Small-molecule inhibition of STAT3 in radioresistant head and neck squamous cell carcinoma

While STAT3 has been validated as a target for treatment of many cancers, including head and neck squamous cell carcinoma (HNSCC), a STAT3 inhibitor is yet to enter the clinic. We used the scaffold of C188, a small-molecule STAT3 inhibitor previously identified by us, in a hit-to-lead program to identify C188-9. C188-9 binds to STAT3 with high affinity and represents a substantial improvement over C188 in its ability to inhibit STAT3 binding to its pY-peptide ligand, to inhibit cytokine-stimulated pSTAT3, to reduce constitutive pSTAT3 activity in multiple HNSCC cell lines, and to inhibit anchorage dependent and independent growth of these cells. In addition, treatment of nude mice bearing xenografts of UM-SCC-17B, a radioresistant HNSCC line, with C188-9, but not C188, prevented tumor xenograft growth. C188-9 treatment modulated many STAT3-regulated genes involved in oncogenesis and radioresistance, as well as radioresistance genes regulated by STAT1, due to its potent activity against STAT1, in addition to STAT3. C188-9 was well tolerated in mice, showed good oral bioavailability, and was concentrated in tumors. Thus, C188-9, either alone or in combination with radiotherapy, has potential for use in treating HNSCC tumors that demonstrate increased STAT3 and/or STAT1 activation.


INTRODUCTION
Signal transducer and activator of transcription 3 (STAT3) is a member of a family of seven closely related proteins responsible for transmission of peptide hormone signals from the extracellular surface of cells to the nucleus [1]. STAT3 is a master regulator of several key hallmarks and enablers of cancer [2] including cell proliferation, resistance to apoptosis, metastasis, immune evasion, tumor angiogenesis, epithelial mesenchymal transition (EMT), response to DNA damage, and the Warburg effect [3][4][5][6]. STAT3 also is a key mediator of oncogene addiction [7] and supports the self-renewal of tumor-initiating cancer stem cells that contribute to cancer initiation, cancer maintenance, and relapse [8,9] in several types of tumors. STAT3 activity is increased in ~50% of all cancers [10], due, in many instances, to activation of signaling molecules upstream of STAT3, including receptor tyrosine kinases (RTK; e.g. epidermal growth factor receptor, EGFR), tyrosine kinase-associated receptors (e.g. the family of IL-6 cytokine receptors or G-protein coupled receptors, GPCR) [11,12], and Src kinases (e.g. Src, Lck, Hck, Lyn, Fyn, or Fgr) [12,13]. Thus, STAT3 is an attractive target for drug development to treat many types of cancer including head and neck squamous cell carcinoma (HNSCC). www.impactjournals.com/oncotarget HNSCC was the first human cancer demonstrated to depend on constitutively activated STAT3 for growth [3,14]. Inhibition of the EGFR or TGFα depletion in HNSCC resulted in growth inhibition and diminished STAT3 DNA-binding activity [14]. Moreover, direct inhibition of STAT3 resulted in growth inhibition of HNSCC cell lines [14,15]. The reported incidence of activated EGFR in HNSCC tumors and cell lines varies from 5% to 90% [16,17], however, suggesting that STAT3 activation in HNSCC may occur independently of EGFR activation. Specific inhibition of EGFR failed to prevent STAT3 activation in some HNSCC cell lines; rather, an autocrine/ paracrine IL-6/gp130 loop was demonstrated in these cell lines [18]. In addition to its contribution to HNSCC oncogenesis, there is increasing evidence suggesting a role for STAT3 and, more recently, STAT1 in resistance of HNSCC tumors to ionizing radiation (IR) [19][20][21][22][23].
STAT3 has been targeted in HNSCC xenograft models using the small-molecule, STAT3 inhibitor, Stattic, which was shown to enhance IR-sensitivity [19,24]. However, it has a low maximum tolerated dose in mice due, most likely, to off-target effects mediated by its covalent mechanism of action; consequently, the pathway for Stattic to enter the clinic is uncertain. EGFR inhibitorscetuximab, gefitinib or erlotinib-combined with IR with or without chemotherapy showed encouraging results [25]; however, well-defined markers for either patient selection or prediction of prognosis did not emerge from these studies. Thus, there remains a need for a STAT3 inhibitor suitable for clinical use either alone or in combination with IR or chemotherapy to improve treatment outcomes in HNSCC.
Ligand-engagement of receptors for growth factors or cytokines [11,26] causes receptor oligomerization and activation of intrinsic or receptor-associated tyrosine kinases, respectively. These activated kinases phosphorylate receptor tyrosine residues creating docking sites for recruitment of cytoplasmic STAT3 [11,26]. STAT3 docks to receptor phosphotyrosyl (pY) peptide sites through its Src-homology (SH) 2 domain, which leads to its phosphorylation on Y 705 followed by STAT3 tail-to-tail homodimerization (SH2 domain of each monomer binds the pY 705 peptide domain of its partner). STAT3 homodimers accumulate in the nucleus, where they bind to specific STAT3 response elements in the promoter of target genes and regulate their transcription. Several small molecule drug-development programs have emerged directed at identifying drug-like compounds that target STAT3 at one or more stages of its activation [27][28][29].
Using virtual ligand screening, we docked 920,000 small molecules from 8 chemical libraries into the pYbinding pocket of the STAT3 SH2 domain and identified 3 hits-C3, C30, and C188-as direct STAT3 inhibitors. C188 demonstrated the greatest activity of the three [9,29,30]. Using C188 as the scaffold, we performed 2D similarity screening and 3D pharmacophore analysis and identified C188-9, which in the studies outlined herein proved to be more potent in all assays for STAT3 inhibitory activity tested including inhibition of growth of HNSCC xenografts. C188-9 also has a high maximum tolerated dose, is orally bioavailable, and has great potential for clinical use either alone or in combination with IR or chemotherapy for the treatment of HNSCC.

