DHX9-mediated pathway contributes to the malignant phenotype of myelodysplastic syndromes

Summary DHX9 is a member of the DEAH (Asp-Glu-Ala-His) helicase family and regulates DNA replication and RNA processing. DHX9 dysfunction promotes tumorigenesis in several solid cancers. However, the role of DHX9 in MDS is still unknown. Here, we analyzed the expression of DHX9 and its clinical significance in 120 MDS patients and 42 non-MDS controls. Lentivirus-mediated DHX9-knockdown experiments were performed to investigate its biological function. We also performed cell functional assays, gene microarray, and pharmacological intervention to investigate the mechanistic involvement of DHX9. We found that overexpression of DHX9 is frequent in MDS and associated with poor survival and high risk of acute myeloid leukemia (AML) transformation. DHX9 is essential for the maintenance of malignant proliferation of leukemia cells, and DHX9 suppression increases cell apoptosis and causes hypersensitivity to chemotherapeutic agents. Besides, knockdown of DHX9 inactivates the PI3K-AKT and ATR-Chk1 signaling, promotes R-loop accumulation, and R-loop-mediated DNA damage.


INTRODUCTION
Myelodysplastic syndrome (MDS) is a highly heterogeneous myeloid neoplastic disease, characterized by ineffective hematopoiesis and a high risk of acute myeloid leukemia (AML) transformation. 1 Recently, massively parallel sequencing in MDS has revealed some high-frequency mutations including RNA splicing-related genes such as SF3B1, U2AF1, and SRSF2. [2][3][4] Functional studies of these genes have furthered our understanding of MDS pathogenesis and the development of more effective targeted treatment strategies. 5-7 In a previous study, we performed whole-exome sequencing of three MDS patients and found a DHX9 mutation in one patient. 8 Further validated sequencing showed that most of DHX9 mutations were defined as single nucleotide polymorphism according to the comparison of sequencing in oral mucosa epithelial cells (unpublished data). However, we unexpectedly found that DHX9 mRNA is over-expressed in MDS patients, indicating the close association between DHX9 and MDS development. Functionally, DHX9 as an NTP-dependent DExH/D-box helicase can unwind both RNA and DNA strands. 9 It regulates gene expression and RNA selective splicing by interaction with transcription factors or complexes. 10 MDS are diseases with abnormal RNA splicing, which prompts us to investigate the role of DHX9 in MDS. Additionally, two recent studies have shown that dysfunction of DDX41 and DDX3X (DHX9 members) are frequently found in leukemia and natural killer/T-cell (NKT) lymphomas, respectively, which further suggests that abnormalities in the RNA helicase family contributes to the development of hematological malignancies. 11,12 To date, no DHX9-related studies have been reported in MDS. Here, we investigated the DHX9 expression and its relationship with the clinical characteristics. We also conducted a series of functional experiments in myeloid tumor cell lines to investigate the effects of DHX9 on the biological characteristics of MDS.

Patients' characteristics
The MDS group was composed of 120 MDS patients with a median age of 55 years (11-84 years), which included 11 RCUD (refractory cytopenia with unilineage dysplasia), 4 RARS (refractory anemia with ringed sideroblasts), 53 RCMD (refractory cytopenia with multilineage dysplasia), 2 MDS-U (myelodysplastic Increased DHX9 expression is common in MDS and associated with poor clinical outcome DHX9 mRNA levels were determined by qRT-PCR in 120 MDS patients (at the time of diagnosis) and 42 control individuals. DHX9 expression was significantly higher in the MDS patients compared with the control individuals (p < 0.001) ( Figure 1A). According to 2008 WHO classification, both low-grade and high-grade MDS group showed higher expression of DHX9 than the control individuals (p = 0.001; p = 0.009) (Figure 1B). The patients with abnormal karyotype showed higher expression of DHX9 than non-MDS controls (p = 0.031) ( Figure 1C). Grouped according to disease progression, the patients with tMDS (finally transformed into RAEB or AML) showed higher DHX9 expression than those with sMDS (stable MDS, not transformed into AML) (p < 0.001) ( Figure 1D). To investigate the influence of DHX9 expression on overall survival and AML transformation , we divided patients into two groups: low DHX9 expression (% median value) and high DHX9 expression (> median value). Survival analysis showed that the group with high DHX9 iScience Article expression had shorter overall survival and more risk of AML transformation compared to the group with low DHX9 expression (p < 0.001, Figure 1E; p = 0.008, Figure 1F).

