Cerium Oxide Nanoparticles Sensitize Pancreatic Cancer to Radiation Therapy through Oxidative Activation of the JNK Apoptotic Pathway

Side effects of radiation therapy (RT) remain the most challenging issue for pancreatic cancer treatment. Cerium oxide nanoparticles (CONPs) are currently being tested in pre-clinical trials as an adjuvant to sensitize pancreatic cancer cells to RT and protect normal tissues from the harmful side effects. CONPs were not able to significantly affect RT-induced DNA damage in cancer cells, thereby ruling out sensitization through increased mitotic catastrophe. However, activation of c-Jun terminal kinase (JNK), a key driver of RT-induced apoptosis, was significantly enhanced by co-treatment with CONPs and RT in pancreatic cancer cells in vitro and human pancreatic tumors in nude mice in vivo compared to CONPs or RT treatment alone. Further, CONP-driven increase in RT-induced JNK activity was associated with a marked increase in Caspase 3/7 activation, indicative of apoptosis. We have previously shown that CONPs increase reactive oxygen species (ROS) production in cancer cells. ROS has been shown to drive the oxidation of thioredoxin 1 (TRX1) which results in the activation of apoptosis signaling kinase 1 (ASK1). The increase in ASK1 activation following the co-treatment with CONPs followed by RT suggests that the increased JNK activation is the result of increased TRX1 oxidation. The ability of CONPs to sensitize pancreatic cancer cells to RT was mitigated when the TRX1 oxidation was prevented by mutagenesis of a cysteine residue or when the JNK activation was blocked by an inhibitor. Taken together, these data demonstrate an important mechanism for CONPs in specifically killing cancer cells and provide novel insights into the utilization of CONPs as a radiosensitizer and therapeutic agent for pancreatic cancer.


Introduction
One in two men and one in three women in the United States will develop cancer at some point in their lifetimes [1]. Of the nearly 600,000 cancer related deaths in the USA each year [2],

CONPs Do Not Impact RT-Induced DNA Damage in Pancreatic Cancer Cells but Inhibit It in Normal Pancreatic Epithelial Cells
To determine if CONPs sensitize pancreatic cells to RT by enhancing mitotic catastrophe, DNA damage analysis was performed. L3.6pl and Panc1 pancreatic cancer cells exposed to RT, CONPs, or the combination of CONPs followed by RT. They all displayed some degree of DNA damage, as indicated by the ratio of the amount of DNA in the comet head to the amount of DNA in the comet tail ( Figure 1A,B).
Cancers 2018, 10, x FOR PEER REVIEW 3 of 17 rescue the apoptotic mechanism and RT-sensitivity in the otherwise RT-resistant pancreatic cancer cells.

CONPs Do Not Impact RT-Induced DNA Damage in Pancreatic Cancer Cells but Inhibit It in Normal Pancreatic Epithelial Cells
To determine if CONPs sensitize pancreatic cells to RT by enhancing mitotic catastrophe, DNA damage analysis was performed. L3.6pl and Panc1 pancreatic cancer cells exposed to RT, CONPs, or the combination of CONPs followed by RT. They all displayed some degree of DNA damage, as indicated by the ratio of the amount of DNA in the comet head to the amount of DNA in the comet tail ( Figures 1A-B). However, the combination therapy did not display additive effects; namely, CONPs followed by RT did not induce a greater amount of DNA damage than RT alone. Panc1 cells, acknowledged to However, the combination therapy did not display additive effects; namely, CONPs followed by RT did not induce a greater amount of DNA damage than RT alone. Panc1 cells, acknowledged to be more radiation resistant, showed less DNA damage demonstrated by the migration of DNA into the Cancers 2018, 10, 303 4 of 17 tail, than L3.6pl, which is a comparatively radiation-sensitive cell line. This indicates that the amount of DNA damage that resulted from exposure to RT correlated with the sensitivity of the cancer cell line to radiation-induced death. As this assay was completed immediately following RT exposure, the effects of DNA repair in response to CONP treatment can be excluded. These results suggest that CONP-induced RT-sensitization does not involve the mitotic catastrophe mechanism. In contrast to the lack of effect of CONPs on RT-induced DNA damage in pancreatic cancer cells, normal pancreatic epithelial hTERT-HPNE cells were responsive to pre-treatment with CONPs prior to RT (Figure 1). The cells that were pre-treated with 10 µM CONPs before exposure to 5 Gy RT displayed significantly less DNA damage, indicated by the tail moment, compared to cells exposed to RT alone immediately following RT exposure. Further, CONPs alone did not induce any DNA damage in normal cells.
These results indicate that CONPs in fact protect the normal cell from suffering mitotic catastrophe when exposed to RT.