2-D fingerprint screening followed by 3-D pharmacophore sorting identified potent second generation STAT3 probes
We previously [29] performed virtual ligand screening (VLS) of 920,000 compounds and identified three compounds C3, C30 and C188 with promising reversible inhibitory activities [29] in a STAT3 (pY)-peptide ligand binding assay and in a ligand-induced STAT3 phosphorylation assay. C188, in particular, was highly active in inducing apoptosis of the breast cancer cell line MB-MDA-468 in vitro (EC 50 = 0.7 μM). To identify more potent STAT3 probes, we used the scaffold of C188 as a reference structure (Table 1) and performed 2-D fingerprint screening of 490,000 compounds in a Life Chemicals compound database. We used 100 percent similarity as cutoff and employed the Tanimoto coefficient as the similarity method (Unity/Sybyl/Tripos) [31], which identified 207 compounds. The molecular features of these compounds were combined and the 3D pharmacophore structures were compared using a matrix that incorporated the structural features as distance bins. Compounds were ranked in decreasing order of pharmacophore similarity and the top 39 compounds with pharmacophore scores greater than 70 were purchased for further testing in our assays for STAT3 inhibition (Supplementary Table S1). Of the first twenty, nine compounds (C188-1, C188-7, C188-8, C188-9, C188-15, C188-16, C188-17, C188-18, C188-19) showed improved inhibition of STAT3 binding to pY-peptide (Table 1 and  Supplementary Table S1, representative binding curves shown in Figure 2A), while two compounds (C188-10 and C188-14) had activity similar to C188. The remaining nine compounds had activity less than C188. None of compounds that ranked below 21 showed potency in inhibiting STAT3 binding to pYpeptide that was similar or improved over C188.
We performed SAR analysis using C188 and its 39 derivatives to gain insight into structural features critical for binding of C188 and its derivatives to STAT3. All 39 C188-like compounds, including C188 itself, are derivatives of N-naphth-1-yl benzenesulfamide. SAR revealed that 37 of the 39 C188-like compounds could be divided into three structural groups (I, II and III) with decreasing activity ( Figure 1A). Addition of a variety of groups (the -R group highlighted in red in the general structure of Group I in Figure 1A), such as a triazole-3-yl-mercapto (C188-15) or a binapthalyl group (C188-9), to the 3-position of the naphthylamine ring yields the Group I compounds, www.impactjournals.com/oncotarget which are the most potent group of STAT3 probes. The -R group appears to be the most important contributor to the inhibitory activity of group I probes: a total of eight 3-substituents are found in Group I compounds, which enhanced the activity by one or more orders of magnitude. Most STAT3 probes in Group II contain a 5-membered ring that combines the 3-R and 4-OR 2 groups, such as a furan (C188-11). However, the compounds in this group are, on an average, ~5x less active than the Group I compounds, which suggests the H atom of the 4-hydroxy group (highlighted in blue in the general structure of Group I in Figure 1A) is important, e.g., involved in a favorable H-bond with the protein. Lack of ability to form the H-bond might attribute to the weaker activities of Group II probes.
To understand the SAR of these compounds in a more quantitative and predictive manner, we performed  Phase H-bond donor fields, superimposed with the aligned structure of C188-9, blue favorable, orange disfavored; E. hydrophobic fields, red favorable; F. electron-withdrawing fields, yellow favorable, light green disfavored; G. negative ionic fields, pink favorable. a 3-D quantitative structure activity relationship (QSAR) study, using the program Phase in Schrödinger (version 2010). These 40 compounds were built and their energy and geometry minimized using the OPLS-2005 force field in Maestro (version 9.1 in Schrödinger. They were then aligned using the "flexible ligand alignment" module in Maestro, which recognizes common features within these molecules (e.g., similar partial charge, hydrophobicity, aromaticity and H-bond donor/acceptor). The aligned ligands ( Figure 1B) were imported into the Phase program and a partial least squares (PLS) method was applied to correlate the STAT3 inhibitory activities (pIC 50 , or -log 10 IC 50 ) of these compounds with the Phase field data calculated based on their aligned 3-D structures ( Figure  1C). As shown in Supplementary   of ≤ 0.24. 3-D QSAR was used to visualize the 4 major factors contributing to activity: H-bond donor ( Figure  1D, blue favorable, orange disfavored), hydrophobic fields ( Figure 1E, red favorable), electron-withdrawing fields ( Figure 1F, yellow favorable, light green disfavored) and negative-ionic Phase fields ( Figure 1G, pink favorable).
We next examined the effect of C188 and C188-9 on growth of tumor xenografts, which revealed that the greater growth inhibitory activity of C188-9 vs. C188 extended to UM-SCC-17B cell line xenografts. While established UM-SCC-17B xenograft tumors continued to grow in nude mice that received C188 (50 mg/kg/day; Figure 5A), xenograft growth was markedly reduced in mice that received C188-9 (50mg/kg/day; Figure  5B, p=0.027). The ability of each compound to inhibit tumor growth correlated with its ability to reduce levels of pSTAT3 within the tumors. Levels of pSTAT3 in tumors from mice treated with C188 were not reduced significantly ( Figure 5C, 5D) but pSTAT3 levels were reduced significantly in tumors from mice treated with C-188-9 by 57% ( Figure 5E, 5F; p =0.017).

