miR-135b-5p enhances doxorubicin-sensitivity of breast cancer cells through targeting anterior gradient 2

The pro-oncogenic anterior gradient 2 (AGR2) is involved in tumor growth and drug resistance of breast cancer. Mechanisms that regulate expression of AGR2 still need to be elucidated. In this study, expression levels of AGR2 and miR-135b-5p were analyzed in different breast cancer cell lines as well as in clinical breast cancer tissues. The in vitro and in vivo functional effect of AGR2 and miR-135b-5p were also investigated. A luciferase reporter assay was applied to confirm the interaction between miR-135b-5p and AGR2 mRNA. We identified AGR2 as a target of miR-135b-5p. Expression of AGR2 was up-regulated in doxorubicin-resistant breast cancer cells. AGR2 mediated doxorubicin-sensitivity of breast cancer cells both in vitro and in vivo. miR-135b-5p negatively regulated AGR2-expression of breast cancer cells increasing doxorubicin-sensitivity. However, miR-135b-5p was down-regulated in doxorubicin-resistant breast cancer cells as well as during treatment with doxorubicin, which might be a probable reason for over-expression of AGR2. Up-regulation of miR-135b-5p increased doxorubicin-sensitivity of breast cancer cells in vivo. In addition, levels of AGR2 negatively correlated with levels of miR-135b-5p in clinical breast cancer tissue samples. Our results highlight the potential of miR-135b-5p as a target for treating AGR2-expressing breast cancer with doxorubicin-resistance.


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 treat-ments [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.

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.

Mice
BALB/c Nude mice were purchased from Vital River (Charles River, Beijing, China). Mice were bred in a special pathogen free room. 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.

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 (GTGTAGGA GAGGGCCACAAG and CGACTCACACAAGG CAGGT) and ACTB (GTTGTCGACGACGAGCG and GCACAGAGCCTCGCCTT).

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.

Xenograft experiments
BALB/c Nude mice were anesthetized and inoculated subcutaneously with 5 × 10 6 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.

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).

Senescence of cells was analyzed using a β-Galactosidase
Staining Kit per manufacturer's instructions (KGPA G001, 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).

Lentivirus mediated over-expression of miR-135b
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.

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.

Knockdown of AGR2 increased doxorubicin-sensitivity of breast cancer cells in association with enhanced apoptosis
To assess the correlation between expression of AGR2 and doxorubicin-sensitivity in breast cancer cells, we analyzed expression of AGR2 mRNA in MDA-MB-231 and MCF-7 cells as well as doxorubicin-resistant MCF-7/DOXR cells. MCF-7 cells expressed higher levels of AGR2 than MDA-MB-231 cells at both mRNA and protein levels. Expression of AGR2 protein was not observed in MDA-MB-231 cells, although we could detect low level of AGR2 mRNA in this cell line ( Fig. 1a  and b). Interestingly, expression of AGR2 was up-regulated in MCF-7/DOXR cells as compared with MCF-7 cells ( Fig. 1a and b). IC50 of doxorubicin in MDA-MB-231, MCF-7 and MCF-7/DOXR cells was 48.16 nM, 147.10 nM and 1.87 μM, respectively (Fig. 1c).
Next, we examined if AGR2 had an impact on doxorubicin-sensitivity in breast cancer cells, through modulating expression of AGR2 with AGR2-shRNA or AGR2-over-expression vectors. Knockdown and over-expression of AGR2 increased and decreased sensitivity to doxorubicin in MCF-7 cells, respectively ( Fig. 2a and b). Since MDA-MB-231 cells did not expression AGR2 protein, we only constructed over-expression model in this cell line and found that forced expression of AGR2 increased resistance to doxorubicin (Fig. 2c). Impact of AGR2 on apoptosis was further analyzed. Knockdown of AGR2 increased level of cleaved caspase-3 in MCF-7 cells treated by doxorubicin. Knockdown of AGR2 increased level of pro-apoptotic Bak and Bim, but decreased level of anti-apoptotic Mcl-1 (Fig. 2d). These findings indicated that AGR2 mediated drug resistance to doxorubicin in breast cancer cells.
In addition, we also tested the effect of AGR2 in treatment with other chemotherapeutics including paclitaxel, docetaxel and cyclophosphamide in MCF-7 cells. Knockdown of AGR2 increased sensitivity to paclitaxel and docetaxel. However, over-expression of AGR2 did not decrease sensitivity to paclitaxel and docetaxel. In addition, modulating expression of AGR2 had no impact on drug sensitivity to cyclophosphamide (Additional file 1: Figure S1).
Levels of AGR2 mRNA showed negative correlation with levels of miR-135b-5p and miR-194-5p in these breast cancer cells (Fig. 4b). Since reduced expression of miRNAs lead to up-regulation of target genes [15], we further analyzed correlation between miR-135b-5p or miR-194-5p and AGR2. Short term exposure to doxorubicin induced up-regulation of AGR2 mRNA both in MCF-7 and MDA-MB-231 cells. MCF-7 cells , and viability of control was set as 100%. Cells were treated with doxorubicin for 48 h (right panel), followed by measuring for cell viability (n = 6), and viability of untreated cells was set as 100%. d MCF-7 cells, stably expressing a control shRNA vector or an AGR2-shRNA vector, were treated with doxorubicin (100 nM) for 48 h. Protein levels were detected by western blotting with indicated antibodies. Experiment was repeated three times. Data are shown as mean ± SD, compared using unpaired t test.*, p < 0.05; n.s., no significance expressed higher levels of AGR2 mRNA than MDA-MB-231 cells after doxorubicin treatment (Fig. 4c). Doxorubicin reduced levels of miR-135b-5p both in MCF-7 and MDA-MB-231 cells (Fig. 4d). Doxorubicin reduced levels of miR-194-5p of MCF-7 cells, but increased levels of miR-194-5p of MDA-MB-231 cells (Fig. 4e). These findings indicated that down-regulation of miR-135b-5p correlated with up-regulation of AGR2 in breast cancer cells.

