CircPAN3 mediates drug resistance in acute myeloid leukemia through the miR-153-5p/miR-183-5p–XIAP axis

•CircPAN3 mediates drug resistance in acute myeloid leukemia (AML).•The effect of circPAN3 depends on the circPAN3–miR-153-5p/miR-183-5p–X-linked inhibitor of apoptosis protein (XIAP) axis.•CircPAN3 may serve as a potential target for reversing drug resistance in AML. The contribution and role of circular RNAs (circRNAs) in mediating chemoresistance in acute myeloid leukemia (AML) are still poorly understood and need further investigation. In this study, we established a doxorubicin (ADM)-resistant THP-1 AML cell line (THP-1/ADM). A high-throughput microarray was used to identify circRNA expression profiles of THP-1/ADM cells and naive THP-1 cells. The identified potential functional circRNA molecule was further validated in THP-1/ADM cells and bone marrow (BM) specimens from 42 AML patients. The interactions with target microRNAs (miRNAs) and downstream messenger RNAs (mRNAs) were also explored. As a result, 49 circRNAs that are significantly differentially expressed between THP-1/ADM and THP-1 cells were identified. Of these circRNAs, downregulation of circPAN3 by small interfering RNA significantly restored ADM sensitivity of THP-1/ADM cells. Furthermore, BM samples from patients with refractory and recurrent AML showed increased expression of circPAN3. A detailed circRNA/miRNA/mRNA interaction network was predicated for this circRNA. Subsequent mechanistic experiments showed that downregulation of circPAN3 could decrease the expression of X-linked inhibitor of apoptosis protein (XIAP), but this effect was counteracted by miR-153-3p or miR-183-5p specific inhibitors. Luciferase experiments further demonstrated that these molecules are involved in the circPAN3 regulatory network. Our results revealed that circPAN3 may be a key mediator for chemoresistance of AML cells, which may depend on the circPAN3–miR-153-5p/miR-183-5p–XIAP axis. Our findings provide evidence that circPAN3 can be a valuable indicator for predicting clinical efficacy of chemotherapy in AML patients and also can serve as a potential target for reversing drug resistance in AML. The contribution and role of circular RNAs (circRNAs) in mediating chemoresistance in acute myeloid leukemia (AML) are still poorly understood and need further investigation. In this study, we established a doxorubicin (ADM)-resistant THP-1 AML cell line (THP-1/ADM). A high-throughput microarray was used to identify circRNA expression profiles of THP-1/ADM cells and naive THP-1 cells. The identified potential functional circRNA molecule was further validated in THP-1/ADM cells and bone marrow (BM) specimens from 42 AML patients. The interactions with target microRNAs (miRNAs) and downstream messenger RNAs (mRNAs) were also explored. As a result, 49 circRNAs that are significantly differentially expressed between THP-1/ADM and THP-1 cells were identified. Of these circRNAs, downregulation of circPAN3 by small interfering RNA significantly restored ADM sensitivity of THP-1/ADM cells. Furthermore, BM samples from patients with refractory and recurrent AML showed increased expression of circPAN3. A detailed circRNA/miRNA/mRNA interaction network was predicated for this circRNA. Subsequent mechanistic experiments showed that downregulation of circPAN3 could decrease the expression of X-linked inhibitor of apoptosis protein (XIAP), but this effect was counteracted by miR-153-3p or miR-183-5p specific inhibitors. Luciferase experiments further demonstrated that these molecules are involved in the circPAN3 regulatory network. Our results revealed that circPAN3 may be a key mediator for chemoresistance of AML cells, which may