circCRKL, a circRNA derived from CRKL, regulates BCR-ABL via sponging miR-877-5p to promote chronic myeloid leukemia cell proliferation

Human chronic myeloid leukemia cell lines (K562, K562/G01, and KCL22), acute monocytic leukemia cell line (THP-1), human lymphoblastic cell line (TK6), and 293 T were all maintained in Roswell Park Memorial Institut-1640 (RPMI-1640, Gibco, USA) or Dulbecco's modified eagle medium (DMEM, Gibco, USA) with 10% fetal bovine serum (FBS, Gibco, USA) respectively. All cells were cultivated at 37 °C in a suitable atmosphere containing 5% CO₂. The Chinese Academy of Science's Cell Bank provided all of the cell lines utilized in this work.

The bone marrow samples of newly diagnosed CML patients and normal donors were obtained from the Department of Hematology, the Second Affiliated Hospital of Chongqing Medical University. The information was illustrated in Additional file 3: Table S1. The Human bone marrow mononuclear cells (BMMCs) separation Kit (TBD, China) was used to isolate BMMCs from the samples. Before the trial began, all patients signed informed permission forms. The ethics committee of Chongqing Medical University gave its approval to this investigation.

TRizol reagent (Accurate, China) was used to extract total RNA from patient samples and cultured cells. To reverse transcribe circRNA and messenger RNA (mRNA) into complementary DNA (cDNA), a reverse transcription kit (Accurate, China) was employed. The stem-loop primer was used to reverse transcribe miR-877-5p. The quantitative PCR was carried out through SYBR Green Premix Pro Taq HS qPCR Kit (Accurate, China) following the product’s instructions. Actin or U6 was used as RT-qPCR internal parameters. The 2^−ΔΔCT method was utilized to examine the relative expression of RNA. All specific primers sequences used in this investigation were listed in Additional file 3: Table S2.

The 1 μg of total RNA extraction of K562, and K562/G01 cells were digested at 37 °C for 10 min after adding 2U/mg of RNase R (Geneseed, China), then the mixture was hatched at 70 °C for another 10 min to inactive the RNase R. Subsequently, the relative RNA enrichment was detected through RT-qPCR.

The 5 μg/mL actinomycin D was mixed with K562 and K562/G01 cells and reacted for 4, 8, and 12 h respectively, the total RNA was then extracted. The relative RNA expression was then identified using RT-qPCR.

The specific circCRKL probe was labeled with cy3 and synthesized by RiboBio (Guangzhou, China). The assay was conducted in accordance with the manufacturer’s orders. In brief, K562 and K562/G01 cells were washed and fixed with 4% paraformaldehyde at room temperature (RT) for 10 min before being permeabilized with 0.5% Triton X-100 at 4 °C for 5 min. Then circCRKL probe was hybridized with cells in a dark moist chamber at 37 °C overnight. To stain the nuclei, 4′,6-diamidino-2-phenylindole (DAPI) was used. A confocal microscope was utilized to acquire photos.

For circCRKL down-regulation, small interfering oligos targeting circCRKL (the sequences are illustrated in Additional file 3: Table S3) were designed and constructed into lentivirus (GeneChem, China), which were subsequently infected into BCR-ABL⁺ and THP-1 cells. Stable transfection was established through culturing cells in RPMI-1640 medium supplemented with 2 μg/ml puromycin (Solarbio, China) for 7 days, and maintained with 1 µg/ml of that for another 7 days.

The cell viability was determined using a CCK-8 kit (TargetMol, China). Cells were seeded at the number of 3000 per well in a 96-well plate and cultured in an incubator for 24 h before receiving treatment of 10 μl CCK-8 reagent for 3 h. Subsequently, the absorbance was assessed at 450 nm. Three duplicates were set in each group.

Cells were seeded at a density of 100 per well into a 96-well plate. After being stilling cultured using RMPI-1640 with 10% FBS for 7 days, photos were taken and the number of colonies was counted. At least three duplicates were set in each group.

For cell cycle assay, at least 1 × 10⁶ infected K562 and K562/G01 cells were immobilized through pre-cooled 70% ethanol at 4 °C overnight. After that, each sample was left in a lucifugal chamber at RT for 30 min after receiving 500 μl of propidium iodide (PI) staining reagent. Then cell cycle was detected and analyzed through a flow cytometer (CytoFlex, Beckman coulter).

For cell apoptosis assay, at least 5 × 10⁵ infected K562 and K562/G01 cells were collected and washed, then cells were suspended with 100 μl pre-cooled PBS and treated with 10 μl Annexin V APC and 5 μl DAPI in the dark at RT for 15 min before being determined with a flow cytometer (CytoFlex, Beckman coulter).

