TAK1 regulates the tumor microenvironment through inflammatory, angiogenetic and apoptotic signaling cascades

Transforming growth factor beta-activated kinase 1 (TAK1) has been implicated for its role in inflammatory signaling and as an important mediator of cellular apoptosis and necroptosis in various cell types. Our recent discovery of a first-in-class, potent and selective TAK1 inhibitor, takinib, represents a novel pharmacological tool to evaluate TAK1’s role in cancer. In this study we evaluated the potential therapeutic capacity of TAK1 inhibition on tumor growth and on tumor microenvironment remodeling. In a screen of 16 cancer cell lines, takinib in combination with tumor necrosis factor (TNF) was found to induce cell death (>20%) in 6 out of 16 cell lines. Furthermore, knocking out of TAK1 in MDA-MB-231 cells dramatically increased their sensitization to TNF-mediated apoptosis. In vivo xenographs of MDA-MB-231 TAK1KO tumors displayed delayed tumor growth and increased overall survival compared to TAK1WT controls. Histological and proteomic analysis of TAK1KO tumors showed altered angiogenic signaling and inflammatory signaling via immune cells. Overall, these findings suggest that the targeting of TAK1 in immune mediated cancers may be a novel therapeutic axis.


INTRODUCTION
The role of immune cells and inflammation has been widely associated with aggressiveness and survival rates in many tumors, wherein pro-inflammatory signaling within the tumor microenvironment stimulates tumor cell growth and metastasis [1]. Both the innate and adaptive immune system have been implicated in tumorigenesis and maintenance of solid tumors [2]. Specifically, tumorassociated macrophages (TAMs) have been shown to create an immense tumor burden in breast cancers, with greater TAM burden leading to poor disease prognoses [3].
TAMs can influence the tumor microenvironment through secretion of biologically active molecules that promote tumor growth, angiogenesis and metastasis while limiting critical anti-tumor immune responses (immunosuppression) [4,5]. TAMs arise from circulating monocytes, which are recruited to the tumor microenvironment largely due to tumor cell chemokine secretion. Specifically, TNF has been implicated in immune cell migration into the tumor [6]. TAMs enhance the tumor microenvironment in large part due to hyperactivating nuclear factor kappa-light-chain-enhancer of activated B cell (NF-κB) signaling, leading to downstream pro-inflammatory, pro-survival and metastatic phenotypes [7,8]. Cytokines such as IL-6 and TNF aid in tumor development by activating pro-survival signaling and antiapoptotic pathways in cancer cells [9,10]. Furthermore, TAMs enhance the tumor microenvironment through neovascularization via vasculature signaling mechanisms, such as VEGF and angiogenin [11]. The constant exposure to inflammatory signals not only enhances the survival/ growth mechanisms in tumor cells, but also educates T and dendritic cells to adopt an immunosuppressive phenotype and aid in disease progression [12]. Immunotherapies aimed at "re-educating" immune cells to adopt tumor

Research Paper
surveillance phenotypes have shown great promise in immune responsive tumors [13,14]. Remodeling of TAM phenotypes have shown potential in reducing tumor burden and in sensitizing tumors to an alternative therapeutic axis. Thus, targeted therapies aimed at modifying TAMs as well as targeting pro-survival mechanisms in tumor cells represent a novel, double edged, therapeutic intervention.
A key signaling element in pro-survival/ inflammatory response pathways is the protein kinase TAK1 (transforming growth factor β-activated protein kinase 1) [15]. Upregulation of TAK1 can be seen in up to 30% of breast cancers, where they enhance tumor burden through increasing activity of NF-kβ and mitogenactivated protein kinases (MAPKs), which are important for tumorigenesis and inflammation [16]. Therefore, inhibitors of TAK1 constitute a means to block release of pro-inflammatory cytokines as well as cell migration. In addition to its anti-inflammatory properties, TAK1 has been shown to play an integral role in pro-survival/proapoptotic signaling [16][17][18][19]. Following TNF Receptor 1 activation by TNF, TAK1 stimulates downstream signaling cascades activating inflammatory and pro-survival proteins NF-kβ, cJun, JNK and p38 [16]. In contrast, it have been shown that inhibition of TAK1 in the presence of TNF induces caspases 3, 7 and 8 activation leading to apoptosis [16].
The duality of TNF stimulation makes TAK1 a key target in mediating between growth and apoptosis in cancer cells. With inhibition of TAK1 via small molecule inhibitors, we are able to bypass the inflammatory action of TNF while isolating and harnessing the apoptotic capabilities of TNF. Our recent discovery of the takinib scaffold has identified a potent, highly specific inhibitor of TAK1 (IC 50 9.5 nM) that we hypothesize can act as a novel small molecule therapy against immune responsive tumors. The present study investigates the therapeutic potential of takinib as an anti-cancer therapy as well as the role of TAK1 in inflammatory signaling within the tumor microenvironment.

