Therapeutic inhibition of GAS6-AS1/YBX1/MYC axis suppresses cell propagation and disease progression of acute myeloid leukemia

From January 2014 to September 2015, bone marrow samples were collected from AML inpatients of Union Hospital, Tongji Medical College, Huazhong University of Science and Technology. Seventy-six patients had AML, including 41 males and 35 females, with an average age of 45.8 (17.3–76.6) years. The diagnostic type of AML was in accordance with the World Health Organization (WHO) classification criteria [28]. Clinical characteristics of the AML patients were listed in Table S1. Twenty bone marrow samples were obtained from healthy donors with a mean age of 38.7 years (ranging from 17.5 to 52.1) and a sex ratio of 11/9 (male/female).

Gene expression data were downloaded from the Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases (GSE85030, GES103828, GSE37642 and GSE96535). The probe sequences were downloaded from GEO or microarray manufacturers, and bowtie was used to re-annotate probes according to GENCODE Release 19 annotation for lncRNAs.

Human AML cell lines Kasumi-1, U937, MOLM-13, THP-1 and HL-60 were maintained in RPMI 1640 medium (HyClone, South Logan, UT) containing 10% fetal bovine serum (Gibco Cell Culture, Melbourne, Australia). HEK-293 T cells were maintained in RPMI DMEM medium (HyClone) containing 10% fetal bovine serum (Gibco). Kasumi-1, U937, MOLM-13, THP-1 and HL-60 cells were authenticated by karyotype, morphology, and PCR analysis. HEK-293 T cells were identified by morphology and capability of virus production.

qRT-PCR was used to detect expression levels of GAS6-AS1 and other genes in AML cells, following the manufacturer’s instructions (TaKaRa, Japan). The β-actin was used as the control. Primers were listed in Table S2.

The full-length cDNA of human GAS6-AS1, YBX1, MYC, and their truncations were synthesized by Invitrogen (Shanghai, China) and cloned into the lentiviral expression vector pWPXL. The small hairpin RNA (shRNA) constructs of GAS6-AS1, YBX1, and MYC were provided by GenePharma (Shanghai, China) and were cloned into the pLKO.1 shRNA lentiviral vector. For stable expression assays, all the vectors were packaged into lentiviruses with packaging plasmids psPAX2 and pMD2G in HEK-293 T cells. The resulting constructs were verified by sequence analysis. Finally, the lentiviruses were transfected into Kasumi-1, U937, and HL-60 cells, and the transformed cells were then selected with puromycin. All shRNA sequences were listed in Table S3.

Cell proliferation ability was examined by CCK-8, colony formation, EdU and cell cycle assays. CCK-8 assay was conducted using a CCK-8 assay kit (Dojindo Japan). Ethynyl deoxyuridine (Edu) assay were conducted with EdU detection kit (KeyGEN BioTECH, Nanjing, China). Colony formation assays were performed according to standard protocols [29]. The FITC-Annexin V and propidium iodide (PI) double staining apoptosis assays were performed with flow cytometer (FACScan; BD Biosciences, USA).

The Cy3-labeled and FITC-labeled oligonucleotide probes were synthesized by Focobio Corporation (Guangzhou, China). Cells were incubated with FISH probes at 37 °C overnight. The nuclei were counterstained with DAPI. The slides were observed using fluorescent microscopy. For distribution assay, cytoplasmic and nuclear RNAs were isolated by the PARIS kit (Life Technologies, Carlsbad, CA) following the manufacturer’s protocols. The expression levels of GAS6-AS1 were measured by qRT-PCR.

Biotin-labeled RNAs were in vitro transcribed using Biotin RNA Labeling Mix (Roche) and T7 RNA polymerase, treated with RNase-free DNase I, and purified with RNeasy Mini Kit (Qiagen). Nuclear extracts were harvested, resuspended in freshly prepared proteolysis buffer, and incubated with biotin-labeled RNA and streptavidin-agarose beads (Invitrogen). Precipitated components were separated using SDS-PAGE. Differential bands were harvested for mass spectrometry analysis.

