Introduction
Breast cancer is the most common form of cancer among women today [
1]. The prognosis of breast cancer patients varies depending on the breast cancer subtype. Clinical breast cancer classification is based on expression of various immunohistochemical markers, with the hormone receptors being the most important. One of the worst prognosis subtypes is the triple-negative (TN) breast cancer subtype, where the malignant cells lack expression of the hormone receptors, estrogen receptor (ER) and progesterone receptor (PR), and human epidermal growth factor receptor 2 (Her2) (ER
−PR
−Her2
−). The treatment options are few for patients with TN breast cancer [
2‐
4].
Toll-like receptors (TLRs) are a family of receptors that are expressed on innate immune cells [
5]. They are part of the pattern recognition receptor (PRR) family and recognize molecular patterns from pathogens (pathogen-associated molecular patterns; PAMPs) or from endogenous stress-induced proteins (damage-associated molecular patterns; DAMPs) [
6‐
9]. Signaling via TLRs leads to activation of nuclear factor kappa B (NFκB) and a subsequent expression of pro-inflammatory genes [
10]. There are 10 different TLRs (TLR1-10) in humans, and these are divided into two subgroups depending on cellular localization; on the surface of the cell (TLR1, TLR2, TLR4, TLR5 and TLR6), or in vesicles such as endoplasmic reticulum, endosomes or lysosomes (TLR3, TLR7, TLR8 and TLR9). Lately, expression of different TLRs has been described in various malignancies, although their function is as yet unclear [
5,
11,
12].
TLR2 and TLR4 respond to the typical PAMP from Gram-negative bacteria, lipopolysaccharide (LPS). Different variants of LPS (from
Escherichia coli and
Salmonella typhimurium) induce different TLR-intracellular signals [
13]. DAMPs can also bind to and activate TLR2 or TLR4, and two endogenous ligands that are well-described are HMGB1 and S100A9 [
14‐
19]. To signal via TLR2 or TLR4, different ligands may also require the co-receptors CD14 or MD2 [
20‐
23]. All TLR ligands initiate activation of NFκB, but also mitogen-activated protein kinase (MAPK) pathways that affect protein translation and processing rather than transcription can be activated [
24]. TLR4 has previously been shown to be expressed in breast cancer [
25,
26].
The transcriptional factors ERα and NFκB are synergistically interrelated, although their exact interactions are unknown [
10,
27‐
31]. NFκB is a transcriptional factor that induces a wide array of pro-inflammatory mediators and is also related to several oncogenic processes [
32]. Both ER and NFκB have previously been shown to attenuate each other in different ways. In line with this observation, ER
− breast cancers have a stronger pro-inflammatory phenotype and microenvironment. NFκB has even been shown to downregulate ERα expression in breast cancer cells [
29], but there is no direct proof that constitutive NFκB would generate ER
− breast cancers in general. On the other hand, a recent positive synergy between ER and NFκB was published, where TNFα and estrogen were shown to remodulate the ERα-promoter landscape in an NFκB and FoxA1 dependent manner resulting in an altered gene expression pattern [
33].
In this study we performed an analysis of TLR expression patterns and function in breast cancer. Using a carefully validated TLR4-specific antibody for immunohistochemistry (IHC), we found that TLR4 protein expression was primarily present in breast cancers of ER/PR-negative phenotype. Using three cell lines of ER+ phenotype and four cell lines of the TN phenotype, we further showed that the expressed TLR4 was biologically active and hence responding to both PAMPs and DAMPs, primarily in the TN breast cancer cell lines. Finally, TLR4 protein expression correlated with a decreased survival in a cohort of 144 primary breast cancer patients. We propose that novel therapies targeting TLR4 may be of value, in particular in ER/PR-negative breast cancers.
