Circ-Foxo3 is positively associated with the Foxo3 gene and leads to better prognosis of acute myeloid leukemia patients

Background The Foxo3 gene, belonging to the forkhead family, is one of the classes of transcription factors characterized by a forkhead DNA-binding domain, which usually considered being a cancer suppressor gene. Circ-Foxo3 is a circular structure which connects the 3’end to the 5’end. Scholars detected that circ-Foxo3 could compete with Foxo3 for binding to some miRNAs. Methods In this study, we will test the expression of Foxo3 and circ-Foxo3 in de novo acute myeloid leukemia (AML) patients to explore the relationship between Foxo3 gene and circ-Foxo3. All the de novo AML samples and normal control samples was measured by real-time quantitative PCR. A receiver operating characteristic curve was conducted to differentiate AML patients from control people. Association of Foxo3 expression and overall survival was conducted by Kaplan-Meier survival analysis. Results We found that the expression of Foxo3 gene in de novo patients was significantly lower than control samples (P = 0.009). Meanwhile, circ-Foxo3 also expressed lower in de novo AML patients than in control samples (P = 0.040). In different classifications, this trend could be observed more remarkably. In non-M3 patients, the Foxo3 high patients’ survival time was longer than Foxo3 low patients (P = 0.002). Besides, in non-favorable risk groups, patients with low expression of Foxo3 had longer survival time than Foxo3 high patients (P = 0.004). Furthermore, in normal Karyotypic patients, the overall survival time of patients with high-expressed Foxo3 was significantly longer than those with low expression (P = 0.034). Besides, Pearson analysis was also conducted between these two genes in AML patients. Results revealed that they were positively correlated (R = 0.63, P < 0.001). Conclusion In conclusion, we found that low expression of circ-Foxo3 and Foxo3 were frequent in AML patients, and patients with high expression of Foxo3 often had a trend of better prognosis.

The Foxo3 gene, belonging to the forkhead family, is one of the classes of transcription factors characterized by a forkhead DNA-binding domain. This family includes four genes: Foxo1, Foxo3, Foxo4 and Foxo6, but the Foxo3 was widely studied. It was a key regulator in the insulin/insulin-like growth factor-1signaling pathway [5][6][7]. It can promote people's health through regulating stress resistance, metabolism, cell cycle and cell apoptosis. Meanwhile, it could make positive response to environmental stimuli and prevent people from suffering the disease which related to aging, such as cancers, cardiovascular disease (CVD) etc. [5,8]. Researchers considered it as a longevity gene [9].
In cancer development, it has been supervised that elevation of AKT activity or deficiency of PTEN often resulted in down-regulation of Foxo3, and accelerated the formation of tumors [10,11]. Therefore, Foxo3 gene was taken as a tumor suppressor gene.
Circular RNA (circRNA), one of the non-coding RNAs, is a circular structure that connects the 3′ end to the 5'end [12]. It was first observed in the cytoplasm of eukaryotic through electron microscopic in 1979 [13]. In 2012, after a statistical analysis of RNA-Seq data and subsequent biochemical analysis, Salzman J. et al. found that circRNA molecules transcribed and spliced from exons in protein and noncoding genes were ubiquitous in the human and mouse genomes [14]. Moreover, studies also reviewed that the copy number of circRNA was 10 times larger than that of linear RNAs. So, they inferred that this was not a coincidence but a potential biological function in cells. Whereas, some studies have shown that some circRNA had multiple binding sites of microRNAs (miRNAs), sponging micro-RNA, and served as a competitive inhibitor for microRNA [15]. For example, the circ-ITCH, containing binding sites for miR-7, miR-17 and miR-214, inhibited the activity of these microRNAs. In 2015, a research group in Toronto found a circRNA called circ-Foxo3, and it promoted cell apoptosis through p53 and puma signal pathway. Scholars detected that ectopic-expressed circ-Foxo3 could compete with some miRNAs and next adjust Foxo3 expression [16]. It was widely accepted that the post-transcriptional repression of Foxo3 expression was regulated by both cir-cRNAs and microRNAs [11,17].
In this study, we tested the expression of Foxo3 and circ-Foxo3 in de novo AML patients, and conducted survival analysis on the expression level and prognosis.