C188-9 targets both STAT3-and STAT1regulated genes in UM-SCC-17B xenografts
To determine the effect of C188-9 treatment on STAT3 gene targets, especially pro-oncogenic genes [5,26], we isolated total RNA from tumor xenografts harvested from mice treated with vehicle (n=5), C188 (n=4), or C188-9 (n=6) and used it for RNA sequencing and analysis (Supplementary Table S3). Of the approximately 13,528 discernible genes, levels of 37 gene transcripts were altered by C188 (17 down and 20 up-regulated, fdr <0.01, fold change ≥ 1.5), of which 7 were known STAT3 gene targets (Supplementary Table S4). In comparison, C188-9 affected a much greater number of genes involved in oncogenesis (384 total, 95 down-and 289 up-regulated), including 76 genes previously reported as regulated by STAT3 (38 down-regulated and 38 up-regulated; Table 3). Among the 38 genes previously shown to be upregulated by STAT3, 24 (63%) genes were downregulated by C188-9 treatment, as expected. Unexpectedly, however, 14 genes (37%) were downregulated by C188-9, including OASL, IFIT3, MX2, and IRF7, previously reported to be negatively regulated by STAT3 (Table 3). Further analysis revealed that many of these 14 genes were reported to be positively regulated by STAT1, as were 16 of the 24 genes previously reported to be upregulated by STAT3. Additionally, we found 10 more genes downregulated by C188-9 (fdr <0.01, fold change ≥ 1.5. Table 3) that previously were shown to be upregulated by STAT1. Thus, 40 of 48 (83.3%) genes downregulated by C188-9 previously were shown to be positively regulated by STAT1 (Table 3), including sixteen genes shown to be co-regulated by STAT3 and STAT1. Many of these genes have been implicated in radioresistance. This analysis raised the possibility that the effect of C188-9 on gene transcript levels in HNSCC tumors was mediated by its effects on both STAT3 and STAT1.
To explore the hypothesis that the activity of C188-9 extended to STAT1, we examined the ability of C188-9 to inhibit STAT1. Examination of Kasumi-1 cells revealed that C188-9 is a potent inhibitor of STAT1 activation by were incubated with a constant concentration (80 nM) of fluorescently labeled STAT3 (aa residues 127-722). Fluorescence was measured continuously before and after application of an infrared laser. The change in fluorescence (F norm ) was calculated from the ratio of fluorescence immediately before heating and 30 seconds after heating and plotted against the logarithm of the different concentrations of C188-9 or EGFR pY-1068 (A and B); the sigmoidal binding curve was used to calculate the dissociation constant K D . www.impactjournals.com/oncotarget   Figure 2B, 2C), as well as by G-CSF (IC 50 =4.1 μM, Table 1, Figure 2D). In addition, C188-9 was effective at reducing levels of constitutively activated STAT1 in UM-SCC-17B cells (IC 50 =19.1 μM, Table 2) as well as SCC61 and HN30 (IC 50 =28.5 and 5 μM respectively, Table 2) but not in SCC-35, which did not have significantly higher pSTAT1 levels compared to the non-tumor line HEEpiC (Table 2, Supplementary  Figure S1B. Importantly, levels of pSTAT1 in UM-SCC-17B tumor xenografts from mice treated with C188-9 were reduced by 80% compared to tumor xenografts from mice that received vehicle control (p = 0.000003; Figure 5E, 5F and Table 1).

