Cryptotanshinone Breaks ERα-dependent and independent BCRP Dimerization to Reverse the Multidrug Resistance in Breast Cancer

Not only a long-term anti-estrogen therapy, but also estrogen receptor-negative breast cancer are generally prone to induce resistance, causing poor prognosis in clinic. Breast cancer resistance protein (BCRP) plays an important role in multidrug resistance. Here it is to elucidate the mechanism that a natural compound cryptotanshinone inhibits BCRP. HPLC was for analyzing the special compound concentration and molecular docking assay for the anity of compound with protein. Non-reducing gradient gel electrophoresis and uorescence resonance energy transfer (FRET) microscopy imaging were used to detect polymer and stain the membrane protein. Immunouorescence staining, plasmids transfection, real-time PCR and western blot were also used. The MX uorescence accumulation was detected in MCF-7 cells (E) and MDA-MB-231 cells (G) treated with CPT for 8h by the ow cytometry FL-4 channel, and the uorescence intensity represents the activity of BCRP eux. The quantitated results (H)


Background
Chemotherapy is one of the foremost methods to treat cancer nowadays, but the occurrence of multidrug resistance (MDR) has made its current clinical application become increasingly limited 1 . The MDR of breast cancer is prominent in all tumor resistance, especially in estrogen receptor α-positive (ERα+) breast cancer patients. After receiving endocrine therapy such as tamoxifen, about 70% patients will have recurrence of drug resistance in the late stages 2 . In recent years, many studies have shown that there is a close relationship between the occurrence of breast cancer MDR and the ATP-binding cassette (ABC) transporter family, especially P-glycoprotein (P-gp/ABCB1), multidrug-resistance-associated protein 1 (MRP1/ABCC1) and the breast cancer resistance proteins (BCRP/ABCG2) play key roles in the development of breast cancer MDR 3,4 .
ABCG2, the subfamily G of the human adenosine triphosphate (ATP)-binding cassette (ABC) transporter superfamily, was known as BCRP 5 . Its structure is similar to P-gp and MRP1, certain homologous sequences and the role of "drug pump" in function 6 . However, BCRP still has its own unique conformation, a nucleotide binding domain (Nucleotide-binding domain (NBD) at the C-terminus and a hydrophobic transmembrane domain (TMD) at the N-terminus, which determined BCRP as a semitransporter 7 . In general, P-gp and MRP1 that have two NBDs and TMDs structures called full transporters 8 . Semi-transporters are commonly localized in cytoplasm, but BCRP was the rst reported semi-transporter localized on the cell membrane 9 . Several studies indicated that BCRP was likely to form homodimers, tetramers, dodecamers, and even larger oligomer structures by intramolecular disul de linkages. It signi cantly increases the e ciency of external pumping by increasing the formation of the outer channel cavity 10 . Thus, inhibition of BCRP and blocking the e ux of therapeutic drugs has been considered a feasible strategy for eliminating the MDR, this boosts the development of BCRP inhibitors 11 . Cryptotanshinone (CPT) is a natural diterpenoid from the Salvia miltiorrhiza. Since CPT was rstly demonstrated to exhibit anti-cancer activity dependent on inhibition of STAT3 dimer 12 , its molecular mechanism attracted much concern. We also made much attempt on it, and found that CPT can not only inhibit mTOR signaling 13 , but also activate MAPK pathways 14 , leading to cell death. However, CPT is not an mTOR inhibitor like rapamycin and its derivatives, but inhibits mTOR signaling via activating AMPK-TSC2 axis 15 . Recently, we found that CPT indicates a signi cant inhibition on MCF-7 but not MDA-MB-231 cells by an ERα-dependent way. Furthermore, MCF-7/ADR, a doxorubicin-induced multidrug-resistant cell line, is also sensitive to CPT, and CPT can distinctly enhance the inhibitory effect of tamoxifen on MCF-7/ADR 16 . As we knew, MCF-7/ADR cells with high expression of ABC protein family but negative ERα expression are induced by doxorubicin to acquire multidrug resistance. Therefore, we judged that the ERα and BCRP are the essential factors mediating CPT reversing MDR.
In this study, we for the rst time found the natural compound CPT can inhibit the dimerization of BCRP in the membrane by an ERα-dependent way in ERα-positive breast cancer cells. However, it can directly block BCRP dimerization in ERα-negative resistant breast cancer cells to reverse the resistance. It can be a potential agent for application in both ERα-positive and ERα-negative breast cancer with high expression and decreasing the possibility of resistance.

