Introduction
The accelerated growth and erratic angiogenesis of solid tumors induce a lack of oxygen and nutrients in parts of the tumor that are distal to functional blood vessels. This is known to have important repercussions for treatment sensitivity and prognosis of cancer patients [
1‐
3]. Variation in the duration and severity of hypoxic stress differentially activates different "do or die" programs and leads to substantial phenotypic variations amongst otherwise identical tumor cells. Best known in this perspective is the Hypoxia-Inducible Factor 1 (HIF-1) pathway, which is induced during hypoxia and has often been associated with poor prognosis in solid tumors including breast cancer [
4‐
6]. Under more anoxic conditions another hypoxia-related program is activated, that is the unfolded protein response (UPR) [
7], which results from endoplasmic reticulum (ER) stress caused by misfolding of proteins in the ER. One of the mechanisms that are activated by the UPR is autophagy [
8], which can temporarily relieve ER stress by reutilization of cellular components. ER stress-induced expression of activating transcription factor 4 (ATF4) and CHOP (C/EBP homologous protein) also leads to transcription of Tribbles homolog 3 (TRIB3) [
9]. TRIB3 is known to inhibit phosphorylation of AKT/protein kinase B (AKT) [
10]. AKT is a phosphoinositide-dependent serine/threonine protein kinase that plays a critical role in the signal transduction of growth factor receptors. Furthermore, TRIB3 has been described to have a role in the mitogen activated protein kinase (MAPK) pathway [
11] and nuclear factor (NF) κB activated apoptosis [
7]. However, the literature is not conclusive on the role of TRIB3 in cell fate; it has been described to have pro-apoptotic [
9] as well as anti-apoptotic features [
12]. TRIB3 is strongly upregulated by hypoxia, nutrient starvation and ER stress-inducing agents [
13‐
15] and has been implicated in ER stress-induced autophagy [
16]. Interestingly, TRIB3 was shown to be upregulated compared with normal tissue in tumors of the human lung, colon, and breast [
14,
17,
18].
We hypothesized that TRIB3 could be relevant for breast cancer prognosis, because in breast cancer, several ER stress, UPR, and/or hypoxia-associated markers have been found to be related to prognosis [
19‐
22]. Furthermore, the involvement of TRIB3 in the other pathways described above makes it more interesting than other proteins in the UPR and other hypoxia pathways, because, for example, the growth factor receptor-induced phosphorylation cascades are also known to be relevant for breast cancer treatment [
23]. We hypothesized that TRIB3 could be an important protein by determining the tumor cell fate under stressed conditions. To this end,
TRIB3 mRNA expression levels were correlated with disease characteristics and patient survival in a large breast cancer patient cohort. In addition, we determined if
TRIB3 expression could be induced in breast cancer cells by a variety of stressors, and we assessed the expression and localization of TRIB3 in breast cancer xenografts and patient tissues. Finally, we determined the effect of
TRIB3 knockdown on the hypoxia response of breast cancer cells.
Materials and methods
Patient samples
Coded tumor tissues were used in accordance with the Code of Conduct of the Federation of Medical Scientific Societies in the Netherlands ('Code for Proper Secondary Use of Human Tissue in the Netherlands' [
24]). The study adhered to all relevant institutional and national guidelines, and was reported according to REMARK guidelines [
25]. A series of 247 patients with unilateral, resectable breast cancer who had undergone resection of their primary tumor between November 1987 and December 1997 were selected based on the availability of frozen tissue in the tumor bank of the Department of Laboratory Medicine of the Radboud University Nijmegen Medical Centre. This bank contains frozen tumor tissue from breast cancer patients of seven different hospitals of the Comprehensive Cancer Centre East in the Netherlands. The patients, inclusion and exclusion criteria, and their treatment have been described earlier [
26]. To summarize, the median age was 59 years (range 31 to 88 years). Patients had undergone modified radical mastectomy (
n = 178) or a breast-conserving lumpectomy (
n = 69). Postoperative radiotherapy to the breast after an incomplete resection or after breast-conserving treatment, or parasternal radiotherapy when the tumor was medially localized, had been administered to 179 patients. Adjuvant systemic therapies were given almost exclusively to the 128 patients with lymph node involvement, according to standard practice at that time [
26]. During follow up, 95 patients experienced a recurrence, either local or distant. A representative part of each tumor was macroscopically selected by a pathologist. The material was frozen in liquid nitrogen and determination of estrogen receptor and progesterone receptor status was performed by ligand binding assay according to the dextran-charcoal method [
27]. Aliquots of tissue were pulverized using a microdismembrator (Braun, Melsungen, Germany) and kept in liquid nitrogen until RNA isolation.
