DNMT3A Low-Expression is Correlated to Poor Prognosis in Childhood B Cell Precursor Acute Lymphoblastic Leukemia and Confers Resistance to Daunorubicin on Leukemic Cells

Background Little is known about DNMT3A expression and its prognostic signicance in childhood B cell precursor acute lymphoblastic leukemia (BCP-ALL). Methods We determined DNMT3A mRNA expression in 102 children with BCP-ALL. Correlations with relapse-free survival (RFS) and common clinical characteristics were analyzed. DNMT3A was stably knocked out by CRISPR/Cas9 gene editing technology in 697 cell line. Cell proliferation activity after treated with daunorubicin was determined by CCK8 assay in DNMT3A KO 697 cell line. Results DNMT3A expression in BCP-ALL patients who were in CCR was higher than in those who got relapse (P=0.0111). Receiver operating characteristic curve showed prognostic signicance of DNMT3A expression (P=0.003). Low expression of DNMT3A (<0.197) was signicantly correlated with poor RFS (P<0.001) in children with BCP-ALL. Knock-out of DNMT3A in 697 cell line signicantly increased IC50 of daunorubicin (P=0.0057), indicating elevated resistance to daunorubicin.


Introduction
B cell precursor acute lymphoblastic leukemia (BCP-ALL) is the most common childhood malignancy.
Although cure rate of childhood BCP-ALL has been greatly improved with risk-adjusted therapy [1,2], relapsed leukemia is still a leading cause of death for children mainly due to therapy resistance [2][3][4].
Thus, it is of importance to clarifying the mechanisms of therapy resistance and relapse of BCP-ALL.
DNA methyltransferase 3A (DNMT3A) catalyzes de novo DNA methylation and plays important roles in the pathogenesis of malignancies including leukemia. Furthermore, DNMT3A mutations in acute myeloid leukemia (AML) and T cell ALL are associated with poor prognosis of the patients [5][6][7][8][9][10][11]. Our previous studies have shown that DNMT3A mutations can be found in a few of children with BCP-ALL, and are correlated with poor prognosis [12]. However, the expression level of DNMT3A and its prognostic signi cance in BCP-ALL remains unclear.
In this study, we assessed the relationship between expression level of DNMT3A and prognosis in Chinese childhood BCP-ALL. Moreover, CRISPR/Cas9 has been used to knock out DNMT3A gene in leukemic 697 cell line in order to explore the role of DNMT3A expression playing in resistance to chemotherapeutic drugs. We showed that low expression of DNMT3A was correlated with poor treatment outcome, knock-out of this gene resulted in obvious resistance to daunorubicin (DNR), a common chemotherapeutic drug in treatment of ALL.

Patients
A total of 102 BCP-ALL patients with available diagnostic bone marrow (BM) samples were enrolled in this study. These patients were diagnosed and treated in accordance with the Chinese Children's Leukemia Group ALL 2008 Protocol (CCLG-ALL 2008) from July 2010 to May 2014 at Beijing Children's Hospital [4]. There were 64 boys and 38 girls, aged from 1 to 13 years with a median age of 4. Thirty-four patients carried 4 types of fusion genes including TEL-AML1, E2A-PBX1, BCR-ABL, and TLS-ERG. The details of strati cation and treatment according to CCLG-ALL 2008 were described previously [4,13]. Ninety-four patients were in continuous complete remission (CR), 8 patients relapsed 2 to 62 months after diagnosis. The follow-up time ranged from 1 to 92 months (median, 59 months). Additionally, BM samples from 11 patients at continuous CR were collected and used as control.
MRD at d33 (the end of induction of remission) and d78 (before consolidation therapy) were detected using RQ-PCR targeted at Ig/TCR (immunoglobulin and T cell receptor gene rearrangements) according to European MRD laboratory guidelines [14][15][16][17].
Informed consents were obtained from all the children's parents or legal guardians.

Cell lines
Human BCP-ALL cell line 697, as a kind gift from Dr. Suning Chen at the rst a liated Hospital of Soochow University (Suzhou, China), was cultured in RPMI 1640 (GIBCO, USA) supplemented with 10% fetal bovine serum (FBS, AusGeneX, Brisbane) and 1% penicillin/streptomycin. HEK293T cell, kindly provided by Dr. Fen Chang at Peking University Health Science Center (Beijing, China), was cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. All cells were maintained at 37°C in a humidi ed atmosphere containing 5% CO2.