Cell viability assay
MCF-7 cells, MDA-MB-231 cells, MCF-7/ADR cells were seeded in a 96-well plate at a density of 1 × 10 4 /well. After treating with drug or reagent, MTS (1:10 dilution in serum-free medium) was added and incubated at 37 °C for 4 h. Finally, the formazan was dissolved in DMSO and cell viability was evaluated through measuring the optical density (OD) at 490 nm using the BioTek Synergy2 microplate reader (BioTek Instruments, VT, USA).

HPLC analysis
CPT concentration was measured using HPLC. Cell lysates were extracted in the extraction buffer (containing methanol: water (1:1, v/v)) in the cold room for 15 min, followed by scraping and centrifuging at 17000 g for 10 min. CPT analysis was performed by HPLC (Waters E2695, Separations Module). Samples were injected into a 4.6mm × 250 mm Stable Bond column (ZORBAX Eclipse Plus C18; Agilent, CA, USA). The chromatography was run started with 45% solution A (methanol) and 55% solution B (H 2 O), and the volume (%) of solution A was raised to 50, 90, 100 at 10 min, 30 min, and 35 min respectively. Finally, the situation comes back to 45% solution A when the time of 45 min, and then holding up until stopped. Data were collected and analyzed by Analyst Software (AB Sciex).

Molecular docking assay
The three-dimensional structure of cryptotanshinone and mitoxantrone was achieved from PubChem Compound database (https://www.ncbi.nlm.nih.gov/pccompound/). Meanwhile, the structure of BCRP/ABCG2 (Protein Data Bank (PDB) ID: 6FFC with resolution of 3.56 Å) was retrieved from the Research Collaborator for Structural Bioinformatics PDB (Anonymous, www.rcsb.org). The molecular docking between the two compounds and BCRP/ABCG2 were evaluated by Discovery Studio (DS) 3.5 using the CDOCKER Protocol under the protein-ligand interaction section after preparing protein and ligands. The poses were scored by CDOCKER interaction energy, and the binding sites were also be showed.

MX/Rh123/DOX/TOPO e ux experiment
The mitoxantrone (MX)/rhodamine/doxorubicin (DOX)/Topotecan (TOPO) e ux experiment were performed following the modi ed method as reference described 17 . Brie y, breast cancer cells (3 × 10 5 cells per well) were seeded in 6-well plates and incubated overnight. When the cell con uence reached about 80%, the drug was added. After the administration, the cells were collected by centrifugation in 2 ml tubes, and each tube was added with 1 ml of serum-free medium to homogenize the cells. Then, all the cells except the blank group were added with the corresponding compounds and incubated in the dark under 37 °C for 30 min. (The MX positive group given KO143 10 µmol/L for 15 min in advance). After the end of time, all cells were undergone 1500 rpm, 4 °C, 5 min centrifugation, and the supernatant was discarded. Then washed cells with pre-cooled PBS twice. Finally, resuspending cells with 400µL precooled PBS to test. The uorescence accumulation of MX/Rh123/DOX/TOPO is detected by BD Accuri C6 Flow Cytometry (Becton, Dickinson and Company, NY, USA). The detection channel was FL-4/FL-1/FL-2.

Non-reducing gradient gel electrophoresis
The non-reducing gradient gel electrophoresis was performed following the modi ed method as reference described 18 . After the end of the cell administration, membrane protein and cytoplasmic protein are extracted as described in 2.11. The protein samples were denatured with loading buffer that containing no reducing agents such as DTT or 2-ME. Samples were boiled at 100 °C for 15 min. And the remaining steps were essentially identical to western blotting. When detecting the BCRP polymer, membrane proteins were separated by 6% Tris-glycine SDS-PAGE and molecular weight markers were used multicolor broad range protein ladder ranging from 10 kDa to 260 kDa (Thermo Scienti c, Waltham, MA, USA).