DISCUSSION
We used the scaffold of C188, a small-molecule STAT3 inhibitor previously identified by us using virtual ligand screening, in a hit-to-lead program to identify a more potent small-molecule STAT3 inhibitor. These studies identified C188-9, which binds to STAT3 with high affinity and is more potent than C188 in inhibiting STAT3 binding to its pY-peptide ligand, in inhibiting cytokine-stimulated pSTAT3, in reducing constitutive pSTAT3 activity in UM-SCC-17B, a radioresistant HNSCC cell line, and in inhibiting anchorage dependent and independent growth of these cells. In addition, treatment of nude mice bearing UM-SCC-17B xenografts with C188-9, but not C188, prevented tumor xenograft growth. RNA-seq analysis of tumor xenografts revealed that C188-9 modulated many STAT3-regulated genes involved in oncogenesis, as well as genes involved in chemoresistance and radioresistance that previously were shown to be regulated by STAT3 and STAT1. Phosphoflow and Luminex assays of cells treated with IFN-γ or G-CSF and immunoblotting of lysates of UM-SCC-17B xenografts revealed that C188-9 was equally potent at targeting STAT1 as STAT3. Thus, C188-9, either alone or in combination with radiotherapy, has potential for use in treating HNSCC tumors that demonstrate increased STAT3 or STAT1 activation.
Several of small-molecule STAT3 inhibitors, as well as other classes of agents, have been shown to be    effective in targeting STAT3 and inhibiting tumor growth in preclinical models, including HNSCC, [33,40]. In particular, Stattic [19], STAT3 anti-sense plasmid [14], STAT3 decoy oligonucleotide [41], erlotinib [42], and most recently, a cyclic version of oligonucleotide decoy [43], have shown promise in pre-clinical models of HNSCC. However, an agent that directly targets STAT3 has yet to be approved by the FDA and it remains to be mice were immunoblotted for pSTAT3, total STAT3, β-actin, pSTAT1, total STAT1. Whisker plots of β-actin-normalized pSTAT3 (left panels) and pSTAT1 values (right panels) for C188 treatment F. are shown with the differences in means compared using t test. seen whether C188-9 or any of these agents will progress beyond Phase 0 clinical trials for HNSCC. Towards this end, we demonstrated that C188-9 has a favorable pharmacokinetic and safety profile in mice (Table 1); C188-9 was well tolerated in mice to 100 mg/kg/day for 14 days, demonstrated plasma bioavailability by the RNA-sequence data was analyzed as stated in methods. Identification of all differentially expressed genes was based on a cutoff false detection rate fdr < 0.01, and a fold change FC > 1.5. The absolute value of FC is the magnitude of up-or down-regulation for each gene/homolog after C188-9 treatment. FC > 1.5 indicates up-regulation, and < −1.5 indicates down-regulation. The genes in this table are arranged in decreasing order of FC, which are shown in column 5. It also shows literature info on whether a gene is reported to be positively (Pos) or negatively (Neg) regulated by STAT3 (Column 5) and/or STAT1 (Column 6).
