Background
Tumor is a complex heterogeneous biomass containing cancer cells and infiltrated leukocytes as well as extracellular matrices and surrounding stromal cells, collectively regarded as “tumor microenvironment (TME)” [
1‐
6]. Notably, cancer cells often expose to various TMEs composed of different host cells (e.g. macrophages and fibroblasts) and secreted soluble factors (e.g. interferon, cytokines and BMP), where these TMEs play critical roles in various developmental stages as well as therapeutic response of cancer [
7‐
11]. These relations are clearly illustrated by that the initiation and maintenance of tumors dependent largely on their ability to adapt to TME changes [
12‐
15]. For example, epithelial-to-mesenchymal transition (EMT) that is highly regulated by cellular niches convert epithelial cells into migratory/invasive mesenchymal cells and TME-induced pathological EMT also contributes cancer progress, especially metastasis. Moreover, the fact that paracrine interactions within TMEs also greatly impact on tumorigenesis suggests TMEs not only bi-directionally communicate with tumor cells but crosstalk to each other [
4,
5,
16‐
18]. Furthermore, it has been shown that adaptive responses in glucose metabolism contribute to macrophage migratory capacity during hypoxic condition and the later acquired therapeutic resistance of tumor cells depends largely on their ability to interact with TMEs [
5,
6,
8,
19,
20]. Therefore, TMEs might present as novel anticancer targets in addition to oncogenic alterations in tumors.
Perhaps the most direct evidence that TME affects cancer development is that patients suffered from chronically inflamed tissues generally exhibit a high cancer incidence [
21,
22]. Patients with ulcerative colitis (UC), one form of inflammatory bowel diseases (IBD), develop colon cancer with a 20-fold overall risk in comparison with population controls [
23]. Chronic stimulation of innate immunity is known to promote inflammation-driven tumorigenesis and induce cancer progression. Through releasing a variety of cytokines, chemokines, and cytotoxic mediators such as reactive oxygen species, metalloproteinases, interleukins (ILs), and interferon (IFN), these immune components are considered key pathological inflammatory factors promoting tumor progression [
22,
24,
25]. In addition, hypoxia is also a key feature of microenvironments in inflammatory conditions, for example, arthritis [
17,
26] and solid tumors often situate in a hypoxic inflammatory TME due to the rapid growth [
27‐
30]. Notably, both hypoxic and inflammatory responses could have pro- and/or anti-tumorigenic influences on oncogenic development, allowing cells to adjust the optimal cellular context via “outside-in” signaling with regulated transcription programming directed by master factors like
hypoxia-
inducing
factors (HIFs, mainly HIF-1α),
signal
transducers,
activators of
transcription family (STAT, such as STAT1) and NF-κB [
21,
22]. These two TMEs communicate to each other as well as oncogenic programs as evidenced by (i) interactions of HIF-1α-directed hypoxic responses with the NF-κB-mediated inflammatory programs [
7,
31,
32]; (ii) HIF-1α directs tumorigenic EMT; and (iii) HIF-1α interacts with oncogenic factors such as Myc and Ras oncogenes [
6,
11,
33‐
38]. Though molecular mechanisms underlying the crosstalk of TMEs with oncogenic programs remain largely unclear, modulations of hypoxic inflammatory responses have been seriously considered as an attractive novel target for cancer therapy [
29,
39].
Under hypoxia, HIF-1α protein is stabilized and its activity is increased mainly due to blockage of hydroxylation and proteasome degradation, followed by activation of a defined set of transcriptional programming. This set of target genes encodes proteins functioning in cellular processes such as erythropoiesis, glycolysis, EMT, metastasis, angiogenesis, therapy resistance and poor prognosis [
29,
39‐
43]. Similar to HIF-1α, NF-κB is a master transcription factor (TF) inducing expression of diverse target gene sets, thus playing an instrumental role in immune, stress and pathological inflammatory TME responses [
44]. Despite of the fact that NF-κB has identified as a redox-sensitive TF, it also binds to the promoter of the
HIF-1α gene and mediates HIF-1α expression under IL-1β, INF-γ, TNF-α and other cytokine treatments in normoxia [
45‐
48], providing a hint that inflammatory and hypoxic transcription programs are linked. In addition, the master TF STAT3 not only mediates inflammatory IFN response and regulates expression of AKT but also involves in the growth signal-induced HIF-1α expression [
49].