DHX9 is indispensable for the maintenance of malignant proliferation of leukemia cells
qRT-PCR analysis showed that the DHX9 mRNA level was higher in SKM-1 and HEL cells than that in the normal CD34 positive cells ( Figure 2A). Two effective DHX9-shRNAs and control shRNA were transfected into leukemia cells, respectively. Over 60% decreases in DHX9 expression were observed ( Figure 2B). Cell counting assay indicated that DHX9-depleted cells exhibited reduced cell growth compared to the control cells ( Figure 2C). The EdU uptake assay indicated that the cells with DHX9 knockdown had a reduced uptake capacity of EdU compared with the control cells after treatment with 10 mM EdU for 2 h ( Figure 2D). Methylcellulose assay showed that reduced colony formation was observed in the DHX9-knockdown cells compared with the control cells ( Figure 2E). Together, these assays revealed that knockdown of DHX9 may impair the cell proliferation capacity in the malignant hematopoietic cells.

DHX9 suppression increases cell apoptosis and causes hypersensitivity to chemotherapeutic agents
It is well-known that MDS cells could overcome apoptosis, obtain a proliferative advantage, and then evolve into AML cells. To investigate whether MDS cells overcome apoptosis through DHX9, annexin V-FITC/PI assay was performed and then we observed that knockdown of DHX9 increased the apoptosis in leukemia cell lines ( Figure 3A). The representative flow graphics are shown in Figure 3B. Further, to understand the role of DHX9 in apoptosis triggered by chemotherapeutic agents azacytidine and ABT-199, we investigated the apoptosis sensitivity of leukemia cells to these agents based on DHX9 knockdown. As shown in Figures 3C and 3D, we observed that the SKM-1 and HEL cells with DHX9 knockdown showed a significant increase in cell apoptosis after treated with azacytidine and ABT-199. Taken together, the suppression of DHX9 increases cell apoptosis and caused leukemia cells to be more sensitive to apoptosis induced by chemotherapeutic agents. To further explore the mechanism of DHX9 in regulating malignant biological behaviors, we tried to identify the target genes of DHX9 and the related biological pathways through the gene expression profiles (GEP) from RNA sequencing. We performed RNA sequencing in DHX9-depleted SKM-1 cells and control cells and identified 3,793 differential genes ( Figure 4A and Data S1). Then, we performed pathway analysis of these differential genes. Significant pathways in SKM-1 included oxidative phosphorylation, P53 signaling, PI3K-AKT pathway signaling, etc. (all p < 0.05) ( Figure 4B).
Knockdown of DHX9 inhibits the activation of the PI3K-AKT signaling pathway As described above, the association between DHX9 and its potential pathways such as PI3K-AKT needs to be validated. Compared with the control cells, the DHX9-depleted SKM-1 and HEL cells showed significantly decreased phosphorylation level of GSK3a/b-Ser21/9 and P70S6K-Thr421/Ser424 ( Figure 4C). DHX9 knockdown also led to reduced expression of PI3K-AKT signaling-targeted genes such as CCND2  Figure 4F) and MYC ( Figure 4G) expression in MDS patients. Our data and GEO data suggested that suppression of DHX9 significantly inhibits the PI3K-AKT signaling pathway in cell lines and MDS patients. Taken together, DHX9 is indispensable for the activation of PI3K-AKT pathway.