Quantitative analysis ofDNMT3Aexpression
Real-time quantitative polymerase chain reaction (RQ-PCR) was performed using Power SYBR™ Green PCR Master Mix (Applied Biosystems 4367659) by an ABI Prism 7500 Sequence Detection System (Applied Biosystems, Foster City, CA, USA). GUS (β-Glucuronidase) expression was used as an internal control. The cycling condition included pre-denaturation at 95°C for 30 seconds, followed by 40 cycles of 5 s at 95°C, 30s at 55°C and 30s at 72°C. Primers are shown in Table 1. We used the cDNA samples obtained from 697 cell line as a calibrator. The relative expression of DNMT3A was calculated by the method of 2 −ΔΔCt . The levels of DNMT3A and GUS were tested in triplicates.

Lentivirus production and infection
To produce lentivirus, 6.0µg of transfer plasmid lentiDNMT3A-sgRNA-Cas9 or control plasmid lenti-CRISPR-v2 were co-transfected into HEK 293T cells with 4.5µg of packaging plasmids psPAX2 (AddGene 12260) and 3.0µg of VSV-G (AddGene 8454) using FuGENE® 6 Transfection Reagent (Promega E2692) according to the manufacturer's instructions. After incubation for 48h or 72h, the culture supernatants containing lentivirus were harvested and ltered with 0.45µm lter and stored at − 80°C. The 697 cell line (5 × 10 5 ) was infected with the lentivirus at an M.O.I. of 40, using spin-transduction (centrifuging the plate coated with 8µg/ml polybrene (SANTA CRUZ) at 1200g for 2 hours at 25°C), then were cultured for 24 hours in the incubator. On the next day, the medium was changed with fresh RPMI 1640 complete medium and the cells were cultured for another 24 hours.

T7EN1 assays for quantifying frequencies of indel mutations
Lentivirus-infected cells were selected by 1µg/ml puromycin for 2 days. Genomic DNA was extracted and used to amplify the genomic region anking the DNMT3A sgRNA target site with KAPA2G Robust HotStart ReadyMix (KAPA BIOSYSTEMS KK5702) and PCR primers listed in Table 1. Then T7EN1 assay was performed using T7 Endonuclease I (NEB #M0302L) according to the Instruction Manual. The digested DNA was analyzed on electrophoresis system using a 2% agarose gel.

Statistical analysis
Receiver Operating Characteristic (ROC) curve was used to decide the cut-off value of low-and highexpression of DNMT3A (DNMT3A low and DNMT3A high ) in leukemic cells of children with BCP-ALL. Fisher's exact test was used to test the differences in clinical characteristics and relapse rates between DNMT3A low and DNMT3A high patients. Relapse free survival (RFS) was de ned as the date of leukemia diagnosis to the date of recurrence. Survival estimates were calculated using the Kaplan-Meier method, and the groups were compared using the log-rank test. The independent prognostic signi cance of DNMT3A expression and the common clinical features was analyzed by Cox proportional hazards model (Method: Enter). All data were analyzed with the SPSS 16.0 software package and a P value < 0.05 was considered statistically signi cant. The tting curves of inhibitory effects of DNR on cell proliferation were plotted by GraphPad Prism 8, and half maximal inhibitory concentration (IC50) was also calculated by the software.

DNMT3A expression in childhood BCP-ALL
Firstly, we determined DNMT3A expression in 102 newly diagnosed (ND) BCP-ALL patients and 11 patients with non-malignant hematological disease (control) by relative quantitative PCR. As a result, DNMT3A expression in ND patients with BCP-ALL, ranged from 0.0006594 to 1.713 with a median of 0.4363, was signi cantly higher than that in control patients (range: 0.08055 to 0.1865, median: 0.1147; P = 0.0004, Fig. 2a). Interestingly, DNMT3A expression in ND BCP-ALL patients who got relapse was signi cantly decreased compared with that in patients who were in continuous complete remission (CCR) at the last follow-up (P = 0.0111, Fig. 2b).
Low expression of DNMT3A indicated poor prognosis in ND BCP-ALL patients ROC curve was used to examine the prognostic value of DNMT3A expression in BCP-ALL patients. It was found that the area under curve (AUC) was 0.819 (P = 0.003, Fig. 3a), which indicated that DNMT3A expression could be a potential prognostic biomarker for ND BCP-ALL patients.
A cut-off of 0.197 was chosen as both the sensitivity (75.0%) and the speci city (87.2%) were acceptable. We then divided 102 patients into two groups, 22 cases with low DNMT3A expression (≤ 0.197, DNMT3A low ) and 80 cases with high DNMT3A expression (> 0.197, DNMT3A high ). There was a signi cantly higher relapse rate in DNMT3A low group (6 out of 22 vs 2 out of 88, Fisher's exact test, P = 0.001). Moreover, poor RFS was observed in the patients of DNMT3A low group (P < 0.001) (Fig. 3b).
Furthermore, Cox proportional hazards regression analysis indicated the independent prognostic signi cance of DNMT3A expression with common prognostic factors as multivariates (HR = 19.195, 95% CI: 3.159 ~ 116.651, P = 0.001, Table 2). These ndings indicated that low expression of DNMT3A in leukemic cells at diagnosis could be a useful indicator for disease relapse in childhood BCP-ALL. In the next step, we analyzed the correlation of DNMT3A expression with common clinical characteristics such as age, gender, white blood cell (WBC) count at diagnosis and fusion genes. However, no correlation was found between DNMT3A expression and above clinical characteristics (Table 3). We further analyzed the association of DNMT3A expression with MRD at d33 and MRD at d78 respectively, but no signi cant correlation between them was found (Fisher's exact test, P > 0.05, Table 3).
Knock-out of DNMT3A enhanced resistance of 697 cell line to DNR To con rm the correlation of low expression of DNMT3A with poor prognosis of children with BCP-ALL, rstly, we disrupted DNMT3A in 697 cell line. T7 endonuclease I (T7EN1) assay showed high e ciency of the sgRNA to direct Cas9-mediated ablation of DNMT3A (Fig. 4a). Furthermore, as expected, Western blotting indicated that DNMT3A expression was remarkably reduced after infection with DNMT3A-sgRNA lentivirus (Fig. 4b).
DNR is one of the main chemotherapeutic drugs in induction therapy of BCP-ALL. We next tested whether knock-out of DNMT3A gene could cause 697 cell line to be tolerant to DNR by CCK8 assay. These cells were treated by different concentrations of DNR for 24h. It was shown that IC50 was signi cantly increased in the DNMT3A-knockout cells, indicating decreased cell viability (Fig. 5, Control vs. DNMT3A KO, 0.06449 vs. 0.1052 µg/ml, P = 0.0057). These results demonstrated that sgRNA mediated Cas9 knock-out of DNMT3A can causes 697 cell line to be resistant to DNR, implying that DNMT3A expression plays an important role in the sensitivity of leukemic cells to chemotherapeutic drugs such as DNR.