Cytokines such as IFN-α, through receptor interactions and subsequent induction of IFN-stimulated genes (ISGs) expression, play critical roles in inflammation [
36]. IFN-α signaling pathways include the classical JAK–STAT and other auxiliary pathways such as the PI3K/mTOR/AKT and MAPK-P38 axes, and dysfunction in signaling of PI3K/PTEN/AKT/mTOR, Wnt/GSK-3 and/or Ras/Raf/MEK/ERK axes is associated tightly with cancer progression and therapeutic resistance [
50]. Notably, inflammatory hypoxia, mainly through expression of HIF genes, contributes significantly to tumor malignance and metastasis in various cancer types. In sum, above findings provide a mechanistic involvement of the IFN-induced signaling and transcription programming with cancer development and metastasis that operate cooperatively with interactions with extracellular constituents of TME. With progressive release of pathological inflammatory cytokines and growth-induced tumor hypoxia, the transformed and infiltrated inflammatory cells of TME facilitate tumor growth and metastasis [
10,
18,
46,
51]. It is therefore reasonable that IFN might, through activation of these above pathways, play a critical role in hypoxia and tumorigenesis.
Previously, our group has demonstrated a novel ISGylation of HIF-1α (a form of posttranslational modification), which leads to a negative feedback loop of hypoxic response during inflammatory IFN stimulation [
36]. Here, we presented experimental data supporting that IFN-α promotes tumorigenic propensities through up-regulation of HIF-1α functions: (i) IFN-α induced expression of HIF-1α at transcriptional level; (ii) IFN-α activated the JAK/PI3K/PTEN/mTOR/AKT and Ras/p38/MEK/ERK signaling pathways to induce HIF-1α expression; (iii) HIF-1α-mediated expression of EMT genes and elevated wound-healing, invasion, EMT and anti-apoptotic abilities were observed upon IFN-α exposure; and (iv) correspondingly, pharmacological modulations of JAK, PI3K and MAPK-p38 significantly reduced the IFN-α-promoted tumorigenic and metastatic propensities. Thus, our results link the IFN-α-mediated inflammatory response to the HIF-1α expression and to tumorigenic and metastasis progression, providing evidence for a novel metastasis-promoting mechanism of cancer when situates in inflammatory hypoxia microenvironments. Pharmacological targeting of cancer cells on the reliance of supporting TMEs would seem a promising therapeutic avenue and our study might present such an example.
