Background
Despite the remarkable clinical success of TMX therapy, patients with metastatic disease develop therapeutic resistance after receiving the hormonal therapy due to various mechanisms, including amplification of Her2, increased Mitogen-activated protein kinases (MAPK) signaling, alterations of ER or cytochrome (CYP) gene expression [
1‐
3]. Notably, about 30–40% of patients who receive adjuvant TMX ultimately recur and die from their disease, presenting a huge clinical challenge [
4,
5]. Therefore, identification of drug combinations for therapy of non-responder patient’s group is of utmost importance.
TMX is a prodrug metabolized by products of CYP genes to a more active metabolite 4-hydroxy-tamoxifen (4OH-TMX) with high binding affinity to ER [
6]. Patients with CYP activity compromised either by specific single nucleotide polymorphisms (SNPs) of CYP2D6 or mutation in CYP genes have worst outcome after TMX treatment [
7], suggesting that detoxification enzymes could be potential biomarkers. Another detoxification enzyme Sulfotransferase 1A1 (SULT1A1) is a member of the sulfotransferase enzymes which can eliminate 4OH-TMX [
8]. It was shown that patients with low activity of SULT1A1 due to SNP in the
SULT1A1 gene and who received adjuvant TMX or chemotherapy display better survival [
9]. In contrast, other studies suggest that SNP conferring normal SULT1A1 activity is associated with better survival upon TMX [
10,
11]. Therefore, it appears important to resolve this controversy and to establish an association between SULT1A1 and TMX.
In this study, we have identified SULT1A1 to be upregulated in relapsed metastatic breast tumors in patients who received TMX therapy. We reasoned that SULT1A1-dependent drugs (or their metabolites) might overcome resistance to TMX. We found that the tumor suppressor effect of three anticancer compounds, RITA [
12‐
14], aminoflavone (AF; (5-amino-2-(4-amino-3-fluorophenyl)-6,8-difluoro-7-methylchromen-4-one; NSC 686288) [
15], and derivative of oncrasin-1 (ONC-1; (1- {(4-chlorophenyl)methyl}-1H-indole-3-carboxaldehyde) [
16], is dependent on the expression of SULT1A1, in line with previous reports [
17‐
19]. Recently, we have identified cancer cell-specific oxidative-dependent inhibition of the transcription of several oncogenes by RITA, AF, and ONC-1 [
20]. Moreover, we identified a common target for these compounds, TrxR1, and demonstrated that targeting TrxR1 by the three compounds is SULT1A1-dependent. We found that RITA and AF can overcome TMX resistance. Our findings can open the way to new treatment modalities for relapsed breast cancer patients.
Methods
Cell lines
MCF7 (ATCC), MCF7 TMX
R spontaneously obtained in our lab and tamoxifen-resistant MCF7/LCC2 (kindly provided by Nils Brünner, University of Copenhagen) were cultured in phenol-red-free DMEM supplemented with 10% FBS (Hyclone), 2 mM
l-glutamine, 100 U/ml of penicillin, and 100 mg/ml of streptomycin (Sigma-Aldrich). The TMX-resistant MCF7/LCC2 cells were selected stepwise against increasing concentrations of 4-OH-TMX. Selection began with 1 nM and increased by half a decade after three consecutive passages and the final concentration used was 1 μM 4-OH-TMX), and maintained in 1 μM 4-OH-TMX [
21]. HCT116 (ATCC), A375 (ATCC), H1299 (ATCC), GP5d (ATCC), A431 (ATCC), and MDAMB-231 (ATCC) were grown in DMEM supplemented with 10% FBS, and antibiotics. Primary patient-derived KADA line (kindly provided by Rolf Kiessling, Karolinska hospital) was cultured in IMDM. SJSA-1 (ATCC), U2OS (ATCC), and SKMEL28 (kindly provided by Lars-Gunnar Larsson, Karolinska Institutet) were cultured in RPMI 1640 with 10% FBS and antibiotics. The pretreatment (96 h) of 50 nM sodium selenite (Sigma-Aldrich) was performed in the cell lines only when TrxR1 activity measurement was performed. CRISPR/Cas9-mediated SULT1A1 deletion was performed in stable Cas9-expressing MCF7 and HCT116 cells using gRNAs targeting exon 4 - ATCTGGGCCTTGCCCGACGA and exon 7 - AATTGAGGGCCCGGGACGGT. Cas9 expressing plasmid was provided by Vera Grinkevich, Welcome Trust Sanger Institute, Cambridge, UK.
