Background
Breast cancer is one of the most common malignant tumors worldwide [
1,
2]. Although surgical treatments are curative for some early stage cases, adjuvant systemic therapies are important for improving survival of patients especially those who with advanced stage diseases [
3]. Doxorubicin, a topoisomerase II inhibitor, is a backbone drug in most chemotherapeutic regimens [
4]. However, intrinsic or acquired drug resistance to doxorubicin limited the efficacy of doxorubicin-based treatments [
5,
6]. Mechanisms mediating drug resistance to doxorubicin still need to be elucidated.
The human anterior gradient 2 (AGR2), a member of protein disulphide isomerases (PDIs) family, regulates protein folding in endoplasmic reticulum and normal mammary gland development [
7,
8]. Over-expression of AGR2 is involved in pathogenesis of breast cancer including growth, drug resistance and metastasis of tumors, which is associated with poor prognosis [
9]. Intrinsic or acquired over-expression of AGR2 was shown to mediate resistance to hormone therapies in estrogen receptor (ER)-positive breast cancers [
10]. AGR2 was also implied to mediate doxorubicin resistance in breast cancer cells [
11]. These findings suggest that AGR2 is an important mediator of drug resistance in breast cancer.
AGR2 was shown to be a target of ER, which regulates expression of AGR2 in both normal mammary gland and breast cancer [
12,
13]. However, over-expression of AGR2 is not restricted to ER-positive breast cancer. High AGR2 expression could be observed in ER-negative breast cancers, while some ER-positive cases showed low levels of AGR2 suggesting that mechanisms other than ER might control expression of AGR2 in breast cancer [
10].
MicroRNAs (miRNAs) are single strand non-coding RNAs which regulate expression of genes at post-transcriptional level through binding 3′-untranslated region (3′-UTR) of mRNA. Some reports had shown that decreased levels of miRNAs led to over-expression of specific oncogenes promoting pathogenesis of cancers [
14,
15]. Aberrant levels of miRNAs were also recognized as predictive factors of drug resistance in breast cancer [
16]. Based on the important roles of AGR2 and miRNAs in breast cancer, we interrogated how miRNAs regulate expression of AGR2 in breast cancer cells. In this study, we found AGR2 was up-regulated in doxorubicin-resistant breast cancer cells. miR-135b-5p negatively regulates expression of
AGR2 which increased sensitivity to doxorubicin in breast cancer cells both in vitro and in vivo. Our finding is indicative for an important role of miR-135b-5p/AGR2 pathway in regulating doxorubicin-sensitivity of breast cancer cells.
Methods
Clinical breast cancer specimens
Twenty-eight breast cancer samples were collected at the Affiliated Hospital of Xuzhou Medical University between October 2017 and April 2018. Subject and disease related variables are shown in Table
1. All the patients have not being treated before resection.
Table 1
Clinical and pathological information of patients
1 | 1,595,476 | 35 | 30 | 0 | 3 | 2A | N | N | N | Triple negative |
2 | 1,597,702 | 44 | 25 | 0 | 3 | 2A | N | N | N | Triple negative |
3 | 1,613,380 | 39 | 25 | 0 | 3 | 2A | N | N | N | Triple negative |
4 | 1,561,558 | 47 | 30 | 0 | 2 | 2A | N | N | N | Triple negative |
5 | 1,593,213 | 71 | 30 | 0 | 1 | 2A | N | N | N | Triple negative |