Methods
Chemicals, plasmids, antibodies, cell lines and small interference siRNA
All chemicals and antibodies unless described otherwise were from Sigma and Cell signaling, respectively. Manumycin A and rapamycin were from ENZO, FH535 and JAK inhibitor I were from Calbiochem, while SP600125 and 2-methoxy-estrodiol were from Cayman. AKT inhibitor IV, IFN-α-2a and IFN-γ were from Santa Cruz, Roche and Peprotech, respectively.17-AAG was from Medchem Express. Plasmids pshHIF-1α, pXP2-Twist-HRE and pRL-tk were from Dr. KJ Wu (CMU, Taiwan), and both pHA-GSK3β-WT and pHA-GSK3β-S9A plasmids were kindly provided by Dr. J. Sadoshima (PSU, USA). GST-Raf plasmid was a kind gift from Dr. TL Shen (NTU, Taiwan). IkBα-M mutant was a gift from MR Chen (NTU, Taiwan). The plasmids including pHA-HIF1α-DM and pBabe-HA-VHL were obtained from Addgene. Antibodies against fibronectin, vimentin, actin (Sigma), HIF-1α, HIF-1β, E-cadherin, N-cadherin, β-catenin, Bcl-2, MCL-1 (BD Bioscience), STAT1Y701 (Invitrogen), CA9, Glut1, PGK1, 4EBP1, MDR1, S6K, S6 and mTOR, Survivin (Genetex), IFN-alpha antibody (R&D systems, MN, USA), Bmi1 (Millipore Inc.) were obtained commercially. STAT1 antibody was from Dr. CK Lee (NTU, Taiwan). Human 769-P, Caki-1 renal carcinoma, MDA-MB-231, MDA-MB-453, MCF7 breast cancer, SKOV3 ovarian cancer, SW480, DLD-1, LoVo, COLO205, HT29, RKO, Ls 174 T colorectal cancer cell lines were from ATCC and cultivated in DMEM media (except Ls174T in MEM media) with 10% fetal bovine serum (FBS), 2 mM glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin in a 37 °C incubator. Simulated hypoxia was achieved by either exposure to 260 μM desferoxamine or placing cells in a hypoxia incubator (1% O2, 5% CO2 and 94% N2, ASTEC). Ambion STAT1 (s278), RELA/NF-κB (s11914, Life Technologies) siRNAs and ON-TARGETplus SMARTpool siRNAs targeting Akt or mTOR (GE Dharmacon) were purchased commercially.
RNAi knockdown, transfection and immunoblotting assay
RNAi-mediated knockdown of gene expression was achieved by either transfection of siRNAs or pshHIF-1α plasmid. Transfection of siRNAs (0.1–0.3 μM) and plasmids were conducted respectively using Lipofectamine RNAiMAX (Invitrogen) and Lipofectamine 2000 (Life Technology) with the manufacturer’s protocols. After 48 h, cells were harvested and lysed for immunoblotting assay as described [
36].
Reverse transcription-polymerase chain reaction, luciferase reporter and colonogenic assays
RT-PCR and luciferase reporter assay were employed to determine gene expression. Trizol-isolated RNA were reversely transcribed by random primers and SuperScript III reverse transcriptase (Invitrogen) and cDNA pools were amplified with following primers (MB Mission Biotech): 5-‘ATGGATCCAATGGTTCTCACAGATGAT-3’ and 5’-ATGATATCTTATACGTGA ATGTGGCCTGT-3′ for HIF1α; 5’-TCCACCGCTAT GGGGAG AGCA-3′ and 5’-AGCTCCAGGATGGTGACCTTC-3′ for Glut1; 5′-ATG TCCACCAGGTCCGTG -3′ and 5’-GGATTTCCTCTTCGTGGAGT-3′ for Vimentin; 5’-ATGCCGCGCTCCTTCCTGGTC-3′ and 5’-TAATGTGTCCTTGAAGCAACCA GG-3′ for Slug; 5’-TCATGGCCAACGTGCGGGAGC-3′ and 5’-CCAGACCGAGA AGGCGTAGCT-3′ for Twist;; 5’-AGCTACTGCCATCCAATCGC- 3′ and 5’-GGGC GAATCCAATTCCAAGAG-3′ for VEGF-A; 5’-GCTGGAAGGTGGACAGCGAG- 3’and 5’-TGGCATCGTGATGGACTCCG-3′ for β-actin as a normalization control. PCR products were separated on 2% agarose gel, stained and visualized with a Dolphin-Doc system (Wealtec). Both luciferase reporter and clonogenic assays were done as described [
36].
Wound-healing and invasion assays
Wound-healing and invasion assays were used to determine cellular migratory and invasion abilities of cells. Cells (~ 80% confluent) were exposed with or without inhibitors for 30 min and treated with IFN-α for another 24 h. For wound-healing assay, cells were scrapped smoothly to generate gap and images were captured. For invasion assay, 1.25 × 105 cells were seeded into Matrigel-coated transwell inserts (8-μm pore size) with different treatments in serum-free medium, and 10% FBS media in lower chamber acting as chemoattractant. After 48 h, cells those did not invade were wiped out with cotton swab and those invaded into the underside of membranes were fixed in methanol, stained with 0.5% crystal violent and scored under a microscope (100X objective). Six fields of each sample were captured to evaluate the average number of invaded cells.