A375 and SJSA-1 cells, stably expressing SULT1A1 cDNA (OriGene, #RC201601L1), were generated by lentivirus transduction using standard procedure [
22].
Clinical material
Between November 2017 and May 2018, fresh breast cancer specimens from 11 patients were collected at the Karolinska University Hospital and Stockholm South General Hospital. Experimental procedures and protocols were approved by the regional ethics review board (Etikprövningsnämnden) in Stockholm, Sweden, with reference numbers 2016/957-31 and 2017/742-32. The material was obtained according to Stockholm Medical Biobank approval number Bbk1730.
Compounds
RITA (NSC652287) and aminoflavone (NSC686288) were obtained from the National Cancer Institute (NCI), oncrasin-1 was from Santa Cruz Biotechnology, and 4OH-TMX and resveratrol were purchased from Sigma-Aldrich. We have tested different concentrations of 4OH-TMX (from 10 nM to 1 μM) in ex vivo samples and from 100 nM to 6 μM range of concentrations in MCF7 cells in a short-term viability experiment. The concentration of 4OH-TMX which we used is consistent with several reports in which 4OH-TMX was used in a short-term experiment [
23‐
25]. The TMX-resistant MCF7-LCC2 cells were treated with ≥1 μM 4OH-TMX. The compound concentrations and durations of treatment are mentioned in the figure legends.
3D ex vivo model
Our 3D ex vivo model is based on the study of Vaira et al. [
26], in which they established an organotypic culture model that maintains original tumor microenvironment in the presence of 20% inactivated FBS. We further modified this protocol by collecting the breast cancer clinical samples with superficial scraping, instead of tumor tissue section, which allows us to culture all the components from parental tumors maintaining tumor heterogeneity and epithelial-stromal interactions [
27].
Primary cancer cells were collected by superficial scrapings from surgically resected breast tumors [
27]. The cell smears were immediately processed by lysis of red blood cells, followed by trypsinization (Thermo Fisher Scientific, MA, USA) and filtration (Miltenyi Biotec, Bergisch Gladbach, Germany) into single-cell suspensions and three time washing with PBS. The last cell pellet was re-suspended with selective DMEM F/12 medium supplied with 20% FBS and Antibiotic-Antimycotic (all from Thermo Fisher Scientific, MA, USA), then seeded at the density of 1500 cells in 60 μL medium per well into 96-well plate (Sigma Aldrich, MO, USA) using the MultiDrop Combi dispenser (Thermo Fisher Scientific, MA, USA). Cells were then divided into either vehicle group or tamoxifen treatment group, where DMSO or 1 μM of 4OH-TMX- (both from Sigma Aldrich, MO, USA) were supplied to the culture and replenished every 48 h. RITA or equivalent volume of DMSO was added to both experiment groups at day 6. The cell viability was assessed using CellTiter-Glo 3D assay (Promega, WI, USA) according to the manufacturer’s instruction, and reading luminescence by a Tecan spark 10M microplate reader (Tecan, Männedorf, Switzerland) at day 9 as the experiment end point.