CONPs Drive RT-Induced JNK Activation in Cancer Cells
To determine which pathway is activated in response to RT in cells pre-treated with CONPs, a multi-analyte ELISA for activation of proteins in key pathways associated with human cancer cell survival was performed on L3.6pl human pancreatic cancer cells exposed to CONPs, RT, and the combination therapy. Of the screened targets, only JNK activation was significantly increased by the combination of CONPs followed by RT, none of the other targets (AKT, Erk1/2, p53) displayed any alteration in response to the pre-treatment with CONPs compared to RT alone ( Figure 2A).
The ability of CONPs to drive RT-induced JNK activation, as determined by JNK phosphorylation, was further assessed by western blot analysis. The activation of both JNK1 ( Figure 2B) and JNK2 (Supplementary Figure S1) was significantly increased in both L3.6pl and Panc1 pancreatic cancer cell lines by 72 h following exposure to RT. In contrast, pre-treatment with CONPs induced a slight decrease or no change in JNK phosphorylation in the hTERT-HPNE normal pancreatic epithelial cells. CONPs alone did not induce significant JNK phosphorylation in either the pancreatic cancer cells or the normal epithelial cells.
When basal expression and phosphorylation of JNK in untreated cells were determined, both L3.6pl and Panc1 pancreatic cancer cell lines had higher total JNK expression than the corresponding normal hTERT-HPNE cells ( Figure 2C). Based on the ratio of phosphorylated JNK (p-JNK) to total JNK, nearly all of the JNK present in normal pancreatic epithelial cells was phosphorylated, i.e., activated. However, there was some difference in amount of phosphorylated and total JNK in pancreatic cancer cells, indicating that not all of the JNK present is activated in un-treated pancreatic cancer cells. Interestingly, ratio of active to total JNK correlated with cellular expression of thioredoxin 1 (TRX1), a redox sensing protein upstream of JNK. TRX1 overexpression by cancer cells compared to the corresponding normal cells was confirmed by western blot. Additionally, the normal cells in which a greater percentage of the basal JNK is phosphorylated express lower levels of TRX1 compared to the pancreatic cancer cells which expressed higher levels of TRX1 and lesser degree of basal JNK phosphorylation. These results suggest a possibility that the TRX1-JNK pathway could play a role in CONPs-induced pancreatic cancer cell sensitization to RT. pancreatic epithelial hTERT-HPNE cells were responsive to pre-treatment with CONPs prior to RT ( Figure 1). The cells that were pre-treated with 10 µ M CONPs before exposure to 5 Gy RT displayed significantly less DNA damage, indicated by the tail moment, compared to cells exposed to RT alone immediately following RT exposure. Further, CONPs alone did not induce any DNA damage in normal cells. These results indicate that CONPs in fact protect the normal cell from suffering mitotic catastrophe when exposed to RT. CONP treatment led to activation of JNK, but not AKT, ERKs or p53 in L3.6pl cells. Cells were treated with 10 μM CONPs for 24 hrs followed by exposure to 0 or 5 Gy RT and allowed for 72 h prior to PhosphoELISAarray analysis. Levels were normalized to untreated group. * p < 0.05. (B) Indicated cells were treated similarly for the period of time indicated and subject to p-JNK1 analysis by Western blot (See Figure S1 for similar change in p-JNK2 levels). (C) Basal expression of pJNK, JNK, and TRX was determined in the cells by western blot, with cancer cells displaying higher levels of all three proteins.