Methods
Cell culture
The human breast cancer cell lines MCF-7, T47D, MDA-MB-231 and MDA-MB-468 were purchased from ATCC and were cultured in RPMI 1640 medium supplemented with 10 % fetal bovine serum (FBS) (Biosera, Boussens, France), 1 % sodium pyruvate, 1 % HEPES and penicillin/streptomycin (100 U/ml and 100 μg/ml respectively); CAMA-1 (also purchased from ATCC) was cultured in MEM/EBSS supplemented with 10 % FBS and penicillin/streptomycin, and SUM-149 and SUM-159 were cultured in F-12 HAM’S medium supplemented with 5 % FBS, 1 mM L-Glutamine, 1 μg/ml hydrocortisone (BD BioScience, San Diego, CA, USA) and 5 μg/ml insulin (Novo Nordisk A/S, Måløv, Denmark). The SUM-149 and SUM-159 cell lines were produced by Professor S Ethier. Media and supplements were purchased from Thermo Scientific HyClone (South Logan, UT, USA) unless otherwise stated.
Compounds and cytokine analysis
LPS was purchased from Sigma Aldrich (St Louis, MO, USA) and originated from
S. Typhimurium (LPS1) and
E. Coli (LPS2)
, respectively. All stimulations were performed for a total of 6 h except for rhS100A9 (20 h). IL-1β and HMGB1 was from R&D Systems. Recombinant human S100A9 (rhS100A9) was a gift from Active Biotech AB and a detailed description on endotoxin-free S100A9 generation and purification has been published previously [
15] and was used in the presence of calcium and zinc (Ca
2+ ≥200 μM; 10 μM ZnCl
2 [
34,
35]). Supernatants from stimulated or siRNA transfected cells were harvested and analyzed using human inflammatory cytokine cytometric bead array (CBA; BD Biosciences, San Diego, CA, USA) according to the manufacturer’s instructions or using IL-6 and IL-8 Quantikine ELISA (R&D Systems, Minneapolis, MN, USA). Annexin V-allophycocyanin (APC) and propium iodide (PI) staining was performed according to the manufacturer’s instructions (BD Biosciences). The cycloheximide (CHX) experiments (Sigma Aldrich) where performed by adding 10 μg/ml CHX, with or without 100 ng/ml LPS for 6 h.
Preparation of necrotic cell supernatant (NCS)
Confluent monolayers of MDA-MB-231 cells were harvested by trypsinization and 3.2 × 106 cells were resuspended in 2 ml serum-free RPMI-1640 medium. Necrosis was induced by performing three freeze-thaw cycles and NCS was separated from the necrotic cell pellet by centrifugation.
Tissue microarray (TMA) and immunohistochemistry
The breast cancer cohort analyzed in this study consists of 144 patients diagnosed with invasive breast cancer at Skåne University Hospital, Malmö, Sweden, between 2001 and 2002. The cohort and TMA have previously been described in detail [
36‐
38] and [
39]. TMA sections of 4 μm thickness were mounted onto glass slides and deparaffinized followed by antigen retrieval using the PT-link system (DAKO, Glostrup, Denmark) and stained in an Autostainer Plus (DAKO) with the EnVisionFlex High pH-kit (DAKO). Antibody used for TLR4 IHC was anti-TLR4 NB100-56566 at 1:250 (Novus Biologicals, Littleton, CO, USA). TLR4 expression in TMA tumor samples was estimated as cytoplasmic staining intensity (0 = negative, 1 = weak, 2 = moderate, 3 = strong intensity and 4 = very strong intensity).
Ethical considerations
Ethical permit was obtained from the regional ethical committee at Lund University (Dnr 447/07), waiving the requirement for signed informed consent. Patients were offered to opt out of research. Ethical permission for using blood from healthy blood donors was obtained from the regional ethical committee at Lund University (Dnr 2012/689).
Gene expression profile array
The publicly available database R2: microarray analysis and visualization platform [
40]; Tumor breast EXPO-351 was used for gene expression profile analysis.