Acute TNF exposure induces TAK1 mediated apoptosis in various cancer cell lines
We first investigated the effects of TAK1 pharmacological inhibition in 16 cell lines from various cancer tissues. Following 24 hours of TAK1 inhibition with takinib, minimal cell death was found in most cell lines ( Figure 1). However, previous groups, including ours, have shown that TAK1-induced cell death is mediated by TNF signaling mechanisms [16,20]. To further test this hypothesis, we treated cancer cells with takinib + TNF (30 ng/mL) combination therapy. We also examined the sensitivity of the cells to TNF alone and found that the combination therapy induced apoptosis (<80% cell survival) in 6 cell lines, compared to 3 by TNF alone. Furthermore, the data revealed cells that show initial sensitivity to TNF alone experience increased cell death after inhibition of TAK1 with takinib ( Figure 1). We next tested the effects of takinib + TNF on MCF10A, a noncancerous cell line. Cellular survival was not affected after the takinib + TNF combination therapy, with minimal cell toxicity observed at a 10 μM dosage of takinib ( Figure 1).
Here, we show that TAK1 KO cells show significantly higher sensitivity to TNF-induced apoptosis compared to TAK1 WT with an ED50 of ~0.5nM to TNF ( Figure 3B).

TAK1 KO suppresses in vivo tumor growth
To test the effects of TAK1 inhibition in tumor growth and overall survival, nude mice were orthotopically injected in the mammary fat pad with either MDA-MB-231 TAK1 WT with Cas9 control, or TAK1 KO cells. Days from injection to 100 mm 3 tumor volume and survival time were evaluated. TAK1 KO tumors, on average, took 14 days to grow to 100 mm 3 , compared to 11 in TAK1 WT with Cas9 control (Supplementary Figure 3A). Furthermore, overall tumor growth was significantly reduced in TAK1 KO compared to TAK1 WT (p < 0.001) ( Figure 4A). These data translated to an overall increase in survival of KO treated mice (p < 0.0001) ( Figure 4B). Cell death was evaluated by TUNEL staining of the tumors, which showed increased TUNEL + cells in TAK1 KO tumors; results are consistent with our cellular studies ( Figure 4C).
Following in vitro cell assays showing increased cell death of MDA-MB-231 TAK1 inhibited cells to TNF, we next sought to test this hypothesis in vivo. We first determined the in vivo MTD of TNF delivered via tail vein injection. Mice were dosed 3 times a week at 10, 30, 60 and 100 μg/kg showed no significant weight change (Supplementary Figure 3B). We next tested the ability of exogenous TNF to increase in vivo death of