6 | 1,558,369 | 67 | 45 | 2 | 3 | 2B | N | N | P | Her2 overexpressing |
7 | 1,562,676 | 46 | 22 | 0 | 3 | 2A | N | N | P | Her2 overexpressing |
8 | 1,568,640 | 72 | 25 | 3 | 3 | 2B | N | N | P | Her2 overexpressing |
9 | 1,538,312 | 73 | 15 | 0 | 2 | 1A | P | P | N | Luminal A |
10 | 1,538,856 | 59 | 40 | 3 | 3 | 2B | P | P | N | Luminal A |
11 | 1,540,806 | 54 | 30 | 1 | 2 | 2B | P | P | N | Luminal A |
12 | 1,557,688 | 61 | 15 | 0 | 2 | 1A | P | P | N | Luminal A |
13 | 1,563,656 | 43 | 30 | 1 | 3 | 2B | P | P | N | Luminal A |
14 | 1,594,881 | 44 | 20 | 0 | 2 | 1A | P | P | N | Luminal A |
15 | 1,558,066 | 49 | 15 | 0 | 2 | 1A | P | P | N | Luminal A |
16 | 1,547,013 | 62 | 15 | 0 | 1 | 1A | P | P | N | Luminal A |
17 | 1,551,621 | 49 | 20 | 0 | 2 | 2A | P | P | N | Luminal A |
18 | 1,554,600 | 74 | 20 | 2 | 3 | 2B | P | P | N | Luminal A |
19 | 1,540,752 | 48 | 20 | 1 | 2 | 2B | P | P | P | Luminal B |
20 | 1,545,637 | 44 | 50 | 13 | 3 | 3C | P | P | P | Luminal B |
21 | 1,589,813 | 35 | 18 | 1 | 2 | 2A | P | P | N | Luminal A |
22 | 1,595,773 | 64 | 27 | 0 | 1 | 2A | P | P | N | Luminal A |
23 | 1,595,546 | 47 | 30 | 13 | 3 | 4 | P | P | P | Luminal B |
24 | 1,596,351 | 45 | 63 | 0 | 3 | 2B | P | P | P | Luminal B |
25 | 1,608,794 | 39 | 25 | 0 | 3 | 2A | P | P | N | Luminal A |
26 | 1,609,339 | 52 | 30 | 12 | 3 | 3C | P | P | N | Luminal A |
27 | 1,612,095 | 67 | 20 | 3 | 2 | 2B | P | P | N | Luminal A |
28 | 1,595,081 | 39 | 30 | 22 | 3 | 3C | P | P | N | Luminal A |
Mice
BALB/c Nude mice were purchased from Vital River (Charles River, Beijing, China). Mice were bred in a special pathogen free room.
Cell culture
MCF-7 cells (ATCC HTB-22) were cultured in DMEM medium (Thermo Fisher Scientific, Waltham, MA, USA) supplied with 10% FBS (Biowest, Nuaillé, France), penicillin and streptomycin. MDA-MB-231 (ATCC HTB-26) cells were cultured in Leibovitz’s L-15 medium (Thermo Fisher Scientific) supplied with 10% FBS, penicillin and streptomycin. MDA-MB-231 cells were maintained without CO2 equilibration.
Doxorubicin-resistant MCF-7 cells (MCF-7/DOXR) were selected as previously described [
17]. MCF-7 cells were sequentially exposed to increasing doses of doxorubicin (0.1, 0.5, 1.0, 2.0 and 5.0 μM). Cells were initially cultured in DMEM medium with 0.1 μM doxorubicin for 1 d, followed by culture with doxorubicin free DMEM medium for 4 d. Selection with the same concentration of doxorubicin was repeated twice before moving to selection with the next dose.
Reagents
Doxorubicin, paclitaxel, docetaxel and 4-hydroperoxy cyclophosphamide were purchased from ApexBio (Houston, TX, USA). Puromycin was purchased from Sigma-Aldrich (Shanghai, China).
Quantitative polymerase chain reaction (qPCR)
Relative expression level of
AGR2 mRNA was detected using qPCR as described previously [
18]. Total RNA was isolated using TRIzol reagent (Invitrogen, Thermo Fisher Scientific). cDNA was synthesized with a PrimeScript cDNA Synthesis Kit (Takara Bio Inc., Shiga, Japan) followed analysis with a LightCycler 480 SYBR Green I Master qRT-PCR kit (Roche, Mannheim, Germany).
ACTB was used as a normalization gene. The following primers were synthesized from Invitrogen (Thermo Fisher Scientific, Shanghai, China):
AGR2 (GTGTAGGAGAGGGCCACAAG and CGACTCACACAAGGCAGGT) and
ACTB (GTTGTCGACGACGAGCG and GCACAGAGCCTCGCCTT).