Discussion
In recent two decades, DNMT3A mutations have been found in approximately 20% of adult AML patients, 9% of adult T-ALL and 0 ~ 1.4% of childhood AML, and the hotspots of mutations are mainly located in exon 23 which encodes the catalytic methyltransferase domain [5][6][7][8][9][10][11]. DNMT3A mutations are associated with poor prognosis and used for risk strati cation in AML [5][6][7][8][9][10], and is associated with increased age and adverse outcome in adult T-ALL [11]. However, few studies focused on the role of DNMT3A in BCP-ALL. Our previous study has shown that DNMT3A mutations can be found in exon 23 and its adjacent intron regions in a few of children with BCP-ALL (5/182, 2.7%), and may have adverse impact on prognosis [12].
As there are only a few BCP-ALL patients with DNMT3A mutations, we sought to determine the prognostic signi cance of DNMT3A expression in BCP-ALL. As expected, low expression leading to decreased methyltransferase activity was associated with relapse in 102 patients with BCP-ALL. Furthermore, knockout of DNMT3A increased IC50 of DNR in 697 cell line, indicating the relationship of low expression of DNMT3A and chemoresistance.
It was reported that Dnmt3a loss in HSCs leads to hypomethylation of genes with a causal role in cancer, such as Runx1 and Gata3. Runx1 promotes murine erythroid progenitor proliferation and inhibits differentiation by preventing Pu.1 downregulation [19]. Gata3 targets Runx1 in the embryonic hematopoietic stem cell niche [20]. Thus, previous studies and ours suggest that deletion or lowexpression of Dnmt3a result in differentiation inhibition of HSCs and allow HSCs to be propagated inde nitely in vivo [21][22][23][24], which may play an important role in leukemogenesis and resistance to chemotherapy. This may provide us with an alternative target of therapy for childhood BCP-ALL.
It has been reported that DNMT3A expression is directly transactivated by transcription factor WT1 in Wilms' tumor cells [25] and is negatively regulated by p53 at the transcriptional level in lung cancer [26]. In fact, overexpression of the WT1 transcript was demonstrated in children with B-ALL at diagnosis [27], which may contribute to increased expression of DNMT3A in leukemic cells. However, increased expression of TP53 by 2 to 20-fold higher in pediatric primary B-ALL than in healthy controls [28] would inhibit DNMT3A expression. Thus, the regulation mechanism of DNMT3A expression is quite complicated in different types of cancer cells and needs to be clarifed especially in childhood BCP-ALL.
In summary, we associated low expression of DNMT3A with poor prognosis in Chinese pediatric patients with B-ALL and resistance to DNR. Furthermore, successful disruption of DNMT3A in 697 cell line may facilitate the studies on mechanism of relapse and chemotherapeutic resistance for childhood BCP-ALL.

Declarations
Ethics approval and consent to participate: Yes Consent for publication: Yes Availability of data and materials: Emails to corresponding author Zhigang Li