Defects in DHX9 promotes R-loop-dependent DNA damage
The PI3K-AKT signaling pathway plays an important role in maintaining genomic stability by involving in DNA replication and cell cycle regulation. 13 It is reported that PI3K inhibition induces DNA damage through nucleoside depletion. 14 Moreover, impaired PI3K-AKT signaling pathway can inhibit DNA double-strand break (DSB) repair, increase radiation-induced DSB, and improve radiosensitivity. 15 Based on the above findings, we evaluated the DNA damage by detecting the level of phosphorylated histone H2AX (gH2AX). The results showed that knockdown of DHX9 increased gH2AX expression in SKM-1 and HEL cells ( Figure 5A), which led us to further investigate the role of DHX9 in this process.
Recently, R-loop accumulation has been proved to be potential mechanism of MDS, which comprises an RNA/DNA hybrid and displaced single-stranded DNA and plays a vital role in the maintenance of genomic iScience Article stability and cancer development. [16][17][18] Quantitative mass spectrometry (MS)-based proteomics has been used to detect proteins associated with R-loop. 19 Helicases including DHX9, DDX41, DDX5 were candidate proteins of the R-loop interactome. [20][21][22] Thus, we then explored whether the DNA damage was associated with R-loop formation in DHX9-knockdown cells. We performed immunofluorescence with the S9.6 antibody to identify and quantify the RNA-DNA hybrid component of R-loop. DHX9 suppression increased the level of RNA-DNA hybrids that were sensitive to RNase H1 overexpression ( Figures 5B and 5C). Further, we carried out DNA/RNA immunoprecipitation coupled to sequencing (DRIP-seq) and observed that R-loop was enriched in most type of genic regions including intergenic, gene bodies, and terminal regions in DHX9-depleted SKM-1 cells ( Figure 5D). The signal of R-loop peaks was sensitive to RNase H1 overexpression ( Figure 5D). Western blotting showed that RNase H1 overexpression can rescued the impact of DHX9 suppression on gH2AX in some degree ( Figure 5E), revealing that defects in DHX9 promotes R-loop-dependent DNA damage. Previous research have reported that pathological R-loop formation elicited ATR response in MDS with splicing factor mutation. 17 To determine whether R-loop induced by DHX9 knockdown can activate ATR-Chk1 pathway, we measured the level of phosphorylated Chk1 (p-Chk1) in cells and found that p-Chk1 expression level diminished in DHX9-knockdown cells ( Figure 6A). The results showed that iScience Article DHX9 was required for activation of ATR-Chk1 pathway. In addition, we implied that DHX9 suppression may further increase DNA damage through inhibiting ATR-Chk1 pathway.