K562/G01 cells were used to conducted RIP assay with the BersinBio™ RNA-Binding Protein Immunoprecipitation Kit (BersinBio, China) in accordance with the manufacturer’s protocols. In a nutshell, at least 2 × 10⁷ cells were harvested, washed, and lysed using 900 μl RIP lysis buffer, then cell lysates were added with 4 μg of anti-AGO2 (Abcam, USA) or the equal amounts of anti-IgG (BersinBio) and incubated at 4 °C overnight. The mixture was subsequently added with 40 μl protein A/G magnetic beads and incubated at 4 °C for 2 h. Finally, the RNAs were extracted and purified using phenol–chloroform and the relative abundance of miR-877-5p and circCRKL was examined through RT-qPCR.

To lyse cells, a radio immunoprecipitation assay (RIPA) lysing solution including phosphatase and protease inhibitor was employed, and total protein was then extracted and quantified through a bicinchoninic acid (BCA) protein extraction assay kit (Beyotime, China). For western blot assay, 40 μg of protein samples were loaded and isolated with 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Subsequently, blocking the polyvinylidene fluoride (PVDF) membranes containing protein blots with 5% nonfat milk powder at RT for 2 h before incubating the primary antibody (1:1000) at 4 °C overnight. After that, the membranes were incubated with relevant secondary antibodies (1:5000) at RT for 90 min. Finally, the Bio-Rad Gel Imaging System was used to visualize and analyze the signals.

For the prediction of target miRNAs directly cooperating with circCRKL, three online databases including ENCORI (http://​starbase.​sysu.​edu.​cn/​), circInteractome (https://​circinteractome.​nia.​nih.​gov/​), and circBank (http://​www.​circbank.​cn/​) were employed. Another two databases TargetScan (http://​www.​targetscan.​org/​vert_​72/​), miRwalk (http://​mirwalk.​umm.​uni-heidelberg.​de/​), and ENCORI were utilized to search the miRNAs interacting with ABL. The overlapping target miRNAs were analyzed and visualized using VennDiagram tools.

The specific probe targeting the back-splicing site of circCRKL and NC oligo probe was coupled by biotin and purchased from RiboBio (Guangzhou, China). The probe sequences were demonstrated in Additional file 3: Table S4. Briefly, approximately 4 × 10⁷ K562/G01 cells were harvested and fixed using 1% formaldehyde, then the cells were lysed, sonicated, and centrifuged at 16,000g at 4 °C for 15 min. The biotin-coupled circCRKL probe or NC oligo probe was incubated with the supernatant obtained from lysates at RT for 4 h. After that, streptavidin-coated magnetic beads (MCE, USA) were utilized to pull down the biotin-labeled RNA mixture. Finally, with the elution of the beads, the RNA was obtained using TRizol, and the relative abundance of circCRKL and miR-877-5p was evaluated with RT-qPCR.

Female NOD/SCID mice aged 5 to 6 weeks were irradiated with an X-ray at a dose of 250 cGy before injection. K562/G01 cells were collected and washed with sterile PBS at a concentration of 2.5 × 10⁷/ml after being infected with sh-NC or sh-circCRKL lentivirus. The mice were then split into two groups at random and intravenously injected with 200 μl cells, respectively. The body weights of each mouse were monitored weekly. The number of white blood cells in the peripheral blood was measured three weeks after injection. The bone marrow cells, livers, and spleens were harvested while mice were closely culled at an ethical endpoint. The animal experiments were given approval by the Ethics Committee of Chongqing Medical University.

GraphPad Prism 8 (GraphPad, USA) was used for analyzing data in this research. The student’s t-test was applied for calculating the main effect in two groups and One-Way ANOVA was used for that in three and more. p-values < 0.05 (*), p < 0.01 (**), p < 0.001 (***) and p < 0.0001 (****) were regarded as statistically significant.

To uncover the hidden role of circCRKL in CML, the circCRKL expression in BMMCs from CML patients (6 cases) and normal donors (3 cases) was detected by RT-qPCR. CircCRKL was substantially expressed in CML patients relative to normal donors, as demonstrated in Fig. 1A. Furthermore, when comparing BCR-ABL⁺ cell lines (K562, K562/G01, KCL22, and SupB15) to BCR-ABL⁻ cell lines (TK6 and THP-1), it was surprising to find that the circCRKL expression was significantly higher in BCR-ABL⁺cell lines (Fig. 1B). The BCR-ABL expression level of cell lines was evaluated by western blot assays (Additional file 1: Figure S1A). These findings suggested that a link may exist between circCRKL and BCR-ABL. Then we conducted a bioinformatics analysis of the CRKL transcript. It was found that circCRKL is a 466-nt circRNA and was back spliced by exon 2 of pre-CRKL, Sanger sequencing was applied to confirm the cyclization site (Fig. 1C). The results of nuclear and cytoplasm RNA separation assays with RT-qPCR uncovered that circCRKL expression level in the cytoplasm was higher than that in the nucleus in CML cells (Fig. 1D, E), which was further supported by fluorescence in situ hybridization (FISH) assays (Fig. 1F). Subsequently, to determine the circular form of circCRKL, the RNase R analysis was performed and the linear CRKL expression diminished dramatically while circCRKL had no changes, indicating that circCRKL was stable in CML cells (Fig. 1G, H). In addition, the half-life of circCRKL and linear CRKL was determined with the treatment of actinomycin D. It was found that circCRKL had a longer half-life compared with that of linear CRKL (Fig. 1I, J). These findings imply that circCRKL, located in the cytoplasm in CML cells and highly expressed, is a stable circRNA, indicating a potential role in CML.