For detecting expression levels of mature miRNAs, cDNA was synthesized from total RNA using a miScript II RT Kit (QIAGEN, Shanghai, China). qPCR was performed using a miScript SYBR Green PCR Kit (QIAGEN) with U6 as a normalization gene. The following primers were used: miR-342-3p (Forward: TCTCACACAGAAATCGCACCCGT), miR-217 (Forward: TACTGCATCAGGAACTGATTGGA), miR-135b-5p (Forward: TATGGCTTTTCATTCCTATGTGA), miR-194-5p (Forward: TGTAACAGCAACTCCATGTGGA), miR-543 (Forward: AAACATTCGCGGTGCACTTCTT), miR-24-3p (Forward: TGGCTCAGTTCAGCAGGAACAG), miR-377-3p (Forward: ATCACACAAAGGCAACTTTTGT), miR-3158-3p (Forward: AAGGGCTTCCTCTCTGCAGGAC), miR-216b-3p (Forward: ACACACTTACCCGTAGAGATTCTA), miR-124-5p (Forward: CGTGTTCACAGCGGACCTTGAT), miR-1267 (Forward: CCTGTTGAAGTGTAATCCCCA), miR-624-3p (Forward: CACAAGGTATTGGTATTACCT) and U6 (CTCGCTTCGGCAGCACA and AACGCTTCACGAATTTGCGT). The results were analyzed by using the method of comparison on –ΔΔct values. qPCR was performed on an LC480 cycler (Roche).
Western blot analysis
Proteins were extracted from cells or tissues using Cell Lysis Buffer (Cell Signaling Technology, Danvers, MA, USA). Western blot analyses were performed with the following antibodies: Cleaved Caspase-3 (Asp175), Mcl-1 (D2W9E), Bcl-2 (D17C4), Bcl-xL (54H6), Bim (C34C5), Bak (D4E4), Cyclin D1 (92G2), Cyclin E1 (D7T3U), CDK2 (78B2), CDK4 (D9G3E), CDK6 (D4S8S), AGR2 (D9V2F) and GAPDH (D16H11), all of which were purchased from Cell Signaling Technology.
Cell viability assay
Cell viability was detected using a Cell Counting Kit-8 (CCK-8) from DOJINDO (CK04, Tokyo, Japan). Briefly, cells were incubated with CCK-8 reagent for 1 h followed by reading optical density at 450 nm.
Short hairpin RNA (shRNA)-mediated knockdown of AGR2
Three pairs of human AGR2-specific shRNAs (AGR2-shRNA) were synthesized according to the human AGR2 gene sequence (NM_006408). An unrelated negative shRNA was used as control. All shRNAs were cloned to a pLVX-shRNA2-Puro lentiviral vector (Genecreate, Wuhan, China). Lentiviral shRNA particles were obtained by transfecting 293FT cells with the constructed lentiviral vector packed in a ViraPower HiPerform Lentiviral Expression System (Invitrogen, Thermo Fisher Scientific). MCF-7 cell line, stably expressing AGR2-shRNA, was established by transducing cells with lentiviral shRNA particles, followed by selection with puromycin. Our preliminary experiment identified one of the AGR2-shRNA as the most efficient one in silencing AGR2-expression.
Over-expression of AGR2
CDS fragment (528 bp) of human AGR2 (NM_006408) was synthesized and cloned to pCDH-CMV-MCS-EF1-GFP-Puro lentiviral vector (Genecreate). Lentiviral AGR2 particles were obtained as described above. MCF-7 and MDA-MB-231 cell lines, over-expressing AGR2, were established as described above. A blank vector was used as control.
Xenograft experiments
BALB/c Nude mice were anesthetized and inoculated subcutaneously with 5 × 106 gene modified MCF-7 cells. Mice were injected intraperitoneally (i.p.) with PBS or doxorubicin (5 mg/kg) once on day 12, 15 and 18 after cell inoculation. Tumor size was measured continuously. Tumor volume was calculated with the formula: Tumor volume (mm3) = length (mm) × width (mm) × width (mm) × 0.5. Tumors were collected on day 32 for histological analyses.
H&E staining and immunohistochemistry
Tissue slides were prepared from tumors collected from xenograft experiments. H&E staining was performed as previously described [
19].
Immunohistochemistry was performed to detect expression of Ki-67 and cleaved caspase-3 in tumor tissues. Anti-Ki-67 and peroxidase-conjugated goat anti-rabbit IgG were from ZSGB-BIO (Beijing, China). Anti-cleaved caspase-3 (Asp175) was from Cell Signaling Technology. DAB substrate was used for visualization (ZSGB-BIO).