To obtain patient material containing the exogenous hypoxia marker pimonidazole, patients with newly diagnosed breast cancer were enrolled in the period 1997 to 1999 in a tumor hypoxia study in accordance with a research protocol approved by the Institutional Review Board at the University of North Carolina Hospitals. The patients provided signed informed consent prior to their participation in the study. Prior to tumor biopsy, patients received an intravenous infusion of pimonidazole hydrochloride (0.5 g/m
2, Hypoxyprobe-1™, NPI Inc, Belmont, MA, USA) diluted in 100 ml NaCl 0.9% for 20 minutes. Between 16 to 24 hours later, biopsies were obtained from primary tumors. After biopsy, fresh tumor samples were placed in cold 10% neutral buffered formalin, held at 4°C for 12 to 24 hours, and processed into paraffin blocks. Four
μm thick sections were sectioned and mounted on glass slides in preparation for immunohistochemical staining. One slide per block was stained with hematoxylin and eosin for pathologic review to confirm the presence of tumor. Based on the hypoxia scoring of the tumors according to a calibrated scoring system [
28] three tumors with a high percentage of hypoxia (> 15% of the viable area) were chosen for analysis of TRIB3 expression.
Xenograft tissue
Xenografts of MDA-MB-231 cells were obtained after subcutaneous injection of 1 × 10
6 cells suspended in RPMI (MP Biomedicals, Illkirch, France) in six-week-old female athymic mice (BALB/c nu/nu, BonholdGard, Denmark). Animal housing and experimental procedures were in accordance with international guidelines and approved by the local ethical committee for animal use, respectively. At a mean tumor diameter of 6 to 8 mm at approximately six weeks after seeding that time mice were injected intravenously with 0.1 ml of 0.9% NaCl containing 2 mg of the hypoxic cell marker pimonidazole hydrochloride (1-((2-hydroxy-3-piperidinyl)propyl)-2-nitroimidazole hydrochloride, Natural Pharmaceuticals, Inc., Research Triangle Park, NC, USA) 60 minutes before harvesting the tumors. Pimonidazole is a bioreductive chemical probe with an immuno-recognizable side chain, which was described previously as a marker for hypoxia [
29‐
31]. The animals were killed by cervical dislocation and the harvested xenograft tissues were immediately frozen in liquid nitrogen.
Cell lines
MCF7 and MDA-MB-231 (ATCC, LGC Promochem, London, UK) human breast cancer cells were cultured for a limited number of passages in standard culture medium ((DMEM, MP Biomedicals, Amsterdam, the Netherlands) with 10% dialyzed FCS (Invitrogen, Breda, the Netherlands), 2 mM L-glutamine, 20 mM HEPES, 10 U/ml penicillin, 10 μg/ml streptomycin, and nonessential amino acids (NEAA, MP Biomedicals)) at 37°C with 5% CO
2, unless stated otherwise. Knockdown of
TRIB3 was performed using siRNA transfection reagent SAINT-RED (Synvolux Therapeutics BV, Groningen, the Netherlands). siRNAs MISSION
® siRNA Universal Negative Control #1 (SIC001), TRIB3 (1) (SASI_Hs01_00197510) and TRIB3 (2) (SASI_Hs01_00197511) were acquired from Sigma-Aldrich (St. Louis, MO, USA). The knockdown of HIF-1α, PRKR-like endoplasmic reticulum kinase (PERK), inositol-requiring 1 (IRE1), activating transcription factor 6 (ATF6), ATF4 and CHOP was performed in MCF7 cells as described previously for HCT116 cells [
32].
