survival of patients downstream activation of various kinases and transcription k with advanced disease is approximately 6 months and the factors including JNK, AP-1, NF- B, and p38 mitogen-acti- survival rate at 5 years is 6% and 45% for Stage III patients (1, 2). vated protein kinase (MAPK), leading to the production of a fl Treatment failure is largely attributed to melanoma's resis- mixture of chemokines and proin ammatory cytokines a tance to all existing forms of cancer therapies. including TNF- , IL-1, IL-6, and IL-8 (12). IRAK signaling Recent reports indicate that Toll-like receptors (TLR) can also induce the expression of several proteins involved in signaling within nonimmune cells, including several types cell survival and division (13). of human cancers, can contribute to cancer progression In the current study, we examined the TLR and TLR signal- fi (4–7). TLRs recognize infectious microorganisms as well as ing-related protein expression pro le on various melanoma endogenous signals released by dying or stressed cells. The cell lines as well as the expression of the activated (phosphor- engagement of all known TLRs, except TLR3, initiates inter- ylated) form of IRAK-4 on patient biopsies. Cytokine production and cell survival in response to stimulating or inhibiting IRAK-1,-4 in melanomas were examined in vitro. In vivo, the Authors' Affiliations: 1Marlene and Stewart Greenebaum NCI Cancer therapeutic efficacy of combinatorial treatment with IRAK-1,-4 2 Center, University of Maryland; Department of Otorhinolaryngology, Head inhibitor plus chemotherapy was investigated in mice with an and Neck Surgery, Baltimore, Maryland; and 3Stanley S. Scott Cancer Center, Louisiana State University, New Orleans, Louisiana established human melanoma tumor. Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Materials and Methods

R. Srivastava and D. Geng contributed equally to this work. Melanoma cell lines Human melanoma cell lines were obtained from American Corresponding Author: Eduardo Davila, University of Maryland/Green- ebaum Cancer Center, Room 10-041, Bressler Research Building, 655 Type Culture Collection (ATCC) within 2 years of manuscript West Baltimore Street, Baltimore, MD 21201-1559. Phone: 410-706-5051; submission. Melanoma cell lines were initially expanded and Fax: 410-706-2484; E-mail: [email protected] cryopreserved within 1 month of receipt. Cells were typically doi: 10.1158/0008-5472.CAN-12-0337 used for 6 months, at which time a fresh vial of cryopreserved 2012 American Association for Cancer Research. cells was used. Malme-3M cells were maintained in Iscove's

www.aacrjournals.org OF1

Srivastava et al.