CONP-Driven JNK Activation Mediates RT-Induced Apoptosis in Pancreatic Cancer Cells
As JNK activation can be both pro-and anti-apoptotic, we next analyzed caspase 3 and caspase 7 activation as markers of apoptotic cells in vitro. Caspase 3/7 activation increases over time in response to RT in both L3.6pl and Panc1 pancreatic cancer cells. In L3.6pl cells, which are inherently more sensitive to RT-induced cell death, CONP-enhanced caspase activation is not evident until 72 h post RT ( Figure 3A). However, in Panc1 cells which are inherently more resistant to RT-induced cell death, the CONP-induced increase apoptotic signaling is discernible by 48 h post RT and persists through 72 h post RT ( Figure 3B). The data indicate enhanced, stable activation of apoptosis in both pancreatic cancer cell lines by the combination therapy. response to RT in both L3.6pl and Panc1 pancreatic cancer cells. In L3.6pl cells, which are inherently more sensitive to RT-induced cell death, CONP-enhanced caspase activation is not evident until 72 hours post RT ( Figure 3A). However, in Panc1 cells which are inherently more resistant to RTinduced cell death, the CONP-induced increase apoptotic signaling is discernible by 48 hours post RT and persists through 72 hours post RT ( Figure 3B). The data indicate enhanced, stable activation of apoptosis in both pancreatic cancer cell lines by the combination therapy.  Conversely, CONPs appear to inhibit or delay caspase activation in normal pancreatic hTERT-HPNE cells. At 48 hours post RT, caspase 3 and caspase 7 activation are significantly decreased in cells pre-treated with CONPs compared to cells treated with RT alone ( Figure 3C). Even at 72 hours post RT, there is no increase in the caspase activation in cells treated with CONPs prior to RT compared to RT alone. These results indicate that the CONPs enhance apoptosis in the RT-treated pancreatic cancer cells selectively.

CONPs Drive RT-Induced ASK1 Activation in Pancreatic Cancer Cells
The results described above also indicate a pro-apoptotic role of TRX1-JNK pathway in the CONPs and TR treated cancer cells. Apoptosis signaling kinase 1 (ASK1), a protein responsible for activating apoptotic signaling through JNK, is directly downstream of TRX1. The reductive form of TRX1 binds and masks ASK1 in its inactive state. Upon TRX1 oxidation, ASK1 dissociates and autophosphorylates become activated. Therefore, ASK1 phosphorylation provides the link between ROSinduced TRX1 oxidation and the activation of JNK to induce apoptosis [19]. ASK1 phosphorylation was determined after RT exposure in the pancreatic cells ( Figures 4A & 4B).

RT
- Conversely, CONPs appear to inhibit or delay caspase activation in normal pancreatic hTERT-HPNE cells. At 48 h post RT, caspase 3 and caspase 7 activation are significantly decreased in cells pre-treated with CONPs compared to cells treated with RT alone ( Figure 3C). Even at 72 h post RT, there is no increase in the caspase activation in cells treated with CONPs prior to RT compared to RT alone. These results indicate that the CONPs enhance apoptosis in the RT-treated pancreatic cancer cells selectively.

CONPs Drive RT-Induced ASK1 Activation in Pancreatic Cancer Cells
The results described above also indicate a pro-apoptotic role of TRX1-JNK pathway in the CONPs and TR treated cancer cells. Apoptosis signaling kinase 1 (ASK1), a protein responsible for activating apoptotic signaling through JNK, is directly downstream of TRX1. The reductive form of TRX1 binds and masks ASK1 in its inactive state. Upon TRX1 oxidation, ASK1 dissociates and auto-phosphorylates become activated. Therefore, ASK1 phosphorylation provides the link between ROS-induced TRX1 oxidation and the activation of JNK to induce apoptosis [19]. ASK1 phosphorylation was determined after RT exposure in the pancreatic cells ( Figure 4A,B).  L3.6pl cancer cells pre-treated with CONPs showed greater than a 1.5-fold (p < 0.0001) increase in ASK1 phosphorylation over RT alone. RT alone was not able to induce significant ASK1 activation in either pancreatic cancer cell line. CONPs alone, however, did induce mild ASK1 activation in the same cell. Similar results were obtained from the Panc1 cell line (data not shown). Conversely, the normal hTERT-HPNE cells showed roughly a 4-fold decrease in ASK1 phosphorylation when pretreated with CONPs compared to RT alone. CONPs alone did not significantly alter ASK1 activation in normal pancreatic epithelial cells compared to mock-treated control levels. Of note, none of the treatments had a significant impact on TRX1 expression ( Figure 4C). L3.6pl cancer cells pre-treated with CONPs showed greater than a 1.5-fold (p < 0.0001) increase in ASK1 phosphorylation over RT alone. RT alone was not able to induce significant ASK1 activation in either pancreatic cancer cell line. CONPs alone, however, did induce mild ASK1 activation in the same cell. Similar results were obtained from the Panc1 cell line (data not shown). Conversely, the normal hTERT-HPNE cells showed roughly a 4-fold decrease in ASK1 phosphorylation when pre-treated with CONPs compared to RT alone. CONPs alone did not significantly alter ASK1 activation in normal pancreatic epithelial cells compared to mock-treated control levels. Of note, none of the treatments had a significant impact on TRX1 expression ( Figure 4C).