Fluorescence resonance energy transfer (FRET) microscopy imaging
The FRET method referred to several references and performed after appropriate modi cation 19 . The pCFP-ABCG2 plasmid and pYFP-ABCG2 plasmid were a gift from M.S. Jun Wang, Drug Discovery and Design Center, Shanghai Institute of Materia Medica, Chinese Academy of Sciences (Shanghai, China).
The breast cancer cells were seeded in a 35 mm diameter confocal culture dish, and when the cell con uence reached to 70%, the plasmid transfection was performed. The cell transfection steps were primarily based on the protocol provided by manufacturer (Invitrogen, NY, USA). Plasmid pCFP-ABCG2 (191.3 ng/µL), pYFP-ABCG2 (399.3 ng/µL) and 3µL lipofectamine 2000 reagent (Life Technologies, NY, USA) were respectively mixed with 50µL Opti-MEM medium. After incubating for 5 min at room temperature, the above two mixtures were lightly mixed and cultured for 10 min at room temperature. Finally, the 100µL mixture was added dropwise to 1 ml serum-free medium, and 100µL FBS was added 6 hours later, then incubation was continued for 16 hours at 37 °C. After successfully transfection, the drug could be administrated. Finally, the living cell FRET image were collected by Leica inverted uorescence microscope (Leica Microsystems, Solms, Germany) and analyzed by Image J software.

Immuno uorescence staining
Cells were plated on glass coverslips in 6-well plates and then treated with corresponding drug or reagents. First, stain the cell membrane with 10 µmol/L DiI (Beyotime Biotechnology, Shanghai, China) for 10 min at 37 °C. After that, cells were xed in 4% paraformaldehyde for 20 min, blocked in BSA (1% BSA dissolved in PBS). Cells were incubated with corresponding primary antibody (1:100 dilution) overnight at 4 °C, followed by incubation with Goat Anti-Rabbit IgG H&L FITC (1:1000 dilution) for 2 h and Hoechst nuclear dye for 10 min in dark. The images were obtained from laser scanning confocal microscope (Leica TCS SP5 X, Solms, Germany).

Cell membrane protein and cytoplasmic protein extraction
Cell membrane protein and cytoplasmic protein were extracted according to the protocol provided by manufacturer (Beyotime Biotechnology, Shanghai, China). In brief, cells were seeded into a culture dish (Lab services, Waltham, MA, USA) with diameter of 150 mm, and treated with compounds when the cell coverage area reached about 80%. After the time of the administration, it was washed once with ice PBS, and then the cells were scraped off with scraper, collected in a centrifuge tube, 4 °C, 600 g, 5 min. Discarding supernatant and resuspending cell pellet with reagent A added with PMSF (1:100 dilution), and placed in ice bath about 10-15 min. Then, the cells were disrupted by liquid nitrogen, freezing and thawing twice, and centrifuged at 4 °C, 700 g, 10 min. The supernatant was collected, 4 °C, 14,000 g, 30 min to precipitate cell membrane fragments. The supernatant was cytoplasmic protein. The bottom cell pellet was resuspended with reagent B, vortexed 5 s, ice bath 5-10 min. Finally, 4 °C, 14,000 g, 5 min, the supernatant was cell membrane protein, which was stored at -80℃. Cells were centrifuged by Beckman Microfuge® 20R (Beckman Coulter, MA, USA).

RNA isolation and real-time PCR
The RNA isolation experiment was mainly carried out according to manufacturer's guidelines. Total RNA was extracted from MCF-7 cells or MDA-MB-231 cells with Trizol (Vazyme, Nanjing, China) and then reversely transcribed to cDNA by HiScript® II Reverse Transcriptase (Vazyme, Nanjing, China). Real-Time PCR was executed of the ChamQTM SYBR® qPCR Master Mix (Vazyme, Nanjing, China), using Applied Biosystems 7500 Real-Time PCR Systems (Thermo Scienti c, Waltham, MA, USA). GAPDH was considered as an invariant control, and mRNA levels were expressed as fold changes after normalizing to GAPDH. Table 1 listed out the primers (Sangon Biotech, Shanghai, China) used.

Statistical analysis
The results were determined using Student's t test (two-group comparison) and ANOVA test by GraphPad Prism 5.0 software. P < 0.05 was considered to be statistically signi cant.