Cell viability and growth suppression assays
For colony formation assay, 25,000 MCF7 cells were seeded in a 12-well format, pretreated with 4OH-TMX for 6 days. After pretreatment, the cells were co-treated with indicated low dose of RITA, AF, and ONC-1 for another 6 days and stained with crystal violet (CV). For HCT116 cells, 100,000 cells were seeded in 12-well plates, treated with indicated concentration of drugs for 3 days and followed by CV staining. For short-term viability assay, 3000 cells/well were plated in a 96-well plate and treated with indicated concentration of drugs for 72 h, and cell viability was assessed using resazurin assay according to the manufacturer’s instructions. No additional selenium was added to culture medium in different cellular assays.
Immunofluorescence staining
Primary breast cancer cells were cultured in low-adherent plate in the presence or absence of 1 μM 4OH-TMX. 4OH-TMX was replenished every 48 h for 6 days, and cytospin was performed at speed of 50×
g for 10 min. Cells were then fixed at − 20 °C with ice-cold 1:1 mixed methanol and acetone (Sigma-Aldrich, MO, USA) for 20 min. Subsequently, cells were incubated in IF buffer (4% bovine serum albumin, 0.05% saponin in PBS) for 1 h and stained in IF buffer with the primary antibodies at 1:200 dilution: α-SULT1A1 ab191069 (green staining) or ab124011 (red staining), both from Abcam. DNA was detected using 4,6-diamidino-2-phenylindole (DAPI, 1 mg/ml: Sigma). Analyses were performed with an automated Olympus IX73 inverted microscope. Quantitative immunofluorescence analysis was performed using Fiji/ImageJ software (
https://imagej.net/Fiji).
Real-time quantitative PCR assay
The total RNA was extracted using Aurum™ Total RNA Mini Kit (Bio-Rad), according to the manufacturer’s protocol, followed by amplification and cDNA synthesis using MessageBOOSTER™ cDNA Synthesis Kit (Lucigen) or iScript cDNA synthesis kit (Bio-Rad) for qPCR, as described by the manufacturer. Real-time PCR was conducted with SsoAdvanced™ Universal SYBR® Green Supermix (Bio-Rad). GAPDH, RPL13A, and RPLP0 were used as housekeeping genes. The SULT1A1 primer sequences are FP- CGGCACTACCTGGGTAAGC and RP- CACCCGCATGAAGATGGGAG.
Correlation sensitivity analysis
NCI-60 Analysis Tools provided by CellMiner [
28] (
https://discover.nci.nih.gov/cellminer/) and the NCI-60 cell line panel with associated drug screens were used to calculate the drug sensitivity Pearson correlation coefficient between SULT1A1 mRNA expression and RITA, oncrasin-1, and aminoflavone. The 50% growth inhibitory concentration (GI
50) values of compounds were used for calculations.
shRNA screen
Genome-wide pooled short hairpin RNA (shRNA) screen to identify genes which confer resistance to RITA was performed in MCF-7 cells. Briefly, cells infected with lentivirus library of 27,290 shRNAs targeting 5046 human genes were selected on Puromycin, allowed to propagate for 1 week before treatment with 1 μM RITA. Sequencing of the barcodes from survived cells gave 20 million individual scores of shRNAs. To identify shRNAs whose abundance was significantly different between control (DMSO) and RITA-treated cells, we used the following criteria: (i) P value < 0.1, (ii) FDR < 0.3, and (iii) P values of the weighted Z-scores, P (wZ) < 0.1, which integrate the information from multiple shRNAs targeting a single gene, thus minimizing the impact of possible off-target effects.
These procedures allowed us to identify shRNAs which confer resistance to RITA. The detailed experimental procedure and the data processing are described in [
29].
RNA-seq analysis
SULT family gene expression was analyzed using RNA-seq data obtained in primary breast cancer and liver metastatic tumors from three patients treated with endocrine therapy for 5 years [
30]. Read count data was downloaded using the accession number GSE58708 and normalized with DESeq2 [
31] for clustering. Hierarchical clustering was computed and visualized using GENE-E software (
https://software.broadinstitute.org/GENE-E/index.html).