Western blot analysis according to standard protocols as described previously [29]. The antibodies including YBX1(ab76149, Abcam Inc.), MYC (ab32072), IL1R1 (ab106278), SRC (ab109381), RAB27B (ab76779), Histone H3 (ab18521), and β-actin (ab8226).

The EZ-Magna RIP Kit (Millipore) was used following the manufacturer’s protocol. Treated cells were lysed in complete RIP lysis buffer, and the cell extract was incubated with magnetic beads conjugated with specific antibodies or control IgG (Millipore). Beads were washed and incubated with Proteinase K to remove proteins. Finally, purified RNA was subjected to qRT-PCR analysis.

ChIP experiments were performed using the Magna ChIP kit (Millipore) according to the manufacturer’s instructions. Real-time qPCR was undertaken with a SYBR Green PCR kit (TaKaRa) and primers targeting gene promoters were listed in Table S2.

Co-IP assay was conducted as described previously [30], with antibodies specific for YBX1 (ab76149) and MYC (ab32072). The bead-bound proteins were released and tested by Western blot.

The promoter fragments of IL1R1 and RAB27B were amplified from genomic DNA by PCR and subcloned into pGL3-Basic (Promega). Primers were listed in Table S4. Dual-luciferase assay was performed according to manufacturer’s instruction (Promega). Luciferase reporters of MYC, EP300, TP53, and TFAP2A were obtained from Qiagen Inc. and Stratagene (La Jolla, CA). Luciferase activity was measured with a luminometer (Lumat LB9507, Berthold, Germany).

Total RNA of cells was extracted according to manual of TRIzol reagent (Life Technologies, Gaithersburg, MD). Library preparation and transcriptome sequencing was performed on an Illumina platform by Annoroad Gene Technology (Beijing, China). Sequencing data were deposited in SRA database under NCBI accession PRJNA737043.

Six- to eight-week old NOD-SCID mice were purchased from Beijing Vital River Laboratory Animal Technology (Beijing, China). The xenograft leukemia in mice was generated by injecting empty vector or sh-GAS6-AS1 transfected cells subcutaneously into a single side of each mouse. Three weeks after injection, xenografted mice were euthanized for analysis. For lentiviral based in vivo knockdown experiment, U937 cells were subcutaneously implanted into both sides of each mouse. Then, empty vector LeV-Scb and LeV-sh-GAS6-AS1 was intratumorally injected, twice a week for 2 weeks, at different side of each mouse. For systemic in vivo leukemia model, NOD-SCID mice were randomly divided in three groups and xenograft leukemia in mice was established by injecting 5 × 10⁶ sh-Scb or sh-GAS6-AS1 transduced U937 cells in 150 μL of phosphate-buffered saline (PBS) or PBS alone into the tail vein. Three weeks after inoculation, xenografted mice were euthanized for analysis. Human cell engraftment in bone marrow and spleen was examined by flow cytometry or hematoxylin and eosin (H&E) staining, as we have described previously [24]. The remaining mice were estimated using the survival analysis method.

The mice were fasted and anesthetized on the 21st day after engraftment. Approximately 200 ± 20 μCi of 18-fluoro-6-deoxy-glucose (18F-FDG) was injected via the tail vein. Static PET images were started 1 h after the injection using a Trans-PET BioCaliburn LH system (Raycan Technology, Suzhou, China). A volume-of-interest analysis was conducted using the AMIDE software package (Free Software Foundation, Boston, USA).

All results were depicted as mean ± standard error of the mean (s.e.m.). Student’s t-test, analysis of variance, and χ² analysis were applied to compare difference. Fisher’s exact test was applied to analyze statistical significance of overlap between two gene lists. The survival curves are drawn using Kaplan–Meier survival plot and tested using log-rank tests. All statistical analyses were performed using SPSS 19 software (IBM, Somers, NY). P < 0.05 was considered to indicate statistical significance. Kaplan-Meier survival curves for mice and P values were calculated using a log-rank (Mantel-Cox) test.