To further explore the biological involvement of circCRKL in CML, a loss-of-function test was performed. We constructed circCRKL knockdown CML cells using shRNA specifically targeting the back spliced junction site of circCRKL. The expression of circCRKL was markedly restrained in CML cells, whereas the linear CRKL expression was unaffected (Fig. 2A, B). Subsequently, CCK-8 and colony formation assays uncovered that circCRKL suppression drastically reduced CML cell growth (Fig. 2C−F). Flow cytometry assay showed that circCRKL silencing reduced the number of cells in the S phase (Fig. 2G, H). The effect of circCRKL on apoptosis in CML cells was then investigated, and we discovered that the second shRNA of circCRKL silencing engendered cell apoptosis rate increasing (Additional file 1: Figure S1B).

The differential expression of circCRKL in BCR-ABL⁺ and BCR-ABL⁻ cell lines raises the question of whether circCRKL specifically actives in cells expressing BCR-ABL. Hence, the effects of circCRKL silencing on cell viability were investigated in two other BCR-ABL⁺ cell lines (KCL22 and SupB15) as well as BCR-ABL⁻ cell lines (THP-1 and TK-6). The knockdown efficiency of circCRKL was confirmed in these cell lines (Additional file 1: Figure S1C). Next, CCK-8 assays showed circCRKL suppression substantially impaired cell viability of BCR-ABL⁺ cell lines but had no influence on BCR-ABL⁻ cells (Additional file 1: Figure S1D–G). Collectively, these findings imply that circCRKL is specifically required in BCR-ABL⁺ cells, and dampening circCRKL inhibits proliferation.

To better investigate the function of circCRKL, a CML mouse model was created. Sh-NC or sh-circCRKL lentivirus-infected K562/G01 cells were delivered intravenously (i.v.) into immunodeficient NOD/SCID mice, respectively. Prior studies have noted that the features of CML mice containing increased white blood cell count and histopathological variation such as bone marrow, liver, and spleen infiltration caused by leukemic cells [32]. As expected, the white blood cell count of the control group mice increased more than that of the knockdown group, but no distinct difference exists (p = 0.0586) (Fig. 3A). In addition, compared with sh-NC group mice, knockdown of circCRKL ameliorated the splenomegaly. However, there was no significant defect for weights of livers in the sh-circCRKL group compared to the NC group (Fig. 3B, C). To evaluate the infiltration of leukemic cells in more depth, we set out to perform Wright’s staining in murine bone marrow, liver, and spleen. The results showed severer infiltration in the sh-NC group xenograft models (Fig. 3D), similarly confirmed by hematoxylin/eosin (HE) staining (Fig. 3E). The BCR-ABL level in mouse BM, liver, and spleen was then assessed with immunofluorescence, and the results revealed a much lower level in circCRKL silencing mice (Fig. 3F). Overall, these findings indicated that circCRKL can promote the malignant progression of CML in vivo.

Since we discovered that circCRKL is essential in BCR-ABL⁺ cells, it’s critical to figure out how circCRKL and BCR-ABL are linked. Notably, western blot showed a reduction of the BCR-ABL and c-ABL protein levels in circCRKL suppressed cells (Fig. 4A). The BCR-ABL expression level was further examined using an immunofluorescence test, and the results were in concert with those previously mentioned (Fig. 4B). Furthermore, RT-qPCR demonstrated that knocking down circCRKL reduced BCR-ABL mRNA expression (Fig. 4C), suggesting circCRKL may regulate BCR-ABL at the mRNA level. The mechanism by which circCRKL controls BCR-ABL, however, is yet unknown.