Patients and samples
The bone marrow (BM) samples of experimental group were collected from patients in the Affiliated Hospital of Jiangsu University who were initially diagnosed with acute myeloid leukemia, while the control group were from healthy BM donors or patients with chest trauma. We tested Foxo3 gene in 122 de novo AML patients and 30 control samples. Because some of these samples had run out and new samples were collected, the circ-Foxo3 gene was tested in 116 de novo AML patients and 24 control samples. These patients were accepted standard treatment after diagnosis of AML. The AML patients were classified by FAB classification and the 2008 WHO criteria. The therapeutic regimen and results of laboratory and equipment inspection were recorded in corresponding physician's order sheet and medical record. All the BM donators had signed the informed consents. This study was approved by the Review Committee of the Ethics Department of the People's Hospital of Jiangsu University. For non-APL patients, who were below 60 years old, induction therapy for them consisted of 4 treatment options: a. Standard-dose cytarabine 100-200 mg/m 2 continuous infusion × 7 days with idarubicin 12 mg/m 2 or daunorubicin 60-90 mg/m 2 × 3 days (for patient ≤45 y). b. Standard-dose cytarabine 200 mg/m 2 continuous infusion × 7 days with daunorubicin 60 mg/m 2 × 3 days and cladribine 5 mg/m 2 × 5 days (for other age groups). c. High-dose cytarabine (HiDAC) 2 g/m 2 every 12 h × 6 days or 3 g/m 2 every 12 h × 4 days with idarubicin 12 mg/m 2 or daunorubicin 60 mg/ m 2 × 3 days. d. Fludarabine 30 mg/m2 IV days 2-6, cytarabine 2 g/m 2 over 4 h starting 4 h after fludarabine on days 2-6, idarubicin 8 mg/m 2 IV days 4-6, and G-CSF SC daily days 1-7 [18].
Definition of AML risk classification [18] Cell lines and cell culture In our laboratory, human hematological cell lines (K562, U937, NB4, SHI-1, and HEL) were purchased from American Type Culture Collection Manassas, VA, USA. The cell line information can be queried from two databases: American Type Culture Collection (ATCC) and Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ). The ATCC number of K562, U937 and HEL were CCL-243, CRL-1593.2 and TIB-180. Details can be found on ATCC Official website. The number of SHI-1 and NB4 were ACC 645 and ACC 207, and details can be queried on DSMZ official website. The use of cell lines was approved by the Review Committee of the Ethics Department of the People's Hospital of Jiangsu University. These cell lines were cultured in RPMI-1640. The mycoplasma contamination of cell lines was negative by using quick mycoplasma test kit. Before we conducted this subject, these cell lines were tested by PCR.

RNA isolation, reverse transcription and real time quantitative PCR
The mononuclear cells of AML patients and healthy donors were isolated from bone marrow (BMMC) using the Ficoll-Hypaque gradient. According to the manufacturer's instructions, Trizol reagent (Invitrogen, Carlsbad, CA, USA) was employed to extract total RNA from BMMC and leukemia cell lines. The cDNA was composed through reverse transcription on the iCycler Thermal Cycler (Eppendorf, Hamburg, Germany) using a reaction mixture, which contains 2 μg of total RNA, dNTPs 10 mM, random hexamers 10 μM, RNAsin 80 units, and 200 units of MMLV reverse transcriptase (MBI Fermentas, Hanover, USA). The system of reverse transcription was incubated for 10 min at 25°C, 60 min at 42°C, and then stored at − 20°C. The primers of Foxo3 were 5′-GCAAGAGCTCTTGGTGGA TCATCAA-3′ (forward) and 5′-TGGGGCTGCCAGGC CACTTGGAGAG-3′ (reverse), and the primers of circ-Foxo3 were 5′-GTGGGGAACTTCACTGGTGCTAAG-3′ (forward) and 5′-GGGTTGATGATCCACCAAGAGCTC TT-3′ (reverse). A 20 μL volume of reaction system (20 ng of cDNA, 0.8 μM of primers, 10 μM AceQTMqPCR SYBR Green Master Mix (Takara Shuzo Co, Ltd., Nojihigashi 7-4-38,Kusatsu,Shiga,Japan) and 0.4 μM ROX Reference Dye1(Invitrogen).) was used to perform Real-time RT-PCR. Amplification for circ-Foxo3 was carried out at 95°C for 30 min, followed by 45 cycles at 95°C for 5 s, 68.7°C for 30 s 72°C for 30 s and 80°C for 31 s. Meanwhile, amplification for Foxo3 was carried out at 95°C for 30 min, followed by 45 cycles at 95°C for 5 s, 63.6°C for 30s 72°C for 30 s and 80°C for 31 s. All reactions were performed on a 7500 Thermo cycler (Applied Biosystems, CA, and USA). Positive and negative controls were included in all tests. Following real-time RT-PCR, a melting curve analysis was carried out to demonstrate the specificity of the PCR product as a single peak. The relative levels of Foxo3 transcript were calculated by the following equation: N Foxo3 = (E Foxo3 ) ΔCT Foxo3 (control-sample) ÷ (E ABL ) ΔCT ABL (control-sample) . The parameter efficiency (E) was counted by the formula E = 10 (− 1/slope) (the slope referred to CT versus cDNA concentration plot) .
Foxo3 and circ-Foxo3 were detected by high-resolution melting analysis (HRMA) as reported previously. DNA direct sequencing was used for confirming positive samples.