Detection of RNA
Total RNA was isolated with the RNeasy RNA isolation kit (Qiagen, Hilden, Germany) with on-column DNAse treatment. For the reverse transcriptase quantitative PCR (RT-qPCR) Reverse Transcription System from Promega Benelux B.V. (Leiden, the Netherlands) was used and cDNAs were amplified with specific primers (TRIB3 forward: att agg cag ggt ctg tcc tgt g, reverse: agt atg gac ctg gga ttg tgg a; VEGF forward: ccg cag acg tgt aaa tgt tcc t, reverse:cgg ctt gtc aca tct gca agt a; PAI-1 forward:ggc cat gga aca agg atg aga, reverse:gac cag ctt cag atc ccg ct; CAIX forward:gag gcc tgg ccg tgt tg, reverse:aat cgc tga gga agg ctc ag; MIF forward:cag ccc gga cag ggt cta c, reverse:tct tag gcg aag gtg gag ttg; ATF4 forward: cct tca cct tct tac aac c, reverse: gta gtc tgg ctt cct atc t; GRP78/BiP forward:tct atg aag gtg aaa gac cc, reverse:ctg tca ctc gaa gaa tac ca) using Sybr Green Master Mix (Applied Biosystems, Nieuwerkerk a/d lJssel, the Netherlands) on an ABI Prism 7700 Sequence detection system (Applied Biosystems, Nieuwerkerk a/d lJssel, the Netherlands). All samples were normalized for levels of hypoxanthine-guanine phosphoribosyltransferase (HPRT) expression. In the experiment in which the pathways involved in TRIB3 regulation were determined, RNA was reverse-transcribed using I-Script (Bio-Rad Laboratories BV, Veenendaal, the Netherlands). Furthermore, these samples were normalized to 18S rRNA.
Detection of protein
Protein localization of TRIB3 and other relevant proteins or markers in xenograft and human breast cancer tissue was determined using immunohistochemical staining. The frozen xenografts or frozen tumor tissues were sectioned at 5 μm thickness after embedding in Tissue-Tek (Sakura Finetek Europe, Zoeterwoude, the Netherlands). Sections were mounted on poly-L-Lysine coated microscopic slides (Menzel, Braunschweig, Germany), fixated in acetone and rehydrated in 0.1 M phosphate buffered saline pH 7.4 (PBS, Klinipath, Duiven, the Netherlands). Between all antibody incubations, sections were rinsed three times in PBS (Klinipath, Duiven, the Netherlands). The following primary antibodies were dissolved in primary antibody diluent (PAD, AbD serotec, Oxford, UK) and stained with the appropriate labeled secondary antibodies; rabbit-anti-TRIB3 (Calbiochem, San Diego, CA, USA), PAL-E for blood vessels in human tissue (Euro Diagnostica, Arnhem, the Netherlands), rat monoclonal against mouse endothelium (9F1, kind gift from Dr. G. van Muijen of the Department of Pathology, Radboud University Nijmegen Medical Centre, Nijmegen, the Netherlands), and rabbit-anti-pimonidazole for hypoxic regions in xenograft tissue. Slides were enclosed using Fluorostab (ICN pharmaceuticals, Inc. Zoetermeer, the Netherlands). Paraffin embedded breast tissue sections (4 μm) were deparaffinized and re-hydrated in histosafe (Klinipath, Duiven, the Netherlands) and graded alcohols (100%-96%-70%). For antigen retrieval, slides were heated (90°C) in 10 mM citrate buffer pH 6.0 (DAKO, Glostrup, Denmark). Endogenous peroxidase was blocked with 3% H2O2 in methanol before the incubation with polyclonal rabbit-anti-TRIB3 (Novus Biologicals, Cambridge, UK) or rabbit-anti-pimonidazole. Subsequently, slides were incubated with Powervision (Lab Vision Products, Thermo Fisher Scientific, Cheshire, UK) and visualization of peroxidase was performed using PowerDAB (Lab Vision Products, Thermo Fisher Scientific, Cheshire, UK). A counterstain with hematoxylin (Klinipath, Duiven, the Netherlands) was performed before dehydration and mounting using Permount (Thermo Fisher Scientific, Cheshire, UK).