To identify the lncRNAs involved in leukemogenesis, we performed an integrative analysis of three gene expression profiles comprising GSE85030, GES103828, and GSE96535 datasets. All these datasets have enrolled bone marrow samples from both AML patients and normal persons. We identified 57 lncRNAs misregulated in both GSE96535 and GES103828 datasets, 118 in both GSE96535 and GSE85030 datasets, and 50 in both GSE85030 and GSE103828 datasets (fold change > 2.0, P < 0.05; Fig. 1A-B). Overlapping analysis revealed 6 lncRNAs upregulated and 13 downregulated in all datasets. Next, we focused on the most significant upregulated GAS6-AS1 as potential carcinogenic drivers or treatment targets. The Cancer Cell Line Encyclopedia (CCLE) data showed GAS6-AS1 expression was higher in AML cell lines than most of the other cancer types (Fig. 1C). We also studied the GAS6-AS1 expression by analyzing TCGA and Genotype-Tissue Expression (GTEx) database, as tumor and normal controls, respectively. The GAS6-AS1 expression was higher than all the other cancers (Fig. 1D). Additionally, comparing AML and normal controls (TCGA vs GETx), GAS6-AS1 was significantly upregulated in AML (Fig. 1E). Further, we examined GAS6-AS1 expression in an AML cohort in our Institute of Hematology of Union Hospital (IHUH), which revealed that GAS6-AS1 was significantly upregulated in AML patients and that 19.7% (15 of 76) patients showed more than 2-fold increase (Fig. 1F). Survival analysis showed that high GAS6-AS1 expression was significantly correlated with poor overall survival (Fig. 1G). Furthermore, we analyzed survival data from TCGA and GSE37642 and observed high GAS6-AS1 expression was associated with poor overall survival in AML (Fig. 1H-I). Collectively, these results indicated that GAS6-AS1 was overexpressed in AML, and it might contribute to leukemogenesis.

To assess the effect of GAS6-AS1 on AML cells, we tested the endogenous expression levels of GAS6-AS1 in various AML cell lines by qRT-PCR. The results revealed that compared with mononuclear cells (MNC) from bone marrow of healthy donors, the expression of GAS6-AS1 was significantly increased in U937, Kasumi-1, HL-60, MOLM-13 and THP-1 cell lines (P < 0.05), with U937 cell line exhibiting the highest level of GAS6-AS1 (Fig. S1). Accordingly, we chose U937 and HL-60 as loss of function experimental cell lines and Kasumi-1 as gain of function experimental cell line. Then EdU assays were performed and the results revealed that knockdown of GAS6-AS1 decreased HL-60 and U937 cells proliferation compared with the respective controls, whereas ectopic GAS6-AS1 overexpression promoted cell growth in Kasumi-1 (Fig. 2A). Colony-formation assays indicated that clonogenic survival was significantly declined following knockdown of GAS6-AS1 in HL-60 and U937 cells, but markedly increased in GAS6-AS1 overexpressed Kasumi-1 cells (Fig. 2B). Consistently, the EdU assay demonstrated that enforced expression of GAS6-AS1 had a positive impact on the leukemia cell proliferation (Fig. 2C). To strictly demonstrate the biological functions of GAS6-AS1 in different AML cell lines, we also conducted the cell viability, colony-formation and EdU assay in GAS6-AS1 overexpressed HL-60 and U937 cells and in GAS6-AS1 down-expressed Kasumi-1 cell. These results (supplementary data Fig. S2) were consistent with the results observed in GAS6-AS1 overexpressed Kasumi-1 cell and in GAS6-AS1 down-expressed HL-60 and U937 cells, respectively. To further investigate the disease progression effect of GAS61-AS1 in vivo, we subcutaneously inoculated NOD-SCID mice with xenograft leukemia tumor (Fig. 2D). Stable lentivector-based knockdown of GAS6-AS1 into U937 cells led to a significant augment in tumor weight, volume, and Ki-67 proliferative index of xenograft leukemia (Fig. 2E-I). Together, these results suggested the leukemogenic role of GAS6-AS1 in AML.