Statistical analysis
All the data analysis was conducted on SPSS 20.0 software package (SPSS, Chicago, IL). Pearson chisquare analysis or variance test was used to distinguish differences in categorical variables, and the differences between two groups of continuous variables were compared by Mann-Whitney U test. In order to make a difference between the expression of AML patients and controls, Receiver operating characteristic curve (ROC) and area under the ROC curve were performed. We plotted ROC curves of AML and healthy people, and next a table of sensitivity and (1specificity) of circ-Foxo3 and Foxo3 expression was drawn. We used these data to calculating the Yoden index (sensieivity+specificity-1) which could indicate the capacity of true patients and non-patients. The number which corresponding to the largest Yoden index was chosen as the cut-off value to separate AML patients into high and low groups. We used COX regression model, also known as the "proportional hazards model" (Cox model), to test if some common factors had effect on survival outcome, and if they were independent variables, This statistical method can simultaneously analyze the influence of many factors on the survival period. The impact of different Foxo3 and circ-Foxo3 expression on overall survival (OS) of AML patients was analyzed by Kaplan-Meier analysis. For all analyses, two-tailed P-values of 0.05 or less were determined statistically significant.

The diagnostic value of Foxo3 and circ-Foxo3 expression
The diagnostic value of Foxo3 and circ-Foxo3 expression was analyzed by ROC curve. AML can be remarkably distinguished from controls with an AUC of 0.655 (95% CI: 0.556-0.753; P = 0.009) (Fig. 2a). According to the result of ROC curve analysis, we determined that the cut-off value of Foxo3 expression was at 0.856, the sensitivity and the specificity were 44.7 and 87.7%, respectively. Similarly, the circ-Foxo3 could differentiate AML patients from control people with an AUC of 0.633 (95% CI: 0.523-0.746; P = 0.041) (Fig. 2b). In a similar way, the cut-off value of circ-Foxo3 expression was at 0.233, the sensitivity and the specificity were 62.1 and 75%, respectively.

The correlation between Foxo3 and circ-Foxo3
In order to analysis the correlation of Foxo3 and circ-Foxo3, Pearson analysis was conducted on the expression of Foxo3 and circ-Foxo3 in cell lines (K562, U937, NB4, SHI-1, and HEL). The result showed that Foxo3 and circ-Foxo3 were positively corrected (R = 0.98, P < 0.0021) (Fig. 3a). Besides, Pearson analysis was also Fig. 1 Expression of Foxo3 and circ-Foxo3 in BMNCs was measured via using RQ-PCR in healthy controls and the whole AML samples. Horizontal lines represent the median, and each dot represents an individual sample. Statistical analysis was performed using Wilcoxon tests, and significance was defined as P < 0.05 conducted between these two genes in AML patients. Results revealed that they are positively corrected as well (R = 0.63, P < 0.0001) (Fig. 3b).

Clinical and laboratory characteristics of AML
Based on the cut-off value of 0.856 of Foxo3 expression, AML patients was divided into two groups, high expression group (Foxo3 high ) and low expression group (Foxo3 low ). The high level was over 0.856 while the low expression was below. In the same way, according to the expression of circ-Foxo3, at the cut-off value of 0.233, the patients were also separated into two groups, circ-Foxo3 high and circ-Foxo3 low group.  (Table.2).