All microscopic images were acquired using IP-Lab for Macintosh software (Scanalytics Inc., Fairfax, VA, USA) in combination with a monochrome CCD camera (Retiga SRV, 1392 × 1040 pixels) and a RGB filter (Slider Module; QImaging, Burnaby, BC, Canada) attached to a motorized microscope (Leica DM 6000, Wetzlar, Germany). For comparison between TRIB3 and pimonidazole expression, whole tumor sections were scanned with a 10× objective at 100× magnification [
33]. The individual colors (DAB (brown) and hematoxylin (blue) signals) were extracted and unmixed from the bright fields images [
33].
Treatment of cells
For anoxia experiments, treatments included addition of 1.2 U/ml oxyrase (Oxyrase, Inc., Mansfield, OH, USA), a reagent that removes all oxygen from culture medium [
34,
35] or 100 μM CoCl
2 (Sigma, Zwijndrecht, the Netherlands), a hypoxia mimetic, to the standard culture medium. The effect of CoCl
2 on HIF-1α protein levels stabilization was confirmed by performing a standard western blot analysis [
36] with minor adjustments using mouse anti-HIF-1α (BD Biosciences, Erembodegem, Belgium) antibody and the effect on HIF-1 activity by measuring CAIX expression on mRNA level.
For less than 0.01% O
2 exposure, MCF7 cells were transferred to a hypoxic culture chamber (MACS VA500 microaerophilic workstation, Don Whitley Scientific, West Yorkshire, UK) as described previously [
32].
For 0.1% to 0.5% O2 exposure MDA-MB-231 cells were transferred to a hypoxic culture chamber (H35 hypoxystation, Don Whitley Scientific, West Yorkshire, UK).
In nutrient starvation experiments, either glucose-free DMEM including the same additions as the standard culture medium, or the standard culture medium in which the NEAA were omitted, were used. To induce ER-stress, 0.5 μM thapsigargin (Sigma, Zwijndrecht, the Netherlands) was added to the standard culture medium.
Cell proliferation assay
To measure cell survival we used the CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS, Promega Benelux BV, Leiden, the Netherlands). Assays were performed in a 96-well plate format in which cells were transfected with siRNA 24 hours after seeding. Treatment started 24 hours after transfection. Directly after treatment for 48 hours 20 μl of MTS solution was added to 100 μl cell culture medium per well. After three hours of incubation with MTS absorbance at 492 nm was measured using a Multiskan Ascent Photometric plate reader (Labsytems, Helsinki, Finland).