depend on the circPAN3–miR-153-5p/miR-183-5p–XIAP axis. Our findings provide evidence that circPAN3 can be a valuable indicator for predicting clinical efficacy of chemotherapy in AML patients and also can serve as a potential target for reversing drug resistance in AML. Acute myeloid leukemia (AML) is one of the most common hematological malignancies [1Siegel R Ma J Zou Z Jemal A Cancer statistics, 2014.CA Cancer J Clin. 2014; 64: 9-29Crossref PubMed Scopus (10810) Google Scholar, 2DeSantis CE Siegel RL Sauer AG et al.Cancer statistics for African Americans, 2016: Progress and opportunities in reducing racial disparities.CA Cancer J Clin. 2016; 66: 290-308Crossref PubMed Scopus (513) Google Scholar]. Despite the application of new molecular targeted drugs and progress of allogeneic hematopoietic stem cell transplantation, chemoradiotherapy is still the mainstay for the treatment of AML. However, AML cells are demonstrated to unavoidably develop primary or secondary chemoresistance, thereby resulting in refractory and recurrent disease in patients. So far, most clinical trials of chemotherapy have shown very limited benefits for refractory and recurrent AML [3Orlowski RJ Mangan JK Luger SM Approach to patients with primary refractory acute myeloid leukemia.Curr Opin Hematol. 2015; 22: 97-107Crossref Scopus (11) Google Scholar]. Therefore, it is necessary to identify the potential molecular targets and novel pathways underlying the occurrence and development of AML drug resistance. Development of drug resistance in cancer cells involves interactions among multiple genes and signaling pathways. Currently, the mechanisms of drug resistance in AML include ATP-binding cassette (ABC) transporter-mediated multidrug resistance, apoptosis tolerance, FLT3 mutation, DNA repair abnormalities, and the bone marrow (BM) microenvironment [4Du Y Chen B Detection approaches for multidrug resistance genes of leukemia.Drug Des Devel Ther. 2017; 11: 1255-1261Crossref Scopus (14) Google Scholar]. In addition to these mechanisms, growing evidence shows that noncoding RNAs (ncRNAs) are associated with drug resistance in some solid tumors [5Majidinia M Yousefi B Long non-coding RNAs in cancer drug resistance development.DNA Repair (Amst). 2016; 45: 25-33Crossref PubMed Scopus (100) Google Scholar], leading to more attention being focused on targeting ncRNAs as a new strategy for overcoming multidrug resistance of cancer. ncRNAs, generally classified as transcribed ultraconserved regions, microRNAs (miRNAs), small nucleolar RNAs, PIWI-interacting RNAs, long noncoding RNAs, and circular RNAs (circRNAs) account for almost 95% of the total RNAs transcribed from eukaryotic genomes and are being increasingly recognized to function in gene regulation and to contribute to the pathogenesis of various human disorders [6Warner JR The economics of ribosome biosynthesis in yeast.Trends Biochem Sci. 1999; 24: 437-440Abstract Full Text Full Text PDF PubMed Scopus (1363) Google Scholar]. As a large proportion of the ncRNA family, circRNAs have attracted growing interest in recent years. circRNAs have been recognized as a class of molecules with variable biological functions. They can compete for the pool of the same miRNA response elements (MREs) to influence the activities of miRNAs in the regulation of gene expression [7Salmena L Poliseno L Tay Y Kats L Pandolfi PP A ceRNA hypothesis: the Rosetta Stone of a hidden RNA language?.Cell. 2011; 146: 353-358Abstract Full Text Full Text PDF PubMed Scopus (4369) Google Scholar]. With the structure of covalently closed continuous loop without 5′ to 3′ polarity and polyadenylated tail, circRNAs show a higher stability than linear transcripts and are not affected by ribonuclease R or RNA exonuclease [8Suzuki H Tsukahara T A view of pre-mRNA splicing from RNase R resistant RNAs.Int J Mol Sci. 2014; 15: 9331-9342Crossref PubMed Scopus (292) Google Scholar]. Furthermore, circRNAs exhibit evolutionary conservation as well as tissue and developmental stage specificity [9Salzman J Chen RE Olsen MN Wang PL Brown PO Cell-type specific features of circular RNA expression.PLoS Genet. 