The implication of lncRNA expression pattern and potential function of lncRNA RP4‐576H24.2 in acute myeloid leukemia

Abstract Background Recent studies have revealed that long noncoding RNAs (lncRNAs) may hold crucial triggers of the pathogenesis of hematological malignancies, while the studies evaluating the expression pattern of lncRNA in acute myeloid leukemia (AML) are few. Thus, this study aimed to investigate the implication of lncRNA expression pattern in AML development and progression. Methods Bone marrow samples from four AML patients and four controls were subjected to lncRNA sequencing. Then, bone marrow samples from 110 AML patients and 40 controls were proposed to real‐time quantitative polymerase chain reaction (RT‐qPCR) validation for 10 candidate lncRNAs. Clinical data and survival profiles were recorded in AML patients. Furthermore, lncRNA RP4‐576H24.2 expression in AML cell lines and its effect on AML cell proliferation and apoptosis were detected. Results LncRNA expression pattern by sequencing clearly distinguished AML patients from controls, and 630 upregulated and 621 downregulated lncRNAs were identified in AML patients compared to controls, which were mainly enriched in AML oncogene‐related biological process and pathways (such as neutrophil degranulation, leukocyte transendothelial migration, and hematopoietic cell lineage). RT‐qPCR validation observed that six lncRNAs correlated with AML risk, one lncRNA associated with risk stratification, and three lncRNAs correlated with survivals, among which lncRNA RP4‐576H24.2 was the only one correlated with AML susceptibility, risk stratification, and survivals. Further in vitro experiments showed that lncRNA RP4‐576H24.2 was upregulated in AML cell lines compared to normal bone marrow mononuclear cells (BMMCs), and promoted proliferation while inhibited apoptosis in HL‐60 and KG‐1 cells. Conclusions LncRNA expression pattern is closely involved in the development and progression of AML, and several specific lncRNAs exhibit potential to be biomarkers for AML risk and prognosis. Besides, lncRNA RP4‐576H24.2 might be a potential oncogene in AML pathogenesis.


| INTRODUCTION
Acute myeloid leukemia (AML), a complex hematological malignancy characterized by its heterogenetic cytology, presents poor long-term outcome in the majority of adult patients. 1 Although AML patients' prognosis has improved, the proportion of patients achieving long-term survival is still insufficient referring to that only 40% in younger adult and 10% in the elderly obtain long-term survival. 1,2 Therefore, management of AML patients pivots around enhancing recovery and survival, in which discovering assistant prognostic factors to enhance the management of AML patients plays a relatively crucial role. Besides the chromosomal dysregulations and gene mutations that have been used in risk stratification, several other genetic factors also present with potentials in AML risk and prognosis prediction, which include diversified kinds of noncoding ribonucleic acid (ncRNA). 3,4 Long noncoding RNA (lncRNA) is a class of RNAs with more than 200 nucleotides which have less protein-coding capacities, those molecules function as key regulators in cellular activities (such as the process of cell cycle, differentiation, and imprinting). [5][6][7][8][9] Recent studies have revealed that lncRNAs may hold crucial triggers of the development and progression of hematological malignancies. 10 In AML, the studies aiming at investigating the role of lncRNA are limited, while the findings are minimal yet intriguing. Previous studies reveal that some specific lncRNAs might play crucial roles in regulating the AML cell functions or chemoresistance, for instance, a previous study reports that lncRNA urothelial carcinoma-associated 1 (UCA1) knockdown inhibits chemoresistance via repressing glycolysis by mediating the microRNA (miR)-125a/hexokinase 2 signaling pathway. 11,12 However, the existed studies mostly focus on investigating the roles of several specific/individual lncRNAs in AML, such as lncRNA antisense noncoding RNA in the INK4 locus (ANRIL), lncRNA UCA1 and so on, while the studies assessing the lncRNA expression pattern in AML are few. 11,13 Herein, we conducted this study to investigate the implication of lncRNA expression pattern in AML pathogenesis and the potential of several specific candidate lncRNAs as markers for AML risk and prognosis, and further explore the effect of lncRNA RP4-576H24.2 on regulating AML progression.

