Introduction
The introduction of aromatase inhibitors (AIs), drugs that block the enzyme that synthesizes estrogens, to treat women with estrogen receptor (ER)-positive breast cancer marked a significant advance in the treatment of this disease, with a reduction in recurrence of approximately 50% [
1]. However, AI therapy can also result in drug-induced musculoskeletal pain as a major side effect that can result in the termination of AI therapy [
2]. For example, in the Arimidex, Tamoxifen, Alone or in Combination breast cancer clinical trial, up to 28% of women treated with AIs developed musculoskeletal pain, and approximately 10% discontinued therapy because of this adverse drug reaction [
3]. Changes in circulating estrogen levels in women have long been associated with musculoskeletal symptoms. Arthritis of the menopause was described by Cecil and Archer over 85 years ago [
4], and joint pain was a major complaint among participants in the Women's Health Initiative study after the withdrawal of estrogen therapy [
5].
We recently performed a case-control genome-wide association study (GWAS) of participants in the NCIC-CTG MA.27 clinical trial of AI adjuvant therapy in postmenopausal women with ER-positive breast cancer in an attempt to identify biomarkers and define mechanisms responsible for musculoskeletal pain associated with pharmacologic blockade of estrogen synthesis. That GWAS identified a SNP signal on chromosome 14 that mapped near the 3' end of the T-cell leukemia 1A (
TCL1A) gene [
6], and the SNP with the lowest
P value (rs11849538,
P = 6.67 × 10
-7) created a functional estrogen response element (ERE). We also observed that TCL1A expression was induced by estrogen exposure, and that it was significantly elevated in lymphoblastoid cell lines (LCLs) that carried variant sequences for the chromosome-14 SNPs; that is, in cell lines with DNA encoding the SNP-related ERE. The present study was performed to pursue possible mechanisms by which these SNPs might be associated with musculoskeletal pain in response to reduced estrogen levels during AI therapy, mechanisms that might also have broader implications for the role of estrogens in musculoskeletal pain [
6].
TCL1A is a member of a TCL1 family of proteins that includes TCL1A, TCL1B and TCL6 [
7]. This protein is expressed in activated T lymphocytes and B lymphocytes as well as thymocytes, can interact with Akt and can enhance Akt kinase activity [
8‐
11], but little is otherwise known about TCL1A function. In follow-up of our original GWAS, we reported that TCL1A expression was estrogen dependent and was correlated with expression of the cytokine receptor IL-17RA [
6]. In the present study, we set out to determine whether TCL1A expression - expression that is estrogen dependent but is altered by the SNPs that were associated with AI-induced musculoskeletal pain - might also be associated with variation in the expression of other cytokines and/or cytokine receptors. Many of the experiments described subsequently were performed with U2OS cells because those cells express TCL1A and have been stably transfected with ERα, and with a powerful genomic data-rich LCL model system that includes cell lines with known
TCL1A SNP genotypes. The availability of these LCLs, also stably transfected with ERα, made it possible for us to link the SNPs that we observed during the clinical GWAS for AI-induced musculoskeletal pain with variation in the expression of a series of cytokine and cytokine receptor genes that have been implicated in arthritis. Specifically, we observed, as described in detail subsequently, that estrogen-dependent, SNP-modulated expression of TCL1A is not only associated with variation in IL-17RA expression but also with the expression of IL-17, IL-1R2, IL-12 and IL-12RB2 as well as variation in NF-κB transcriptional activity.
Materials and methods
Human Variation Panel lymphoblastoid cell lines
The Human Variation Panel of LCLs from 100 healthy European-American, 100 African-American and 100 Han Chinese-American subjects was obtained from the Coriell Institute (Camden, NJ, USA). These cell lines were generated from blood samples obtained by the National Institute of General Medical Sciences. We genotyped DNA from these cell lines for genome-wide SNPs using the Illumina 550K and 510S SNP BeadChip (Illumina, San Diego, CA, USA). The Coriell Institute also genotyped DNA from the same cell lines using the Affymetrix SNP Array 6.0 (Affymetrix, Santa Clara, CA, USA) for a total of ~1.3 million unique SNPs per cell line [
12]. We also generated basal Affymetrix U133 2.0 Plus GeneChip expression array data for all of the cell lines. This LCL genomic model system has been described in detail elsewhere [
12]. The microarray data and SNP data for these LCLs have been deposited in the NCBI Gene Expression Omnibus [
13] under SuerSeries [GEO:GSE24277].