CONPs Increase JNK Activation and Apoptotic Signaling In Vivo
To corroborate the apoptotic signaling observed in vitro, we next tested the effect of combination therapy on JNK activation and apoptosis in pancreatic tumors established by implanted L3.6bl cells in nude mice. JNK activation ( Figure 5A

CONPs are Unable to Induce Radiosensitization of Pancreatic Cancer Cells in the Absence of JNK Activity
To determine if JNK activation was the molecular driver of CONP-induced sensitization, we next used JNK inhibitors to block JNK activity in the cells exposed to RT, CONPs, or their combination. In the presence of a JNK inhibitor (SP600125), caspase 3/7 activation and sensitization to RT-induced cell death were completely mitigated in L3.6pl (Figures 6A & 6C)   To determine if JNK activation was the molecular driver of CONP-induced sensitization, we next used JNK inhibitors to block JNK activity in the cells exposed to RT, CONPs, or their combination. In the presence of a JNK inhibitor (SP600125), caspase 3/7 activation and sensitization to RT-induced cell death were completely mitigated in L3.6pl ( Figure 6A,C) and Panc1 ( Figure 6B,D) pancreatic cancer cells. Similar results were demonstrated with an independent JNK inhibitor (BI 78D3) (data not shown) as well as JNK silencing (Supplementary Figure S2), demonstrating that JNK function is essential for CONP-induced radiosensitization in pancreatic cancer cells.
Cancers 2018, 10, x FOR PEER REVIEW 9 of 17 demonstrating that JNK function is essential for CONP-induced radiosensitization in pancreatic cancer cells. Results were normalized to untreated control group. The control viability value was set as 100% viability. The % cell death was calculated by subtracting % viability from 100%. * p < 0.001. (see Figure S2 for similar results obtained when JNK expression was knocked down).

CONPs Fail to Induced Radiosensitization of Pancreatic Cancer Cells in the Absence of TRX1 Oxidation
To confirm if the TRX1-ASK1 pathway is the critical activator of JNK and apoptosis in mediating radiosensitization by CONPs, we generated L3.6pl and Panc1 cell lines that stably expressed HAtagged wild type TRX1, its C32S or C35S mutant, or the empty lentiviral vector. Stable TRX1 protein expression was confirmed by anti-HA western blot ( Figure 7A). The C32 and C35 are required for TRX1 oxidation and subsequent release and activation of ASK1. Unlike the wild-type TRX1, the C32S and C35S mutants are unable to activate ASK1 and downstream JNK apoptotic signaling [20]. Indeed, these mutations significantly blocked the caspase activation ( Figures 7B & 7C) and apoptotic induction (Figures 7D & 7E) by CONPs pre-treatment in both the L3.6pl and Panc1 cancer cells. The control viability value was set as 100% viability. The % cell death was calculated by subtracting % viability from 100%. * p < 0.001. (see Figure S2 for similar results obtained when JNK expression was knocked down).