Modified Dulbecco's Medium, SK-MEL-2, WM115, C32, and that mice received with IRAK-1,-4 inhibitor. Tumor sizes were RPMI-7951 in Eagle's Minimum Essential Medium, A375 in analyzed using a mixed model approach for repeated measure- Dulbecco's Modified Eagle's Medium, and G361 in McCoy's 5a ments and mouse survival data were analyzed with the exact Medium Modified. All media were purchased from Invitrogen log-rank test. Life Technologies and supplemented with FBS, penicillin, and streptomycin according to culture media recommended by TLR agonists, IRAK inhibitor, chemotherapies, and flow ATCC. cytometry The TLR1-TLR2 ligand tripalmitoyl-S-(bis(palmitoyloxy) Western blotting and immunohistochemical staining propyl)-Cys-Ser-(Lys)3-Lys (Pam3CysK4) was purchased from Total cell extracts were prepared from melanoma cells. Invivogen. In some experiments we used the IRAK-1/4 inhib- Proteins (20–30 mg/lane from cell lines) were resolved in itor (EMD Millipore), which is a cell-permeable benzimidazole Tris-glycine SDS gels and transferred to polyvinylidene difluor- compound that selectively inhibits IRAK-1 and -4 and exhibits ide membranes. The membrane was blocked for 4 hours with little activity against a panel of 27 other kinases. In some 5% milk in PBS and 0.05% Tween-20, followed by incubation experiments cells were treated with the indicated concentra- with antibodies against TLR1 (N-23), phosphorylated (p)- tions of IRAK-1,-4 inhibitor and vinblastine, cisplatin, and 5- IRAK-1 (Ser 376), total IRAK-1 (H-273; Santa Cruz Biotechnol- fluorouracil (Sigma-Aldrich), and apoptosis was quantitated by ogy); TLR2, TLR4, TLR7, TLR9, TRIF, IRAK-M, IRAK-4, Tollip, flow cytometry after staining cells with fluorescein isothiocy- glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 14C10), anate-labeled Annexin-V (BD Pharmingen) and propidium and b-actin (Cell Signaling); TLR3/CD283, TLR5 (IMGENEX); iodide (PI; Sigma-Aldrich). TLR6 (3D10H11), TLR8 (44C143), TLR10, MyD88 (Abcam), PARP, caspase-3 (Cell Signaling), overnight at 4C and, sub- Results and Discussion sequently, incubated with horseradish peroxidase-conjugated TLR expression profiles on melanoma cell lines secondary antibody, and detected using enhanced chemilumi- Previous reports have shown the presence of TLR mRNA nescence (ECL Plus; Amersham Pharmacia Biotech). Melano- transcripts and a limited number of TLRs by flow cytometry ma tissue arrays were purchased from US Biomax, Inc. For on human melanomas (4, 6). We examined in greater detail immunohistochemical staining, samples were stained with the expression profile of TLR1 through TLR10 and several mouse monoclonal anti-human phosphor-IRAK-4 antibody TLR-related signaling proteins on various melanoma cell (Abnova; dilution, 3.5 mg/mL) or normal mouse serum. Tissue lines by Western blotting. With the exception of C32 and histology sections (5-mm-thick) were deparaffinized, hydrated, Malme-3M, all melanoma cells expressed relatively high but heated in boiling 10 mmol/L sodium citrate (pH 6.0), for variable levels of TLR1 (Fig. 1A). Appreciable levels of TLR2 antigen retrieval, for 30 minutes, and washed in Tris buffer. and TLR3 were detected in SK-Mel-2 and A375. TLR3 was Peroxide blocking was done with 3% H2O2 in methanol at room also moderately expressed on RPMI-7951 and G361 cells. temperature for 15 minutes, followed by 10% FBS in TBST for Most cell lines expressed variable levels of TLR4 and TLR7. 30 minutes at room temperature. Primary antibody incubation TLR5 was strongly expressed on SK-Mel, WM115, and A375, was done overnight at 4C. Secondary antibody incubation moderately expressed on G361 and Malm-3M, but weakly with horse anti-mouse (1:200; Vector Laboratories) was done detected on C32 and RPMI-7951. TLR8 levels were low on for 1 hour, followed by application of diaminobenzidine chro- SK-Mel-2 and A375. TLR9 was strongly expressed on SK-Mel- mogen for 5 minutes. The slides were then counterstained 2 and WM115 and moderately expressed on Malm-3M and with haematoxylin and topped with a coverslip. Immunohis- G361. TLR10 expression was weak and variable on the tochemistry (IHC) specimens were analyzed using "quick different cell lines. All cell lines expressed the TLR adapter score" system in which the intensity of the immunohistochem- molecules MyD88 and TRIF. ical reaction was scored by multiplying the intensity 1 (weak), 2 IRAK-1 and IRAK-4 play a central role in TLR-mediated (moderate), and 3 (strong) and the proportion of cells staining signaling. All of the melanoma lines expressed high levels of positively 0 (<10%), 1 (10%–25%), 2 (26%–50%), 3 (51%–75%), or IRAK-4 and variable levels of IRAK-1 (Fig. 1A). Interestingly, 4(>75%). In some experiments, A375 cells were transfected via total IRAK-1 as well as the activated form of IRAK-1 (phos- electroporation (Amaxa, Program X-001, Lonza buffer L) with phorylated at serine 376; p-IRAK-1) was strongly expressed in siRNA-hIRAK-1 and siRNA-hIRAK-4 (5 mg/mL; InVivogen). Malme-3M, SK-MEL-2, and A375, specifically in the absence of exogenous TLR agonists. Similarly, variable levels of p-IRAK-4 A375 melanoma xenograft model (at threonine 345) were detected in melanoma cells (Fig. 1A). NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ IL2RG (NSG) mice (Jackson We also examined whether TLR stimulation could augment p- Laboratories) were subcutaneously injected with 2.5 106 IRAK levels in cells that expressed IRAK or induce p-IRAK in A375 melanoma cells delivered in PBS. The Institutional Ani- cells deficient in this protein. However, p-IRAK levels in A375 mal Care and Use Committee approved use of mice. When cells, which express relatively high p-IRAK-1 and p-IRAK-4, tumors reached a size of approximately 50 mm2 mice were remained unchanged following TLR1-TLR2 stimulation sug- injected intratumorally with 35 mg/kg of IRAK-1,-4 inhibitor or gesting that phosphorylated levels may already be at or near an equal volume of vehicle (dimethyl sulfoxide; DMSO). Mice the maximum, Supplementary Fig. S5B. In G361 cells, which were injected via intraperitoneal injection with vinblastine express low levels of p-IRAK-1 (and which express TLR5), (0.25 mg/kg) every 2 days for 5 days starting on the same day neither TLR1-TLR2 agonist (Pam3CysK4) nor TLR5 (Flagellin)

OF2 Cancer Res; 72(23) December 1, 2012 Cancer Research

IRAK-4 Signaling and Melanoma Progression

A B

VEGF CXCL1 IL-8 IP-10 600 6,000 45 5000 **

SK-Mel-2 * * * G361 Malme-3M WM115 A375 C32 RPMI-7951 TLR1 400 4,000 30 2500 TLR2 200 2,000 15