oral route similar to the IP route, and was concentrated in tumors to levels nearly 3-fold that of simultaneously harvested plasma. Also, in GLP-compliant safety studies, C188-9 demonstrated no clinical, laboratory, or pathological abnormalities in rats or dogs up to a dose of 200 mg/kg/d or 100 mg/kg/d, respectively, for 28 days (data not shown).
In addition to promoting many hallmarks and enablers of cancer, STAT3 activation has been linked to resistance to radiation therapy in cells obtained from several normal tissues and cancers, including HNSCC [19], through several mechanisms, including EGFR signaling [44], HER2-STAT3 cross-talk [45], and activation of the LIF-STAT3 axis [46]. Addition of STAT3 inhibitors [19,47] to radiation therapy increased clearance of cancer xenografts in mice. Thus, we were not surprised to find modulation of STAT3-regulated genes shown previously to contribute to resistance to radiation within tumor xenografts of mice treated with C188-9 vs. vehicle. However, RNAseq analysis of these tumor xenografts also uncovered modulation of many STAT1-regulated genes, including many, involved in radiation resistance, raising the possibility that C188-9 targeted STAT1, in addition to STAT3, in these tumors. In fact, examination of the effects of C188-9 on STAT1 in each of our in vitro assays, as well as in tumor xenografts, demonstrated that C188-9 was as effective at targeting STAT1 as it was in targeting STAT3. This is not entirely surprising, given the high degree of similarity between the SH2 domains of STAT1 and STAT3 [29].
The role of increased STAT1 activity in tumors resistant to IR [20,48], as well as chemotherapy e.g. doxorubicin and topoisomerase-II inhibitors [48], now is well established. Khodarev et al developed an IR-resistant, HNSCC cell line, Nu-61, from a xenograft in nude mice that grew out after multiple rounds of implantation and irradiation of xenografts starting with the IR-sensitive HNSCC cell line, SCC-61 [20]. Comparison of gene expression profiles of the two lines, established that the IFN/STAT1 pathway is responsible for this acquired IR resistance. A signature list of 25 such genes, termed the IFN-related damage signature (IRDS [23]), was also shown to be induced by IR therapy in xenograft tumor models of head and neck, breast, and colon cancer [21]. Furthermore, STAT1 silencing in IR-resistant Nu-61 cells rendered them IR sensitive with concurrent reduction of these IRDS genes [21,23]. Our RNA-seq data shows that the same set of signature genes is upregulated in xenografts of UM-SCC-17B, a cell line derived form a tumor from a patient who failed multiple rounds of radiotherapy [49]. These findings suggest that adjuvant use of C188-9, a dual inhibitor of STAT1 and STAT3 in HNSCC, may overcome IR resistance in this tumor system. Recently, both STAT3 and STAT1 were discovered among the top transcription factors activated in IR-resistant HPV-negative HNSCC and combined siRNA mediated inhibition of STAT3 and STAT1 had more pronounced effect on cell growth of STAT-activated H0N1 cells [50]. Our study describing the dual anti-STAT3 and anti-STAT1 action of C188-9 on HNSCC cells with activated STAT3/1 thus becomes particularly relevant.