Intracellular accumulation of CPT is possibly related to BCRP
In order to further investigate the reasons why MCF-7 and MDA-MB-231 cells show different sensitivity to CPT except dependence on the ERα, we rst detected the intracellular and extracellular CPT levels in both cell lines by HPLC. As indicated in (Fig. 1A), following the increase of CPT concentration, the intracellular and extracellular CPT level have a sharp increase and almost equal in MCF-7 cells, while in MDA-MB-231 cells the intracellular CPT level just indicates a slight uctuation at a low level and is greatly lower than the extracellular level. And the difference is signi cant when the CPT concentration is up to more than 10 µmol/L (Fig. 1B). This result suggested that there is a certain transporter which regulates the intake and pumping process of CPT in cells. The most studied transporter proteins are ATP-binding transporters nowadays, and the main family members are P-gp, MRP1, BCRP and so on. We analyzed the expression of these three transporters in many cell lines from multi human tissues with the Human Protein Altas Database. The analysis indicated that BCRP/ABCG2 was highly expressed in MCF-7 cells (Fig. 1C), whereas P-gp/ABCB1 was almost not expressed and MRP1/ABCC1 was lowly expressed (Fig. 1D, E). Upon the abovementioned results, we speculated that BCRP may play an important regulatory role in transporting CPT across the membrane.

CPT inhibits e ux function of BCRP dependent on ERα
The close relationship between CPT and BCRP encouraged us to further explore the speci c regulation of CPT on BCRP. Firstly, CPT has no signi cant effect on the total protein expression of BCRP in both cells ( Fig. 2A, B). In addition, no signi cant effect was found on ABCG2 mRNA levels in MCF-7 and MDA-MB-231 cells by q-PCR (Fig. 2C, D; Table S1). These results prompted us to continue thinking whether CPT possibly in uences the function of BCRP. Considering the most important ability of BCRP is to transport substrate, so it is necessary to examine this function. We evaluated the e ux function of BCRP through the mitoxantrone e ux experiment 20 . The results showed that the uorescence peak of CPTadministered group in MCF-7 cells was signi cantly shifted to right compared with the control group (Fig. 2E, F). The uorescence abundance of intracellular mitoxantrone increased represents the e ux function of BCRP was inhibited. However, the same phenomenon was not observed in MDA-MB-231 cells (Fig. 2G, H). So we suspected that the difference between the two cells would be related to the expression of ERα? Thus, we silenced the ERα expression in MCF-7 cells and then co-incubated with CPT for 8 hours (Fig. 2I). We found that the original e ux inhibition effect was partially reversed (Fig. 2J, K). These results veri ed that CPT does not affect the total protein and transcription level of BCRP/ABCG2, but can speci cally inhibit the e ux function of BCRP in MCF-7 cells, and this effect is closely relevant with the presence of ERα.

CPT reduces membrane expression of BCRP
Particularly, BCRP is the rst semi-transporter found to be localized on the cell membrane to function by an oligomer 21 . So it is necessary to know whether CPT affects the membrane BCRP and its oligomer. We extracted the membrane proteins and cytoplasmic proteins from MCF-7 and MDA-MB-231 cells for western blot analysis. The results showed that CPT can signi cantly inhibit the membrane protein expression of BCRP, but not the cytoplasm expression in MCF-7 cells (Fig. 3A, B). And the BCRP expression in cell membrane or in cytoplasm in MDA-MB-231 cells was not affected (Fig. 3C, D). It undoubtedly further illustrates the effect of CPT on BCRP function. In addition, we also used LCSM to colocalize BCRP protein with cell membrane. The ndings were consistent with western blot results, indicating that CPT can signi cantly inhibit the enrichment of BCRP on MCF-7 cell membrane. As seen in Fig. 3E, when high concentrations of CPT were used (10 µmol/L, 20 µmol/L), The enrichment of FITC-BCRP on the cell membrane dyed with Dil were obviously inhibited. But this phenomenon was also not observed in MDA-MB-231 cells (Fig. 3F). Additionally, we con rmed the role of ERα during this process again. When ERα was silenced in MCF-7 cells, the enrichment of BCRP on the cell membrane increased, and there was still no signi cant change in cytoplasm (Fig. 3G, H). In hence, CPT inhibits the e ux function of BCRP through decreasing its expression in MCF-7 cell membrane, and ERα is essential.

