Screening of potential miRNA therapeutics for the prevention of multi-drug resistance in cancer cells

Chemotherapy, a major cancer treatment approach, suffers seriously from multidrug resistance (MDR), generally caused by innate DNA repair proteins that reverse the DNA modification by anti-cancer therapeutics or trans-membrane efflux proteins that pump anti-cancer therapeutics out of the cytosol. This project focused on finding microRNAs that can regulate MDR proteins by managing corresponding mRNA levels through post-transcriptional regulation based on nucleotide sequence matching. Screening was done with bioinformatics databases for unpublished/unexplored microRNAs with high nucleotide sequence correspondence to two representative MDR proteins, MGMT (a DNA repair protein) and ABCB1 (an efflux protein), revealing microRNA-4539 and microRNA-4261 respectively. To investigate the enhancement of chemotherapeutics in cancer cells, high MGMT expressing glioblastoma (T98G) and a high ABCB1 expressing triple-negative breast cancer cell line (MDA-MB-231-luc) were treated with varying concentrations of chemotherapeutics and corresponding miRNAs. Newly identified MDR-related miRNAs (MDRmiRs) enhanced the response to anti-cancer therapeutics and resulted in effective cell death. In this study, we demonstrated that therapeutic miRNAs could be identified based on the nucleotide sequence matching of miRNAs to targeted mRNA and the same approach could be employed for the screening of therapeutic candidates to regulate specific target proteins in diverse diseases.

Bioinformatics database search. Out of public, open source microRNA databases, Diana-TarBase, miRDB, miRbase, and microRNA.org were chosen for the quantitative screening of microRNAs that have complementary (matching) nucleotide sequences to the target mRNAs of chosen MDR proteins. Two major MDR proteins in cancer, O 6 -Methylguanine-DNA Methyltransferase (MGMT) and ATP Binding Cassette Subfamily B Member 1 (ABCB1), were chosen to be investigated for a complementary microRNA search [12][13][14][15] . The screening criteria of the miRNA sequences were chosen to be either matching nucleotide sequences in: (1) the 3′-UTR region in target mRNAs; or (2) the translated region sequence in target mRNAs. For this project, the thresholds for the lower boundary of effectiveness of the sequence matching were determined with a mirSVR score of −0.8 or lower, or a Target score of 80 or higher (with different score types being given in different databases). MiRNA candidates for each MDR mRNA were screened based on the screening thresholds. Each miRNA candidate was explored to confirm its relevance to cancer based on the prior literature. Finally, the miRNA hits that showed the highest score in the screened databases were selected for further investigation.
Preparation of MN-miRNA. Amine-derivatized iron oxide nanoparticles (MN-NH 2 ) were prepared from dextran-coated iron oxide nanoparticles through modification with epichlorohydrin and ammonium hydroxide. The nanoparticles were labeled with Cy5.5, a near infrared (NIR) dye, by reacting 1 ml of MN-NH 2 (120 μM, 14.4 mg Fe/ml) with 1 mg of Cy5.5-NHS ester (886 nmole, 7.5 eq.) in 100 μl DMSO. The reaction was carried out overnight at 4 °C, then the resulting Cy5.5-conjugated nanoparticles (MN-Cy5.5) were purified with a size exclusion column using nuclease free PBS buffer as an eluent. The MN-Cy5.5 was treated with excess amounts of SPDP (250 eq.) for 4 hrs at 4 °C to form MN(Cy5.5)-SPDP, which was again purified with a size exclusion column using nuclease free PBS buffer as an eluent. MC6 modified microRNAs were synthesized and provided by IDT (reference materials) and activated by TCEP to reduce disulfide bonds to thiol before conjugation. Activated miRNA-4539 and miRNA-4261 were combined with MN(Cy5.5)-SPDP to form MN-miR4539 and MN-miR4261, respectively (Fig. 1A). The molar ratio of miRNA to NP in conjugation was 10 to 1. The final products were stored in nuclease free PBS buffer at 4 °C until use.