DISCUSSION
DHX9 is a member of the DEAH helicase family and plays important roles in various biological processes such as DNA replication and RNA processing. 23,24 DHX9 interacts with cancer-associated genes such as CBP, BRCA1 and EGFR to regulate their transcription and translation. [25][26][27] Studies have shown that the expression level of DHX9 mRNA was significantly increased in a variety of cancers, such as ewing sarcoma, colorectal cancer, and hepatocellular carcinoma. [28][29][30] Recent studies have reported the role of several DHX9-like genes including DDX41, DDX3X, and DHX32 in hematological diseases. [31][32][33] These reports suggest that the RNA helicase family may play a vital role in the development of hematological tumors.
In this study, we investigated the clinical significance and biological function of DHX9 in MDS for the first time. We determined the expression of DHX9 in MDS patients and performed a series of functional experiments to investigate the effects of DHX9 on biological characteristics of MDS/leukemia cells. Higher DHX9 expression was observed in MDS patients compared with the control individuals and associated with poor survival and high AML transformation, suggesting the potential role of DHX9 as an oncogene in promoting clonal proliferation in MDS. In vitro experiments, DHX9 was found to be essential for the proliferation of leukemia cells, and the knockdown of DHX9 reduced cell growth, induced cell apoptosis and sensitivity to chemotherapeutic drugs. These results may explain why the MDS patients with DHX9 overexpression show high AML transformation and suggest that DHX9 acts as an oncogene in MDS disease progression. The oncogenic characteristic of DHX9 was also reported in other studies. [34][35][36] A recent study showed that inhibition of DHX9 expression is lethal to human cancer cell lines and lymphoma cells, while its suppression rarely affected the normal tissues. 37 DHX9 family members, such as DDX1, DDX3, and DDX5 are considered as oncogenes to promote proliferation of tumor cells. [38][39][40] Mechanistically, DHX9 is indispensable for the activation of PI3K-AKT signaling in this study. Similar to our findings, DHX9 family member DDX5 was also a crucial regulator for activation of AKT signaling in oncogenesis. 40,41 Previous studies exhibited the PI3K-AKT signaling pathway was over-activated and associated with disease progression in high-grade MDS. 42,43 DHX9 maintains the activation of PI3K-AKT signaling, which explains the reasons that overexpression of DHX9 is associated with poor survival and high risk of AML transformation in MDS patients.
Genomic instability is one of the hallmarks of tumors; various endogenous and exogenous stress can induce DNA damage and lead to genomic instability. In order to pass on correct genetic information and survive, cells have developed extremely harmonious network called the DNA damage response (DDR). 44 Previous studies have revealed that DHX9 is a DDR protein, both DHX9 and PI3K have effects on DNA damage repair pathways. 15,45,46 These results enabled us to further explore the function of   16,17 Accumulation of R-loops impairs transcription, causes replication stress and chromosome fragility, which are linked to compromised proliferation of MDS progenitor cells. A recent survey revealed the ability of DDX47 to unwind R-loop, while another study showed that restoring DDX5 levels in hypoxia lead to further increased replication stress and R-loop. 22,50 Our study indicated that knockdown of DHX9 promotes R-loop accumulation and R-loopmediated replication stress. Given that R-loop accumulates frequently upon transcription-replication collisions, and DHX9-depleted cells had reduced DNA synthesis, we speculated that DHX9 may alleviate transcription-replication conflicts to maintain genomic stability.
ATR is PI3K-related protein kinases and central regulators of the DDR. 51 Once activated, ATR phosphorylates a variety of substrates particularly Chk1, so as to arrest the cell cycle, stabilize and repair stalled replication forks. 52 Studies reported pathological R-loop formation elicited ATR response in MDS with splicing factor mutation. 17 However, Dr. Chakraborty et al. indicated that knockdown of DHX9 decreased the level of p-ATR and p-Chk1 in U2OS cells. 46 Consistent with their findings, we investigated p-Chk1 expression was diminished in DHX9-depleted MDS/leukemia cells, suggesting that DHX9 maintain ATR-Chk1 activity in response to DNA damage.
Finally, in high-grade MDS, overexpression of DHX9, on the one hand, may promote excessive proliferation, and on the other hand, may produce apoptotic resistance to exogenous anti-tumor agents. It is well-known that high-grade MDS and MDS-derived AML show multi-drug resistance. It could be imagined that the application of chemotherapeutic agents in combination with a DHX9 inhibitor will greatly improve the therapeutic effect in MDS.

Conclusions
In summary, overexpression of DHX9 is frequent and associated with poor survival and more risk of AML transformation in MDS. Mechanically, DHX9 maintains the activation of PI3K-AKT and ATR-Chk1 pathways, suppress R-loop accumulation and R-loop-mediated DNA damage, and eventually induced excess cell proliferation and apoptosis resistance ( Figure 6B). DHX9 may serve as a novel prognostic marker for AML transformation and therapeutic target in MDS.