TLR3 pg/mL 0 0 0 0 TLR4 GM-CSF MCP-1 IL-6 Fractal TLR5 400 60 20 3,000 TLR6 * 40 –TLR1/2 Lig TLR7 200 2,000 10 TLR8 pg/mL 1,000 20 +TLR1/2 Lig TLR9 0 0 0 0 TLR10 TRIF C MyD88 A375 C32 G361 RPMI-7951 IRAK1 p-NF-κB (65 KDa) p-IRAK1 (S376) NF-κB (total) IRAK4 p-IRAK-4 (T345) + + + –––– + IRAK1/4 Inhib GAPDH

D VEGF CXCL1 MCP-1 PDGF-A FGF-2 450 6,000 400 30 ** ** 3,000 ** ** ** 300 4,000 20 – IRAK1/4 Inhib 1,500 200 150 2,000 10 + IRAK1/4 Inhib pg/mlL 0 0 0 0 0

Figure 1. TLR and TLR signaling-related protein expression profiles on melanoma cell lines. A, the expression levels of TLR1-TLR10 and various TLR-related proteins were examined by Western blotting in samples from human melanoma cell lines. These data are representative of more than 3 independent experiments, each showing similar trends. B, the TLR1-TLR2 ligand Pam3CysK4 (2.5 mg/mL) was added to A375 cells for 48 hours. Cytokine and chemokine production was determined using a Milliplex cytokine/chemokine array. C, IRAK-1,-4 inhibitor or vehicle alone (DMSO) was added to melanoma cells for 48 hours. The level of phosphorylated and nonphosphorylated p65 subunit of NF-kB was determined by Western blotting. D, cytokine and chemokine levels in the supernatant A375 cell culture were examined using a Milliplex cytokine/chemokine array 48 hours after adding IRAK-1, -4 inhibitor or DMSO. These data are representative of at least 3 experiments, each showing similar trends. ANOVA; , P < 0.001; , P < 0.05. stimulation increased or induced p-IRAK-4 or p-IRAK-4 as factors capable of promoting tumor growth such as angio- expression levels. Inexplicably, however, the TLR5 agonist genenic and inflammatory cytokines. We compared the cyto- flagellin reduced total and p-IRAK-4 levels in both cell lines. kine/chemokine profile between A375 cells stimulated with the This is the first report showing the expression of constitu- TLR1-TLR2 agonist Pam3CysK4 and untreated cells. TLR stim- tively phosphorylated IRAK-1 and IRAK-4 on human cutane- ulation significantly augmented the production levels of var- ous melanoma cells. These data also represent a comprehen- ious factors including those associated with angiogenesis such sive protein expression profile of TLRs and TLR-signaling as VEGF, the melanoma growth factor chemokine ligand-1 proteins on melanoma cells and highlight the differences in (CXCL1), and IL-8, which promote cell survival and prolifer- the expression of these proteins in different melanoma lines. It ation (Fig. 1B; P < 0.05; ANOVA; refs. 14–16). The levels of is worth noting however, that Western blotting was used to granulocyte macrophage colony-stimulating factor (GM-CSF) detect total TLR proteins levels versus flow cytometry, which and IP-10 were also increased following addition of the TLR1- detects surface TLRs. TLR2 agonist (Fig. 1B; P < 0.001; ANOVA). TLR1-TLR2 also increased MCP-1 and IL-6 levels but appeared to reduce Cytokine/chemokine production by melanoma cells fractalkine concentrations. To further confirm that the TLR- following activation or inhibition of IRAK-1,-4 IRAK signaling pathway was intact in melanoma cells and that The stimulation of TLR-MyD88 or IL-1/18/33–MyD88 acti- changes in cytokines/chemokines are a result of activating this vates IRAK-1,-4 resulting in the expression of various chemo- pathway, we transiently overexpressed IRAK-1 in G361 mela- kines and cytokines involved in cell survival and division as well noma cells and compared changes in cytokine/chemokine

www.aacrjournals.org Cancer Res; 72(23) December 1, 2012 OF3

Srivastava et al.