CPT inhibits BCRP oligomer formation
We continued to evaluate the CPT effect on BCRP oligomer using the assay of non-reducing gradient gel electrophoresis. As shown in the Fig. 4A, BCRP on the membrane of MCF-7 was mainly in the form of dimers and oligomers (molecular weight over 140 kDa). Oligomer formation was signi cantly inhibited when treated with CPT (10, 20 µmol/L). The Fig. 4B shows that although BCRP exists in the cytoplasm of MCF-7 cells, its main form is a non-functional monomer with a molecular weight of 70 kDa. Besides, we also applied FRET assay to further ensure the effect of CPT on the BCRP oligomer formation on cell membrane surface. We transfected the gene sequence ABCG2 labeled with CFP and YFP into MCF-7 or MDA-MB-231 cells. After con rming the pCFP-ABCG2 and pYFP-ABCG2 successfully translated and distributed in the cell membrane (Fig. 4C), CPT was given for 8 h and then photographed. If BCRP can form oligomers, FRET phenomenon will occur (Fig. 4D), and we can judge the formation of BCRP oligomers by FRET e ciency. Finally, the experimental results are consistent with non-reducing electrophoresis. CPT can destroy BCRP oligomer formation in MCF-7 cells (Fig. 4E, F), but has no signi cant effect in MDA-MB-231 cells (Fig. 4G, H). CPT can inhibit the membrane BCRP oligomer formation, ultimately stop functional e ux.

CPT inhibits BCRP in resistant breast cancer cells independent on ERα
Previously, it was found that doxorubicin-resistant breast cancer cells MCF-7/ADR are sensitive to CPT. As indicated in Fig. 5A, BCRP expresses highly in MCF-7/ADR, but lowly in MDA-MB-231 cells with a signi cance upon the interactive comparation. Combined with current researches, we try to explore whether CPT inhibits MCF-7/ADR cells also dependent on BCRP. Firstly, HPLC analysis found that the concentration of intracellular CPT nearly equals to the extracellular CPT in MCF-7/ADR cells (Fig. 5B).
This result made us convinced that the transporters were involved in the inhibition of MCF-7/ADR cell proliferation by CPT. Due to the high expression of P-gp and MRP1 in MCF-7/ADR cells, we also found that CPT does not inhibit P-gp, MRP1 membrane protein and cytoplasmic protein expression by western blot (Fig.S1A, B, C, D) and also not affect the P-gp, MRP1 e ux function by the ow cytometry (Fig.S1E, F,  G). So, excluding P-gp and MRP1 proteins, we paid attention to BCRP. Veri ed by a series of experiments, we knew that CPT also has no signi cant effect on the total protein expression of BCRP in MCF-7/ADR cells (Fig. 5C), but strongly inhibits the membrane protein expression of BCRP even at low concentration of 5 µmol/L (Fig. 5D). The cytoplasm has less BCRP content, but it could be observed that CPT can increase the BCRP content in the cytoplasm to some extent (Fig. 5E). Similarly, ow cytometry results showed that CPT could inhibit the e ux function of BCRP in MCF-7/ADR cells, especially at the high concentration of 20 µmol/L (Fig. 5F, G). FRET experiments also demonstrated that CPT inhibits the formation of BCRP oligomers in MCF-7/ADR cells too (Fig. 5H, I). It is worth noting that functional inhibition of BCRP by CPT requires ERα mediation in MCF-7 cells, whereas MCF-7/ADR cells have undergone many changes in morphology, characteristics and functions compared with parental MCF-7 cells. Above all, the most obvious change is the absence of ERα expression in MCF-7/ADR cells. This information suggested CPT may have a diverse approach to functional inhibition of BCRP. ERα expression or deletion is not the only criterion for judgment.
BCRP consists of a nucleotide binding domains (NBDs) and a transmembrane domains (TMDs) 7 , TM1-3 are substrate binding regions (Fig. 5J). To further determine whether CPT can be recognized by BCRP and extracellularly pumped out, we used molecular docking technology to dock mitoxantrone, the speci c substrate of BCRP, with the protein 3D structure (Fig.S2A, B). After con rming the drug binding pocket, the CPT 3D structure was embedded for docking (Fig. 5K, L). As shown in Fig. 5K & Fig.S2A, through the intermolecular force comparison and calculation, the common binding sites were found to be VAL401, PHE439, LEU405, PHE432, PHE439. These results indicated CPT can be recognized by BCRP and extracellularly pumped out. The whole transport simulation process is shown in the Fig. 5M. When the substrate is bound into the cavity 1, the conformation of the protein changes with the transition from inward open to outward open. NDB combines with ATP hydrolysis to provide energy for conformational changes in the protein, while the substrate is transferred into the cavity 2, which has a relatively weak binding force, and the drug nally be pumped out. Taken together, CPT can bind to BCRP but plays an antagonized role for external pumping and the process is blocked leading to an accumulation of CPT in MCF-7 cells.