2-D fingerprint screening
Using C188 scaffold as a reference structure, 2-D fingerprint screening was performed using the Life Chemicals database, which contained over 490,000 compounds. We used 100 percent similarity as cutoff and employed Tanimoto coefficient/Unity/Sybyl/Tripos as the similarity comparing method.

3-D pharmacophore sorting
The 207 compounds resulted from 2-D fingerprint screening were converted into 3-D structures for analysis as described in Supplemental Information.

QSAR studies
3-D quantitative structure activity relationship (QSAR) study was performed as described in Supplemental Information.

STAT3/pY-peptide binding using a surface plasmon resonance (SPR) assay
Binding of STAT3 (200nM in 20 mM Tris buffer, pH 8) pre-incubated without or with C188/C188-9 to phosphorylated and control non-phosphorylated biotinylated EGFR derived dodecapeptides based on sequence surrounding Y1068 [51] was measured using a Biacore 3000 biosensor (Biacore inc., Piscataway NJ) as described in Supplemental Information.

Phosphoflow assay
Phosphoflow analysis was done as described in Xu et al [29]. Briefly, Kasumi-1 cells, were serum-starved, pre-treated with compound or DMSO (1 hour, RT) and then stimulated with 20 μl of GCSF (10 ng/ml) for 15 minutes at 37°C. The cells were then permeabilized and stained with Alexa647-pSTAT3, Alexa488-pSTAT5 and PE-pSTAT1 antibodies and analyzed using the BD LSR2. FCS files were exported and uploaded to Cytobank for determination of phosphoproteins as earlier [5]. Gating was done on live, single cells and then gates selected for positive ligand-activated pSTAT3 and pSTAT1. IC 50 values were calculated using the GraphPad Prism.

Luminex bead-based assay
Luminex bead-based assays were used to determine levels of pSTAT1, pSTAT3, and GAPDH, as described [5] as detailed in Supplemental Information.

Microscale thermophoresis
Binding of C188-9 or EGFR Tyr(P)-1068 phosphopeptide to WT STAT3β (127-722) was measured by microscale thermophoresis (MST). C188-9 was titrated between 0.305 and 10,000 nM and EGFR Tyr(P)-1068 was titrated between 0.025 and 800 nM to a constant amount (~80 nM) of fluorescently labeled STAT3 (127-722). Movement of STAT3 under a temperature gradient was measured by recording the change of fluorescence signal as the heated molecules moved away from the point of application of the IR-laser used for heating. This movement was traced in the fluorescence time trace. The change in fluorescence F norm =F hot /F cold where F cold is the homogeneous fluorescence distribution observed inside the capillary before the IR-Laser is switched on and F hot is the steady low fluorescence state after the IR-Laser is switched on for 30s. F norm was calculated and plotted against the logarithm of the different concentrations of the peptide or C188-9 dilution series, to obtain a sigmoidal binding curve. This binding curve was fitted with the nonlinear solution of the law of mass action, and the dissociation constant K D was calculated.