levels with control G361 cells. Overexpressing IRAK-1 vinblastine, 50fluorauracil, and cisplatin. The data in Fig. 2A increased the levels of various cytokines/chemokines including show increased sensitivity of both cell lines to vinblastine VEGF, CXCL1, G-CSF, and IL-12p40. IRAK-1 expression also (left panels). The addition of 50-fluorouracil and IRAK-1,-4 induced the expression of IP-10, G-CSF, and PDGF-AA but had inhibitor also appeared to enhance the apoptosis of Mal- no effect on EGF production, as shown in Supplementary Fig. me3M cells but had no effect on A375 cells (Fig. 2A, middle S1. Collectively, these data indicate that melanoma cells panels). IRAK-1,-4 inhibition did not increase cisplatin's express a functional TLR-IRAK signaling pathway and that cytotoxicity of either cell line (Fig. 2A, right panels). We the activation of this pathway might play a role in promoting examined in greater detail the role that IRAK had in medi- cell survival or proliferation in part through the production and ating the chemoprotective effects. A375 cells engineered to chemokines/cytokines. knockdown IRAK-1 expression were treated with vinblas- On the basis that melanoma cells exhibited increased levels tine, and apoptosis was measured 48 hours later. psi-RNA of phosphorylated IRAK-1 and IRAK-4 and IRAK signaling reduced IRAK protein levels more than 95%; a representative results in the activation of various transcription factors, we image is shown in Fig. 2B. Reducing IRAK-1 or IRAK-4 examined the outcome of inhibiting IRAK signaling in mela- expression levels augmented the toxic effects of vinblastine, noma cells. Melanoma cells cultured in the presence of an as compared with cells transfected with the control IRAK-1,-4 inhibitor showed marked reduction of phosphory- plasmid, Fig. 2B (right panel). Cells coexpressing psi-RNA- lated NF-kB (p-NF-kB) in all 4 melanoma cell lines tested as hIRAK-1 and -4 exhibited a higher percentage of death than compared with cells treated with vehicle alone (DMSO), Fig. cells expressing either of these vectors alone, Fig. 2B (right 1C. Furthermore, IRAK-1,-4 inhibition reduced the production panel). It is worth noting that IRAK inhibitor alone or cells of VEGF more than 90% and diminished CXCL1, monocyte expressing psiRNA-IRAK moderately increased melanoma chemotactic protein-1 (MCP1), platelet-derived growth factor apoptosis (in the absence of vinblastine), suggesting that alpha (PDGF-A), and fibroblast growth factor (FGF-2) levels in IRAK signaling plays an important role in cell survival. A375 cells (Fig 1D, P < 0.05). The fact that that IRAK-1,-4 IRAK-1,-4 inhibition resulted in reduced levels of activated inhibition reduced the levels of these molecules in the absence NF-kB (Fig. 1C). Therefore, we examined whether IRAK–NF- of exogenous TLR agonists, suggests that IRAK-1,-4 contributes kB signaling is linked to chemoresistance by treating A375 greatly to cell function through the production of various melanoma cells with and without NF-kB inhibitor and in the factors. It is worth highlighting that the addition of IRAK-1,- absence and presence of vinblastine. Interestingly, NF-kB 4 inhibitors also decreased the effects of TLR1-TLR2 agonist inhibition did not alter A3750s sensitivity to vinblastine (Supplementary Fig. S2), further confirming that changes in (Supplementary Fig. S3A; P ¼ 0.14). Furthermore, because chemokines/cytokines occurred in a TLR-MyD88-IRAK the toxic effects of doxorubicin have been shown to involve fashion. NF-kB signaling, we also examined the effects of inhibiting Collectively, these findings support the contention that the NF-kB in A375 melanoma cells (18). In sharp contrast to the activation of IRAK-1,-4 signaling on melanoma cells in vivo combinatorial effects of NF-kB inhibition plus vinblastine, might contribute to cancer progression by inducing the expres- NF-kB inhibition increased the toxic effects of doxorubicin sion of various chemokines and cytokines beneficial to cancer on A375, Supplementary Fig. S3A. These data suggest that cell survival, division, and/or angiogenesis. Furthermore, the although TLR-IRAK signaling can activate NF-kBandinhi- inhibition of IRAK-1,-4 drastically reduced the production of biting IRAK signaling can reduce p-NF-kB levels, the che- several cytokines/chemokines, highlighting the possibility that moprotective effects observed appear to be mediated via a IRAK-1,-4 signaling plays a central role in cytokine/chemokine mechanism independent of NF-kB. Because IRAK signaling production even in absence of exogenous TLR agonists. How- results in the activation of various transcription factors ever, these data do not exclude the possibility that TLRs including c-jun/Fos, Elk-1, and cAMP-responsive element recognize endogenous TLR agonists or that other cytokines binding protein (CREB), we postulate that perhaps IRAK produced in response to IRAK signaling might further poten- signaling is linked to chemoresistance via the activation of tiate this signaling pathway. one of these factors. The possibility that these transcription factors could impart melanoma cells a prosurvival signal is Combining IRAK inhibition with certain chemotherapies substantiated by several studies highlighting that these augments melanoma cell apoptosis in vitro and in vivo transcription factors play a critical role in melanoma pro- Melanoma cells become resistant to a variety of che- gression (19–22). motherapies by altering their survival signaling pathways Alternatively, cytokines/chemokines produced in response during cancer progression (17). Various studies have docu- to TLR-IRAK signaling might also contribute to the chemo- mented the prosurvival effects that TLR signaling have on protective effects observed. Therefore, supernatant from different cell types (13). Considering the impact that inhibiting TLR1-TLR2-stimulated or unstimulated A375 melanoma cells IRAK-1,-4 had on NF-kB activation and chemokine/cytokine was added to the C32 melanoma cells in the presence or production, we examined whether IRAK-1,-4 inhibitor might absence of vinblastine. Another group of cells were treated be used as a means to augment the toxic effects of certain with TLR1-TLR2 ligand and apoptosis was examined by flow chemotherapies, which alone are not very effective. The mel- cytometry. The Supplementary Fig. S3B show that supernatant anoma cell lines A375 and Malme-3M were cultured in the from TLR-stimulated A375 cells moderately reduced vinblas- presence of IRAK-1,-4 inhibitor and various concentrations of tine-induced apoptosis as compared with untreated cells (P <