Lymphoblastoid cell line transfection and culture
Three of the European-American Human Variation Panel LCLs with variant genotypes for the chromosome-14 SNPs rs7158782, rs7159713, rs2369049 and rs11849538, and three with wild type (WT) sequences were stably transfected with a pcDNA4.1-ERα construct provided by Dr Thomas Spelsberg (Mayo Clinic, Rochester, MN, USA). Genotypes for the rs11849538 SNP in the cell lines were confirmed by performing the PCR with genomic DNA as the template using the following primers: forward, 5'-GTGACAAGAAAGCTGTGGACTAGAGACACA-3'; and reverse, 5'-TTGGAGGCATACGTTGAGAACCATTGGAGTAA-3'. Genotypes for the other three chromosome-14 SNPs in these cells had been determined during GWAS genotyping. These six stably transfected LCLs were cultured with or without estradiol (E2), as described previously [
6]. In some experiments, the ERα inhibitor ICI-182,780 (Tocris, Baldwin, MO, USA) was added to culture medium containing 0.01 nM E2 for an additional 24 hours after the initial 24 hours at final E2 concentrations of 10
-10, 10
-9, 10
-8 and 10
-7 μM, and total RNA was isolated from the cells with the RNeasy mini kit (Qiagen, Valencia, CA, USA). Two hundred nanograms of this RNA was then used to perform quantitative RT-PCR with appropriate primers. Expression levels were normalized on the basis of ERα expression in each cell line.
NF-κB transcriptional activity
U2OS-ERα cells [
14] were seeded in triplicate in 12-well cell culture plates, with 10
5 cells/well. After 24 hours, the cells were transfected using Lipofectamine 2000 (Invitrogen, Grand Island, NY, USA) with two plasmids; one encoding an NF-κB promoter-luciferase construct (SABioscience, Foster City, CA, USA), and the other encoding a Renilla luciferase vector (Promega, Madison, WI, USA). Co-transfection with the Renilla construct made it possible to correct for possible variation in transfection efficiency. Transfected cells were incubated overnight and were then grown for 24 hours in DMEM with 5% charcoal-stripped FBS, followed by incubation in DMEM without FBS for 24 hours, either with ethanol or with 0.1 nM E2 dissolved in ethanol. The cells were then harvested and analyzed for luciferase activity (Promega).
The six LCLs with differing genotypes for the four chromosome-14 SNPs that had been stably transfected with ERα were also transiently transfected with 2 μg NF-κB reporter plasmid and 500 ng Renilla luciferase or 2 μg empty reporter plasmid plus 500 ng Renilla luciferase construct. Specifically, 2 × 106 cells were suspended in Cell Line Nucleofector Kit V solution (Lonza, Cologne, Germany) with 2 μg purified NF-κB reporter plasmid and 500 ng Renilla reporter plasmid, and were electroporated with the T-030 program using the Amaxa Nucleofector II (Amaxa Biosystems, Gaithersburg, MD, USA). Cells from six electroporation procedures per LCL were pooled to obtain 1.2 × 107 cells. Electroporated cells were then plated in RPMI 1640 medium supplemented with 15% FBS, were allowed to recover from electroporation for 24 hours, and were cultured for 24 hours in RPMI 1640 media containing 5% (vol/vol) charcoal-stripped FBS, followed by incubation for 24 hours in the same media containing increasing concentrations of E2. At that time, cells treated with 0.01 nM E2 were exposed to increasing concentrations of ICI-182,780 for a final 24 hours. Luciferase assays were performed, and those values were corrected for possible variation in transfection efficiency by use of the Renilla luciferase values.
Western blot analysis
Proteins were isolated from U2OS-ERα cells after lysis with CelLytic M Cell Lysis buffer (Sigma-Aldrich, St. Louis, MO, USA) and were subjected to electrophoresis on 15% SDS-PAGE gels, followed by transfer to polyvinylidene fluoride membranes. The polyvinylidene fluoride membranes were probed with appropriate antibodies, and protein bands were visualized using enhanced chemiluminescence (Thermo Scientific, Rockford, IL, USA).