| Patients and controls
About 110 de novo AML patients between July 1, 2015 and June 31, 2018 were consecutively enrolled in our study. The inclusion criteria were: (a) Diagnosed as de novo AML patients according to the World Health Organization (WHO) Classification of Tumors of Hematopoietic and Lymphoid Tissues (2008); (b) Age above 18 years. The exclusion criteria were: (a) Secondary or mixed AML; (b) History of other solid tumor or hematological malignancies; (c) Previous treatment with chemotherapy or radiotherapy. Meanwhile, 40 participants who were adult bone marrow donors or adult patients with nonmalignancy diseases (such as thrombocytopenic purpura and myelofibrosis) who underwent bone marrow biopsy were enrolled as controls. This present study was approved by the Ethics Committee of our hospital, all participants (AML patients and controls) provided written informed consents.

| Data collection, sample acquisition, treatment, and assessment
Age, gender, French-American-British (FAB) classification, white blood cell (WBC), cytogenetics, molecular genetics, and risk stratification (according to National Comprehensive Cancer Network [NCCN] Clinical Practice Guidelines in Oncology of AML [Version 2, 2013]) of AML patients were documented. And, bone marrow samples were obtained from AML patients during biopsy. Patients were treated in accordance with NCCN Clinical Practice Guidelines in Oncology of AML (Version 2, 2013). Treatment remission status of induction chemotherapy was assessed, and complete remission (CR) was defined as bone marrow with at least 20% cellularity and BM blasts below 5% at steady state after treatment, without cytological evidence of leukemia, no transfusion requirement, leukocyte count above 1 × 10 9 /L, and platelet count above 100 × 10 9 /L. Patients were then followed up, the last follow-up date was June 31, 2018 and the median follow-up duration was 17.0 (range 1.0-36.0) months, and event-free survival (EFS) as well as overall survival (OS) were calculated. The EFS was defined as the time interval from initiation of treatment to disease recurrence, progression, or death, and the OS was defined as the time interval from initiation of treatment to death. Besides, bone marrow samples were also acquired from controls by biopsy as well.

| RNA sequencing process
Four de novo AML patients and four age-and gender-matched controls were randomly selected from total participants, and their bone marrow samples were subjected to RNA sequencing by Genergy Bio Company. In brief, (a) Total RNA was extracted from bone marrow using PureZOL RNA isolation reagent (Bio-Rad), and then concentration, purity, and integrity were assessed and adjusted; (b) Ribosomal RNA (rRNA) was removed by Epicentre Ribo-zero™ rRNA Removal Kit (Epicentre) and the remaining RNA was proposed to generate sequencing library according to the methods in a previous study 14 ; (c) The library was then sequenced on Illumina Hiseq X10 platform (Illumina); (d) Trimmed reads were mapped to the human genome Hg38 by HISAT2 with the default parameters; (e) The gene (lncRNAs and mRNAs) read count was calculated using FeatureCounts.

| Bioinformatics
The following bioinformatic analyses were carried out based on the sequencing data of lncRNA and mRNA expression patterns: (a) Principal component analysis (PCA) of gene expression pattern (lncRNA and mRNA individually) was performed by Stats package; (b) Heatmap analysis of gene expression pattern (lncRNA and mRNA individually) was performed by Pheatmap package; (c) Dysregulated genes (lncRNA and mRNA individually) were analyzed by DeSeq2 package and showed as volcano spot with the statistical significance defined as P value <.05 and the biological significance defined as a difference of fold change (FC) above 2.0 times; (d) Heatmap analysis of dysregulated genes (lncRNA and mRNA individually) was performed by Pheatmap package; (e) Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of dysregulated lncRNAs were performed using DAVID web server according to their regulated mRNAs, afterward, GO and KEGG enrichment analysis of dysregulated mRNAs were also performed; (f) Regulatory network of top 40 dysregulated lncRNAs (20 upregulated and 20 downregulated) was drawn by Igraph package based on their regulated mRNAs; (g) Circos graph for transcription and regulation information was drawn by RCircos package. R Software (Version 3.5.3) (Lucent Technologies) was used for bioinformatic analysis. The genes (lncRNAs and mRNAs) which were identified in 50% or above samples were included in the bioinformatics analysis.