OF4 Cancer Res; 72(23) December 1, 2012 Cancer Research

IRAK-4 Signaling and Melanoma Progression

A 80 80 80 Figure 2. IRAK-1,-4 inhibition ) augments vinblastine-mediated + 60 60 60 apoptosis of melanoma cells. A, 40 40 40 A375 and Malme-3M melanoma cells A375 20

Annexin 20 20 were cultured with or without 2.5 + mmol/L of IRAK-1,-4 inhibitor in the 0 0 0 +IRAK1/4Inhib presence varying concentrations 0 100 200 0 25 50 75 100 02550 of vinblastine, cisplatin, or –IRAK1/4Inhib 0 40 40 40 5 Fluorouracil. Forty-eight hours later apoptosis was evaluated by 30 30 30 staining cells with Annexin-V and PI 20 20 20 and analyzed by ﬂow cytometry. The result from 1 of 5 experiments is 10 10 10 shown. B, A375 cells were 0 0 0 Malme3M Percentage apoptosis (PI transfected via electroporation with 0 100 200 0 25 50 75 100 02550 siRNA-IRAK-1,-4. IRAK-1 protein levels were examined by Western Vinblastine (nmol/L) 5’Fluoro Cisplatin (nmol/L) blotting. A375-control, A375- psiRNA-hIRAK-1, A375-psiRNA- B A375 melanoma hIRAK-4, or cells expressing both 60 psiRNA-hIRAK-1 and -4 were 50 Con plasmid cultured in the presence of IRAK-1 40 vinblastine (100 nmol/L) for 48 hours (78 KDa) 30 psiRNA-IRAK-1 ﬂ and apoptosis was examined by ow 20 psiRNA-IRAK-4 cytometry. C, A375 melanoma cells 10 were cultured in the presence of psiRNA-IRAK-1 + 0 vehicle alone (DMSO) or various Percentage apoptosis psiRNA-IRAK-4 0 100 concentrations of IRAK-1,-4. Forty- siRNA IRAK1 eight hours later cell lysates were Control vector Vinblastine (nmol/L) used to analyze the expression of levels of cleaved PARP or caspase-3 C D by Western blotting. D, A375 PARP (cleaved) PARP melanoma cells were cultured with or Casp-3 (cleaved) without 2.5 mmol/L of IRAK-1,-4 and Casp-3 in the presence or absence of β vinblastine for 48 hours. PARP and -Actin Vinblastine +–– + caspase-3 levels were determined by DMSO 11020 IRAK-1,-4 Inhibitor ––+ + Western blotting. μmol/L IRAK-1,-4 In

0.05; ANOVA), whereas TLR1-TLR2 ligand had little effect. PARP and caspase-3. In contrast, treatment with vinblastine Furthermore, neither IL-1 or CXCL1 appeared to be the cyto- or IRAK-1,-4 inhibitor alone had little effect on the levels of kines responsible for the observed chemoprotective effects, these molecules. Supplementary Fig. S3C. These data suggest that cytokines/ We also sought to determine whether inhibitor-treated cells chemokines produced in response to TLR-IRAK stimulation exhibited changes in the levels of apoptosis-related molecules. can contribute to cell survival; however, other IRAK-mediated We compared the expression of various apoptosis-related factors appear to play a more important role. It is also worth genes in A375 melanoma cell line treated with DMSO vehicle noting that different melanoma cell lines might be more or less (control), IRAK inhibitor, and vinblastine and observed a gene responsive to factors produced in response to TLR-IRAK profile that favored apoptosis in the presence of inhibitor alone stimulation based on the expression levels of different cyto- or vinblastine (Supplementary Fig. S4). Taken together with kine/chemokine receptors. data presented in Fig. 2, these gene array data confirm that As an independent biochemical assay to confirm that inhibiting the IRAK pathway in melanoma cells is important A375 melanoma cells triggered apoptosis, we also conducted for their survival and indicate that interfering with this path- Western blot analysis using antibodies specific for caspase-3 way can sensitize melanoma cells or at lease function in and PARP. Active caspase-3, which is cleaved to yield cat- concert with certain chemotherapies to enhance their toxic alytically active subunits, was detected following the addi- effects. These data also show that some chemotherapeutic tion of 10 mmol/L IRAK-1,-4 inhibitor (Fig. 2C). Enhanced drugs, such as vinblastine, that have been deemed ineffective accumulation of cleaved PARP, which is targeted for cas- might be rendered therapeutically useful when combined with pase-dependent proteolysis, was also observed in IRAK-1,-4 a TLR/IRAK inhibitor. inhibitor-treated cells. Increased levels of cleaved PARP and The antitumor effects of combinatorial therapy using vin- caspase-3 occurred in a concentration-dependent manner blastine and IRAK-1,-4 inhibitor were tested in vivo. A375 cells (Fig. 2C). As shown in Fig. 2D, the combination of vinblastine were subcutaneously injected into NSG mice and grown to 30 and IRAK-1,-4 inhibitor also increased the levels of cleaved to 50 mm2. Mice were injected intraperitoneally with

www.aacrjournals.org Cancer Res; 72(23) December 1, 2012 OF5

Srivastava et al.