Discussion
The use of adjuvant AI therapy to treat ER-positive breast cancer patients represents a major advance in the treatment of that disease [
26]. A recent report demonstrating that an AI (exemestane) was highly efficacious in preventing breast cancer highlights the importance of understanding mechanisms responsible for the musculoskeletal side effects of AI therapy [
27]. These side effects limit patient adherence to therapy with this important class of drugs for either breast cancer treatment or prevention. Equally important, however, may be the fact that AI therapy represents a medically-indicated form of pharmacologic estrogen deprivation that might provide a window on mechanisms by which estrogen withdrawal can cause musculoskeletal symptoms. Our GWAS performed with DNA samples from patients enrolled in the MA.27 clinical trial identified four SNPs on chromosome 14 near the 3' end of
TCL1A that were associated with increased risk for musculoskeletal adverse events in women receiving adjuvant AI therapy for the treatment of ER-positive breast cancer [
6]. In the present study, we pursued those observations and linked E2-dependent induction of TCL1A to the expression of a series of cytokines and cytokine receptors, including IL-17RA, IL-17, IL-12RB2, IL-12 and IL-1R2, with SNP-dependent variation in this induction (Figure
1). Obviously, results obtained for other cell lines might identify additional or different cytokines/cytokine receptors, but the purpose of this study was to take the first step in the elucidation of a novel pathway for estrogen-dependent,
TCL1A SNP-dependent regulation of cytokine and cytokine receptor expression. Our results also demonstrated a relationship between estrogen and SNP-dependent variation in TCL1A expression and NF-κB transcriptional activity (Figure
6).
TCL1A expression is associated with CD4
+ and CD8
+ T-cell activation through the phosphoinositide 3-kinase/Akt signaling pathway [
28], and TCL1A enhances Akt kinase activity [
29]. However, there have been no previous reports of the regulation of TCL1A by estrogens or of an association of TCL1A expression with cytokine or cytokine receptor expression. A recent study did report that another member of the TCL1 gene family,
TCL1B, is E2-inducible because of an ERE located near the 3' end of that gene [
30]. The SNP with the lowest
P value in our GWAS, rs11849538, created an ERE near the 3' terminus of
TCL1A, and cell lines that carried the variant SNP genotype displayed increased TCL1A expression after estrogen exposure (see Figures
1 and
6A).
In the present study, we found that increased expression of TCL1A upregulated IL-17RA expression and downregulated the expression of IL-17, IL-12, IL-12RB2 and IL-1R2 (Figure
1). IL-17 has been reported not only to drive the T-helper type 17 immune pathway, but also to regulate the T-helper type 1 pathway by decreasing IL-12 and IL-12RB2 subunit expression, especially in patients with rheumatoid arthritis [
31]. The E2-dependent regulation of cytokine and cytokine receptor expression that is mediated by TCL1A might help explain the association of TCL1A with musculoskeletal symptoms in patients treated with AIs. TCL1A can also influence NF-κB transcriptional activity (Figure
6), suggesting that, after estrogen withdrawal, increased NF-κB activity might contribute to AI-induced musculoskeletal pain. We also showed that cell lines containing variant chromosome-14 SNP genotypes had significantly elevated TCL1A expression after exposure to increasing concentrations of estrogen (Figures
1 and
6A). After ER blockade with ICI-182,780, however, TCL1A expression dropped precipitously in LCLs with variant SNPs, while it was elevated in cells with the WT SNP genotypes (Figure
6A). Conversely, NF-κB transcriptional activity increased after ER blockade in cells carrying variant SNP genotypes (Figure
6B).
Acknowledgements
The authors thank Luanne Wussow for her assistance with the preparation of this manuscript. Funding support: NIH grants R01 GM28157, R01 CA132780, U01 HG05137, U19 GM61388 (The Pharmacogenomics Research Network), P50 CA116201, U10 CA77202, the Mayo Clinic Center for Individualized Medicine, the Biobank Japan Project funded by the Ministry of Education, Culture, Sports, Science and Technology, CCS 015469 from the Canadian Cancer Society, and the Breast Cancer Research Foundation. This study was supported, in part, by the NIH Pharmacogenomics Research Network (PGRN) - RIKEN Center for Genomic Medicine (CGM) Global Alliance. PEG is supported by the Avon Foundation, New York, USA.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
ML, LW, JNI, DJS, JRH, SNM, PEG, TM, MK, YN and RMW designed the research. ML, DJS, PEG, TM, MK, YN, NK and RMW performed the research. ML, LW, TB, DJS, SNM and RMW analyzed the data. ML, LW, TB, JNI, DJS, JRH, SNM, PEG and RMW wrote the paper. All authors read and approved the final manuscript.