Statistical analyses
Statistical analyses were carried out using SPSS 10.0.5 software (SPSS Benelux BV, Gorinchem, the Netherlands). Normality of distribution of variables was tested using Kolmogorov-Smirnov testing. Differences in TRIB3 expression in cells with different treatments were evaluated using analysis of variance (ANOVA) and post-hoc Tukey's testing. Differences in levels of TRIB3 expression in samples from patients categorized by clinicopathological characteristics, used as grouping variables, were assessed with non-parametric Mann-Whitney U tests (for two groups) or with Kruskall-Wallis tests (for more than two groups) where appropriate. Non-parametric correlations were established using Spearman Rank correlation testing. Disease free survival (DFS) time (defined as the time from surgery until diagnosis of recurrent or metastatic disease) and overall survival (OS) time (defined as the time between date of surgery and death by any cause) were used as follow-up endpoints. Survival curves were generated using the method of Kaplan and Meier, after patients were categorized by TRIB3 expression in either two or three equally sized groups, thus either at the p50, or at the p33 and p66. Equality of survival distributions was tested using log-rank testing, with Mantel-Cox test for trend when more than two groups were analyzed, and using Cox univariate and multivariable regression analyses. Variables were selected for the multivariable survival analyses by backward stepwise selection, with removal testing based on the probability of the likelihood-ratio statistic, at a P > 0.10. Two-sided P-values below 0.05 were considered to be statistically significant. Cases with more than 96 months of follow up were censored at 96 months, because of the rapidly declining number of patients thereafter, although data on some patients was available for up to 169 months after primary surgery. This censoring was done because after a certain period of observation patients are frequently redirected to their general practitioner for checkups and mammography and cease to be among the outpatients of the breast cancer clinics. Further inclusion of the small remaining groups in statistical analyses would be non-informative. Additionally, the data met the proportional hazard assumption, and hazard ratios did not change over time.
Discussion
Here, we find that TRIB3 is associated with poor prognosis of breast cancer patients, independent of other clinicopathological characteristics. TRIB3 is expressed in hypoxic areas of both breast cancer xenografts and human breast tumor tissues. We found that this TRIB3 up-regulation is induced by ER stress, hypoxia and nutrient starvation, and is dependent on the PERK pathway of the UPR. TRIB3 induction is most pronounced at a moderate oxygen concentration and TRIB3 seems involved in hypoxia response of tumor cells.
Tribbles was originally identified as a delayer of mitosis in
Drosophila Melanogaster [
38‐
40]. One of the three human homologs, TRIB3, has recently been described to be involved in regulation of the cell cycle regulator cdc25A in human cells [
41]. Furthermore, TRIB3 has a role in insulin sensitivity and diabetes [
10,
42‐
44] and has also been described to interact as a scaffold protein in multiple signaling cascades, including v-akt murine thymoma viral oncogene homolog 1 (PKB/Akt) [
10,
13] but also MAPK [
11] pathways. This apparent involvement of TRIB3 in tumor cell proliferation and/or survival and growth factor receptor signaling cascades combined with its role in hypoxia and cell stress pathways [
9,
13‐
15,
45,
46] originally spurred us to investigate its role in breast cancer progression. Our results presented here show that TRIB3 could be involved in the hypoxia response of breast cancer cells; it is induced by cell stressors including hypoxia, ER stress, and nutrient starvation in breast cancer cells, which is in line with earlier observations in other cell types [
9,
13‐
15,
45,
46]. We confirmed here that TRIB3 is part of the UPR [
9,
12], more specifically the PERK/ATF4/CHOP pathway. We find that knockdown of
TRIB3 results in cells that are more sensitive to hypoxia. One pathway involved in hypoxia tolerance of tumor cells is the UPR [
32]. Tumor cells utilize this pathway to counter cell stresses like hypoxia, most likely through induction of autophagy [
32]. By "self-eating" of cellular components, cells can provide for their own energy and constituents, thereby surviving prolonged periods of severe hypoxia [
32]. TRIB3 is known to inhibit PKB/Akt [
10], a putative link between the UPR and autophagy [
16].