GST-Raf pull-down assay for activated Ras
GST-Sepharose beads (20 μl, GE Healthcare) was incubated with bacterial lysates with GST or GST-Raf fusion proteins in a final volume of 300 μl of buffer A (1 mM DTT, 20% glycerol, 50 mM Na2H2PO4 pH 8.0, 300 mM NaCl and 1 mM PMSF) at 4 °C for an hour with rotation, washed 5 times with 0.8 ml of buffer A with 1% Triton X-100 and once with 0.8 ml of buffer B (10 mM Tris pH 7.5, 1 mM EDTA, 210 mM NaCl, 15% glycerol, 0.25% NP-40, 1 mM DTT, 1 mM Na3VO4, 1 mM PMSF and 1X protease inhibitor cocktail. Beads were then reacted with a 300 μl mixture containing cell lysates, 1X protease inhibitor cocktail, 1 mM DTT and 1 mM PMSF at 4 °C for hour with rotation, washed 3 times with buffer B, boiled for 7 min in 1X SDS sample buffer, and then subjected to immunoblotting analysis.
Extraction of cytoplasmic and nuclear fraction protein
Fractionation of 769P cells was performed using NE-PER nuclear and cytoplasmic extraction reagents according to manufacturer protocol (Thermo Scientific). The protein contents of cytoplasmic (supernatant) and nuclear extracts were determined by electrophoresis and immunoblotting analyses.
The vasculogenic mimicry (VM) assay
The VM assay was performed with matrigel (Corning) following the manufacturer’s protocol. Briefly, Matrigel was thawed at least for two hours at 4 °C. 100 μl of the de-thawed matrigel was used to coat the wells of a 96 well plate and was allowed to polymerize for 2 h at 37 °C. After equilibrating the gel with the complete growth medium, 4 × 104 cells were seeded in each well and incubated for 24 h. The tube formation was observed under a phase contrast microscope.
The anchorage-independent cell growth/ soft- agar assay
For the soft agar assay, cells (2 × 104) exposed to pharmacology inhibitors or not were mixed with 2 mL of 0.3% agarose-low EEO (Bionovas)-containing DMEM and then overlaid onto a 2 mL layer of pre-coated 0.6% agarose-containing DMEM in 6-well plates. After cultivation for 3–4 weeks, colonies were stained using 0.005% crystal violet and the number of colonies was scored on 6 random fields triplicate under microscope in three independent experiments.
Here, sphere-formation and xenograft tumor assays were employed to determine the tumorigenic activities of human 769P renal cancer cells with IFN-α treatment and/or HIF-1α knockdown in vitro and on nude mice in vivo [
36,
52]. For the sphere-formation assay, cells were transfected with pSuper vector control or pshHIF-1α plasmid for one day, followed by addition of IFN-α (1000 U/ml, 24 h) and then cells were counted and plated at a cell density of 1000 cells per well in ultra-low attachment 24-well plate (Corning) containing 1 ml of serum-free DMEM/F12 supplemented with 2% B27 (Invitrogen), human recombinant fibroblast growth factor 2 (FGF-2, 20 ng/ml) and epidermal growth factor (EGF, 20 ng/ml, Peprotech). Severn days after plating, total sphere colonies formed in each well were counted under microscope. The protocol for animal experiments was designed in accordance with the guidelines of the Institutional Animal Care and Use Committee and approved by the same (No. 20160242). For the tumorigenic xenograft model, 4- to-5-week-old BALB/c nude mice were supplied by Bio LASCO and the experiment was carried out in the Animal Center at National Taiwan University College of Medicine (NTUCM) and maintained in the specific pathogen-free (SPF) facility. A total of 10
7 cells were re-suspended in 100 μl mixture containing 50 μl of 1X PBS and 50 μl of matrigel, and then injected subcutaneously into the dorsal region of the nude mice. Tumor size was measured every 2-to-4 days using calipers and tumor volumes were estimated by the following formula: V = ab
2/2 (a, length; b, width).