CPT enhances cancer cells sensitivity combined with BCRP e ux anti-cancer drugs
Based upon the above results, we immediately thought whether the synergistic effect would generate from the combination of CPT and anti-cancer drugs of BCRP e ux. To verify this assumption, we selected two of the most classic BCRP e ux drugs, mitoxantrone and topotecan 22 . At rst, we screened the concentration of CPT and the positive control drug Ko143. Then, we carried out a 72-h co-incubation experiment at a concentration of 1 µmol/L which had no signi cant effect on the proliferation of MCF-7/ADR cells (Fig. 6A, B). Compared with the single-incubation group of mitoxantrone (Fig. 6C), topotecan ( Fig. 6D), the co-incubation group with CPT could signi cantly reverse drug resistance at higher doses (mitoxantrone-10 µmol/L, topotecan-0.5 µmol/L and 1 µmol/L), and the inhibition rate of proliferation was under 50% which is close to Ko143 (Fig. 6E-6F). Meanwhile, subsequent ow cytometry showed that when mitoxantrone co-incubated with CPT, the accumulation of mitoxantrone in MCF-7/ADR cells was increased, that is, the functional inhibition of BCRP by CPT made the drugs e ux reduced (Fig. 6G, H).
And the similar signi cant results were repeated with co-incubation of CPT and topotecan (Fig. 6I, J). In conclusion, CPT could enhance the sensitivity of the cells to anti-cancer drugs leading to reverse multidrug resistance.