Quantitative real-time PCR (RT-qPCR)
RNeasy Plus kit was used to extract total RNA according to the manufacturer’s instructions (Qiagen, Hilden, MD, USA). Random hexamers and the M-MuLV reverse transcriptase enzyme (Thermo Scientific) was used and quantitative real-time PCR (RT-qPCR) were performed in triplicates for the genes analyzed using Maxima SYBR Green/Rox (Thermo Scientific) according to the manufacturer’s instructions. RT-qPCR analysis was performed on the Mx3005P QPCR system (Agilent Technologies, Santa Clara, CA, USA) and the relative mRNA expression was normalized to
YWHAZ,
UBC and
SDHA and calculated using the comparative cycle threshold (Ct) method [
41]. For primers see Additional file
1: Table S1.
Transient transfections
siRNA transfections were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA): 2 μM of the following silencer select siRNA oligonucleotides from Ambion (Carlsbad, CA, USA) were used; Silencer Select Negative Control #2: 4390846, siTLR2 #1: s168, siTLR2 #2: s170, siTLR4 #1: s14194, siTLR4 #2: s14195. Analyses were performed 48 h and 72 h post transfection. For luciferase assays, breast cancer cells were co-transfected using Lipofectamine 2000 with a total of 0.6 μg pNFκB-luciferase (BD Biosciences) and 0.06 μg TK-renilla-luciferase (Promega, Madison, WI, USA) plasmids and was subsequently analyzed using Dual-Luciferase Reporter System (Promega). For TLR4 transfections breast cancer cells were transfected using Lipofectamine 2000 with a total of 1.0 μg pDUO-MD2/hTLR4 or pUNOI-hTLR4-GFP (Invivogen, San Diego, CA, USA) per 24 wells for 72 h or 48 h, respectively, and was subsequently analyzed using immunofluorescence (×40 magnification) or ELISA as described in the figure legends.
Statistical analyses
Graph Pad Prism software was used to perform analysis of variance (ANOVA) or Students t test for the in vitro experiments as indicated. Spearman's Rho and the chi-square (χ
2) test was used for correlation analysis and Kaplan-Meier analysis with the log-rank test was used to illustrate differences in survival. All statistical tests were two sided and P ≤0.05 was considered significant. Calculations were performed with IBM SPSS Statistics version 19.0 (SPSS Inc).
Discussion
Breast cancers with an ER-negative phenotype have previously been shown to promote a strong pro-inflammatory microenvironment [
44]. Furthermore, historically there is a negative relationship between ERα and NFκB that has previously been described in depth [
10,
27‐
30]. Despite the fact that ER signaling can inhibit NFκB activity and vice versa, there is no evidence that the development of ER-negative breast tumors are caused by constitutive NFκB activity. Rather, it may be a result of the typical molecular gene landscapes found in luminal A compared to basal breast cancers, respectively. A link between PRR, e.g., TLR-induced activation of NFκB in breast cancer and its relation to expression of ER, has not been described. Both IL-6 and IL-8 can be highly expressed in TN breast cancers and this has partly been attributed to constitutively active NFκB [
44]. In order to investigate whether TLRs, which are known to induce strong activation of NFκB, are expressed primarily in TN breast cancers and if this might affect the expression of pro-inflammatory genes in the same, we investigated the functional role of TLRs and co-receptors in breast cancer.