| RT-qPCR
First of all, the total RNA was isolated from bone marrow samples or cells using PureZOL RNA isolation reagent (Bio-Rad) and was evaluated by a spectrophotometer. Then, the total RNA was reversely transcribed into complementary DNA using the QuantiNova Reverse Transcription Kit (Qiagen). Subsequently, PCR was performed by QuantiNova SYBR Green PCR Kit (Qiagen) and the results were calculated with 2 −△△Ct formula using U6 or GADPH as an internal reference. And, the primers used in RT-qPCR are listed in Table 1.

| LncRNA and mRNA expression patterns in AML patients
PCA plot analysis disclosed that lncRNA expression pattern could differentiate AML patients from controls ( Figure 1A), and heatmap also showed that lncRNA expression pattern distinguished AML patients from controls with a good intra-group correlation in both AML patients and controls ( Figure 1B).  Furthermore, PCA plot disclosed that AML patients and controls could be distinguished by the mRNA patterns as well ( Figure 1C), and heatmap also displayed that mRNA patterns could differentiate AML patients from controls with good introgroup correlation in both groups ( Figure 1D). Then the volcano plot identified 630 upregulated lncRNAs and 621 downregulated lncRNAs in AML patients compared with controls ( Figure 2A), and heatmap for dysregulated lncRNAs displayed stratifying intra-group correlation in both AML patients and controls ( Figure 2B). Similarly, volcano plot identified 1357 upregulated mRNAs and 790 downregulated mRNAs in AML patients than those in controls ( Figure 2C), and heatmap of dysregulated mRNAs also displayed that the intro-group correlation was good in both AML patients and controls ( Figure 2D).
These results indicated that the lncRNA expression profile is implicated in the development of AML.

| Enrichment analysis of dysregulated lncRNAs/mRNAs
Then, for the purpose of evaluating the potential regulatory role of dysregulated lncRNAs and mRNAs in AML pathogenesis, enrichment analysis was performed, and the GO enrichment analysis revealed that the dysregulated lncRNAs were enriched in biological processes (such as neutrophil degranulation, inflammatory response, and immune response), cellular component (including secretory granule membrane, lysosome, and ficolin-1-rich granule lumen), and molecular functions (including voltage-gated cation channel activity, glycoprotein binding, and proteoglycan binding) ( Figure 3A), and the KEGG analysis showed that the dysregulated lncRNAs were mainly correlated with pathways related to AML pathogenesis including leukocyte transendothelial migration pathways, hematopoietic cell lineage pathways, and apoptosis pathways ( Figure  3B). As for the dysregulated mRNAs, GO enrichment analysis presented that they were mainly associated with biological processes (such as neutrophil degranulation, regulation of immune response, and inflammatory response), cellular component (such as extracellular exosome, external side of plasma membrane, and cell surface), and molecular functions (such as metal ion binding, peptide antigen binding, and antigen binding) ( Figure 3C), and the KEGG enrichment analysis disclosed that the dysregulated mRNAs were enriched in several pathways that were correlated with AML pathogenesis, such as hematopoietic cell lineage, graft-versus-host disease, and allograft rejection ( Figure 3D).

| Regulatory network of top 40 dysregulated lncRNAs
Top 20 upregulated lncRNAs and top 20 downregulated lncRNAs were selected based on the rank of the absolute value of Log 2 FC, which are presented in Table 3. Afterward, the regulatory network of the trans-and cis-regulations of these dysregulated lncRNAs with mRNAs is presented in Figure 4. Then, the Circos graph for transcription and

| Correlations of 10 candidate lncRNA expressions with AML risk stratification
Only lncRNA RP4-576H24.2 expression in bone marrow was correlated with AML risk stratification, to be exact, the level of lncRNA RP4-576H24.2 was the highest in patients