2013; 9e1003777Crossref PubMed Scopus (1290) Google Scholar, 10Memczak S Jens M Elefsinioti A et al.Circular RNAs are a large class of animal RNAs with regulatory potency.Nature. 2013; 495: 333-338Crossref PubMed Scopus (4693) Google Scholar], thereby suggesting their important role in mediating gene expression variability. With the development of high-throughput sequencing, aberrant expression of circRNAs has been found in a wide range of solid tumors [11Hansen TB Kjems J Damgaard CK Circular RNA and miR-7 in cancer.Cancer Res. 2013; 73: 5609-5612Crossref PubMed Scopus (722) Google Scholar, 12Bachmayr-Heyda A Reiner AT Auer K et al.Correlation of circular RNA abundance with proliferation–exemplified with colorectal and ovarian cancer, idiopathic lung fibrosis, and normal human tissues.Sci Rep. 2015; 5: 8057Crossref PubMed Scopus (539) Google Scholar, 13Wang X Zhang Y Huang L et al.Decreased expression of hsa_circ_001988 in colorectal cancer and its clinical significances.Int J Clin Exp Pathol. 2015; 8: 16020-16025PubMed Google Scholar, 14Li P Chen S Chen H et al.Using circular RNA as a novel type of biomarker in the screening of gastric cancer.Clin Chim Acta. 2015; 444: 132-136Crossref PubMed Scopus (645) Google Scholar, 15Su H Lin F Deng X et al.Profiling and bioinformatics analyses reveal differential circular RNA expression in radioresistant esophageal cancer cells.J Transl Med. 2016; 14: 225Crossref PubMed Scopus (152) Google Scholar, 16Xie H Ren X Xin S et al.Emerging roles of circRNA_001569 targeting miR-145 in the proliferation and invasion of colorectal cancer.Oncotarget. 2016; 7: 26680-26691Crossref PubMed Scopus (365) Google Scholar] and is associated with tumor differentiation, invasion, and distant metastasis [13Wang X Zhang Y Huang L et al.Decreased expression of hsa_circ_001988 in colorectal cancer and its clinical significances.Int J Clin Exp Pathol. 2015; 8: 16020-16025PubMed Google Scholar, 14Li P Chen S Chen H et al.Using circular RNA as a novel type of biomarker in the screening of gastric cancer.Clin Chim Acta. 2015; 444: 132-136Crossref PubMed Scopus (645) Google Scholar]. In some solid tumors, circRNAs have been reported to be involved in the development of chemo- and radioresistance of tumor cells [15Su H Lin F Deng X et al.Profiling and bioinformatics analyses reveal differential circular RNA expression in radioresistant esophageal cancer cells.J Transl Med. 2016; 14: 225Crossref PubMed Scopus (152) Google Scholar, 17Gao D Zhang X Liu B et al.Screening circular RNA related to chemotherapeutic resistance in breast cancer.Epigenomics. 2017; 9: 1175-1188Crossref PubMed Scopus (113) Google Scholar]. However, the functional role of circRNAs in chemoresistance of AML and the underlying mechanism have been studied only rarely. In this study, using high-throughput circRNA microarray, we compared the expression profile of circRNAs in chemosensitive and chemoresistant AML cells. An important circRNA molecule, circPAN3, was identified to play a crucial role in driving drug resistance of AML and the underlying interaction network and downstream functional components were also investigated for this molecule. THP-1 human AML cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (Thermo Fisher Scientific Inc., Waltham, MA, USA). A doxorubicin (ADM)-resistant cell line (THP-1/ADM) was established by exposing THP-1 cells to a gradually increasing concentration (0.2–2 µg/mL) of ADM for 6 months. The untreated parental cell line was used as a control. BM samples were obtained from 42 AML patients who were hospitalized at Fujian Provincial Hospital (Fuzhou, China) between May 2015 and May 2018. Briefly, mononuclear cells were isolated from BM aspirates by density gradient centrifugation using Ficoll-Paque Plus density gradient media (Pharmacia LKB Biotechnology, Piscataway, NY, USA). Two million cells were collected from each sample for further examination. The patient population consisted of 22 males and 20 females with a median age of 31. Of these patients, 22 (52.4%) had newly diagnosed AML and responded well to chemotherapy, whereas the rest (n = 20, 47.6%) had refractory/recurrent AML and responded poorly to chemotherapy. Pathological diagnosis and confirmation of recurrence and refractoriness of AML were defined according to the published criteria [18O'Donnell MR Tallman MS Abboud CN et al.Acute Myeloid Leukemia, Version 3.2017, NCCN Clinical Practice Guidelines in Oncology.J Natl Compr Canc Netw. 2017; 15: 926-957Crossref PubMed Scopus (272) Google Scholar]. The study was conducted in accordance with the Declaration of Helsinki and the protocol was approved by the ethics committee of Fujian Provincial Hospital. All patients provided written informed consent before participation. Cell Count Kit-8 (CCK-8) reagent (Vazyme Biotech, Nanjing, China) was used to assess cell sensitivity to ADM. THP-1/ADM and THP-1 cells were plated in 96-well plates at a density of 1 × 104 cells/mL and then treated with 200 μL of medium containing various concentrations of ADM (0.01–2.4 µg/mL) for 24 hours. Next, 10 μL of CCK-8 reagent was added to cultured cells, and incubated in a humidified incubator containing 5% CO2 at 37°C for 2 hours. Absorbance was detected at a wavelength of 450 nm. The half-maximal inhibitory concentration (IC50) and inhibitory ratio values were calculated from the concentration–response curve generated for each cell line. A circRNA chip (Arraystar Human Circular RNA Microarray, ArrayStar, Rockville, MD, USA) containing 13617 probes specific for human circRNA splicing sites and Agilent Scanner G2505C (Agilent Technologies Inc., Santa Clara, CA, USA) were used for analyzing circRNA expression profiles of THP-1 and THP-1/ADM cell lines. Specifically, three pairs of samples (THP-1 and THP-1/ADM cells) were analyzed on the chips following hybridization and washing. Exogenous RNAs developed by the External RNA Controls Consortium (ERCC, Stanford, CA, USA) were used as controls. Scanned images were imported into Agilent Feature Extraction software (version 11.0, Agilent Technologies Inc.) for raw data extraction. Quantile normalization and subsequent data processing were performed using the R software package. Those with fold change ≥2.0 and a p value < 0.05 were selected as differentially expressed circRNAs with statistical significance. Total RNA was extracted from cell lines and BM tissue samples using TRIzol reagent and reversely transcribed into cDNA using MLV RTase cDNA Synthesis Kit (Takara Bio Inc., Kusatsu, Japan). Quantitative real-time polymerase chain reaction (qRT-PCR) was conducted with a LightCycler 480II RT-PCR system (Roche Applied Science, Indianapolis, IN, USA) according to the manufacturer's instructions. The reaction condition was 45 cycles of denaturation at 95°C for 15 sec, annealing at 60°C for 30 sec, and extension at 72°C for 10 sec. U6 RNA served as internal control. The circRNA sequences were obtained from circBase (www.circbase.org). The primers used in this study were listed in Table 1. The delta-delta cycle threshold (ΔΔCt) method was used for quantification [19Schmittgen TD Livak KJ Analyzing real-time PCR data by the comparative C(T) method.Nat Protoc. 