In immune cells, TLR expression is generally inhibited by prolonged activation of NFκB [
45]. In contrast, our findings show that TLRs (TLR2, TLR 3, TLR 4) are preferentially expressed in TN breast cancer cell lines with constitutive NFκB activity, suggesting that the TLRs may be responsible for the NFκB activation pathway rather than induced by the same. Although introduction of a functional MD2/TLR4 complex in an ER
+ cell line has been shown to induce expression of pro-inflammatory cytokines, silencing of TLR4 in TN cells only caused a slight decrease in pro-inflammatory mediator release, indicating that the constitutive NFκB activation seen in TN cells in general is caused by another mechanism [
44]. Apart from MDA-MB-468, the TN breast cancer cells were also demonstrated to express the co-receptors CD14 and MD2 meaning that they harbor the necessary proteins for a functional TLR4 signal to occur [
20‐
22]. The exception, MDA-MB-468, only expressed CD14 and in line with this showed no biological TLR function. In the patient cohort we found correlation between TLR4 expression and ER/PR-negative tumors, but not TN tumors. This strengthens the interrelationship between TLR4, ER and NFκB activity, as expression of HER2 was not correlated in the TLR4-expressing primary tumors. We did not perform our in vitro analyses on any Her2
+ breast cancer cell line. Interestingly, the typical membrane staining seen in immune cells was not as obvious in the malignant cells, indicating that different regulation of TLR expression and signaling could be possible in cancers. This was previously described in neuroblastoma cells [
46] and is also supported by our finding that a GFP-tagged hTLR4 primarily showed a vesicular cytoplasmic localization in breast cancer cells. Furthermore and supporting this observation, it was recently reported that the TLR4-specific DAMP, S100A9, needs to be internalized to be able to signal via TLR4 [
15]. Indeed, scoring of membrane TLR4 expression in breast cancer lesions did not reveal as much as that of cytoplasmic staining, and both TLR2 and TLR4 have been reported to be expressed intracellularly as well [
47].
The DAMP, HMGB1, has previously been shown to signal via TLR4 in myeloid cells [
6,
19]. Although we have also previously shown this in primary myeloid cells [
48], we did not see an effect of HMGB1 on breast cancer cells in vitro. This could be due to different culture conditions, or to receptor expression patterns in myeloid as compared to cancer cells which might also reflect the fact that different sources of LPS generate different signals in the different cell lines in this study. Instead, we could show that the DAMP, S100A9, also induced pro-inflammatory proteins in breast cancer cells expressing TLR4.
It has previously been shown that NFκB [
29] and targets (IL-6) [
49] can downregulate ERα. We also investigated whether overexpression of TLR4 would affect ERα expression per se in ER
+ MCF-7 cells. We did see a slight although non-significant decrease of ERα after 72 h (data not shown), a finding that is probably explained by the significantly increased levels of IL-6 we observed in these experiments (Fig.
3h). In spite of this, we suggest that the ER/PR-negative breast cancer subtype probably is not caused by expression of TLRs and their downstream mediators, but rather further affected by them. Perhaps the expression of TLRs is even affected by the ERα/FoxA1/GATA3 network [
50]. We show that both PAMPs and DAMPs induced release of pro-inflammatory mediators in ER/PR-negative breast cancer cells in vitro, a process that was regulated both at the transcriptional and post-transcriptional level. This means that although ER-negative breast cancer cells express high endogenous levels of pro-inflammatory mediators, a functional TLR4 is still likely to enhance their phenotype and surrounding inflammatory microenvironment, and this is also reflected by the decreased recurrence-free survival seen in the patients with tumors expressing TLR4 at high levels. In support of this, previous studies have shown that TLR4 expression promotes metastasis in a breast cancer model, an effect that was even enhanced by Paclitaxel [
25,
26].
Acknowledgements
The authors thank Ms Elise Nilsson for professional technical skills in preparation of TMAs and IHC. This work was supported by grants from the Swedish Research Council, The Swedish Cancer Society, Kocks Foundation, Österlunds Foundation, Gunnar Nilsson Cancer Foundation, MAS Cancer Foundation, and Åke Wibergs Foundation.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MM performed the majority of experiments and analyzed data. RA and CB performed experiments and analyzed data. LHS was involved in revising the manuscript critically for important intellectual content, contributed to analysis and interpretation of data and also was responsible for linguistic correction. SPE produced and provided the TN breast cancer cell lines SUM149 and SUM159, contributed to analysis and interpretation of data and was involved in revising the manuscript critically for important intellectual content. As a clinical pathologist MEJ verified the IHC staining, and helped and mentored RA in scoring of the IHC. KJ was responsible for the breast cancer clinical samples and verified the IHC staining and TMA scoring. KL designed the experiments, wrote the manuscript and interpreted and analyzed the data. All authors read and approved the manuscript and were involved in revising the manuscript critically for important intellectual content.