To further explore whether GAS6-AS1 enhanced AML proliferation by regulation cell cycle progression in AML cells, flow cytometry was performed and demonstrated that HL-60 and U937 cells with sh-RNAs had a clear cell cycle arrest in the G1-S phase and the population of cells in the S phase was decreased (Fig. 3A-B). In Fig. 3A, the sh-GAS6-AS1#1 group showed no obvious apoptotic peak, while the sh-GAS6-AS1#1 group in Fig. 3D clearly showed apoptosis. We suspected that because the cell membrane of apoptotic cells is relatively fragile, it is possible that the apoptotic cells ruptured more severely during processing, resulting in most of them becoming invalid signals. This might lead to no obvious apoptotic peak in the sh-GAS6-AS1#1 group in Fig. 3A. However, forced expression of GAS6-AS1 induced G1-S progression and accumulated S phase (Fig. 3C). Apoptosis analysis indicated that the percentage of early and late apoptotic cells was significantly augmented in HL-60 and U937 cells with sh-GAS6-AS1 than the sh-Scb cells (Fig. 3D-E). Consistently, forced expression of GAS6-AS1 decreased apoptosis rate as compared with the empty vector group (Fig. 3F). These data revealed that GAS6-AS1 accelerated cell cycle progress and reduces apoptosis of AML cells.

To further functionally characterize GAS6-AS1, we analyzed the cellular distribution of GAS6-AS1 in U937 cells. As indicated by FISH assay and qRT-PCR, GAS6-AS1 was mainly located in the nucleus of AML cells (Fig. 4A-B; Fig. S3). We then screened potential GAS6-AS1-binding proteins via biotin-labeled RNA pull-down followed by mass spectrometry, which revealed 71 proteins pulled down by GAS6-AS1 in U937 cells (Fig. 4C; Table S5). Overlapping analysis with RNA-binding proteins defined by RBPDB (http://​rbpdb.​ccbr.​utoronto.​ca) revealed five potential partners of GAS6-AS1, including Nucleolin (NCL), Dyskerin Pseudouridine Synthase 1 (DKC1), YBX1, Splicing Factor Proline And Glutamine Rich (SFPQ), and Non-POU Domain Containing Octamer Binding (NONO). Further validating biotin-labeled RNA pull-down and Western blot assays demonstrated the direct interaction of GAS6-AS1 with YBX1, but not with NCL, DKC1, SFPQ or NONO (Fig. 4D). RIP assay confirmed the specific binding of YBX1 to GAS6-AS1, but not to the control primarily nuclear-retained lncRNA HOX transcript antisense RNA (HOTAIR), in U937 cells (Fig. 4E-F). Further catRAPID analysis revealed potential interaction region of GAS6-AS1 with YBX1 (Fig. S4). Meanwhile, the secondary structure of GAS6-AS1 was predicted by RNAfold webserver (Fig. S5). Thus, we constructed a series of truncated GAS6-AS1 to map its binding fragment with YBX1 (Fig. 4G-H). As verified by RNA pulldown assays, exon 5 of GAS6-AS1 was responsible for its interaction with YBX1 protein. These data indicated that GAS6-AS1 directly interacted with RNA binding protein YBX1 in AML cells.

To identify the putative targets of GAS6-AS1, RNA sequencing (RNA-seq) assay revealed 155 upregulated and 348 downregulated genes (fold change > 1.5, P < 0.05) in U937 cells upon GAS6-AS1 knockdown (Fig. 5A). Further overlapping analysis of transcriptional regulators of these altered genes by ChIP-X program and YBX1-interacting protein in BioGRID database [31] revealed six potential transcription factors (Fig. 5B). Among these six transcription factors, the expression of E1A binding protein p300 (EP300), MYC, transcription factor AP-2α (TFAP2A), or tumor protein p53 (TP53) were reported to facilitate leukemogenesis and significantly associated with survival of AML (Fig. 5B). Notably, stable overexpression or knockdown of GAS6-AS1 modulated the activity of MYC, but not of EP300, TFAP2A, or TP53, in U937 and HL-60 cells (Fig. 5D-F). Among the MYC target genes derived from RNA-seq results and ChIP-X analysis, the expression of interleukin 1 receptor type I (IL1R1), proto-oncogene tyrosine-protein kinase src (SRC), and ras-related protein Rab-27B (RAB27B) was most significantly altered. Stably ectopic expression or silenced expression of GAS6-AS1 increased and decreased the MYC enrichment on target gene promoters in Kasumi-1 and U937 cells, which were rescued by silencing or forced expression of YBX1, respectively (Fig. 5G-H). Furthermore, alteration in expression of IL1R1, SRC and RAB27B induced by silencing of GAS6-AS1 was markedly rescued after forced expression of YBX1. (Fig. 5I-J). These findings implied that GAS6-AS1 enhanced the expression of MYC target genes through YBX1 in AML cells.

