Combined inhibition of Ref‐1 and STAT3 leads to synergistic tumour inhibition in multiple cancers using 3D and in vivo tumour co‐culture models

Abstract With a plethora of molecularly targeted agents under investigation in cancer, a clear need exists to understand which pathways can be targeted simultaneously with multiple agents to elicit a maximal killing effect on the tumour. Combination therapy provides the most promise in difficult to treat cancers such as pancreatic. Ref‐1 is a multifunctional protein with a role in redox signalling that activates transcription factors such as NF‐κB, AP‐1, HIF‐1α and STAT3. Formerly, we have demonstrated that dual targeting of Ref‐1 (redox factor‐1) and STAT3 is synergistic and decreases cell viability in pancreatic cancer cells. Data presented here extensively expands upon this work and provides further insights into the relationship of STAT3 and Ref‐1 in multiple cancer types. Using targeted small molecule inhibitors, Ref‐1 redox signalling was blocked along with STAT3 activation, and tumour growth evaluated in the presence and absence of the relevant tumour microenvironment. Our study utilized qPCR, cytotoxicity and in vivo analysis of tumour and cancer‐associated fibroblasts (CAF) response to determine the synergy of Ref‐1 and STAT3 inhibitors. Overall, pancreatic tumours grown in the presence of CAFs were sensitized to the combination of STAT3 and Ref‐1 inhibition in vivo. In vitro bladder and pancreatic cancer demonstrated the most synergistic responses. By disabling both of these important pathways, this combination therapy has the capacity to hinder crosstalk between the tumour and its microenvironment, leading to improved tumour response.


| INTRODUC TI ON
Reduction oxidation (redox) effector factor 1/ apurinic/apyrimidinic en- preclinical studies in cancer as well as ocular neovascular diseases. [7][8][9] APX3330 slowed tumour growth preclinically and had limited side effects in a phase I clinical trials (NCT003375086). 3,10 The redox activity of Ref-1 diminishes cell growth, causes cell cycle arrest and shrinks pancreatic patient-derived xenografts (PDX) tumours by blocking the tumour cells' ability to regulate key transcriptions factors such as STAT3 (signalling transducer and activator of transcription-3), hypoxia inducible factor-1α (HIF-1α) and NFκB. 3 The redox activity of Ref-1 reduces the oxidized cysteines in STAT3, resulting in increased binding of STAT3 to DNA. 6 Although Ref-1 can regulate STAT3 DNA binding and thus expression of its downstream targets such as survivin, it does not affect overall levels of total or phosphorylated STAT3 protein. 6,11,12 In addition to being under redox control, STAT3 is regulated by phosphorylation which induces homodimerization and promotes translocation of activated STAT3 to the nucleus and subsequent regulation of downstream target genes. 13 Many STAT3 target genes have been shown to promote inflammation, immune-escape, tumour invasion and metastasis by up-regulating cytokines, such as IL-6, making STAT3 a target of interest in cancer therapy. 3,14 Two clinically approved drugs for inhibiting STAT3 include ruxolitinib (Rux) and napabucasin (Napa, BBI-608). Rux blocks JAK signalling upstream of STAT3 and thus inhibits its phosphorylation and activation, and Napa is a cancer stem cell inhibitor that affects STAT3 activation and has been shown to be a substrate of NAD(P) H:quinone oxidoreductase-1 (NQO1) and to a lesser extent P450 oxidoreductase (POR). [15][16][17] Napa also reduces the tumour growth of pancreatic cancer line MIA-PaCa-2 in vivo and blocks expression of stemness genes in pancreatic cancer cells. 18 A tumour engages in crosstalk with its microenvironment (TME) which can be influenced by both Ref-1 and STAT3 signalling pathways. We previously demonstrated that dual targeting of Ref-1 and STAT3 is synergistic and decreases cell viability in pancreatic cancer cells. 6 Here, we extend these studies in multiple cancer cell lines and investigate the effects in pancreatic cancer models in vivo. Associated with the actual tumour epithelial cell is a complex stroma composed of CAFs, immune and endothelial cells, and a rich extracellular matrix (ECM). 19,20 CAFs play a central role in PDAC progression, 21 and many studies report that CAFs secrete tumour-promoting growth factors and cytokines. [22][23][24][25][26][27][28][29][30] CAFs also synthesize and remodel the ECM in the desmoplastic stroma during the progression of the disease. This study examines the one-two punch of blocking Ref-1 redox signalling along with STAT3 activation on tumour growth in the presence and absence of the TME. By disabling multiple key pathways, the combination therapy has the potential to disrupt the crosstalk between the tumour and its microenvironment and lead to improved tumour response as well as broad applicability across multiple tumour types.