Takinib reduces TAM contribution of angiogenetic signaling
To further examine in vivo interactions between cancer and immune cells on tumor growth, we examined tumor slices from the previously mentioned mice injected with MDA-MB-231 TAK1 WT with Cas9 control or TAK1 KO cells. Previous studies, by our lab and others, have shown that TAK1 plays an important role in mediating inflammatory and potentially angiogenetic signaling cascades [22][23][24]. To test whether TAK1 mediated angiogenetic signaling was present in the tumor microenvironment, we stained tumor slices for CD31 (PECAM) expression. In comparison to TAK1 WT , the TAK1 KO tumors showed increased vasculature ( Figure 5A, 5B). We next profiled MDA-MB-231 TAK1 WT , and TAK1 KO cells for angiogenesis factors. Minimal significant biological changes were seen between TAK1 WT and TAK1 KO cells with a 1.19 fold increase in angiopoietin-1 and decreases in expression of HB-EGF (1.23-fold), IGFBP-3 (1.09-fold), IL-8(1.05fold) and thrombospondin (1.16-fold), suggesting that the TAK1 mediated angiogenetic signaling may occur outside of the cancer cells themselves ( Figure 5C). Due to the large contribution of TAMs in the tumor immune microenvironment of many solid tumors, we also tested the ability of TAK1 inhibition to alter angiogenetic signaling in macrophages. THP-1, a human macrophage cell line, was stimulated with LPS, followed by treatment with either vehicle or takinib for 24 hours. Angiogenetic markers were profiled. Significant differences can be observed between the two treatments with a reduction of protein expression of 2.67-fold IGFBP-3 (p < 0.0012), 1.7-fold IL-1B (p < 0.0001), 1.42-fold thrombospondin (p < 0.0005), 1.54-fold VEGF (p < 0.156), 1.88-fold Pentraxin (p < 0.0001), 1.75-fold uPA (p < 0.0001), 1.68fold CXCL16 (p < 0.056) seen ( Figure 5D).
To address whether changes in cytokine production are a consequence of altered TAM infiltration or changes in phenotype we next stained TAK1 KO and TAK1 WT tumors against F4/80 and quantified the number of macrophages present in the tumor stroma ( Figure 6B). No significant changes were seen between the density of macrophages in TAK1 WT versus TAK1 KO tumors.

DISCUSSION
Here we show the ability of takinib to not only induce TAK1 mediated apoptosis in TNF sensitive cancer cells, but to reduce pro-inflammatory and proangiogenic signaling in surrounding TAMs. TAK1 has been seen to play an integral role in mediating many   signaling cascades including inflammatory, angiogenic and apoptotic; however, lack of selective and potent TAK1 pharmacological inhibitors have limited research around this target as a novel therapeutic axis. Here, we demonstrate the role of TAK1 in mediating TNF signal cascades in various cancer cell lines. Although TNF is widely considered a pro-inflammatory/survival signaling cytokine, we show that TAK1 inhibition blocks traditional TNF inflammatory and survival signaling and induces apoptosis in TAK1 dependent cancer cell lines.
In an effort to further validate TAK1 mediation of TNF survival-apoptosis signaling, we developed a TAK1 KO MDA-MB-231 cell line. TAK1 KO recapitulated the effects of TAK1 inhibition, further validating the role of this kinase in TNF signaling and apoptosis. In murine models, these knockout effects translated to tumor growth suppression. These results follow our knowledge of the molecular mechanisms of TAK1 inhibition. When TAK1 is not present, it cannot activate subsequent signaling pathways and is unable to promote cell proliferation. Interestingly we also noticed distinct cytokine profiles in the TAK1 KO tumors. Some of the differentials in cytokine expression between TAK1WT and TAK1KO cells can be used to explain the tumor suppression in the knockouts. For example, Endostatin which is present at higher concentrations in the TAK1KOs, suppresses both tumor growth and angiogenesis [25]. On the other hand, TAK1WTs express higher levels of Fetuin A, Lipocalin 2 and Reg3G, all of which promote tumorigenesis and angiogenesis [26][27][28].
At the time of writing, takinib demonstrated poor pharmacokinetics and bioavailability with rapid plasma clearance, limiting the ability to perform in vivo pharmacological studies with takinib. Further chemical modifications to the takinib scaffold such as the addition of solubilizing groups in the solvent accessible carbon positions of the aminobenzamide may improve bioavailability and provide a novel compound to evaluate TAK1's role in vivo. However, despite lack of a pharmacological in vivo intervention, studies with TAK1 KO cell lines showed significantly reduced tumor burden and increased survival, with changes in inflammatory cytokines present in the tumor microenvironment.
Due to TAK1's ubiquitous expression in most cells in the body, its pharmacological mechanism of action can be quite complex in certain disease contexts, such as cancer. It plays an integral role in inflammatory signaling of leukocyte populations, as well as angiogenic regulation associated with canonical inflammatory signaling [29]. However, previous studies as well as the one herein, have shown in certain cancer cell types TAK1 inhibition induces apoptosis, switching the cell fate from survival to apoptosis [30,16]. There lies the potential that in conjunction with immunotherapy treatments, which stimulate dormant immune suppressive immune cell populations in the tumor immune microenvironment, one may further exacerbate immune cancer clearance by induction of endogenous apoptotic pathways from the TAK1-TNF signaling axis.
Additionally, often genetic instability in the cancer cells allow for mutations to occur, leaving the cells immune to the kinase inhibitor effects or induction of compensatory mechanisms. However, therapies targeting TAK1 appear to have effects not only at the cancer cell level but additionally at modulating the immune microenvironment. Non-cancerous immune cells may provide a more stable "druggable" cell population with TAK1 inhibitors targeting the chronic inflammatory processes that provide a positive feedback loop in tumors, promoting cell proliferation and metastasis. Thus, in the case of highly pro-inflammatory and aggressive cancers TAK1 inhibition may reduce the inflammatory signaling milieu which aids in the pro-survival/growth phenotype associated with highly aggressive cancers. Overall, here we have shown the role of TAK1 in both various cancer cells and TAM populations, showing a potential therapeutic axis in modulating the immune microenvironment of tumors.