Considering GAS6-AS1 bound to YBX1 and regulated MYC target genes, we hypothesized that GAS6-AS1 might act as a lncRNA linking YBX1 and MYC. Endogenous physical interaction between YBX1 and MYC was validated in U937 cells (Fig. 6A). Co-IP and Western blot assay demonstrated that knockdown of GAS6-AS1 suppressed the interaction between YBX1 and MYC (Fig. 6B). For the rescue experiments, both in HL-60 and U937 cells, forced overexpression of YBX1 alleviated the decreased MYC transactivation induced by silencing of GAS6-AS1 (Fig. 6C-D). Consistently, colony formation assay indicated the high proliferative phenotype of GAS6-AS1-silenced AML cells was rescued by overexpression YBX1 and MYC (Fig. 6E-F). In addition, Gene Set Enrichment Analysis (GSEA) demonstrated that the gene sets of regulation of apoptosis and the MAPK pathway were enriched in GAS6-AS1-silencing cells (Fig. 6G). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis suggested the potential implication of cytokine-receptor interaction (Fig. 6H). The GSEA and KEGG results were in accordance with reported main functions of those MYC target genes including IL1R1, SRC and RAB27B. Further, dual luciferase assays showed IL1R1 and RAB27B promoter activity was enhanced by MYC, while reduced by MYC knockdown (Fig. S6). These data indicated that GAS6-AS1 facilitated propagation of AML cells via YBX1-mediated MYC transactivation.

To analyze the therapeutic effect of lentivirus-mediated GAS6-AS1 knockdown on leukemic progression in vivo, subcutaneous tumorigenesis by U937 cells and thereafter scheduled lentiviral vectors injection were performed. The day of subcutaneous implantation was Day 0, and then the lentiviral vectors injection was conducted at Day 6, Day 10, Day 14, Day 18 (five subjects in each group). The PET scanning was performed at 21 days after engraftment. A decrease in the maximal standard uptake value of 18F-FDG was found in GAS6-AS1 shRNA-interfered group (Fig. 7A). Administration of lentivirus-mediated shRNA against GAS6-AS1 (LeV-sh-GAS6-AS1 #1) dramatically reduced the weight and Ki-67 proliferation index in subcutaneous xenograft mice (Fig. 7B-C). Additionally, GAS6-AS6 and downstream gene expression in the tumor tissue from LeV-sh-GAS6-AS1 group were dramatically decreased (Fig. S7). These data indicated that lentivirus based GAS6-AS1 knockdown inhibited leukemia progression.

In addition to the subcutaneous in vivo leukemia model, we also characterized the effect of GAS6-AS1 knockdown using systemic in vivo leukemia model. The NOD-SCID mice were injected via the tail vein with 5 X 10⁶ U937 cells with sh-GAS6-AS1 (knockdown) or sh-Scb (control). We assessed the organ infiltration meidated by GAS6-AS1 via H&E staining 3 weeks after injection. Compared with sh-Scb mice, sh-GAS6-AS1–treated mice exhibited reduced levels of U937 in the bone marrow and spleen, suggesting a lower degree of aggression in the indicated organs (Fig. 7D). Consistently, flow cytometry verified leukemia engraftment in the bone marrow and further demonstrated that sh-GAS6-AS1–treated mice exhibited lower percentages of hCD45⁺ cells in bone marrow than sh-Scb–treated mice (Fig. 7E-F). Remarkably, sh-GAS6-AS1 extended the survival of the recipients from 26 days for sh-Scb–treated mice to 36 days for sh-GAS6-AS1–treated mice (Fig. 7G; P = 0.0050), suggesting that GAS6-AS1 knockdown inhibits leukemia progression.