Cell proliferation and viability were measured with Alamar
Blue assay as previously described. 43 Briefly, cancer cell lines were seeded between 1250 and 2000 cells/well depending on their growth rate. Viability was measured 72 hours after treatment. For GBM10 (10 000 cells/well) and GBM26 cells (8000 cells/well), cells were seeded overnight on 96 well plates in DMEM (Gibco) containing 10% FBS and treated the next day with Ref-1 inhibitors or Napa.
After 5 days of incubation, cell growth was determined by methylene blue staining. 44 Each experiment was conducted in triplicate and repeated at least three times. Final DMSO concentration was ≤0.1%. For isogenic MDA-MB-231 and MIA-PaCa-2 NQO1+ and NQO1− cell lines, the relative survival assays were based on a longterm, DNA content assessments after 7 doubling times post-treatment (relative to vehicle-treated control) performed in 48-well dishes. Indicated cell lines were treated with specified agents at various doses for 24 hours. 45 The relative DNA content (a measure of cell growth -adapted from the method of Labarca and Paigen for each treatment (T) condition was determined by the fluorescence of the DNA dye (Hoescht 33258, Sigma) using a plate reader (Victor X) normalized to the vehicle control (C). 46 All experiments were performed in at least triplicate and the Welch's t-test (two-tailed) was performed for statistical analysis between NQO1+ vs NQO1−. The relative survival values at different treatment conditions and doses were graphed to obtain the LD 50 using GraphPad PRISM 8.4.1.

| Tumour spheroid 3-dimensional (3D) assays
PDAC cells were grown in co-culture as 3-dimensional tumour spheroids as described previously using cancer-associated fibroblasts (EGFP-positive) to mimic stroma and pancreatic cancer cells (TdTomato (red)-positive). The intensity of the red or green signal from the spheroids over time was quantitated as described in our previous studies. 8

| Immunohistochemistry
After euthanasia, tumour tissues were harvested, fixed in 10% neutral buffered formalin (NBF), processed for histological analysis. All tissues were processed through graded alcohols, cleared in xylenes, infiltrated with molten paraffin and then embedded in paraffin blocks.
Five-micron thick sections were cut and mounted on slides for staining. with an Aperio CS2 Scanscope to generate whole slide images. Aperio (Leica Biosystems) Image Analysis software was used to identify the pixels positive for the diaminobenzidine (DAB) substrate, by colour and optic density. The per cent positive pixels was then calculated for each annotation and averaged within each tissue type.

| Statistics
All the experiments were performed at least three independent times and replicates expressed as Average ± Standard Error (SE).
Significance was calculated as per either 2-way ANOVA or unpaired t-test wherever applicable using GraphPad Prism version 8. For qRT-PCR in the 3D spheroids, analysis of covariance models (ANCOVA) was used to test the difference in the Ct of each target gene compared with APX2009, Napa, vehicle (DMSO) and combination treatment after normalization by reference gene (Actin) as previously described. 50 Mixed effect repeated measure regression models with random intercept were used to test tumour growth rate of each treatment (ie the regression slope for a particular treatment) and differences in tumour growth rates between a pair of treatments (ie the difference in regression slopes between two treatments) in the in vivo model. 51 Tumour weights over time were estimated and compared between treatments from the regression models. A P-value of at least <.05 was considered statistically significant. All statistical analysis was conducted using SAS 9.4 (SAS, Inc, Cary, NC, 2016).