miR-135b-5p negatively regulated expression of AGR2 and increased doxorubicin-sensitivity of MCF-7 cells
Through transfecting MCF-7 cells with miRNAs mimics, we found that miR-135b-5p mimic reduced levels of AGR2 mRNA, whereas miR-194-5p had no impact on levels of AGR2 mRNA (Fig. 5a). Next, we analyzed if miR-135b-5p targeted 3′-UTR of AGR2 mRNA. As MCF-7 cells expressed a relative low level of background miR-135b-5p, MCF-7 cells were used as transfection host of a luciferase reporter vector expressing 3′-UTR of AGR2 mRNA. Co-transfection with miR-135b-5p mimic suppressed luciferase activity, compared with a negative control miRNA. We also constructed a mutant AGR2 3′-UTR luciferase reporter vector by deleting the binding site of miR-135b-5p, and found that miR-135b-5p mimic failed to suppress luciferase activity of cells transfected with the mutant vector (Fig. 5b). Doxorubicin-treatment reduced levels of miR-135b-5p of MCF-7 cells (Fig. 4d). To test if doxorubicin induced down-regulation of miR-135b-5p is a probable reason for over-expression of AGR2, cells were transfected with miR-135b-5p mimic before treatment with doxorubicin. Doxorubicin induced up-regulation of AGR2 level was reduced by miR-135b-5p mimic (Fig. 5c).  in panel A). a Tumor size was measured continuously. b-e On day 32, tumors were collected from mice followed by morphological (b) and histological analyses (c). d-e Immunohistochemistry was performed to detect expression of Ki-67 and cleaved caspase-3 in tumor tissues. TUNEL assay was applied to dectec apoptosis. Scale bar in (c and d): 500 μm. e Percent of Ki-67-positive cells, apoptotic cells and cleaved caspase-3-positive cells were counted. Experiment was repeated twice. Data are shown as mean ± SD, compared using one-way ANOVA test. *, p < 0.05; n.s., no significance; Doxo, doxorubicin Furthermore, we up-regulated expression of miR-135b-5p in MCF-7 cells by transducing cells with a lentivirus vector expressing double strand miR-135b precursor. Transduction of miR-135b vecter reduced expression of AGR2 (Fig. 5d). Transduction of miR-135b vecter had no impact on viability of cells in the absence of doxorubicin (Fig. 5e). Forced expression of miR-135b-5p in MCF-7 cells increased doxorubicin-sensitivity as compared with negative control (Fig. 5f). Next, we tested the impact of miR-135b-5p on doxorubicin-sensitivity in the murine xenograft model. The miR-135b vecter-transduced MCF-7 cells showed a similar growth rate as control ( Fig. 6a and b). Transduction of miR-135b vecter reduced level of AGR2 in the tumors (Fig. 6c). After being treated by doxorubicin, miR-135b tumors grew more slowly (Fig. 6a and b) and had less Ki-67-positive cells (Fig. 6d-f) than control. These findings confirmed that miR-135b-5p negatively regulated expression of AGR2 through targeting 3′-UTR and increased doxorubicin-sensitivity of MCF-7 cells.

Levels of miR-135b-5p negatively correlated with levels of AGR2 in clinical breast cancer samples
Clinical breast cancer samples were collected and subjected to analysis for expression of AGR2 protein and miR-135b-5p. ER-positive tumors (n = 20) expressed higher levels of AGR2 than ER-negative ones (n = 8). On the contrary, ER-positive tumors showed lower levels of miR-135b-5p comparing with ER-negative ones ( Fig. 7a  and b). Although expression of AGR2 was reported to be regulated by ER, we observed that expression of AGR2 are at low levels or undetectable in some ER-positive samples. In addition, ER-negative sample #6, #7 and #8 expressed low or middle levels of AGR2 (Fig. 7a). Interestingly, we observed a negative correlation between levels of AGR2 and miR-135b-5p in these samples (Fig. 7c). These findings indicated that level of miR-135b-5p might be a predictive marker for AGR2-expression in breast cancer.