Correlation between circ-Foxo3 expression and clinical outcome
We also studied the clinical data of patients with high and low level of circ-Foxo3 expression. Results showed that patients with high level of circ-Foxo3 expression survived longer (95% CI, 11.28-25.16 months, median value, 8 months) than low level of that (95% CI, 11.93-24.07 months, median value, 7 months), although this was out of statistical significance (P = 0.762). More than that, the difference of overall survival time between high and low level of circ-Foxo3 expression in different patients group also had no statistical significance. Multivariate analysis (COX regression model) was also conducted, applying age (≤ 60 y vs. > 60 y), sex (male vs. female), WBC (≥ 30 × 10 9 /L vs. < 30 × 10 9 /L), HB (< 110 g/L vs. ≥110 g/L), PLT (100 × 10 9 /L vs. 100 × 10 9 /L), karyotype classifications (favorable vs. intermediate vs. poor), gene mutations (mutant vs. wild-type) and Foxo3 expression status (high vs. low) as covariates. The result demonstrates that, in our study, these covariates above did not affect the prognosis of patients.

Discussion
Both circular Foxo3 (circ-Foxo3) and linear Foxo3 (Foxo3 mRNA) are encoded by Foxo3 gene [19]. Recently, Foxo3 gene was widely believed to be a tumor suppressor gene in many cancers, like breast [20,21] and ovarian [22] cancer. Besides, circ-Foxo3 was found to down-regulated in patient tumor samples and in a group of cancer cells [20]. Currently, AML has been internationally recognized to be a malignant clonal disease with multiple prognoses: 5-year overall survival was less than 50%, and only 20% of older patients will survive within 2 years of diagnosis [2,23] while there are many treatment programs for this. Associated with karyotypes and molecular mutations, AML patients can be assorted into diverse prognostic risk groups [24]. To investigate whether Foxo3 and circ-Foxo3 were two factors which affect the prognosis of AML, the expression of Foxo3 and circ-Foxo3 was tested in AML patients. In our results, it showed that expression of Foxo3 (P = 0.009) and circ-Foxo3 (P = 0.04) was both down regulated in AML patients. Then we conducted Kaplan-meier survival analysis, we found that there was a tendency that Foxo3 high group patients had longer survival time than Foxo3 low group in these patients we tested, but it was not statistically significant (P = 0.192) (Fig. 4). As we mentioned above, AML patients can be assorted into different classifications according to different basis for grouping. In non-M3, non-favorable and normal karyotype group, Foxo3 high group patients survived longer than Foxo3 low group obviously (Fig. 5). Similarly, the LFS time was compared between Foxo3 high and Foxo3 low patients. The result showed that Foxo3 high group patients had longer LFS time than that of Foxo3 low group in normal karyotype, non-M3 and non-favorable risk group (Fig. 6). At last, Kaplan-Meier analysis was also conducted on intermediate patients below 65 years. But it showed no statistical significance (Fig. 7).
Kaplan-meier analysis was further performed, and it found out that there was no statistical significance correlation between circ-Foxo3 expression and OS time of AML patients. Maybe it was due to the small amount of specimens we used in the experiments, so that the results could not represent general characteristics. In the future, we will collect more patient samples and test the level of circ-Foxo3 expression. According to our current experimental results, we were unable to conclude that circ-Foxo3 was a prognostic factor that affecting survival time of AML patients.
William W Du et al. discovered that, after treated with H 2 O 2 , Cisplatin and Doxorubicin, the expression of circ-Foxo3 was up-regulated in cancer cell lines (66C14, 4 T1, MDA-MB-468, and MDA-MB-231) [20]. They also found that MB-231 cells transfected with siRNA-targeting circ-Foxo3 could decrease Foxo3 level [20]. Moreover, William W Du et al. discovered that ectopic circ-Foxo3 could increase Foxo3 level in MB-231 cells [20], and which means the expression of the circ-Foxo3 and Foxo3 was positively correlated. Not surprisingly, in our research, we found that same phenomenon in cell lines. The expression of Foxo3 and circ-Foxo3 were positively corrected not only in cell lines (K562, U937, NB4, SHI-1, and HEL) (R = 0.98, P < 0.0021) (Fig. 3), but also in AML patients (R = 0.63, P < 0.0001) (Fig. 3). Results showed that the expression of Foxo3 in AML patients was significantly down-regulated. Previous studies revealed that up- Fig. 4 The impact of Foxo3 expression on overall survival of the in AML patients. Survival analysis was performed via Kaplan-Meier survival analysis, with differences between curves analyzed via a logrank test(P = 0.192). Significance was defined as P < 0. 