Discussion
The role of estrogen receptor-α (ERα) in breast cancer is well understood. ER expression is a critical factor for hormonal therapy and also considered as a prognostic marker. Nowadays, about 75% of breast cancer patients are ER-positive and administrated with anti-estrogen drugs, such as the selective estrogen receptor modulators tamoxifen. However, a part of ERα-positive breast cancers become hormone resistant with ERα-negative expression, and a majority of these patients suffered relapse in ve years 23 .
Furthermore, many researchers also found some chemotherapeutic agents may be less effective in ERα + breast cancer patients than ERα − ones 24 . These bidirectional results caused the complexity of the role of ERα in the resistant breast cancer.
Generally, the breast cancer resistance would be developed to multi-drug resistance, also called cross resistance. Several factors including ATP-binding cassette (ABC) transporters 5 , the mutation of targeted oncogenes, survived cancer stem cells (CSCs) 25 , activated cell growth factors are possibly involved in MDR. Specially, BCRP, one type of the ABC transporters is an important factor controlling the breast cancer MDR. Indeed, the relationship between ERα and BCRP expression has been investigated 26 . The estrogen response element (ERE) and progesterone response element (PRE) exist in the promoter region of BCRP 27 . The excessive transcriptional expression of this type of response element may play a major role in the development of breast cancer MDR 11 . The MDR is a multi-factor, multi-way, multi-stage and comprehensive process 28 . This study focused on BCRP primarily, and expected to look for new applications of CPT to reverse breast cancer MDR.
In the previous research, we have proved that CPT is not a selective estrogen receptor inhibitor though it can bind ERα and produce tamoxifen-like effect on cancers 16 . More importantly, CPT inhibits resistant MCF-7/ADR cells although it can activate MAPK and AKT, suggesting that induction of resistance via activation of MAPK 29 and AKT 30 is unavailable for CPT and there must be other factors involved in. So we judged BCRP is the key molecule mediating CPT reverse of the resistance. At present, a lot of BCRP inhibitors including highly selective inhibitors and non-selective inhibitors have been identi ed, and some highly selective inhibitors should be prospectively applied in clinic for reversing the MDR 31 . However, it is not completely clear to understand the molecular mechanism of BCRP inhibition. Certainly,it is known that some inhibit BCRP through inhibiting its ATPase activity, such as FTC, Ko134 and Ko143, while others as BCRP substrates can bind to BCRP and play a competitive role in suppressing transport function of BCRP 9, 31 .
CPT indicates a different inhibition on BCRP. Primarily, breaking the BCRP dimerization in the membrane and inactivating its function contributes to its e ux blocked. Secondly, in ERα + MCF-7 cells, inhibition of BCRP oligomer is mediated by ERα, downregulation of ERα can decrease the accumulation of MX in cells, indicating the function of BCRP is enhanced. In ERα − MCF-7/ADR cells, BCRP membrane expression and oligomer were both inhibited. However, in MDA-MB-231 cells with negative ERα and low BCRP expression, CPT doesn't affect its proliferation. These data make a conclusion that CPT reversing resistance of breast cancer is dependent on BCRP, the higher expression of BCRP, the stronger inhibition effect on breast cancer cells. Although CPT is an anti-estrogen compound, ERα becomes an unnecessary mediator because CPT can directly bind BCRP to block e ux. Additionally, it was found that breast cancer cells can switch between ERα and ErbB signaling to induce resistance, combined inhibition of both signal can postpone resistance 32 . From MCF-7 to MCF-7/ADR, ERα is nearly silenced but BCRP signi cantly overexpressed. These suggest that there are two modes of CPT inhibition BCRP: ERα-dependent and ERαindependent. CPT switches the targets between ERα and BCRP to attenuate resistance.
Nowadays, more and more researchers thought that drugs invented or designed for individual molecular target cannot generally conquer multigenic diseases such as cancer, diabetes, hypertension, neural diseases and so on 33, 34 . So combined drugs that simultaneously affect multiple targets are more bene cial to control complex disease, reverse drug resistance. For a compound, if it is a multi-target molecule and targets different proteins in different cell circumstance, the occurrence of resistance would also be decreased to minimum extent. We believed CPT is a multi-target compound as targeting the molecules such as STAT3 12 , AMPK 15 , MAPK 14 , ER 16 , NRF2 35 to prevent cancer growth has been proved.
However, these proteins are not the nal targets of CPT but the mediators because just ER and BCRP can be structurally bound with CPT. The Fig. 7 illustrates in detail the molecular process of CPT inhibition of BCRP by an ERα-dependent and -independent manner.
At present, designing the multi-target molecules based on the systems biology, structural biology and chemical informatics has become an optimum way to develop a new drug 34 . CPT as a multi-target compound would be a potential candidate drug plus with its selective BCRP inhibition in ERα − breast cancer. Meanwhile, the combination of CPT and anti-cancer drugs can enhance the sensitivity of cancer cells to chemotherapeutic drugs, and spread the range of their application in clinic. Additionally, although the inhibitory activity of CPT is not dominant compared to some current BCRP inhibitors, we expected CPT as a lead compound from natural herb, and after appropriate optimization of the structure in the later stage, will be develop to a highly effective and low-toxic natural BCRP-speci c inhibitor to serve clinical therapy. Availability of data and materials All data generated or analysed during this study are included in this published article.

Conclusions
WC, YL provided the study design and supervision, and wrote the paper. WN, HF performed experiments and analyses. SW, SH assisted with molecular analysis and interpretation. XZ assisted with data analysis and statistic. FX, YW, XL and AW provided special reagents and/or helped in analyzing the experiments. All authors read and approved the nal manuscript.             with CPT for 8h were respectively detected by the ow cytometry FL-1 channel, and the uorescence intensity represents the activity of P-gp and DOX e ux. RH123 and DOX were respectively the substrate of P-gp and MRP1. (G) The quantitated data of (E) and (F).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.