TUNEL assay
TUNEL assay was performed to detect apoptotic cells in tissue slides using a TUNEL Kit per manufacturer’s instructions (KGA702, KeyGEN BioTECH, Beijing, China).
β-Galactosidase staining
Senescence of cells was analyzed using a β-Galactosidase Staining Kit per manufacturer’s instructions (KGPAG001, KeyGEN BioTECH).
Prediction of miRNA target
Several candidate miRNAs were predicted and selected from literature [
20] and miRNA databases (TargetScan and miRBase). All these candidate miRNAs had a potential targeting 3′-UTR of
AGR2 mRNA (Additional file
1 Table S1).
Luciferase reporter assay
3′-UTR (1097 bp) of AGR2 mRNA was synthesized and cloned to pmirGLO Dual-Luciferase miRNA Target Expression Vector (Promega, Madison, WI, USA). A mutant AGR2 3′-UTR (1090 bp) luciferase vector, lacking the binding site AAGCCAT of miR-135b-5p, was also constructed. Luciferase vector and miRNA mimics (Genecreate) were co-transfected into MCF-7 cells using Attractene Transfection Reagent (QIAGEN). Twenty-four hours after transfection, firefly and renilla luciferase activities of cell lysate were analyzed by using a Dual-Luciferase Reporter Assay kit (Promega). Firefly luciferase is a reporter, and renilla luciferase is a normalizer.
Lentiviral particles, expressing a control miRNA or hsa-miR-135b double strand precursors, were obtained from Genechem (Shanghai, China). MCF-7 cell lines, stably expressing control miRNA or hsa-miR-135b, were established by transducing cells with lentiviral particles, followed by selection with puromycin.
RNA interference
ESR1-targeting small interfering RNAs (siRNA) (ESR1 siRNA-1: CCGGCAUUCUACAGGCCAA, ESR1 siRNA-2: GGAGAAUGUUGAAACACAA, ESR1 siRNA-3: GCUAGAGAUCCUGAUGAUU), synthesized from Genecreate (Wuhan, China), were transfected into MCF-7 cells using HiPerFect Transfection Reagent (QIAGEN). An irrelevant siRNA (Genecreate) was used as negative control.
Statistics
Data are presented as mean ± standard deviation (SD). Comparison of means was performed with unpaired Student t test or one-way ANOVA test followed by Bonferroni post-tests. Correlations were analyzed using Pearson correlation test. Value of p < 0.05 was considered statistically significant.
Discussion
AGR2 is a protein disulphide isomerase regulating protein folding in endoplasmic reticulum which is important for survival of tumor cells, as many tumor cells produce abundant proteins which would cause proteotoxic stress to cells in case of insufficient or inappropriate protein-folding [
7,
21]. Importantly, AGR2 was implied to mediate drug resistance in breast cancer [
7,
10,
11]. Expression level of AGR2 correlated with doxorubicin-sensitivity of breast cancer cells in our study. Up-regulation and down-regulation of AGR2 decreased and increased doxorubicin-sensitivity, respectively (Fig.
2), and this effect is associated with apoptotic process. Knockdown of AGR2 increased apoptosis and reduced proliferation of the xenograft tumors (Fig.
3). AGR2 was shown to induce expression of cyclin D1 in breast cancer cells [
22]. Cyclin D1 is a critical driver of tumorigenesis of breast cancer [
23,
24]. On the other hand, dysregulation of cyclin D1 is a marker of senescence, which counteracts tumor genesis [
25,
26]. However, exposure to doxorubicin had no impact on senescence or expression of cyclin D1 in our study (Additional file
1: Figure S3). Li and colleague had shown that AGR2 stabilized hypoxia inducible factor-1α enhancing hypoxia-induced doxorubicin resistance in breast cancer cells [
11]. Over-expression of AGR2 alone seems insufficient for tumorigenesis of breast cancer [
27], but rather can contribute to drug resistance. Consistently, our results showed that knockdown or over-expression of AGR2 in MCF-7 and MDA-MB-231 cells had no significant impact on cell viability in the absence of doxorubicin. AGR2 mediated drug resistance seems more related with topoisomerase II inhibitor. Knockdown of AGR2 increased sensitivity to paclitaxel and docetaxel. However, over-expression of AGR2 had no impact on sensitivity to these drugs in MCF-7 cells (Additional file
1: Figure S1). AGR2 also mediates resistance to hormone therapy in breast cancer [
9]. All these suggest that manipulating expression of AGR2 might counteract drug-resistance in breast cancer, especially in those who with high level of AGR2.