Oxidative stress measurement
Cells were treated as mentioned in figure legends and incubated 30 min with 10 μM DCF-DA in serum-free medium. Afterward, cells were trypsinized and washed with PBS, and fluorescence was analyzed by a FACSCalibur flow cytometer (BD Biosciences) using FL1-H channel.
TrxR1 enzymatic activity measurement
Total cell lysates (40 μg protein) from cells treated with RITA, AF, and ONC-1 were subjected to TrxR-dependent Trx-coupled insulin disulfide reduction assay as described before [
32].
Western blot analysis
The extraction of total cell lysates and western blot were performed according to standard procedure. The antibodies used for immunoblotting were as follows: SULT1A1 (ab124011 and ab191069, Abcam), p53 (sc-126, Santa Cruz Biotechnology), PARP (#95423, Cell Signaling), and γH2AX (#07-164, Millipore). Anti-β-actin monoclonal antibody (Millipore) was used as loading control. The horseradish peroxidase-coupled secondary antibodies (Jackson ImmunoResearch) and SuperSignal™ West Dura Extended Duration Substrate detection system (Thermo Fisher Scientific), and images were visualized using ChemiDoc Imaging System (Bio-Rad).
Statistical analysis
For all experiments, statistical analyses were performed using GraphPad Prism 5. Results are given as mean ± s.d. To evaluate statistical significance, Student’s t test (unpaired) or one-way ANOVA with Bonferroni’s multiple comparison test were performed and mentioned in the figure legends. P values ≤0.05 were considered statistically significant.
Discussion
Use of tamoxifen as the principal therapy for postmenopausal women diagnosed with estrogen receptor (ER)-positive breast cancer has contributed significantly to the survival of patients. However, around 30% of women experience relapse within the first 5 years of tamoxifen therapy [
4,
5]. Therefore, novel combination therapies as well as biomarkers are needed to help patients. Since monotherapy cannot cure cancer, it is imperative to find those drug combinations which would help in fighting TMX resistance and cancer recurrence. Here, we identified a potential biomarker of TMX resistance, SULT1A1, which is upregulated after TMX treatment in metastatic breast cancers in relapsed patients, in treated breast cancer samples ex vivo and in breast cancer cell lines in vitro. Moreover, we found that increased SULT1A1 level renders selective vulnerability to three anticancer compounds, whose mechanism of action we found to be SULT1A1-dependent. We validated this finding in breast cancer patient samples ex vivo. Thus, we identified a novel therapeutic combination, which could be used in clinic to overcome the problem of TMX resistance and improve the survival of patients.
SULT1A1 is a phase II enzyme, a member of the sulfotransferase family, which has the capability to sulfate xenobiotics and endogenous chemicals as a metabolic step towards neutralizing them [
42]. Sulfotransferase 1A1 performs the detoxification of 4-OH-TMX [
8]. Our finding of the association of increased resistance to tamoxifen with high level of SULT1A1 and that TMX-resistant cells are rendered sensitive to 4OH-TMX treatment after SULT1A1 KO is in agreement with the notion that detoxification of 4-OH-TMX by SULT1A1 is neutralizing its activity.
Previous clinical studies addressed the role of polymorphism in SULT1A1 in the survival of breast cancer patients. However, these studies have produced conflicting results, which indicate either a better survival upon TMX therapy of breast cancer patients carrying SULT1A1 allele encoding enzyme with low activity [
9] or, in contrast, worse survival of such patients [
10,
43], or no impact whatsoever [
35]. It should be noted that the outcome of TMX therapy is not solely determined by a single detoxification enzyme but a combination of several factors, including by uridine diphosphate glucuronosyltransferases (UGTs), polymorphisms associated with the CYP genes, especially the CYP2D6, and others [
44]. This complexity of TMX metabolism may explain the discrepancies observed in different clinical studies mentioned above. However, there has been no clinical study so far addressing the role of the level of SULT1A1 expression in relation to patient survival. More systematic clinical studies, including more patients, and considering several aspects related to TMX metabolizing enzymes, will shed light on this issue.