Characterization of MN-miRNA. The amount of Cy5.5 per nanoparticle was quantified by UV absorption at 675 nm (ε = 250,000). The number of SPDPs per nanoparticle was determined by quantifying pyridine-2-thion using UV spectrometer at 343 nm (ε = 8,080) after incubation of MN(Cy5.5)-SPDP with TCEP for 30 min. The iron content of the nanoparticles was determined by the Total Iron Reagent Set (Pointe Scientific, Canton, MI) using a UV spectrometer at 410 nm. Each experiment was performed three times. The amount of microRNA on MN-miR4539 and MN-miR4261 was determined by semi-quantitative agarose gel electrophoresis. Briefly, magnetic activated cell sorting columns (MACS, Miltenyi Biotech, Auburn, CA) were loaded with 5 μL of MN-miR4539 and MN-miR4261, respectively, and washed in a magnetic column with 100 μL of nuclease free water to remove unconjugated free microRNA. Pure MN-MDRmiR was collected by eluting the column with 20 μL of nuclease-free water. The collected MN-MDRmiR was treated with TCEP for 30 min to release miRNA from the MN. The TCEP-treated MN-MDRmiR, untreated MN-MDRmiR, and miRNA (50, 100, 150, 200 pmole) standard reference solutions were loaded onto a 2% agarose gel in Tris borate-EDTA buffer (Invitrogen, Carlsbad, CA) and electrophoresed at 120 V for 90 min. After electrophoresis, the microRNA residing in the gel was stained with 0.5 μg/ml of ethidium bromide for 30 min and visualized using a Molecular Imager FX scanner (Bio-Rad Laboratories, Hercules, CA). The intensities of miRNA bands in each image were quantified using ImageJ (ver.1.45i) software (NIH). The amount of miRNA released from the nanoparticles following TCEP treatment was determined relative to reference miRNA standard solutions. Relaxivities of nanoparticles were measured using a 20 MHz NMR spectrometer (miniSpec, Bruker, Billerica, MA).
The size of the magnetic nanoparticles (MN) was measured to be 19.87 ± 0.25 nm and increased up to 23.01 ± 0.82 nm after the addition of the fluorescent dye and miRNA. The number of amine groups per MN was found to be 56 (NH 2 /MN) by quantitative analysis of SPDP per MN. The ratio of fluorescent dye (Cy5.5) per MN was 3.3 (Cy5.5/MN). The average molar ratio of miRNA per MN was measured to be 5.7 ± 0.4 (miRNA/MN) by electrophoresis. Considering the reaction ratio of 10 miRNA per MN, the mean loading efficiency was calculated as 57%. Monitoring of in vitro uptake through fluorescence imaging. MN-miRNA uptake by the human cancer cell lines MDA-MB-231 and T98G was assessed through fluorescence imaging. T98G cells (ATCC) were seeded at a density of 1 × 10 5 cells per well in a 12-well glass plate. MN-miR4261 was added into each well at various concentrations ranging from 0, 5 and 30 μM and incubated at 37 °C in a humidified 6% CO 2 atmosphere for 48 hrs. The cells were fixed in 4% paraformaldehyde and mounted on slides with Vectashield Mounting Media with DAPI for nuclear staining. Microscopic images of Cy5.5 and DAPI were obtained using a Nikon Eclipse 50i microscope, and then raw images were imported by ImageJ and processed to generate an overlayed image, showing the subcellular location of the MN-MDRmiR. Following the same procedure, MDA-MB-231 was incubated with MN-miR4539 and the cellular uptake and localization of MN-MDRmiR in the cytosol was visualized following the same procedures described above. www.nature.com/scientificreports www.nature.com/scientificreports/ Quantification of MDR protein expression. The expression of the targeted MDR protein in each cell line was quantified by Western blotting after MN-MDRmiR treatment. Briefly, T98G cells were seeded in 12-well plates (1 × 10 5 cells/well) and cultured for 24 hrs. The cells were incubated with 0, 5 and 30 μM of MN-miR4539 for 48 hrs without changing the media, then washed with HBSS. Next, the cells were harvested and treated with lysis buffer (a RIPA buffer with 100 mM EDTA, 100 mM PMSF and protease inhibitor cocktail). The amount of protein in the supernatant was quantified by a Pierce BCA assay (Thermo Scientific). Denatured protein extracts (40 μg) were loaded onto a polyacrylamide gel and proteins were separated through electrophoresis. After electrophoresis, the protein was transferred onto a polyvinylidene fluoride membrane, which was blocked in 5% non-fat milk for 1 hour. The membrane was incubated with rabbit monoclonal antibody to MGMT (dilution of 1:500) at 4 °C overnight and then labeled with secondary antibody (Goat anti-rabbit IgG (H + L), Dylight 488 pre-adsorbed, dilution of 1:400) for 1 hr at room temperature. The MGMT proteins were detected using the IVIS Spectrum imaging system (Perkin Elmer, Hopkinton, MA) and compared to reference β-actin that was detected using the same procedure. The expression of ABCB1 in MDA-MB-231 was quantified following the same procedure, but using MN-miR4539, the rabbit monoclonal antibody to ABCB1 (dilution of 1:500), and secondary antibody (Goat anti-rabbit IgG (H + L), Dylight 488 pre-adsorbed, dilution of 1:400). Statistical analysis. Data were expressed as mean ± s.d. or s.e.m., where indicated. Statistical comparisons were drawn using a two-tailed t-test (SigmaStat 3.0; Systat Software, Richmond, CA). A value of P < 0.05 was considered statistically significant.