Quantitative measurements and statistical analyses
Quantitative measurements of obtained results were performed using ImageQuant program (Molecular Dynamics) and presented as the mean ± standard error of mean (S.E.M.) of at least three independent experiments (n ≥ 3). Statistical analyses were performed using the simple Student’s t-test. Data were considered to be significant if the P value was less than 0.05 (P < 0.05).
Discussion
Tumor microenvironment (TME) provides supportive niche for tumor progression and establishes communication links with cancer cell survival, stemness-like property, hypoxia, inflammation and progression immunity. It has also been reported as an important step in the metastatic cascade of epithelial tumors and thus suggesting that TME could be constituted with corresponding signaling pathways served as promising target(s) for cancer therapy [
53‐
55]. Despite of great advances in health care programs, basic, translational and clinical medical research in the past decades, cancer remains a daunting threat around the world despite great advances in health care programs, basic, translational and clinical medical research in the past decades [
12,
56]. A global unmet need for understanding fundamental bases of cancer biology and new interventions of anticancer therapeutics is urgently needed [
56,
57]. It has become notably apparent that generating optimal oncogenic responses needs proper signaling crosstalk between inflammatory and hypoxic pathways of TME [
58]. In this regard, solid tumors often situate in hypoxic inflammatory niche containing leukocyte infiltrates in which ~ 20% of them arising in association with chronic inflammation [
1‐
3,
7,
17,
26]. Moreover, inflammatory and hypoxic niches as well as corresponding signaling pathways play a complex role in cancer development through regulating expression of tumorigenic propensities. Thus, targeting the inflammatory interferon-driven and hypoxia-induced pathological TME and corresponding signaling pathways of our study might present as such an example for novel anticancer intervention, particularly identified signaling pathways those regulate the expression of HIF-1α in malignant tumor.
We showed that cellular signaling responses of hypoxic and inflammatory TMEs interact and crosstalk through up-regulation of expression of HIF-1α and associated tumorigenic activities, which are mediated by one inflammatory cytokine IFN-α. This novel communication of above two TMEs, via the JAK pathway with PI3K/PTEN/ mTOR/AKT/GSK3β/β-catenin axis and the Ras pathway with p38/MEK/ERK/JNK axis, further fine-tunes cellular homeostasis of tumor cells, i.e., adapting to hypoxia, enhancing survival as well as promoting migration, invasion and EMT. Our study has thereby not only revealed a new crosstalk between two key inflammatory and hypoxic TMEs, but provided a vicious molecular connection between hypoxic inflammation and tumorigenic programming. In this same line, we have recently reported a new negative feedback loop for the HIF-1α-mediated pathway involving the regulation of HIF-1α by ISG15 and ISGylation of HIF-1α [
36]. Using genetic and pharmacological inhibition, our study has provided new insights into an important involvement of hypoxic inflammatory TME in tumorigenesis. In addition, clinical and/or preclinical inhibitors of signaling pathways involved in regulating HIF-1α expression used in this study also present as new strategies for treatment of human cancer. Consistently, the anti-inflammatory drugs, such as NSAID, have been reported to reduce the risk of several solid tumors [
59].
Our advanced understanding of a cancer-TME interaction in molecular regulation of tumorigenesis shall lead to a rationale guidance for the development of new cancer therapeutics targeting TME-supplemented IFN signaling and HIF-1α programming to prevent disease relapse after initial diagnosis and treatment [
53,
54,
60].From the points of view on interventions for metastatic cancer, HIFs and regulatory pathways of gene expression are obvious targets of interest and targeting disseminated cancer cells, corresponding reliance/molecules and signaling responses/interactions on TMEs could also be promising revenues. Moreover, mounting experimental and clinical evidence have also suggested a central role for the signaling networks operating to promote a metastatic and/or therapeutic resistance cascade in cancer development that involves interactions of AR, TMPRSS2, HGF and c-MET with critical components of TMEs [
61]. In agreement, our preclinical study on that signaling inhibitors including LY297002, JAKi inhibitor, 17-AAG and 2-ME all diminished the tumor propensities and growth also supports the novel usage of these signaling inhibitors in anti-cancer therapy.