| Combination treatment using either ruxolitinib or napabucasin with Ref-1 redox inhibitors has additive and synergistic effects on cytotoxicity in multiple cancer cell lines
We at the higher doses of Napa (Isobolograms in Figure 1, Figure S1A, Table S1 and S2). In bladder cancer cell lines B01 and B02, both 2 and 4 μmol/L APX2014 was also more effective in combination with Napa than Napa alone, however the effect in B01 cells was more dramatic and displayed synergy at all doses of Napa (Figure S1B, Isobologram graph in right panel, Table S1 and S2). Importantly, the B01 line has been characterized as inherently resistant to cisplatin, while B02 is sensitive. 12,31 These data demonstrate that novel com-

| KPC cells that do not express IL-6 are more sensitive to Ref-1 inhibitors and combination therapy
The CRISPR-Cas9 system was used to generate KPC cells that are void of IL-6 as shown in Figure 2A and as previously described. 36 The control KPC cells and the IL-6 knockout (IL-6 KO) cells were then plated and treated with APX compounds as well as Rux and Napa to determine their sensitivity to these agents. We determined that the slopes were significantly different upon comparison of the

| Napabucasin and Ref-1 inhibitors work in concert to prevent PDAC 3D spheroid growth
Napa has previously been reported as a STAT3 inhibitor. 52 To confirm this activity in PDAC cells, Pa03C cells were incubated for 4 hours with 0.5 or 1 μmol/L Napa, and 12.5 μmol/L or 25 μmol/L Rux as a control for blockade of the phosphorylation of Y705 on STAT3 ( Figure 3A). At the end of the treatment, 50 ng/mL IL-6 was added for 15 minutes, and cells were immediately harvested for western blotting. In the presence of IL-6, STAT3 was phosphorylated in Pa03C cells as expected. In response to either Napa or Rux, STAT3 phosphorylation was inhibited dramatically ( Figure 3A).
Expression levels of total STAT3 were unaffected by the treatment.
In Pa03C cells, Napa treatment blocks the activity of STAT3 phosphorylation at Y705.  Figure 3B) and markers down-regulated following Napa treatment (β-catenin, SOX2, Nanog, SMO; Figure 3C). 16,47 With the exception of SIPA1, the Ref-1 responsive genes decreased in expression following treatment with APX2009 ( Figure 3B), while SMO and β-catenin were reduced in expression following Napa treatment ( Figure 3C). is able to kill 3D spheroids at much lower doses than ruxolitinib and has been shown to inhibit the growth of xenografted PDAC tumours. 18 Low passage patient-derived PDAC cells that express the TdTomato-red fluorescent protein were grown in 3D co-culture spheroids with GFP-labelled CAFs for 14 days ( Figure 3D-K).  Figure 4D). In a combination study, Napa and APX2009 or APX2014 gave slightly higher ROS output than Napa alone in all three cell types ( Figure 4E