CONPs Fail to Induced Radiosensitization of Pancreatic Cancer Cells in the Absence of TRX1 Oxidation
To confirm if the TRX1-ASK1 pathway is the critical activator of JNK and apoptosis in mediating radiosensitization by CONPs, we generated L3.6pl and Panc1 cell lines that stably expressed HA-tagged wild type TRX1, its C32S or C35S mutant, or the empty lentiviral vector. Stable TRX1 protein expression was confirmed by anti-HA western blot ( Figure 7A). The C32 and C35 are required for TRX1 oxidation and subsequent release and activation of ASK1. Unlike the wild-type TRX1, the C32S and C35S mutants are unable to activate ASK1 and downstream JNK apoptotic signaling [20]. Indeed, these mutations significantly blocked the caspase activation ( Figure 7B,C) and apoptotic induction ( Figure 7D,E) by CONPs pre-treatment in both the L3.6pl and Panc1 cancer cells. The role of the TRX1-ASK1 pathway was further tested in vivo. As expected, the parental and vector control cells were highly responsive to the combination of CONPs and RT with much reduced tumor burden. When TRX1 was overexpressed, however, such responsiveness was lost ( Figure 8A & 8B), consistent with that seen in vitro. Consequently, the combination treatment failed to extend the survival rate in the mice bearing the tumors with overexpressed TRX1 whereas lifespan of the mice bearing the tumors without overexpression was significantly extended compared the combination treatment to untreated control groups ( Figure 8C). The role of the TRX1-ASK1 pathway was further tested in vivo. As expected, the parental and vector control cells were highly responsive to the combination of CONPs and RT with much reduced tumor burden. When TRX1 was overexpressed, however, such responsiveness was lost ( Figure 8A,B), consistent with that seen in vitro. Consequently, the combination treatment failed to extend the survival rate in the mice bearing the tumors with overexpressed TRX1 whereas lifespan of the mice bearing the tumors without overexpression was significantly extended compared the combination treatment to untreated control groups ( Figure 8C).

Discussion
Our work identified the TRX1-ASK1-JNK apoptosis pathway as a novel CONPs responsive redox signaling that sensitize pancreatic cancer to RT ( Figure 8D). In the absence of CONPs, the reduced form of TRX1 masks the ASKs from phosphorylation and activation. CONPs convert the radical superoxide to RT-produced hydrogen peroxide that oxidizes TRX1 and releases ASK1 for phosphorylation and activation. The subsequent activation of JNK by ASK1 triggers the apoptosis in the cancer cells.