2008; 3: 1101-1108Crossref PubMed Scopus (16258) Google Scholar].Table 1Primers used for qRT-PCR of the top 10 upregulated circRNAscircRNAGene SymbolForward Primer Sequence (5′–3′)Reverse Primer Sequence (5′–3′)hsa_circ_0100989GPC5CAGCCTCCCTCAGCACTACTTTATGAGCGTTTTCCCTCTGAhsa_circ_0102446GPHNCAGCCTCCCTCAGCACTACTTTATGAGCGTTTTCCCTCTGAhsa_circ_0100181PAN3GATTTCCATGCTGGAGGAGAGGAATGAAGAGGGGAAGACChsa_circ_0101868PRPF39ATGAGCAGGGAAACCTGAGATCTGAACTGTTGCCTGTGCThsa_circ_0104204CSNK1G1ACTCCAATGCACCAATCACAGTTGCCATCTCCTCTGGAAAhsa_circ_0001264RAD18CAGCTCATTAAAAGGCACCAGGAAGAAGCAGGAGATTTGGhsa_circ_0102888PP4R3AATGGATCCCGCTACAATCTGTGCCCATGTTTGACAAAGTChsa_circ_0100759DIAPH3CATCTTCCTGATCAAGAGCAATTAGGGACAAAGGATCAATGGAAhsa_circ_0101113FARP1GAGGAAGAGGAGGAGGTCGTTGCGTCTTGCTGAGGTATGThsa_circ_0100891MYCBP2AATTGGACTCCAGGGGCTATTGGTAAATTTCCCTCAAGTGC Open table in a new tab The small interfering RNA (siRNA) targeting the junction region of circPAN3 sequence (siRNA sequence: 5′-UCU GAC CCA AAA CAA CCC CdTdT-3′) and negative control siRNA (NC), as well as the mimics and inhibitors of miR-153-5p and miR-183-5p, were provided by Geneseed Bio-Tech (Guangzhou, China). Hieff Trans Liposomal Transfection Reagent (Yeasen Bio, Shanghai, China) was used for transfection. Briefly, THP-1 and THP-1/ADM cells, as well as leukemic cells from five relapsed AML patients, were cultured in six-well plates and transfected with the aforementioned siRNAs or miRNA mimics or inhibitors (10 nmol/well) for 6 h. The transfected cells were cultured in regular medium for 48 hours and then collected for subsequent experiments. The circRNA–miRNA interaction was predicted using Arraystar's homemade miRNA target prediction software based on TargetScan and miRanda. Cytoscape was applied to build a circRNA–miRNA–messenger RNA (mRNA)/gene interaction network. The predicted gene functions in the network were annotated using Gene Oncology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Protein lysates from cell lines and BM tissue samples were subjected to Western blot analysis using the standard protocol as described previously [20Ohmura M Hishiki T Yamamoto T et al.Impacts of CD44 knockdown in cancer cells on tumor and host metabolic systems revealed by quantitative imaging mass spectrometry.Nitric Oxide. 2015; 46: 102-113Crossref PubMed Scopus (17) Google Scholar]. Antibodies against X-linked inhibitor of apoptosis protein (XIAP) (catalog #AF6368, 1:500), pro-caspase 3 (pro-Cas-3, catalog #AF6311, 1:500), cleaved caspase 3 (CL Cas-3, catalog #AF7022, 1:500), pro-caspase 9 (pro-Cas-9, catalog #AF6348, 1:500), and cleaved-caspase 9 (CL Cas-9, catalog #AF5240, 1:500) were obtained from Affinity Biosciences (Changzhou, China). Antibody against β-actin (Affinity Biosciences, catalog #AF7018, 1:1000) was used as a loading control. Proteins were detected using Pierce ECL Western Blotting substrate (Thermo Fisher Scientific Inc.). The bands were quantified using Image-Pro Plus 6.0 software (Media Cybernetics, Rockville, MD, USA). The wild-type (WT) and mutant sequences of circPAN3 and the 3′-untranslated region (3′-UTR) of XIAP were obtained through whole-genome synthesis. For the mutant sequences of circPAN3 and XIAP 3′-UTR, the putative binding sites for miR-153-5p and miR-183-5p were mutated (Supplementary Figures E1A and E1B, online only, available at www.exphem.org). All sequences were cloned into the XhoI and NotI sites of psiCHECK2 vector (Geneseed Bio-Tech) to construct the corresponding luciferase reporters (psiCHECK2/circPAN3, psiCHECK2/circPAN3-mut153, and