| DISCUSSION
Profiting from the development of RNA sequencing and bioinformatics analysis, investigation of the heterogenetic pathology of AML has become more profound, among which the role of ncRNAs in development and progression of AML has been increasingly revealed in recent decades, including forming regulatory networks that could mediate multiple pathways related to AML. [15][16][17] Nonetheless, previous studies investigating the roles of lncRNAs in AML only begin recently, and are quite limited, even though lncRNAs have been found to be crucial genetic factors in other malignancies. 18,19 In terms of the studies including RNA sequencing in AML, in a study using miRNA sequencing and transcription factor (TF) activity array, 308 dysregulated miRNAs and 84 dysregulated TFs are detected, then 1462 miRNA-target gene pairs, 982 TF-target gene pairs, and 196 TF-miRNA pairs are identified subsequently; after emerging as a regulatory network of these dysregulated miRNAs and TF, the KEGG pathway analysis finds that the network is markedly enriched in 33 pathways with the AML-related pathways the most significant. 20 Another previous study performs whole-genome microarrays in extramedullary infiltration (EMI) AML patients and non-EMI AML patients, and discloses that 253 circular RNAs (circRNAs) and 663 mRNAs are upregulated, but 259 circRNAs and 838 mRNAs are downregulated in EMI AML patients compared to non-EMI AML patients, then further enrichment analysis finds that these dysregulated circRNAs and mRNAs are mainly enriched in cell adhesion, migration, signal transduction, and cell to cell communications. 21 With respect to lncRNA, there are also studies evaluating its expression pattern in AML. However, these studies are very few. A previous study using microarray and bioinformatics analyses reveals that in pediatric AML patients, 372 lncRNAs and 136 mRNAs are found to be dysregulated in patients compared with normal controls, and further RT-qPCR validates that the most dysregulated lncRNAs in pediatric AML patients is lnc RNA ENST00000435695. 22 And, another study assessing the lncRNA expression profile in cytogenetically normal AML patients observes a specific lncRNA expression profile which is dependent on the mutational status of NPM1 gene, and finds 12 lncRNAs being able to distinguish NPM1-mutated patients from NPM1 wild-type patients; in addition, they also discover that lncRNA XLOC_109948 is associated with drug resistance and prognosis. 23 However, these previous studies which investigate lncRNA expression pattern in AML are performed in different patient cohorts, such as the pediatric AML patients (vs healthy children) and cytogenetically normal AML patients (vs cytogenetically abnormal AML patients). And the study focusing on the implication of lncRNA expression pattern in adult AML patients (vs adult controls) is not reported yet. In this study, we found that the lncRNA expression pattern could differentiate adult AML patients from controls, then 630 upregulated lncRNAs and 621 downregulated lncRNAs in AML patients were identified, and further enrichment analysis showed that these dysregulated lncRNAs were enriched in AML pathology-related biological processes (such as neutrophil degranulation, inflammatory response as well as immune response), and AML-related signaling pathways (including leukocyte transendothelial migration, hematopoietic cell lineage, apoptosis and so on), and were predominantly located in cell lysosomes and extracellular exosomes. Our results indicated that these dysregulated lncRNAs might act as crucial players in AML pathology mainly through mediating neutrophil degranulation and pathway of leukocyte transendothelial migration in cell lysosomes and extracellular exosomes. The poor prognosis of AML has highlighted the need for searching more novel and reliable biomarkers which can assist in the management of AML patients, and several specific lncRNAs | 7157 ZHENG Et al. have been proposed to exhibit prognostic value. For instance, the lncRNA LINC00265 is reported to be upregulated in bone marrow in AML patients than that in controls and is associated with worse OS in AML patients. 24 Another recent study elucidates that higher levels of lncRNA CCAT1 and lncRNA PVT1 in peripheral blood mononuclear cells are associated with poor prognosis in AML patients. 25 In addition, a recent study reports that lncRNA in nonhomologous end joining (NHEJ) pathway 1 (LINP1) is upregulated in pediatric and adolescent AML patients, and promotes the malignant behaviors in AML cells via regulating the HNF4α/AMPK/WNT5A signaling pathway. 26 Another study illuminates that lncRNA HOXB-AS3 enhances AML cell proliferation and its elevated expression predicts worse prognosis patients with AML and myelodysplastic syndrome. 