Importantly, hypoxic cells are particularly refractory to treatment and tend to have an increased metastatic potential [
1,
3]. Further, the hypoxia and nutrient starvation-induced UPR-pathway is important for cancer treatment efficacy [
21,
32,
47,
48] and therefore an interesting new target for therapy. In comparison with HIF-1, which is activated over a wide range of oxygen concentrations of around 2%, maximum activation of the UPR requires exposure to more severe hypoxia (reviewed in [
49]). Nevertheless, activation of the UPR has been shown to protect tumor cells against hypoxia-induced cell death both
in vitro and
in vivo over a range of oxygen concentrations [
47,
48,
50]. We are the first to describe the specific oxygen tensions that induce
TRIB3, and found that up-regulation of
TRIB3 does not increase with decreasing oxygen tensions. These results indicate that the up-regulation of
TRIB3 under moderate hypoxia is at least partly due to other pathways than the UPR. PI3K is already described to up-regulate
TRIB3 expression [
13] and could be responsible for the up-regulation of
TRIB3 at the more physiological moderate hypoxia, rather than at severe hypoxia (i.e. anoxia). Severe hypoxia is mainly found in perinecrotic areas within solid tumors and probably has less influence on prognosis and treatment sensitivity than areas with more moderate hypoxia. Within the moderate hypoxic tumor regions cells can adapt, and are eventually more likely to reoxygenate and/or metastasize as they are more proximal to blood vessels.
Here, we show that TRIB3 expression can independently predict disease outcome in human breast cancer patients. The importance of TRIB3 in cancer is supported by the finding that TRIB3 is also involved in the prognosis of colorectal cancer patients [
17]. The data suggest that TRIB3 is induced by hypoxia in an ATF4 dependent manner and supports hypoxia sensitivity, but from the measurements of other hypoxia markers we find that
TRIB3 mRNA abundance from tumor biopsies is not merely a reflection of tumor hypoxia. This suggests that
TRIB3 marks or supports tumor aggressiveness rather than reflecting hypoxia itself. Furthermore, the predictive value of
TRIB3 expression in our study was specifically significant in the patient group that received radiotherapy. Combined with the observation that TRIB3 is mostly expressed in the therapy resistant hypoxic areas of breast tumors this indicates that the prognostic role of TRIB3 could indeed be a consequence of therapy resistance. When this would be solely due to radiotherapy resistance one would expect also a correlation with loco-regional control but this was not the case. However, a correlation with distant control could also not be found, probably both due to low number of events and small group sizes. In addition, as this was a non-randomized retrospective analysis, any predictive value of
TRIB3 mRNA remains to be confirmed in a larger prospective trial.
Conclusions
Examining TRIB3 expression in human breast tumor material revealed that there was an independent association with poor prognosis of breast cancer patients. A relation of TRIB3 with hypoxia was seen by the co-localization with the hypoxia marker pimonidazole in both breast cancer xenografts and human breast tumor tissue. We found that TRIB3 up-regulation after hypoxia, ER stress and nutrient starvation also holds true for breast cancer cells and is HIF-1 independent and UPR dependent. TRIB3 up-regulation after hypoxia was found to be most pronounced at physiological intermediate hypoxia, in contrast to most UPR-induced proteins. Finally, knockdown of TRIB3 revealed an effect on hypoxia response of breast cancer cells. In combination these results indicate that TRIB3 might be associated with tumor cell survival under prolonged intermediate hypoxic stress. This hypothesis warrants further experiments in other (breast cancer) cell lines applying additional knockdown, possibly with rescue, and/or overexpression techniques. Furthermore, the involvement of TRIB3 in multiple important signaling pathways makes it an interesting target for cancer therapy. Further research will provide more insight into the mechanisms of action and possibilities of intervention.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MW performed laboratory experiments, data analysis and participated in the experimental design and drafting of the manuscript. JB participated in the design and coordination of the research and assisted in xenograft assays. BS contributed to the interpretation of the data and critical revised the manuscript. IDN participated in the immunohistochemical stainings and the interpretation of these data. HWML participated in the collection of the patient material and data. JAR and MAV participated in the collection of patient material containing pimonidazole. JJTMH participated in RT-qPCR experiments and interpretation of the results. KMR carried out the MCF-7 knockdown experiments and critically revised the manuscript. FCGJS participated in the design and coordination of the research and assisted in the patient material collection. PNS participated in the design and coordination of the research, performed the statistical analysis and helped to draft the manuscript. All authors read and approved the final manuscript.