| PDAC tumours grown in the presence of CAFs are more sensitive to Ref-1 inhibition in combination with STAT3 pathway inhibition via ruxolitinib
To study the impact of CAF cells on tumour growth and response to combination therapy, mice were implanted with either tumour cells alone or co-implanted with CAFs at varying ratios. After 24 days of growth, Pa03C cells with a ratio of either 1:2 or 1:4 CAF cells had significantly larger tumour volume than Pa03C cells alone ( Figure 5A).
The response in Panc10.05 was similar, yet not as robust, with significantly larger tumours in the ratio of 1:4 ( Figure 5B). If CAFs were implanted alone, there was no tumour formation ( Figure 5B). As expected, the addition of CAFs stimulates the growth of pancreatic cancer cells and accelerates the tumour growth rate significantly.
In a second study of CAF cells in response to combination therapy, Pa03C cells were implanted into mice, either alone or in conjunction with CAF19 cells (1:2 ratio) and allowed to grow for 11 days.
Mice were then given APX3330 (50 mg/kg), Rux (50 mg/kg), or a combination of both via gavage five days a week for two weeks.
At these doses there was no difference in tumour volume in mice implanted with tumour cells alone ( Figure 5C). However, tumours co-implanted with CAFs demonstrated growth inhibition under all conditions. We observed a ~25% decrease in tumour volume with single agents compared to vehicle control, and the combination of APX3330 and Rux resulted in the greatest response, about 50% smaller than vehicle ( Figure 5D, **P < .01, ***P < .001). We also estimated the growth rate per day of the tumour volume among the groups using repeated measure regression model and determined that all the growth rates of the tumours in the graph in Figure 5D were significantly different (P < .01) except for APX3330 and Rux treatment as single agents. Combination treatment was well tolerated as demonstrated by no significant differences in bodyweights ( Figure 5E)

| Inhibition of STAT3 via napabucasin in combination with Ref-1 inhibition is more effective at preventing tumour growth when grown in the presence of CAFs
As a further confirmation of the effects of dual inhibition of Ref-1 and STAT3 in PDAC tumours co-implanted with CAFs, we performed an in vivo study using the stem cell/ STAT3 inhibitor Napa.
Mice were implanted with Pa03C alone or Pa03C and CAF19 cells.
Tumours were allowed to grow to ~150 mm 3 and then treatment with APX3330 (25 mg/kg) and Napa (50 mg/kg) began twice daily.
Similarly, to the tumour growth in Figure 5C, when Pa03C cells were implanted alone, there was no significant difference between tumour growth regardless of treatment conditions ( Figure 6A).
However, in tumours co-implanted with CAFs, treatment with the combination of APX3330 and Napa resulted in tumour volumes ~46% smaller than vehicle control (P < .02, Figure 6B). Napa treatment alone decreased the tumour volume by ~25% (P < .05, Figure 6B). APX3330 at the 25 mg/kg dose ( Figure 6C) was not effective in reducing tumour volume compared to the 50 mg/kg dose shown in Figure 5D. APX3330+Rux, or gemcitabine+APX3330+Rux ( Figure 6D).
Again, we estimated the growth rate per day of the tumour volume among the groups using repeated measure regression model and determined that all the slopes of the lines in the graph in Figure 6D were significantly different (P < .01). The slope difference between the gemcitabine alone and the gemcitabine+APX3330+Rux regimen was also significantly different (P < .01). Mice were killed for tumour weight measurements when tumours reached 2000 mm 3 or after 36 weeks. Tumours were collected and weighed, and in mice treated with gemcitabine+APX3330+Rux regimen or gemcitabine alone, the tumours were significantly smaller than vehicle tumours (P < .01). This tumour size reduction is significant considering that the treated tumours were allowed to grow an additional 13 days past vehicle-treated mice ( Figure 6E). The APX+Rux combination was allowed to grow for an additional 4 days and was nearing significance (P = .07, Figure 6E). These results further implicate that STAT3 and Ref-1 inhibition are effective in PDAC tumour reduction and may enhance the response to standard of care agent, gemcitabine. points along that axis. The other STAT3 inhibitor that we used was Napa (or BBI-608). Napa was shown to inhibit STAT3 transcription leading to a decrease in stem-like properties of pancreatic and colon cancer cells. 18 It also inhibits p-STAT3 levels following treatment in PDAC cells (Figure 2) as well as inhibition of direct STAT3 targets including survivin, c-Myc and Nanog in prostate cancer cells. 16 It can also be bioactivated by NQO1 resulting in ROS production in pancreatic cancer cell lines. 17 Phase III data in colon cancer patients demonstrate that in patients with high levels of p-STAT3 overall survival was greater in Napa-treated patients. 15 Napa is currently being investigated in the clinic for multiple cancer indications where F I G U R E 4 Napabucasin and APX compounds generate ROS in PDAC cells, but only napabucasin is a substrate for NQO1. A, Immunoblotting for NQO1 in MIA-PaCa-2 cells. B, Cell survival in NQO1+ and NQO1− cells using relative long-term survival assays based on DNA content. The Welch's t-test (two-tailed) was performed for statistical analysis between NQO1+ vs NQO1−, *P < .05 was considered significant. C-E, CellRox Green was used to quantitate ROS levels after APX and Napa treatment. ROS levels in three pancreatic cancer lines are expressed as Relative fluorescent units (RFU) when treated with either APX2009, APX2014, Napa or the combination for 2 h (Mean ± SE [Unpaired one sided t-test], n = 3, *P < .05, **P < .01 and ***P < .001). Black asterisks are compared to vehicle and the coloured asterisks are compared to the corresponding dose of APX compound