Combination treatment with MN-MDRmiR and anti-cancer therapeutics.
To observe the potential of combined treatment with MN-MDRmiR and chemotherapy, we performed studies in which we co-administered MN-miR4539 and doxorubicin in the triple negative breast cancer cell line, MDA-MB-231 (Fig. 2D). Specifically, the concentration of doxorubicin was set as 0, 0.2, 0.4, and 0.8 μM, with the addition of varying concentrations of MN-miR4539 (0, 5 and 30 μM as miRNA). From the cell viability assay, no significant cytotoxicity was observed at concentrations of doxorubicin up to 0.8 μM. However, the cell viability was significantly affected by the co-administration of MN-miR4539. Up to 0.4 μM of doxorubicin, the cytotoxic effect was observed only with 30 μM of MN-miR4539 (p = 0.011) but not with 5 μM of MN-miR4539. In the case of 0.6 μM doxorubicin, cell viability was reduced to 84% with 5 μM MN-miR4539 (p = 0.00042) and 59% with 30 μM MN-miR4539 (p = 0.0015). When the concentration of doxorubicin was increased to 0.8 μM, the cell viability was reduced to 59% with 5 μM of MN-miR4539 (p = 0.0005) and 47% with 30 μM of MN-miR4539 (p = 0.0005) (Fig. 2D). Interestingly, with a low dose of doxorubicin, there was no significant difference in cell viability between 5uM and 30uM of MN-miR4539. This result indicated that the downregulation of MDR protein, ABCB1, was not sufficient to induce cytotoxic effects but enhanced the therapeutic efficacy of doxorubicin synergistically.
High MGMT expressing T98G glioblastoma cells were treated with varying concentrations of Temozolomide (0, 0.2, 0.4. 0.6 and 0.8 mM) (Fig. 2E). Up to 0.6 mM of Temozolomide, there was no significant change in the cell viability. However, the addition of MN-miR4261 reduced cell viability significantly, even at a Temozolomide concentration of 0.2 mM. When T98G cells were treated with 0.2 mM of Temozolomide and 5 μM of MN-miR4261, cell viability decreased to 42.8% of the Temozolomide only treatment. With the increase of concentrations, the cytotoxic effects of 30 μM MN-miR4261 showed a plateau at concentrations higher than 0.2 mM of Temozolomide. In contrast, co-administration of 5 μM MN-miR4261 induced additive effects in cell viability. Combined, these results indicated that the MN-miRNA mimics potentiated the effect of chemotherapy in both the glioblastoma and breast cancer cells through the downregulation of MDR proteins.

Discussion
In this manuscript, we describe the identification of new miRNA candidates with relevance to multi-drug resistance in chemotherapy. Two representative MDR proteins were selected based on the key roles played by these effectors in the development of drug-resistance, representing an efflux pump or a DNA repair system. Presently, the screening of miRNAs has been mainly carried out through the analysis of variations in epigenetic profiles in cancer cells. In this approach, the function of irrespective miRNA could be overlooked and hidden behind the major effector miRNAs. We hypothesized that MDRmiR could be identified for the regulation of specific MDR proteins with the criteria of endogenous short non-coding RNAs, sequence matching with target mRNA, and a higher correspondence than mirSVR < −0/8 or Target score > 80, which are expected to be effective regulators for MDR mRNA. Endogenous miRNA could reduce the unexpected immune responses or off-targeting effects compared with exogenous RNA therapeutics, including siRNAs. We chose two representative MDR proteins, ABCB1 and MGMT, and screened tentative therapeutic MDRmiRs.
With an outlook towards therapeutic application of mimics corresponding to miRNAs, we utilized iron oxide nanoparticles (MN) as an image-guided delivery vehicle. MN is an ultrasmall iron oxide nanoparticle coated with amino-dextran with a size of 19.87 nm that is delivered with high efficiency to tumor cells in vivo, because of its long circulating time and enhanced accumulation through the EPR effect. It also permits image-guidance by MRI because it affects T2 relaxation times 42,43 .