IFN-α, which belongs to a family of “biologic response modifiers” and activates a network of signaling molecules, is FDA-approved for hairy cell leukemia, malignant melanoma, AIDS-related Kaposi’s sarcoma, follicular non-Hodgkin’s lymphoma as well as other clinical indications such as renal cell cancer and cervical cancer. The signaling pathways and molecules, such as PI3K, JAK, and HSP90, have also been suggested and developed into new anti-cancer strategies as well as for overcoming drug resistance [
62‐
65]. For example, due to its frequent activation, the JAK/STAT axis is an attractive target for breast cancer therapy and thus clinical trials of JAKi in advanced breast cancer are ongoing [
62]. Functional inhibition of HSP90 causes the degradation of its client proteins and thus subsequently providing a novel anti-cancer intervention to concomitantly disrupt multiple oncogenic signaling cascades especially in in a variety of client protein-driven tumors [
64,
65]. Similarly, the PI3K/AKT/mTOR signaling pathway is commonly deregulated in human malignancy including non-small cell lung cancer (NSCLC) [
63]. Thereby, this pathway is also an excellent target for many therapeutic development and its inhibitors are undergoing heavy clinical evaluation. In agreement, our results also showed that pharmacological inhibitors targeting the IFN-α signaling and pathways could not only modulate HIF-1α expression but also tumorigenic activities such as EMT and tumor invasion. Thus, our study provides a good example from molecular pharmacological modulations to modular tumor therapy. Specifically, we observed a novel regulation of hypoxia TME (mainly HIF-1α expression and functions) by inflammatory IFN-α. Importantly, IFN-α promotes tumorigenic propensities such as EMT, invasion and anti-apoptosis abilities and crosstalk to hypoxic TME mainly through up-regulation of HIF-1α expression [
55,
58,
66‐
69]. Coupled with the importance of hypoxia and inflammation during tumorigenic processes, our reports on physical, functional and genetic interactions among key components of these two TMEs further suggest the critical roles of both the individual pathway of and the interaction between IFN-α and HIF-1α for the underlying tumorigenic mechanism(s) in the context of hypoxic inflammation. The observations that the IFN-induced ISG15 conjugation (ISGylation) pathway can modulate the cancer cell-killing activity of drugs also support the above notion [
70,
71].
Tumor is a complex biomass containing heterogeneous cancer cells and TME with the surrounding stromal, infiltrated immune cells and extracellular matrices. It is known that TME plays critical yet diverse roles in various stages of cancer development, but mechanisms underlying signaling crosstalk and molecular communications of TMEs with oncogenic programs remain unclear. Through studying molecular mechanisms underlying interactions of inflammatory molecule interferon (IFN) with hypoxic TME, we provided the first experimental evidences for a novel communicating mechanisms occurring within TMEs in cancer progression of renal oncogenic development, thus representing as one emerging paradigm of cancer pathology. We unraveled a hypoxic pro-inflammatory role of IFN-α in the HIF-1α mediated TME and then leading to promotion of tumor tumorigenesis, migration/ invasion, EMT, vasculogenic mimicry and drug resistance (schematic illustration in Fig.
7e). Importantly, pharmacological modulations of HIF-1α as well as the JAK/PI3K/PTEN/mTOR/AKT and Ras/p38/MEK/ERK signaling axes all significantly reduced the above IFN-α promoted tumorigenic propensities. In this regards, advances in the understanding of cancer-TME interaction and the drugs targeting TME-associated disease biology and pathology may bring about the TME-guided therapy for various diseases.