| D ISCUSS I ON
STAT3 signalling is believed to be important, which could result in rapid translation in PDAC. We observe dramatic, significant inhibition of 3D co-culture tumour growth upon combination treatment with APX compounds and Napa ( Figure 2). This effect is specific to the cancer cells as the CAFs in the 3D co-culture did not seem to be overly affected by the combination and this could be due to the increase in cancer stemness of the tumour cells in comparison to the CAFs. 18 This is important to note and a strength of our 3D co-culture assay as more studies continue to demonstrate that depletion of the stroma does not decrease the progression of the disease. 78 Our efforts are aimed at discovering a therapeutic combination that is efficacious at killing the tumour in the presence of its protective stroma. However, there is still much to discover regarding the role of STAT3 signalling in CAFs as well as Napa's mechanism of action (as evidenced by the extreme sensitivity of cell lines that express NQO1), and we fully acknowledge that there could be F I G U R E 6 Combination of napabucasin and APX3330 is only effective when the CAFs are present and enhancement of standard of care agent, gemcitabine can be achieved using a tumour model with co-implantation of CAFs. Pa03C tumours alone (A) or co-implanted with CAFs at a 1:2 ratio (B) were treated with APX3330 (25 mg/kg, BID), Napa (50 mg/kg, BID), or APX3330+Napa (A+N; n = 5-6, *P < .05, **P < .01, ***P < .001, black lines indicate times of treatment), with tumor weights in C that correspond to tumor volumes in B. Standard of care agent, gemcitabine was also added to the in vivo regimen (orange arrows) and the three-drug combination grew significantly slower than the other groups (D, *P < .05, ** P < .01, green lines indicate times of treatment, orange and purple stars are compared to vehicle). Pa03C tumours co-implanted with CAFs were treated with APX3330 (50 mg/kg, BID, PO)+Rux (50 mg/kg, SID, PO), Gemcitabine alone (35 mg/kg, days 13, 16, 20, ip) or APX3330+Rux+Gemcitabine (G+R+A; n = 7, *P < .05, **P < .01) with tumour weights in E (*P < .05, **P < .01). Orange and purple stars (*) are compared to vehicle control and black star (*) compared to Gem alone Varying tumour:CAF ratios were tested (n = 5-7 mice, avg ± SE, *P < .05, **P < .01, ***P < .001). Pa03C tumours alone (C) or co-implanted with CAFs at a 1:2 ratio (D) were treated with APX3330 (50 mg/kg, BID), Rux (50 mg/kg, SID), or APX3330+Rux (R+A; n = 7, **P < .01, ***P < .001, black lines indicate times of treatment

ACK N OWLED G EM ENTS
We would like to thank Dr David Tuveson and Dr Christopher Frese for the KPC32043 and KPC32908 cells. We would like to acknowledge the In Vivo Therapeutics Core in the Indiana University Simon Comprehensive Cancer Center for the mice and assistance in dosing the various combination treatments. Writing -original draft (lead); Writing -review and editing (equal).

DATA AVA I L A B I L I T Y S TAT E M E N T
There was no big data or datasets generated in this manuscript.