* * A B 250 100 ) 2 * 200 80 None

150 60 * Vin IRAK-1,-4 In 100 40 Vin + IRAK-1,-4 In 50 Percent survival 20 Tumor size (mm

0 0 02040 02040 Days posttreatment Days posttreatment

Figure 3. Treatment with IRAK-1,-4 inhibitor enhances the therapeutic effects of vinblastine in mice with an established human melanoma tumor. NSG mice (n ¼ 5) were injected subcutaneously with A375 melanoma cells. When tumors reached a size of approximately 50 mm2, mice remained untreated or were injected intraperitoneally with vinblastine (0.25 mg/kg; administered every 2 days for 10 days) or peritumorally with IRAK-1,-4 inhibitor (35 mg/kg) or with vinblastine plus IRAK-1,-4 inhibitor. Tumor sizes were calculated by measuring perpendicular by longitudinal diameter. , P < 0.05. Mouse survival data were analyzed using the exact log-rank test. , P < 0.05; , P < 0.001.

vinblastine or intratumorally with IRAK-1,-4 inhibitor (35 various clinical stages. Although little staining was observed in mg/kg) or with vinblastine or IRAK-1,-4 inhibitor alone. Mice normal skin, p-IRAK-4 was highly expressed in melanoma receiving IRAK-1,-4 inhibitors plus vinblastine showed a samples (representative staining are shown in Fig. 4A). We marked reduction in tumor growth and improved mouse sought to determine the association between p-IRAK-4 expres- survival (median survival: 38 days) as compared with mice sion and the clinical stage. As shown in Fig. 4B, p-IRAK-4 levels receiving vehicle (DMSO) plus vinblastine (median survival: 22 were not linked to melanoma stage nor was there a correlation days) or mice receiving IRAK-1,-4 inhibitor (median survival: 19 between p-IRAK-4 levels and metastasis. Unexpectedly, how- days), Fig. 3A and B. Mice receiving single therapy for IRAK-1,-4 ever, quantification of the staining revealed 2 distinct groups in inhibitor or vinblastine showed modest tumor growth delay melanoma samples in all clinical stages; one group of mela- but similar survival as compared with control mice (median noma samples expressed p-IRAK-4 levels similar to those on survival: 17 days). Despite a tumor growth delay, mice receiving normal skin, and another group expressed significantly higher IRAK-1,-4 inhibitors plus vinblastine succumbed to tumor p-IRAK-4 levels (Fig. 4B; P < 0.005; ANOVA). Samples from stage challenge 40 days after initiation of treatment. Nonetheless, I through IV melanoma also showed a distinct division these results emphasize the antitumor effects of combinatorial between high and low/no expression of phosphorylated therapy and highlight the need for further optimization. Iden- IRAK-4 (Fig. 4B; P < 0.05; ANOVA). Of the 242 melanoma tifying novel signaling pathways that can improve the efficacy samples analyzed, nearly half expressed elevated levels of p- of chemotherapeutic drugs to treat melanoma patients is IRAK-4. It is worth noting that the elevated levels of p-IRAK-4 in critical, considering that advanced melanoma is highly resis- melanoma patients did not correlate with patient age or tant to all known chemotherapies (17). The synergistic effects gender (Fig. 4C). of IRAK-1,-4 inhibitor and vinblastine may reside in the mech- Currently, there is no effective cure for patients with anism of action by vinblastine. Whereas, vinblastine kills cells advanced melanoma. This is due in large part to their high by suppressing microtubule dynamics, cisplatin and resistance to chemotherapeutic drugs. Ongoing studies by our 50fluorouracil damage DNA by inducing DNA (and protein) group are focused on identifying the association between total cross-linking, and by irreversibly inhibiting thymidylate and phosphorylated IRAK-1 and IRAK-4 expression levels with synthase, respectively. Ongoing studies by our group are clinical outcome including response to treatment and survival. focused on identifying additional chemotherapeutic drugs that The immunohistochemical data presented here are in line with are enhanced when used in combination with IRAK-1,-4 the Western blotting data showing that a subset (42%; 3 of 7) of inhibitor. melanoma cell lines express p-IRAK. Ongoing studies are Altogether, these findings corroborate a link between IRAK- focused on determining the use of total and/or p-IRAK-1,-4 1,-4 signaling in melanoma cells and survival and suggest a as biomarkers for prediction of therapy resistance and patient phenotype supportive of melanoma progression in cancer survival. It is important to note that recent studies support our patients. The importance of IRAK-1/IRAK-4 signaling and cell contention that IRAK signaling contributes to cancer progres- survival is further highlighted in immune cells in which TLR sion. Boukerche and colleagues found that IRAK-1 gene tran- engagement enhances the expression of various antiapoptotic script levels in melanoma patients were associated with highly genes (13). metastatic melanoma individuals, whereas Li and colleagues observed that reexpression of miR-146a in pancreatic cancer Phosphorylated IRAK-4 in melanoma biopsies lines downregulated IRAK-1 expression as well as NF-kB Phosphorylated IRAK-4 expression was analyzed by IHC in activity that blocked pancreatic cancer cell invasiveness and normal skin and in melanoma tissue derived from patients at metastasis (23, 24).