Discussion
Our work identified the TRX1-ASK1-JNK apoptosis pathway as a novel CONPs responsive redox signaling that sensitize pancreatic cancer to RT ( Figure 8D). In the absence of CONPs, the reduced form of TRX1 masks the ASKs from phosphorylation and activation. CONPs convert the radical superoxide to RT-produced hydrogen peroxide that oxidizes TRX1 and releases ASK1 for phosphorylation and activation. The subsequent activation of JNK by ASK1 triggers the apoptosis in the cancer cells.
This conclusion is consistent with our previous work showing that CONPs induce H 2 O 2 production in pancreatic cancer cells [14]. It is further supported by our results demonstrating the requirement for the reduction of TRX1 and inactivation of ASK1 and JNK for the tumor insensitivity to RT, and this barrier is breached by CONPs that facilitate oxidation of TRX1 and subsequent activation of ASK1 and JNK for apoptosis (Figures 2-5 and Figure S1). Indeed, blocking TRX1 oxidation or JNK activation prevents CONPs from sensitizing the cancer cells or tumors to RT (Figures 6-8 and Figure S2).
The inherent overexpression of TRX1 by cancer cells ( Figure 4C) indicates a higher than normal volume of total TRX1 pool. Hence, the pool of reduced TRX1 in the cancer cell available to inhibit ASK1 activation allows for the cell ability of resistance to RT. This could explain why forced overexpression of the wild-type TRX1 does not enhance the radiosensitization by CONPs (Figure 7). Nevertheless, the high level of TRX1 expression in the cancer cells does not seem to be a primary contribution to the CONPs-inducted radiosensitization since TRX1 expression in either pancreatic cancer cells or normal pancreatic epithelial cells does not change in response to treatment with RT, CONPs or their combination ( Figure 4C). It is interesting that overexpression of the wild type TRX1 was more sensitive than the C32S and C35S mutants to the RT plus CONPs combination in vitro ( Figure 7D) but the wild type and mutant TRX1 were equally resistant to the treatment in vivo ( Figure 8A-C). This could be due to the difference in the cell growth environment between the in vitro tissue culture and the in vivo tumor microenvironment that differentially affects the oxidation of TRX1 mediated by CONPs. This notion can be tested in a 3-dimensional coculture of tumor cell with stromal cells mimicking tumor growth conditions in vivo with regulated oxygen supplies and pH value although a reliable device for measuring acidity inside tissues remains unavailable. ASK1-mediated sustained activation of JNK has been shown to be required for ROS-induced apoptosis, with ASK1 knockout cells being substantially more resistant to H 2 O 2 [21]. These results suggest that the RT-CONPs combination therapy primarily targets TRX1 oxidation rather than its expression to enhance apoptosis.
In addition to JNK apoptotic pathway, there are a variety of other pro-and anti-apoptotic pathways such as the AKT, ERK, and p53 that play key roles in pancreatic cancer. AKT activation has been linked to cancer cell growth, survival, and proliferation [22], as well as correlated to the invasiveness of pancreatic cancers [23]. ERK1/2 has been shown to protect pancreatic cancer cells from apoptosis and promote progression through the cell cycle [24]. Both AKT and ERK1/2 signaling frequently become over-activated in pancreatic cancers. The P53 tumor suppressor normally plays a key role in inducing cell death in response to significant DNA damage, but becomes inactive in more than 60% of pancreatic cancers [25]. While AKT [26], ERK1/2 [27], and p53 [28] may all become active in response to ROS, our results showed that none of them, unlike JNK, was involved in the cellular response to CONP-driven RT-induced ROS (Figure 1). Although JNK activation can be both pro-and anti-apoptotic, depending upon cellular context [29], it is generally accepted that it induces apoptotic effects in cancer cells exposed to RT [30]. However, while JNK activation appears to be the primary redox-responsive effect in pancreatic cancer, it is possible that apoptotic pathways other than the JNK pathway could play a main role in other types of cancer in response to RT-CONPs combination therapy.
CONPs have been shown to be toxic to various types of cancer cells [14,15,31] by inducing stress oxidation, glutathione oxidation, lipid peroxidation, and membrane damage [14,31]. However, whether CONPs plays a role in DNA damage and whether such a role contributes to radiosensitization of cancer cells have never been reported. Our results clearly demonstrate that CONPs and RT combination does not have a synergistic effect on the cancer cells (Figure 1), excluding mitotic catastrophe as a main radiosensitization mechanism.
Despite the incapability of sensitizing to RT-induced DNA damage-mitotic catastrophe in cancer cells, CONPs alone induce a low level of DNA damage in the cancer cells but not the normal cells.
Further, CONP treatment dramatically decreased the amount of initial DNA damage in normal pancreatic epithelial cells exposed to RT (Figure 1). This difference may be attributable to the differential activity of CONPs at differing pH that determines whether CONPs perform like SOD to attack or catalase to protect the cell [14]. Therefore, only under acidic conditions found in cancer cells, CONPs may be able to directly induce a low level of DNA damage. Combining with the effect of CONPs on activation of the TRX1-ASK1-JNK apoptotic pathway predominantly in the cancer cells, our results further support the potential anticancer therapeutic applications of CONPs as a novel radiation adjuvant.

Generation of Stable Cell Lines
Wild type TRX1 and its C32S and C35S mutants (Addgene 20183, 21284 and 21285) [20] were transferred into the pKH3 vector to add an HA-Tag [32,33] and then into the lentiviral pLVPZ vector encoding puromycin resistance. As previously described [34], the lentivirus was produced by transfecting 293FT cells with the vectors, and the media collected from the transfected 293FT cells were used to infect L3.6pl and Panc1 cells for three repeated times. Positively infected cells were then selected with and maintained in medium supplemented with 1 µg/mL puromycin. Stable TRX1 protein expression was confirmed by anti-HA western blot.

Comet Assay
L3.6pl, Panc1 or hTERT-HPNE cells were plated in 10 cm dishes and left to attach overnight. Cells were then treated with 0 or 10 µM CONPs. 24 h later, cells were exposed to 0 or 5 Gy RT using a 160-kV cell culture and small animal irradiator (Kimtron Inc., Woodbury, CT, USA). Cells were trypsinized immediately following RT and subjected to single cell electrophoresis (CometAssay ® , Trevigen, Gaithersburg, MD, USA) as described previously [35] according to the manufacturer's protocol, including control cells with predetermined amounts of DNA damage purchased from Trevigen with the kit.