05 regulated expression of circ-Foxo3 triggered stressinduced apoptosis and inhibited tumor growth. In normal breast mammary tissues, higher circRNA was observed, and it appears to be inversely correlated with the risk-of-relapse proliferation score for proliferation genes in breast cancer [20]. It has been demonstrated that up-regulation of Foxo3 can promote apoptosis by upregulated pro-apoptotic Bcl-2 family members (Bim), and leading to an increase of cell cycle protein levels in p27kip1 [25][26][27]. Xinbo et al. hold the view that Foxo3 can promote apoptosis through activating pro-apoptotic proteins (like Bim and Bad), death receptor ligands (like Fas); tumor necrosis factor related apoptosis-inducing ligand and cyclin-dependent kinase inhibitors [28]. So we could infer that low level of circ-Foxo3 down-regulated Foxo3 and inhibited these pro-apoptotic factors so that patients with low level expression of circ-Foxo3 have shorter survival time.
Besides, Fei et al. found that low level of Foxo3 was reported to be related to chemotherapy resistance and associated with poor prognosis of ovarian cancer patients [22]. In fact, the recovery of Foxo3 expression had been developed for some mechanismbased anticancer therapies [29,30]. Furthermore, Foxo3 was widely considered to be an important factor affecting the efficacy of various chemotherapy drugs. Coincidently, in our study, patients with high level of Foxo3 had better prognosis than those with low level. On clinical, the chemotherapy drugs that Fig. 5 The impact of Foxo3 expression on overall survival of the different groups of AML patients. a non-M3 patients. b Normal Karyotypic patients. c Non-Favorable patients. Survival analysis was performed via Kaplan-Meier survival analysis, with differences between curves analyzed via a log-rank test. Significance was defined as P < 0. 05 commonly used were Daunorubicin, Cytarabine, Thioguanine, Harringtonine, Mitoxantrone, etc. The patients, in our medical center, were all treated by chemotherapy. Maybe patients with high level of Foxo3 were more sensitive to these drugs. After standard chemotherapy, patients with high level of Foxo3 expression often lived longer than those with low level. We may infer that lower Foxo3 expression contribute patients to drug resistance. In order to validate our conjecture, we will design and complete more experiments in the future. Combined with above consequence, we may speculate that AML  patients with high expression of Foxo3 were sensitivity towards current chemotherapy. Our experimental results were exactly in line with the above results. Moreover, not only circ-RNA and linear gene can affect the outcome of AML cancers, but also micro RNA can make contributions. Some reported that Foxo family transcripts were firmly regulated by the microRNA networks in cancer progression and metastasis [31][32][33]. For example, reduced miR-215 expression predicts poor prognosis in patients with acute myeloid leukemia [34]. As we mentioned above, circular RNAs can sponge microRNA, and serves as a competitive inhibitor for microRNA [15]. After analyzing our results of experiments, we found that Foxo3 and circ-Foxo3 gene were both down-regulated and positively corrected in AML patients. Combined with the existing theory, we speculated that the down-regulated circ-Foxo3 could release some relevant mircoRNA, and contribute to the low expression of Foxo3 correspondingly; so that these pro-apoptotic factors would be suppressed. At last, patients with low level of circ-Foxo3 and Foxo3 would have adverse prognosis.
We had tested some miRNA in AML, but we have not yet found a miRNA to be correlated with circ-Foxo3 and Foxo3. Based on the results, we will study some other microRNAs and analysis the relation of circ-Foxo3, microRNA and Foxo3 gene. In the future, we will conduct more researches in detail to verify current conclusions. We hope our effort could make sense for diagnosis and therapy for AML patients.

Conclusion
In summary, we could conclude that circ-Foxo3 and Foxo3 were negatively correlated and low level of Foxo3 gene was a frequent molecular event which leads to adverse prognosis of AML patients.