OF6 Cancer Res; 72(23) December 1, 2012 Cancer Research

IRAK-4 Signaling and Melanoma Progression

A Melanoma; stage Normal skin I II III IV p-IRAK-4 p-IRAK-4

Figure 4. Expression of p-IRAK-4 in melanoma cells biopsy specimens. A, biopsies obtained from melanoma S100

patients at different clinical stages Ms serum were stained for phosphorylated IRAK-4 or S100 and examined by IHC. Mouse serum was used as a B 300 control to examine nonspecific antibody staining. Representative staining from patients at different 200 stages of cancer or normal skin are fi * shown ( 20 eld sections). B and C, 100 quantification of p-IRAK-4 staining as a function of clinical stage, age, and gender. The number in p-IRAK-4 expression 0 parentheses below the clinical stage Norm skin I II III IV is the number of patient samples Met analyzed. The dotted line in C (left (20) (6) (127) (18) (5) (86) panel) represents the trend line from all samples and the bar line in the C 300 right panel is the average p-IRAK-4 R² = 0.001 300 value generated from samples 250 250 obtained from female and male 200 patients. 200 150 150 100 100 50

p-IRAK-4 expression p-IRAK-4 expression 50 0 0 20 40 60 80 F M Age (Yrs)

The studies presented here highlight the possibility of discovery of more effective therapies to treat melanoma identifying new chemotherapeutic agents that might be patients. currently in use to treat other cancers or other disorders but might be repurposed against melanoma when joined Disclosure of Potential Conﬂicts of Interest No potential conﬂicts of interest were disclosed. with IRAK inhibitors. Although new drugs such as vemurafenib (BRAF enzyme inhibitor), ipilimumab (anti-CTLA-4), Authors' Contributions BMS-936558 (anti-PD1), and BMS-663513 (anti-CD137; Conception and design: Z. Li, E. Davila ref. 25) show antitumor activity, the effects are quite mod- Development of methodology: E. Davila Acquisition of data (provided animals, acquired and managed patients, erate, prolonging survival 2 to 6 months. In addition, provided facilities, etc.): D. Geng, M.A. Joseph, C. McKenna patients eventually develop a resistance to vemurafenib. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, For this reason, Plexxikon and Roche are currently testing computational analysis): Z. Li, E. Davila Writing, review, and/or revision of the manuscript: Y. Liu, A.C. Ochoa, E. vemurafenib in combination with other signaling pathway Davila inhibitors such as mitogen-activated protein inhibitors in an Administrative, technical, or material support (i.e., reporting or orga- attempt to prevent the development of tumor chemoresis- nizing data, constructing databases): N. Bansal Conducting experiments: R. Srivastava, Y. Liu, L. Zheng tance. Unfortunately, no other therapies to date have proven effective at improving overall survival of these patients Grant Support (3, 26, 27), and the development of new therapeutic agents National Cancer Institute 1R01CA140917-01, NIH Center for Biomedical has not kept up with the increased melanoma incidence Research Center Excellence grants 1P20 RR021970, and University of Maryland, (28). We propose that understanding how IRAK inhibition Marlene and Stewart Greenebaum Cancer Center. sensitizes melanoma to chemotherapies (or augments their Received February 2, 2012; revised September 5, 2012; accepted September 5, toxic effect) might provide novel insights and allow the 2012; published OnlineFirst October 4, 2012.

www.aacrjournals.org Cancer Res; 72(23) December 1, 2012 OF7

Srivastava et al.