27 While the systemic investigation of multiple lncRNAs as biomarkers for AML risk and prognosis is not conducted yet. In our study, 10 candidate lncRNAs that were selected from RNA sequencing results were further validated by RT-qPCR in 110 AML patients and 40 controls, and their correlations with disease risk and prognosis were assessed, which elucidated that there were six candidate lncRNAs correlated with AML risk, regulating AML pathogenesis could be found in any previous study. As for the possible explanations to our results, it might include: 1) lncRNA RP4-576H24.2 may promote development and progression of AML through regulating multiple downstream genes as displayed in our RNA sequencing results, in which several genes regulated by lncRNA RP4-576H24.2 are reported to participate in the pathogenesis of leukemia, such as PTPRJ, FCN1, KLF4, G0S2, and LILRB2. [28][29][30][31][32] Therefore, ln-cRNA RP4-576H24.2 might promote the development or progression of AML by mediating the leukemia pathology-related genes, such as PTPRJ, FCN1, KLF4, G0S2, and LILRB2; 2) our further cell experiments disclosed that lncRNA RP4-576H24.2 promoted proliferation while inhibited apoptosis of two AML cell lines, indicating that lncRNA RP4-576H24.2 might aggravate the progression of AML through regulating AML cell functions.
In addition, there were several issues that need to be illustrated in this study. First, there were patients with FAB classification of M1 and M6 in the 110 AML patients in RT-qPCR validation; however, no patients with FAB classification of M1 or M6 were included in the four AML patients in RNA sequencing. This heterogeneity might result from that the sample size was much larger in RT-qPCR validation and all the patients were enrolled consecutively, thus, there were patients with more diverse FAB classifications in RT-qPCR validation. However, we further analyzed the difference of candidate lncRNA expressions among patients with different FAB classifications, which disclosed that candidate lncRNA expressions did not vary among patients with different FAB classifications. These results indicated that the FAB classification was not a factor affecting the candidate lncRNA expressions in AML patients. Second, lncRNA LINC01262 and lncRNA OR7E140P expressions were markedly dysregulated in AML patients compared with controls in RNA sequencing, while did not show any difference between AML patients and controls in RT-qPCR validation, which might derive from that: (a) the results in RNA sequencing might be affected by a singular value due to that the sample size for RNA sequencing was small; (b) the RNA sequencing technique was less accurate than RT-qPCR, which might also contribute to this conflict.
Furthermore, the underlying mechanisms of several specific lncRNAs in AML pathogenesis are also reported previously. For instance, knockdown of lncRNA ZFAS1 represses the AML progression through mediating the miR-150/Sp1 and miR-150/Myb pathways. 33 And lncRNA ANRIL enhances the malignant cell survival and cell glucose metabolism of AML by regulating the AdipoR1/AMPK/SIRT1 pathway. 13 In addition, lncRNA UCA1 expression is increased in AML pediatric patients after adriamycin (ADR)-based chemotherapy, and knockdown of lncRNA UCA1 inhibits chemoresistance to ADR in AML cells via mediating miR-125a/hexokinase 2 pathway. 11 And lncRNA TUG1 suppresses AML cell proliferation while promotes apoptosis through targeting aurora kinase A. 12 Furthermore, a recent experiment reveals that lncRNA SNHG1 promotes cell proliferation through mediating miR-488-5p/NUP205 axis in AML. 34 In this study, due to that lncRNA RP4-576H24.2 was the only candidate lncRNA correlated with AML risk, risk stratification, and prognosis of AML, we further performed in vitro experiments to explore the effect of lncRNA RP4-576H24.2 on AML cell functions, which showed that lncRNA RP4-576H24.2 was upregulated in AML cell lines compared with normal BMMCs, and it promoted cell proliferation as well as repressed cell apoptosis in two AML cell lines (HL-60 cells and KG-1 cells), which elucidated the potential mechanisms of lncRNA RP4-576H24.2 in regulating the AML progression. To the best knowledge of ours, the lncRNA RP4-576H24.2 was reported in AML for the first time by our study, and the probable explanations to the results in our in vitro experiments might be due to that lncRNA RP4-576H24.2 could promote progressive cell functions via targeting multiple mRNAs, which was also shown in the regulatory network of lncRNA-mRNA in our results.
In conclusion, the lncRNA expression pattern is closely involved in the development and progression of AML, and several specific lncRNAs exhibit potential to be biomarkers for AML risk as well as prognosis. Besides, lncRNA RP4-576H24.2 might be a potential oncogene in AML pathogenesis. These data would shed light on the potential role of lncRNA in AML pathology and the application of lncRNA as a novel biomarker for AML management.