While we found an increased level of SULT1A1 after treatment with tamoxifen, the exact mechanism of this phenomenon is unclear. 4-OH-TMX might mediate the induction of SULT1A1 expression on mRNA level, as it has been shown earlier [
35]. It is possible that TMX enhances the recruitment of the transcription factor SP1, required for the activation of SULT1A1 expression [
45], to its promoter, as it has been shown for genes encoding p21 and p27 [
46,
47]. Since we demonstrate here that high expression of SULT1A1 confer resistance to 4OH-TMX, it is also possible that the elevated expression of SULT1A1 in cell population after treatment with 4OH-TMX was due to the preferable survival of cells with intrinsic high SULT1A1 level.
Interestingly, SULT1A1 has also been shown to bio-transform pro-drugs into their active form [
17‐
19,
48]. Our experiments, using a number of approaches and cell models, including genome-wide shRNA screen, CRISPR/Cas9-mediated SULT1A1 deletion versus its overexpression, and analysis of more than a dozen of cancer cell lines, provide a strong evidence of SULT1A1 dependence of growth suppression by three anticancer compounds, RITA, AF, and ONC-1. While this phenomenon was indicated in previous studies [
17‐
19], the reason for it was not clear.
RITA, AF, and ONC-1 cause tumor regression in several cancer cell models and mouse tumor models without noticeable adverse effects, making them promising candidates for anticancer treatment [
13,
49‐
52]. The mechanism of action of these compounds was addressed in several studies, including ours, but their most relevant target(s) remain a subject of debate [
12,
17,
18]. Tumor suppression by RITA, AF, and ONC-1 has been linked to the induction of ROS [
15,
32,
41,
53,
54]. Importantly, in this study, we identified a common target for all three compounds, which is affected by them in a SULT1A1-dependent manner. We found that RITA, AF, and ONC-1 inhibit TrxR1 and induce oxidative stress in cancer cells in a SULT1A1-dependent manner.
Increased oxidative stress, associated with enhanced proliferation and altered metabolism, is one of the vulnerabilities of cancer cells which could be exploited by targeting ROS-neutralizing enzymes. NADPH-dependent selenoprotein thioredoxin reductase (TrxR) is often overexpressed in cancer [
39,
55], which associates with poor survival of patients with different types of cancer, including breast, lung, pancreatic, prostate, and head and neck cancers [
39]. This makes TrxR1 a promising target for the development of anticancer drugs, especially considering that in part due to the selenoprotein nature of this enzyme it is particularly prone to inhibition by electrophilic compounds [
40]. From the therapeutic perspective, it is interesting to note an opposite effect of TrxR1 inhibition on normal and cancer cells: survival of normal cells or even strengthening them against oxidative stress versus death of cancer cells [
39]. In our recent study, we have demonstrated that the three compounds induce exaggerated oxidative stress selectively in cancer cells, which is a prerequisite for targeting the mRNA transcription machinery. Thus, these compounds attack cancer vulnerability—transcriptional addiction. Inhibition of transcription machinery results in preferential inhibition of major oncogenic pathways, thus killing cancer cells [
20].
In summary, the high level of SULT1A1, while conferring resistance to TMX, provides a selective vulnerability to SULT1A1-dependent compounds. Our ex vivo experiments using breast cancer samples demonstrated the efficiency of the combination of TMX with SULT1A1-dependent compounds, both in TMX-sensitive and TMX-resistant patients. Thus, our study provides a strong mechanistic support for novel combinatorial treatment of relapsed patients with SULT1A1-activated compounds, such as RITA and AF. When patients recur on TMX, they are usually treated with aromatase inhibitors (AI), often the main alternative for high-risk patients. However, premenopausal women cannot be treated with AI since the endogenous estrogen production is too high. We speculate that combining TMX with SULT1A1-dependent compounds could provide a therapeutic option for young patients. Our results might pave a way to a strategy to circumvent TMX resistance and suggest that this approach could be developed further for the benefit of patients.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.