Animal care
Female nude mice were bred in-house or purchased from The Jackson Laboratory (Bar Harbor, ME, USA). All experiments were approved and carried out in accordance of the University of North Carolina-Chapel Hill, Institution Animal Care and Use Committee (IACUC) and conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Mice were housed in a temperature and humidity-controlled facility under 12-hour light/dark cycle (lights on at 7 am) and access to food and water ad libitum.

Tumor injections
Cells were mixed 1:1 with matrigel solution prior to injection into the mammary fat pad of female nude mice. Tumors were allowed to develop, and caliper measurements were obtained throughout the study period.

Cell culture
THP-1, MDA-MB-231, COLO 205 and all other cells were obtained from Duke CCF. All cells are tested for mycoplasma and authenticated via Duke CCF prior to use. Cells were incubated at 37° C in 5% CO 2 . THP-1 was cultured in RPMI 1640×, 10% FBS, 1% Penicillin-Streptomycin (PS), HEPES, Pyruvate, Glucose and BME. All cell lines were cultured according to ATCC media guidelines. www.oncotarget.com

Apoptotic, cytokine and chemokine arrays
THP-1 cells were differentiated as previously described in this manuscript. Following differentiation, cells were treated with 10 μM Takinib or DMSO vehicle control. 24 hours after treatment, supernatant was added to Human Cytokine XL proteome array (R&D Systems), or Angiogenesis proteome array. Apoptosis biomarkers were visualized with the Apoptosis Array kit (R&D Systems). All procedures were conducted in accordance with manufacturer protocol. Chemiluminescence was used to visualize protein quantities.

Immunohistochemistry
Briefly, 10 μM cut parrafin embedded tumor slices from TAK1 KO or WT tumors underwent antigen retrieval prior to H&E and TUNEL staining as per manufacturer protocol.

Western blot analysis
Cells were lysed (50 mM Tris, 150 mM NaCl, 1 mM EDTA, 1% Triton-X100, 1 mM DTT, cOmplete protease (Roche) and PhosSTOP phosphatase inhibitor (Roche)) after indicated treatment and run on Criterion XT Tris-HCl gel 4%-15% gradient (Bio-Rad). Following transfer to PVDF membrane and blocking in 5% non-fat dry milk in TBST, membranes were incubated with antibody overnight. After incubation with secondary antibody, chemiluminescence was used to visualize bands.

Drug treatment and cell viability assays
Cells were plated at 80% confluency on day one in media containing 10% FBS. After a 24 h incubation period, cells were pre-treated with either FBS-free media or varying doses of takinib in FBS-free media for 24 h prior to the addition of TNF (300 ng/mL). Cell viability assay was completed 24 h post-treatment and cell death quantified using Cell Titer Glo 2.0 (Promega) according to the manufacturer's protocol.

Quantification and statistical analysis
GraphPad Prism 8 was used for statistical analysis of viability, proteome assays, TAK1 KO analysis, survival analysis. For each analysis, total n and SEM are presented in the figure legend. Curves were plotted using variable slope (four parameters) non-linear fit. An alpha of 0.05 was used for all statistical analysis.

Author contributions
SS and KY designed and carried out the cell assays, data preparation, manuscript writing. PH synthesized and purified takinib. JR performed all animal studies. TH and SS directed the overall experimental studies. All authors have read and approved the final manuscript.