In cancer cells, MN-MDRmiR efficiently engages the RNA-Induced Silencing Complex (RISC) and inhibits the expression of complementary mRNA targets 42 . A key advantage of this delivery approach derives from the fact that MN-MDRmiR incorporates a superparamagnetic nanoparticle core whose accumulation in tissues could be monitored by quantitative noninvasive MRI. This capability is highly significant when designing and optimizing therapeutic protocols in the process of drug development. In addition, MRI could be used during treatment to determine therapeutic concentration in tissue, which could be crucial for patients whose therapy is failing for unknown reasons.
Our results demonstrated efficient uptake of MN-MDRmiR by cancer cells and perinuclear localization that is consistent with the known localization of the RNA induced silencing complex (RISC) to perinuclear sites of RNA storage and processing. Also, the perinuclear localization of MN-MDRmiR implies delivery through predominant lipid raft endocytosis rather than clathrin-mediated endocytosis 41 .
The alignment of miRNA-4539 fell within the 3′-untranslated region of ABCB1 mRNA (Supplemental Fig. 1). There was inhibition of ABCB1 by MN-miR4539 but the effect was not dose-dependent, suggesting that the inhibition of ABCB1 could be compensated at lower doses of doxorubicin by other efflux pump proteins, such as ABCC1 (MRP1) and ABCG2 (BCRP). When MN-miR4539 was co-administered with doxorubicin, there was a dose-dependent decrease in viability independent of doxorubicin concentration. Doxorubicin alone showed no effect on cell viability at low concentrations (below 0.8 μM). At a doxorubicin concentration of 0.8 μM, cytotoxicity increased by 41.3% in the presence of 5 μM MN-miRNA4539 and 53.1% in the presence of 30 μM MN-miRNA4539. This result implies that in a future therapeutic scenario, co-treatment with miRNA4539 could permit the application of lower doses of doxorubicin to potentially minimize the associated adverse effects.
Our studies in a model of glioblastoma focused on MGMT. MGMT is a member of the innate DNA repair systems employed by the cell to correct erroneous DNA methylation. It is implicated in resistance to the anti-glioma chemotherapeutic, Temozolomide, which functions by methylating DNA base to inhibit its participation in cell division or transcription. MiRNA-4261 showed a highest sequence matching to MGMT mRNA and the expected alignment of miR-4261 fell within the 3′-UTR region of MGMT mRNA (Supplemental Fig. 2). In T98G cells, MN-miR4261 localized to perinuclear regions and inhibited the expression of MGMT in a dose dependent manner. The expression of MGMT was inhibited by 48.3% at a concentration of 5 μM MN-miRNA4261 and 72.4% at a concentration of 30 μM MN-miRNA4261. Cell viability showed a dose-dependent decrease with increasing concentrations of MN-miRNA4261 at all doses of Temozolomide. The combination treatment was highly effective, even at a 0.2 mM concentration of Temozolomide. Furthermore, above 0.4 mM Temozolomide, cell viability plateaued and was not further influenced by an increase in MN-miRNA4261 concentration. This suggests the existence of other compensatory mechanisms for resistance to Temozolomide in this cell line that limit the effect of MGMT inhibition. Although there are several DNA repair systems that have been identified, MGMT is known as the key player in removing methyl groups from DNA molecules. Therefore, it has been speculated that other DNA repair systems could make up the function of MGMT partially, but not completely substitute the role of MGMT 44 .
One potential limitation of the approach derives from the possibility that the mutation in the protein sequence of ABCB1 and/or MGMT in patient samples may lead to nonfunctional proteins and lack of target engagement and therapeutic efficacy. In that case, empirical knowledge about the specific sequence of the messenger RNAs in each individual patient will be necessary in order to finetune the rational design of the MDRmiRs. This approach is made more feasible by advancement in Next Generation Sequencing.
In summary, the present study identifies new miRNA candidates for the prevention of MDR caused by efflux protein and DNA-repair protein in breast cancer and glioblastoma. In addition, it describes the design of new nanoparticle-based miRNA mimics that are therapeutically relevant and could be applied to efficiently engage the endogenous RNA interference apparatus in tumor cells, following a simple intravenous injection 42,43,45 . If implemented in the clinic, this new approach can present a significant benefit to patient health by decreasing chemotherapeutic doses, and minimizing side effects, while increasing therapeutic efficacy.