Given that circCRKL is predominantly found in the cytoplasm of CML cells, which is a critical prerequisite for circCRKL to function as a miRNA sponge. Therefore, we screened downstream miRNAs of circCRKL from three databases (circInteractome, circBank, and ENCORI) (Fig. 4D). Subsequently, ENCORI and the other two databases TargetScan and miRwalk were applied to predict miRNAs bound to ABL. Then we overlapped miRNAs of circCRKL and ABL (Fig. 4E). Among candidate miRNAs, miR-877-5p is a unique miRNA containing binding sites for both circCRKL and ABL 3’UTR (Fig. 4F), as confirmed through dual-luciferase reporter assays. The wild-type circCRKL reporter’s luciferase activity was drastically diminished when miR-877-5p was ectopically expressed, however, the mutant reporter’s activity was unaffected, demonstrating that circCRKL may interact with miR-877-5p at specific sites (Fig. 4G), and similar results were noticed between miR-877-5p and ABL (Fig. 4H). The connection between circCRKL and miR-877-5p was further confirmed by RNA pull-down assays. The efficiency of the circCRKL probe was testified by RT-qPCR, then as expected, miR-877-5p was found to be specifically enriched by the circCRKL probe in K562/G01 cells (Fig. 4I). Moreover, prior studies have noted the importance of AGO2 for circRNA to act as a miRNA sponge [33]. Herein, an anti-AGO2 RNA immunoprecipitation was performed, showing both circCRKL and miR-877-5p could interact with AGO2, implying that circCRKL might act as a sponge for miR-877-5p (Fig. 4J).

To figure out the biological relevance of miR-877-5p in CML and whether BCR-ABL is controlled by miR-877-5p, the expression level in BMMCs from CML patients was determined (Fig. 5A), then inhibitor or mimics was utilized to knock down or overexpress miR-877-5p, respectively. The effect of inhibitor or mimics on miR-877-5p in CML cells was examined with RT-qPCR (Fig. 5B, C). The BCR-ABL mRNA expression was significantly up-regulated by inhibiting miR-877-5p, whereas overexpressing miR-877-5p markedly decreased BCR-ABL (Fig. 5D). Likewise, western blot demonstrated that miR-877-5p inhibition augmented BCR-ABL protein level, while miR-877-5p overexpression showed an opposite effect (Fig. 5E). Furthermore, CCK-8 (Fig. 5F, G) and colony formation assays (Fig. 5H, I) revealed that miR-877-5p inhibition stimulated CML cells proliferation and miR-877-5p overexpression had an opposite effect. Altogether, these findings revealed miR-877-5p acts as a negative regulator of BCR-ABL.

To further explore whether circCRKL supports CML cells proliferation by controlling miR-877-5p/BCR-ABL axis. A rescue experiment was performed by co-transfecting sh-circCRKL along with miR-877-5p inhibitor. Knockdown of miR-877-5p distinctly restored the restrained growth induced by suppressing circCRKL in CML cells, according to CCK-8 and colony formation assays (Fig. 6A–D). Moreover, miR-877-5p knockdown markedly ameliorated the reduced BCR-ABL level induced by circCRKL suppression (Fig. 6E). In addition, the function of sh-circCRKL in miR-mimic cells was determined by measuring the BCR-ABL protein level in CML cells coupled with circCRKL knockdown. According to the findings, BCR-ABL protein levels were lower in CML cells co-transfected sh-circCRKL along with miR-mimic than that in sh-circCRKL + mimic-NC cells (Fig. 6F). Collectively, these results proved that circCRKL promotes CML cell proliferation through regulating miR-877-5p mediated BCR-ABL.

Given that the imatinib-resistant cell line K562/G01 had higher levels of circCRKL expression than K562 cells. We speculate that a link may exist between circCRKL and imatinib resistance. The half-maximal inhibitory concentration value (IC-50) of K562 or K562/G01 cells against imatinib was measured using a CCK-8 assay (Additional file 2: Figure S2A). Subsequently, the cell viability of stable knockdown of circCRKL and circCRKL wild-type K562/G01 cells treated with imatinib for 48 h was determined and the IC-50 value was computed. Indeed, the data demonstrated that circCRKL silencing observably reduced cell viability, and the IC-50 value was decreased simultaneously (Additional file 2: Figure S2B). In addition, flow cytometry demonstrated that knocking down circCRKL boosted the apoptosis in K562/G01 cells treated with imatinib (Additional file 2: Figure S2C). We assumed that miR-877-5p might be implicated in imatinib-resistance, hence the impact of suppressing or overexpressing miR-877-5p on the IC-50 value of K562/G01 cells was then investigated. However, no distinct effect on IC-50 value was caught (Additional file 2: Figure S2D). Previous investigations have linked the Wnt pathway with imatinib-resistance [34, 35], the relevant markers including CTNNB1 and c-Myc were detected using a western blot assay, accordingly. The data demonstrated a discernible loss of both CTNNB1 and c-Myc in K562/G01 cells which circCRKL was knocked down (Additional file 2: Figure S2E). Collectively, these findings indicated that circCRKL may regulate the susceptibility of K562/G01 cells to imatinib through mediating the Wnt/β-catenin signaling pathway.