psiCHECK2/circPAN3-mut183; psiCHECK2/XIAP 3′-UTR, psiCHECK2/XIAP 3′-UTR-mut153, and psiCHECK2/XIAP 3′-UTR-mut183) for transfection in 293T cells. Briefly, 3 × 104 293T cells were seeded in 24-well plates at 60% confluence level in triplicate and cotransfected with each of the indicated luciferase reporters (0.5 μg) and either miRNA mimics or NC. Forty-eight hours after cotransfection, the luciferase reporter assay was conducted using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) according to the manufacturer's instructions. Relative luciferase activity was normalized to the Renilla luciferase internal control. The Wilcoxon signed-rank test was performed to analyze significant differences regarding the expression level of circRNAs and miRNAs between samples. Pearson's analysis was applied to determine correlation coefficients for different variables. Other data are presented as means ± standard deviation (SD) and the Student t test (two-tailed) was performed for data analysis. p < 0.05 was considered to be significant. An ADM-resistant THP-1 cell line (THP-1/ADM) was established by continuous exposure to increasing concentrations of ADM for 6 months. To verify resistance, both drug-naive and drug-resistant THP-1 cells were treated with various concentrations of ADM for 24 hours and then the IC50 values were determined. As depicted in Figure 1, naive THP-1 cells had an IC50 value of 0.15 ± 0.05 μg/mL in response to ADM, whereas THP-1/ADM cells had an IC50 value of 2.62 ± 0.83 μg/mL, which was 17.5-fold higher than that of naive THP-1 cells. These results demonstrate the resistance to ADM in THP-1/ADM cell line. Next, the circRNA expression profiles of naive and ADM-resistant THP-1 cells were analyzed using Arraystar Human circRNA Array. After normalization, a total of 4573 circRNA targets were identified in three pairs of THP-1 and THP-1/ADM cell samples. Differentially expressed circRNAs were displayed through fold change filtering (Figure 2A) and those with a statistically significant change in expression between the two groups were identified by a volcano plot (Figure 2B). As a result, 49 distinct circRNAs were found to be differentially expressed in the two cell lines with a fold change >2 and a p value < 0.05. Of these circRNAs, 35 were upregulated and 14 were downregulated in THP-1/ADM cells. The top 10 upregulated and downregulated circRNAs are listed in Table 2. In addition, the heat map of the top 50 upregulated and downregulated circRNAs further revealed distinctive circRNA expression pattern between THP-1 and THP-1/ADM cells (Figure 2C). Taken together, these data indicate that the expression of circRNAs in ADM-resistant AML cells is considerably different from that in naive cells, suggesting a potential role of circRNAs in conveying drug resistance in AML cells.Table 2Top 10 upregulated and downregulated circRNAs between THP-1/ADM and THP-1 cells screened by microarrayCircRNA IDChromosomep valueFold ChangeBest TranscriptGene SymbolTop 10 upregulated circRNAshsa_circ_0100989Chr210.0000000495.522634NM_004466.5GPC5hsa_circ_0102446Chr140.0000000625.095632NM_020806GPHNhsa_circ_0100181Chr130.0000243004.483257NM_175854PAN3hsa_circ_0101868Chr140.0001290004.239404NM_017922.3PRPF39hsa_circ_0104204Chr150.0000005974.219353NM_022048CSNK1G1hsa_circ_0001264Chr30.0000011004.086172NM_020165RAD18hsa_circ_0102888Chr140.0082900003.993946NM_001284281.1PP4R3Ahsa_circ_0100759Chr130.0003520003.695359NM_001042517.1DIAPH3hsa_circ_0101113Chr130.0000062003.635144NM_001001715.3FARP1hsa_circ_0100891Chr130.0009850003.584472NM_015057MYCBP2Top 10 downregulated