Others reported that level of AGR2 positively correlated with expression of ER in breast cancer. ER binds directly to AGR2 promoter to activate transcription of AGR2 in both cell lines and tumor samples of breast cancers [
12,
13]. AKT signaling pathway was also reported to induce AGR2 expression [
28]. Our data on clinical breast cancer samples also showed that ER-positive tumors expressed higher levels of AGR2 than ER-negative ones (Fig.
7). MCF-7 and MDA-MB-231 cells are ER-positive and ER-negative cells, respectively. Thus, expression of ARG2 in these two cell lines might have correlation with expression of ER (Fig.
1). Some clinical data showed that over-expression of AGR2 can also be found in ER-negative breast cancers [
10], which indicates that other mechanisms might also participate in regulating AGR2-expression. Interestingly, we found expression of AGR2 could be further regulated by miR-135b-5p.
miRNAs play important roles in multiple cellular processes such as proliferation and differentiation [
29,
30]. Prognostic miRNAs, such as miR-342 and miR-30e, were reported in breast cancer and some miRNAs were proposed as predictive markers of drug resistance [
16,
31,
32]. However, miRNAs-mediated regulation of AGR2 is still unclear in breast cancer. Since AGR2 mediates doxorubicin-resistance, we further focused on miRNAs-mediated regulation of AGR2. Among these candidate miRNAs, down-regulation of miR-135b-5p was in parallel with up-regulation of AGR2 in breast cancer cells. Doxorubicin-treatment induced up-regulation of AGR2 in association with down-regulation of miR-135b-5p (Fig.
4). miR-135b-5p negatively regulated expression of
AGR2 through targeting 3′-UTR confirmed that AGR2 is a target gene of miR-135b-5p in breast cancer cells (Fig.
5). As expected, miR-135b-5p reduced expression of AGR2 and increased doxorubicin-sensitivity of MCF-7 cells both in vitro and in vivo (Figs.
5 and
6). Doxorubicin treatment induced down-regulation of miR-135b-5p is a probable reason for over-expression of AGR2 in breast cancer (Figs.
4 and
5), which is indicative for a treatment related drug resistance. Intriguingly, we also observed a negative correlation between levels of AGR2 and miR-135b-5p in clinical breast cancer samples (Fig.
7). Level of miR-194-5p also negatively correlated with level of AGR2 in breast cancer cell lines. Interestingly, doxorubicin treatment increased level of miR-194-5p in MDA-MB-231 cells (Fig.
4). Expression of miR-194-5p is regulated by some upstream regulators such as GATA2 and non-coding RNAs [
33,
34]. The difference on levels of miR-194-5p in MCF-7 and MDA-MB-231 might have some relations with ER or those upstream regulators of miR-194-5p. In addition, transfection with miR-194-5p did not down-regulate expression of AGR2 (Fig.
5). miR-194-5p might not be a direct regulator of AGR2 in breast cancers. Thus, in addition to ER-dependant mechanism, we identified that miR-135b-5p is another regulator of AGR2 in breast cancer.
Others reported that increased levels of miR-342-3p and miR-217 positively correlated with expression of ER in breast cancer [
20]. Consistently, we showed MCF-7 had higher levels of miR-342-3p and miR-217 than MDA-MB-231. Their relations with AGR2 were reported by others recently. miR-342-3p targets AGR2 and down-regulation of miR-342-3p is associated with over-expression of AGR2 in non-small cell lung cancer [
35]. Down-regulation of miR-217 was associated with over-expression of AGR2 in chronic myelogenous leukemia cells [
36]. Doxorubicin-treatment had no impact on level of miR-342-3p or miR-217 in breast cancer cells (data not shown). Since levels of miR-342-3p and miR-217 positively correlated with level of AGR2 in breast cancer cells, we deduce that miR-135b-5p is a more dominant regulator of AGR2-expression in breast cancer, comparing with miR-342-3p and miR-217. miR-1291 was shown to regulate expression of AGR2 in pancreatic cancer cell [
37]. The findings suggest that miRNAs mediated regulation on AGR2 differs in different type of cancers.