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 Fig. 4 Decreased levels of miR-135b-5p correlated with up-regulation of AGR2 in breast cancer cells. a cDNA was synthesized from total RNA extracted from MDA-MB-231, MCF-7 and MCF-7/DOXR cells. qPCR was applied to measure levels of mature miRNAs with U6 as a normalization gene. b Correlation between levels of AGR2 mRNA and miRNAs were analyzed using Pearson correlation test. c-e MCF-7 and MDA-MB-231 cells were treated with gradient doses of doxorubicin for 48 h. Expression levels of AGR2 mRNA and miRNAs were measured by qPCR (n = 3). Experiment was repeated twice. Expression levels of RNAs represent fold changes. Data are shown as mean ± SD and are compared using unpaired t test or one-way ANOVA test. *, p < 0.05 compared with MDA-MB-231 cells; #, p < 0.05 compared with untreated cells. 0 nM represents untreated cells 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 Fig. 5 miR-135b-5p negatively regulated expression of AGR2 and increased doxorubicin-sensitivity of MCF-7 cells in vitro. a MCF-7 cells were transfected with mimics of a control miRNA, miR-135b-5p or miR-194-5p. Twenty-four hours after transfection, level of AGR2 mRNA was measured by qPCR (n = 3). b AGR2 3′-UTR (1097 bp) and a mutant AGR2 3′-UTR (1090 bp) fragments were synthesized and cloned to pmirGLO Dual-Luciferase miRNA Target Expression Vector. Luciferase vector and miR-135b-5p mimic were co-transfected into MCF-7 cells. Twenty-four hours after transfection, firefly and renilla luciferase activities of cell lysate were analyzed. Luciferase activity was expressed as ratio of Firefly/Renilla. c MCF-7 cells were transfected with mimics of a control miRNA or miR-135b-5p, followed by treatment with doxorubicin (100 nM) for 48 h. AGR2expression was detected using western blot. d-e MCF-7 cells were stably transduced with lentivirual vector expressing double strand precursors of a control miRNA or miR-135b. d AGR2-expression was detected using western blot. e Cell viability was measured by CCK-8 assay (n = 6). f Cells were treated with doxorubicin for 48 h followed by measuring for cell viability (n = 6). Experiment was repeated three times. Expression levels of mRNA represent fold changes. Data are shown as mean ± SD and are compared using unpaired t test or one-way ANOVA test. *, p < 0.05; n.s., no significance 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. Fig. 6 miR-135b-5p increased doxorubicin-sensitivity of MCF-7 cells in vivo. MCF-7 cells, stably transduced with lentivirual vector expressing double strand precursors of a control miRNA or miR-135b, were inoculated subcutaneously into BALB/c Nude mice at a dose of 5 × 10 6 per mouse (n = 4 in each group). Mice were injected intraperitoneally with PBS or doxorubicin (5 mg/kg) once on day 12, 15 and 18 after cell inoculation (indicated by arrows in panel A). a Tumor size was measured continuously. b-f On day 32, tumors were collected from mice. c Proteins extracted from tumors were analyzed by western blotting with indicated antibodies. d Histological analysis was performed using H&E staining. e Immunohistochemistry was performed to detect Ki-67-expression in tumor tissues. Scale bar in (d and e): 500 μm. f Percent of Ki-67positive cells was counted. Experiment was repeated twice. Data are shown as mean ± SD, compared using one-way ANOVA test. *, p < 0.05; n.s., no significance 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.

Conclusion
In conclusion, AGR2 is a target of miR-135b-5p. Decreased levels of miR-135b-5p correlated with over-expression of AGR2 in breast cancer cells during doxorubicin treatment. Up-regulation of miR-135b-5p suppressed expression of AGR2 increasing doxorubicin-sensitivity of breast cancer cells. Our results highlight the potential of miR-135b-5p as a target for treating AGR2-expressing breast cancer with doxorubicin-resistance.

Additional file
Additional file 1: Table S1. Candidate miRNAs with AGR2-targeting potential. Figure S1. Knockdown of AGR2 increased sensitivity to paclitaxel and docetaxel in MCF-7 cells. Figure S2. Knockdown of estrogen receptor 1 (ESR1) increased level of miR-135b-5p but decreased level of miR-342-3p in MCF-7 cells. Figure S3. Long term exposure to doxorubicin had no impact on senescence. (DOCX 1106 kb) Abbreviations AGR2: Anterior gradient 2; DOXR: Doxorubicin resistant; ER: Estrogen receptor; miRNAs: microRNAs; PDIs: Protein disulphide isomerases; shRNA: Short hairpin RNA; UTR: Untranslated region Fig. 7 Levels of miR-135b-5p negatively correlated with levels of AGR2 in clinical breast cancer tissues. Twenty-eight breast cancer tissue samples were collected from patients. Total RNA and protein were extracted from tissues. a AGR2-expression was detected using western blot. Intensities of AGR2-bands were analyzed with ImageJ software, and were normalized to GAPDH. b qPCR was applied to measure levels of miR-135b-5p with U6 as a normalization gene. Expression levels of RNA represent fold changes. Data are shown as mean ± SD and are compared using unpaired t test. c Correlation between levels of AGR2 protein and miR-135b-5p was analyzed using Pearson correlation test