circRNAshsa_circ_0001819Chr80.00064–3.82166NM_015902UBR5hsa_circ_0001622Chr60.00078–3.79373NM_001137668CASP8AP2hsa_circ_0103108Chr150.00109–3.75534NR_027992NBEAP1hsa_circ_0100888Chr130.00313–3.6837NM_015057MYCBP2hsa_circ_0100310Chr130.00341–3.35805NM_015032PDS5Bhsa_circ_0101416Chr140.01940–3.25243NR_027457LINC00221hsa_circ_0104676Chr150.01430–3.1281NM_015629PRPF31hsa_circ_0104613Chr150.00461–2.82206XM_011522038.1PEAK1hsa_circ_0101922Chr140.00784–2.71451NM_172193.2KLHDC1hsa_circ_0100926Chr130.00406–2.52645NM_001242868SLAIN1CircRNA ID was based on circBase (http://www.circbase.org/). p value was calculated from paired t test. Fold change represents the ratio (log scale) of normalized intensities between two cell lines. Open table in a new tab CircRNA ID was based on circBase (http://www.circbase.org/). p value was calculated from paired t test. Fold change represents the ratio (log scale) of normalized intensities between two cell lines. To validate the microarray analysis, we performed qRT-PCR for the top 10 upregulated circRNAs in three pairs of THP-1 and THP-1/ADM cells. As shown in Figures 2D and 2E, the log2 fold changes calculated for microarray data were highly correlated with those for qRT-PCR data (Pearson's r = 0.682, p = 0.029), thereby indicating overall consistency between microarray and qRT-PCR measurements. Based on Table 2, we found that circPAN3 is in the top 10 upregulated target circRNAs. It has been reported that the gene PAN3 is considered a candidate gene located in amplified chromosome regions in AML patients with a complex karyotype [21Mrózek K Cytogenetic, molecular genetic, and clinical characteristics of acute myeloid leukemia with a complex karyotype.Semin Oncol. 2008; 35: 365-377Crossref PubMed Scopus (117) Google Scholar], whereas the findings of our circRNA expression analysis indicate that circPAN3 may be involved in the resistance to ADM of THP-1 cells. To verify this, we downregulated the expression of circPAN3 in THP-1/ADM cells using a targeted siRNA and then examined the sensitivity of the treated cells to ADM. As a result, we found that downregulation of circPAN3 by siRNA considerably increased the sensitivity of THP-1/ADM cells to ADM compared with the scrambled negative control (IC50 = 0.32 ± 0.06 μg/mL and 2.41 ± 0.23 μg/mL in the si-circPAN3 and si-NC groups, respectively) (Figure 3A), thereby suggesting a role of circPAN3 in the development of drug resistance in AML cells. For further validation, we examined the expression of circPAN3 in BM samples from refractory/recurrent AML patients (n = 20) and those sensitive to chemotherapy (n = 22). As shown in Figure 3B, refractory/recurrent AML patients had a significantly higher expression level of circPAN3 in their BM than chemosensitive patients did (p < 0.01), which is further evidence that circPAN3 is most likely to contribute to postchemotherapy recurrence in AML patients. circRNAs usually act as an miRNA sponge and regulate their own circRNA–miRNA–mRNA networks [22Kulcheski FR Christoff AP Margis R Circular RNAs are miRNA sponges and can be used as a new class of biomarker.J Biotechnol. 2016; 238: 42-51Crossref PubMed Scopus (517) Google Scholar]. To identify the circRNA–miRNA interaction network for circPAN3, we first analyzed the MREs associated with circPAN3 and predicted potential target miRNAs through TargetScan and miRanda. Then, we ranked the predicted target miRNAs of circPAN3 according to their mirSVR scores and the top 10 candidates (miR-153-5p, miR-183-5p, miR-338-3p, miR-346, miR-545-3p, miR-574-5p, miR-599, miR-653-5p, miR-766-3p, and miR-767-3p) were selected for further analysis. In brief, the bioinformatics tool DIANA was used to predict the target genes of these miRNAs, followed by function analysis of the predicted target genes by GO and KEGG. GO