Anchorage-independent and dependent cell growth
Cells were cultured in triplicates in in complete DMEM ± drug, in ultra-low attachment 96 well plates for 72 hrs or cell-culture treated plates for 48 hrs and viable cells were quantitated using MTT. Optical density (OD) was measured at 590 nm using a 96-well multi-scanner (EL-800 universal microplate reader, BioTek Inc, VT, USA). Relative % viability (viability after any treatment ÷ viability of untreated cells x 100) was plotted along Y-axis. At least 2 replicates experiments were performed and were used for IC50 calculation using GraphPad software.

RNA extraction and sequencing using RNA-seq
Tissue was cryosliced (5μ) with Cryotome and total RNA was extracted using RNeasy Microarray tissue kit (Qiagen, Venlo, Limburg) followed by DNase digestion and Qiagen RNeasy column purification (Qiagen, Valencia, CA, USA). The RNA integrity was verified using an Agilent Bioanalyzer 2100 (Agilent, Palo Alto, CA, USA). High-quality RNA (RNA Integrity number or RIN >9.0) was processed using an Illumina TruSeq RNA sample prep kit following the manufacturer's instruction (Illumina, San Diego, CA, USA). Detailed method is provided in Supplemental Info.

PK studies in mice
Pharmacokinetic (PK) studies to assess plasma levels of C188-9 following IP and oral routes were performed in C57BL/6 mice. PK studies to assess tumor vs. plasma levels of C188-9 were performed in nude mice bearing xenograft tumors. Details regarding these studies are in the Supplemental Methods.

Statistical analysis
Student's t-test or a paired t-test was used to compare control-and C188-9 treated groups as indicated.

ACKNOWLEDGMENTS
The authors wish to thank George Bondar and his colleagues at Life Chemicals for performing the 2D fingerprint screening and Joel M. Sederstrom for his expert assistance in flow cytometry analysis. We also thankfully acknowledge receipt of the cell line UM-SCC-17B from Dr. Thomas E Carey from University of Michigan, the cell lines SCC-9, SCC-15, HN5, UM-SCC-1, HN30 and HN31 from Dr. Heath Skinner, at MDA, and the lines SCC-61, SQ-20B, SCC-35 from Dr. Ralph Weichselbaum, Department of Radiation Oncology, University of Chicago, Chicago, IL. We thank the Charecterized Cell Line Core Facility at MDA for STR-genotyping and authentication of the cell lines used in this study.

CONFLICTS OF INTEREST
Please note that DJT has the following conflict of interest: Baylor College of Medicine, with DJT as primary inventor, filed composition and methods of use patents on C188 and C188-9. StemMed, Ltd. currently holds an exclusive license to these compounds. DJT is founding partner, President, and CEO of StemMed and also have majority ownership of StemMed. He has been in compliance with all conflict of interests (COI) policies at Baylor College of Medicine and currently is in compliance with all COI policies at StemMed and at the University of Texas M.D. Anderson Cancer Center, where he relocated on December 1, 2014.

GRANT SUPPORT
This work was supported, in part, by a National Institutes of Health grants CA149783 to DJT, P30 CA125123 to the BCM Cancer Center, AI036211, CA125123, and RR024574 to BCM Cytometry and Cell Sorting Core, BCM, SPORE program 5 P50CA97007 10 awarded to MD Anderson Cancer Center, CA16672 awarded to CCLC at MDA and research grants from the John S. Dunn Gulf Coast Consortium for Chemical Genomics Screening Grant Program to DJT.