Phospho-ELISA
L3.6pl cells were seeded into 10 cm dishes and left to attach overnight. Cells were then treated with 0 or 10 µM CONPs followed by exposure to 0 or 5 Gy RT 24 h later. 72 h post RT, cells were collected and cell lysate was subjected to screening with the PhosphoELISArray Multi-Analyte Kit for key pathways associated with human cancer according to the manufacturer's protocol (SA Biosciences, FEM-5001, Frederick, MD, USA).

Analysis of Caspase Activation and Cell Viability
hTERT-HPNE, Panc1, and L3.6pl cells were seeded onto white-walled, clear-bottomed, 96 well plates (2,000 cells/well) and left to attach overnight. Media was then removed and media containing 0 or 10 µM CONPs was added. 24 h later, cells were exposed to 0 or 5 Gy radiation. At 24-48 h post radiation, caspase 3/7 activation was determined via Caspase-Glo ® 3/7 Assay (Promega, Madison, WI, USA) following the manufacturer's protocol. Caspase activity was normalized to cell viability, which was determined with CellTiter-Glo ® Luminescent Cell Viability Assay (Promega). Data was normalized to control (untreated) caspase levels for graphing. Cell viability was determined using WST-1 assay as previously reported [35,37].

JNK Inhibitor Studies
Panc1 and L3.6pl cells were seeded into 10-cm dishes and left to attach overnight. Media was then removed and replaced with media containing 0 or 10 µM CONPs and 20 µM specific JNK inhibitor (EMD #420119) [38], added at the same time. Radiation treatment (0 or 5 Gy) and the subsequent analysis of caspase activation (Caspase-Glo ® 3/7 Assay, Promega) normalized to viability (CellTiter-Glo ® Luminescent Cell Viability Assay, Promega) were completed by the same protocol as previously described. Additionally, the impact of the JNK inhibitor to mitigate radiation-induced cell death and CONP-driven radiosensitization was also determined from the CellTiter-Glo Luminescent Cell Viability Assay (Promega) data, which was normalized to control levels.

Orthotopic Xenograft of Human Pancreatic Cancer Cells in Athymic Mice
Experiments were performed essentially as described in our previous report [14]. Briefly, female athymic mice of 4 to 6 weeks of age (NCI-nu) were transported to our vivarium, housed and maintained in specific pathogen-free rooms. The project was approved by the institutional animal care and use committee (IACUC protocol #11-22) in accordance with the American Association for Accreditation of Laboratory Animal Care and current regulations and standards of the United States Department of Agriculture, Department of Health and Human Services, and the National Institute of Health. The mice were used in accordance under institutional guidelines with approved IACUC protocol. To develop tumors, L3.6pl cells, as well as L3.6pl cells stably expressing wild type TRX1, C32S mutant TRX1, and the empty pLVPZ vector, were harvested from culture dishes and injected as previously described [39][40][41].

Therapeutic Approach for Human Pancreatic Tumors Bearing Athymic Mice
The L3.6pl human pancreatic cancer cells with or without ectopic overexpression of wild type TRX1, C35S TRX1, and the empty pLVPZ vector were implanted into athymic mice. The mice were then grouped by random and treated with CONPs, RP, CONPs in combination with RP or saline. Tumor monitoring and subsequent analyses by liquid displacement and immunohistochemical staining were done as we previously described [14].

Statistical Analysis
Data are presented as mean ± standard deviation with a minimum of three observations per group. Unpaired, paired or single sample Student's t-test was applied using the GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA, USA) with the Bonferroni correction for the multiple comparisons. Animal survival was analyzed using Kaplan-Meier survival plot. Significance was determined by the alpha level of 0.05.

Conclusions
In summary, this work identified the TRX1-ASK1-JNK redox-sensing pathway to be a novel mechanism used by CONPs to re-activate the apoptotic mechanism in the human pancreatic cancer cells in response to RT. These findings provide further insight into the cellular implications of CONP-based therapies, especially in conjunction with RT or other potential radical inducing agents. Overall, this work supports the pursuit of CONPs as a novel tumor tissue sensitizer to increase the therapeutic index of RT to benefit pancreatic and potentially other types of cancer patients. Funding: This work was supported by a grant from National Institute of Health (R01 CA123977) and institutional funds to J.Z.

Conflicts of Interest:
The authors declare no conflict of interest.