References 1. Barth A, Wanek LA, Morton DL. Prognostic factors in 1,521 mel- 16. Singh RK, Gutman M, Radinsky R, Bucana CD, Fidler IJ. Expression of anoma patients with distant metastases. J Am Coll Surg 1995;181: interleukin 8 correlates with the metastatic potential of human mela- 193–201. noma cells in nude mice. Cancer Res 1994;54:3242–7. 2. Tawbi HA, Buch SC. Chemotherapy resistance abrogation in meta- 17. Soengas MS, Lowe SW. Apoptosis and melanoma chemoresistance. static melanoma. Clin Adv Hematol Oncol 2010;8:259–66. Oncogene 2003;22:3138–51. 3. Eggermont AM, Robert C. New drugs in melanoma: it's a whole new 18. Arlt A, Vorndamm J, Breitenbroich M, Folsch UR, Kalthoff H, Schmidt world. Eur J Cancer 2011;47:2150–7. WE, et al. Inhibition of NF-kappaB sensitizes human pancreatic car- 4. Goto Y, Arigami T, Kitago M, Nguyen SL, Narita N, Ferrone S, et al. cinoma cells to apoptosis induced by etoposide (VP16) or doxorubicin. Activation of Toll-like receptors 2, 3, and 4 on human melanoma Oncogene 2001;20:859–68. cells induces inflammatory factors. Mol Cancer Ther 2008;7: 19. Dobroff AS, Wang H, Melnikova VO, Villares GJ, Zigler M, Huang L, 3642–53. et al. Silencing cAMP-response element-binding protein (CREB) iden- 5. Molteni M, Marabella D, Orlandi C, Rossetti C. Melanoma cell lines are tifies CYR61 as a tumor suppressor gene in melanoma. J Biol Chem responsive in vitro to lipopolysaccharide and express TLR-4. Cancer 2009;284:26194–206. Lett 2006;235:75–83. 20. Gorden A, Osman I, Gai W, He D, Huang W, Davidson A, et al. Analysis 6. Saint-Jean M, Knol AC, Nguyen JM, Khammari A, Dreno B. TLR of BRAF and N-RAS mutations in metastatic melanoma tissues. expression in human melanoma cells. Eur J Dermatol 2011;21: Cancer Res 2003;63:3955–7. 899–905. 21. Woods D, Cherwinski H, Venetsanakos E, Bhat A, Gysin S, Humbert M, 7. Salaun B, Lebecque S, Matikainen S, Rimoldi D, Romero P. Toll-like et al. Induction of beta3-integrin gene expression by sustained acti- receptor 3 expressed by melanoma cells as a target for therapy? Clin vation of the Ras-regulated Raf-MEK-extracellular signal-regulated Cancer Res 2007;13:4565–74. kinase signaling pathway. Mol Cell Biol 2001;21:3192–205. 8. Cao Z, Henzel WJ, Gao X. IRAK: a kinase associated with the inter- 22. Yang S, McNulty S, Meyskens FL Jr. During human melanoma pro- leukin-1 receptor. Science 1996;271:1128–31. gression AP-1 binding pairs are altered with loss of c-Jun in vitro. 9. Li S, Strelow A, Fontana EJ, Wesche H. IRAK-4: a novel member of the Pigment Cell Res 2004;17:74–83. IRAK family with the properties of an IRAK-kinase. Proc Natl Acad Sci 23. Boukerche H, Su ZZ, Kang DC, Fisher PB. Identification and cloning of U S A 2002;99:5567–72. genes displaying elevated expression as a consequence of metastatic 10. Wang Z, Wesche H, Stevens T, Walker N, Yeh WC. IRAK-4 inhibitors for progression in human melanoma cells by rapid subtraction hybridiza- inflammation. Curr Top Med Chem 2009;9:724–37. tion. Gene 2004;343:191–201. 11. ChengH,AddonaT,KeshishianH,DahlstrandE,LuC,DorschM, 24. Li Y, Vandenboom TG, Wang Z, Kong D, Ali S, Philip PA, et al. miR-146a et al. Regulation of IRAK-4 kinase activity via autophosphorylation suppresses invasion of pancreatic cancer cells. Cancer Res 2010;70: within its activation loop. Biochem Biophys Res Commun 2007;352: 1486–95. 609–16. 25. Hodi FS, O'Day SJ, McDermott DF, Weber RW, Sosman JA, Haanen 12. Akira S, Takeda K. Toll-like receptor signalling. Nat Rev Immunol JB, et al. Improved survival with ipilimumab in patients with metastatic 2004;4:499–511. melanoma. N Engl J Med 2010;363:711–23. 13. Li X, Jiang S, Tapping RI. Toll-like receptor signaling in cell proliferation 26. Agarwala SS. Current systemic therapy for metastatic melanoma. and survival. Cytokine 2010;49:1–9. Expert Rev Anticancer Ther 2009;9:587–95. 14. Dhawan P, Richmond A. Role of CXCL1 in tumorigenesis of melanoma. 27. Eggermont AM, Kirkwood JM. Re-evaluating the role of dacarbazine in J Leukoc Biol 2002;72:9–18. metastatic melanoma: what have we learned in 30 years? Eur J Cancer 15. Dhawan P, Richmond A. A novel NF-kappa B-inducing kinase-MAPK 2004;40:1825–36. signaling pathway up-regulates NF-kappa B activity in melanoma 28. Robinson JK. Sun exposure, sun protection, and vitamin D. JAMA. cells. J Biol Chem 2002;277:7920–8. 2005;294:1541–3.

OF8 Cancer Res; 72(23) December 1, 2012 Cancer Research

Augmentation of Therapeutic Responses in Melanoma by Inhibition of IRAK-1,-4

Ratika Srivastava, Degui Geng, Yingjia Liu, et al.

Cancer Res Published OnlineFirst October 4, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-12-0337

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2012/10/03/0008-5472.CAN-12-0337.DC1

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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2012/11/16/0008-5472.CAN-12-0337. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.