Limitations of the study
Although we revealed the connection among DHX9, R-loop, PI3K-AKT, and ATR-Chk1 pathway in MDS/ leukemia cell lines, the reciprocal relationship among them needs to be further investigated.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:   iScience Article plate was incubated with the lentivirus (3310 8 TU/mL) and polybrene (5 mg/ml) in a 1 mL volume. The silencing efficiency of the DHX9 was evaluated using qPCR and western blotting. The sequences of shRNA were shown in Table S3.
Cell Counting Kit-8 assay Cell proliferation was evaluated by Cell Counting Kit-8 (CCK-8) assay at different time point after plating in 96-well plates (4,000 cells/well) in triplicate. The cell growth curve was described according to the absorbance values at 450nm.
EdU proliferation assay 5-ethynyl-2 0 -deoxyuridine (EdU) uptake assay was also used to determine the cell proliferation. In briefly, cells transfected with shDHX9 or shControl were incubated with EdU (10 mM) for 2 hours in complete medium. After treatment with a fixing solution and a permeabilizing solution, the cells were stained with anti-EdU antibody and Propidium Iodide (PI

Colony formation assay
Cells were plated in 35mm dishes with methylcellulose medium (MethoCultä H4435) containing SCF, GM-CSF, IL-3, and erythropoietin (Stem Cell Technologies, CA) at 800 cells/well in triplicate wells for each condition. After 14 days of incubation in a humidified incubator at 37 C, the colonies containing at least 30 cells were counted.

Apoptosis detection
Annexin V-FITC/PI assay (beyotime, CN) was used to quantitatively detect the cell apoptosis according to the manufacturer's instructions. Annexin V-FITC positive cells were considered to be apoptotic cells.

Measurement of apoptosis-sensitive to apoptosis-inducing agents
SKM-1 and HEL cells transfected with shControl or shDHX9 were cultured for 48 hours in the presence and absence of 5-Azacytidine and ABT-199. 5-Azacytidine and ABT-199 were purchased from Selleck Chemicals (US). Cell apoptosis was analyzed using Annexin V-FITC/PI.

RNA sequencing
cDNA libraries were prepared using TruSeq RNA Sample Preparation Kit (Illumina), according to the manufacturer's protocol. One microgram of total RNA was used for sequencing on a HiSeq 2000 instrument (Illumina) with 100 bp paired end reads, according to the manufacturer's protocol. Data analysis was performed using edgeR27 to evaluate the whole transcript expression (false discovery rate <0.05) and with DEXSeq28 to evaluate differential exon usage (false discovery rate <0.05). The GEPs were identified between the SKM-1 cells with DHX9 knockdown (n = 3) and the control cells (n = 3).

Immunofluorescence
The cells were fixed in 4% paraformaldehyde (PFA) for 20 min, then washed once with PBS and transfer to glass slides. Subsequently, cells were permeabilized with 0.5% Triton X-100 and blocking buffer for 30 min successively, and then incubated with S9.6 antibody (Absolute antibody, Ab01137) overnight at 4 C. iScience Article secondary antibody (Cell Signaling Technology, #4412) for 1 hour. The nuclei were stained with 1 ug/ml DAPI (Sigma-Aldrich, D9542) for 10 min. Images were acquired by using laser scanning confocal microscopy.

DRIP-seq analysis
The total nucleic acids were extracted from the SKM-1 cells transfected with lentivirus. DRIP was immunoprecipitated with S9.6 antibody and IP efficiency was assessed by qPCR. The qualified libraries were sequenced on an Illumina NovaSeq 6000 system. The 5 0 , 3 0 -adaptor bases were trimmed using cutadapt software. The trimmed reads were aligned to human reference genome GRCh38 using bowtie2 software. The aligned reads were used for peak calling of the DRIP regions using MACS2. Statistically significant DRIP-enriched peaks were identified at a p-value threshold of 0.001. The peaks were annotated with the overlapping or nearest gene. Differentially accessible peaks were analyzed using DiffBind.

QUANTIFICATION AND STATISTICAL ANALYSIS
Statistical analysis were conducted using SPSS software version 25.0 and Graphpad Prism 8. The association of mutations with clinical characteristics was analyzed by the chi-squared (c2) test. The Kaplan-Meier test was used for univariate survival analysis. Two independent samples were compared using Student's t-test. Three independent samples were compared using one-way ANOVA. The results were considered statistically significant with a p-value less than 0.05. The corresponding statistical analysis relative to the control group is annotated with an asterisk. *: p < 0.05; **: p < 0.01; ***: p < 0.001.

ll
OPEN ACCESS