enrichment analysis revealed that these predicted target genes are involved in a wide range of cellular activities, such as cytosol, protein complex, biosynthetic process, and cellular nitrogen compound metabolic process (Supplementary Figure E2A, online only, available at www.exphem.org), whereas KEGG analysis showed that a few target genes are involved in several important signaling pathways associated with cancer development and progression such as the transforming growth factor-beta, mammalian target of rapamycin, ErbB, and FoxO signaling pathways (Supplementary Figure E2B, online only, available at www.exphem.org). Then, we focused on the KEGG Pathways in Cancer, an important tumor-related pathway in KEGG database. It involved all 10 target miRNAs and 138 target genes of circPAN3 and had a significant p value of 0.01, suggesting an essential role of circPAN3 in cancer-related pathways for AML with the greatest potential. Next, we employed Cytoscape to construct the circRNA–miRNA–mRNA interaction network of circPAN3 within the KEGG Pathways in Cancer (Supplementary Figure E2C, online only, available at www.exphem.org). In the network shown in Supplementary Figure E2C (online only, available at www.exphem.org), XIAP was found to be targeted by both miR-153-5p and miR-183-5p, suggesting that it might be a crucial factor mediated by circPAN3. Overexpression of XIAP in AML has been demonstrated to lead to chemoresistance [23Shaffer BC Gillet JP Patel C Baer MR Bates SE Gottesman MM Drug resistance: still a daunting challenge to the successful treatment of AML.Drug Resist Updat. 2012; 15: 62-69Crossref PubMed Scopus (194) Google Scholar], whereas its downregulation induces activation of caspases and apoptosis of leukemic cells [24Prabhu KS Siveen KS Kuttikrishnan S et al.Targeting of X-linked inhibitor of apoptosis protein and PI3-kinase/AKT signaling by embelin suppresses growth of leukemic cells.PLoS One. 2017; 12e0180895Crossref Scopus (24) Google Scholar]. Considering the functions of XIAP and its connection with circPAN3, we therefore hypothesized that the circPAN3–miR-153-5p/miR-183-5p–XIAP axis may contribute to circPAN3-medicated ADM resistance in THP-1/ADM cells. To validate our hypothesis, we first detected the mRNA and protein levels of XIAP in THP-1/ADM and THP-1 cell lines by qRT-PCR and Western blot. As shown in Figures 4A and 4B, the mRNA and protein levels of XIAP were significantly increased in ADM-resistant cells compared with naive cells. Furthermore, we examined the mRNA and protein expression of XIAP in BM samples from 20 refractory/recurrent AML patients and 22 chemosensitive AML patients. The results were consistent with those of the cell experiments, showing that the mRNA and protein levels of XIAP were dramatically higher in samples from refractory/recurrent patients than in those from chemosensitive patients (Figures 4C and 4D). Collectively, these data showed a significant association between XIAP and chemoresistance of AML. Then, we treated the cells using circPAN3-specifc siRNA to downregulate the expression of circPAN3, which decreased by about 70% as validated by qRT-PCR. We found that the protein level of XIAP was also significantly decreased compared with the mock control (Figure 4F), but this was not the case for XIAP mRNA (Figure 4E). These findings indicated a posttranscriptional regulation mechanism of circPAN3 on XIAP. We also examined the clinical samples from all refractory/recurrent and chemosensitive AML patients and observed a significant positive correlation between circPAN3 level and XIAP protein level (Pearson's r = 0.543, p = 0.0002) (Figure 4G). These results, together with the data shown in Figure 3A, suggest that XIA