The gene expression RNA sequencing data and clinical information of 132 patients with AML from TCGA (28) were obtained from UCSC Xena (https://xenabrowser.net/datapages/?dat-aset=TCGA-LAML.htseq_counts.tsv). The dataset ID in UCSC Xena was TCGA-LAML.htseq_counts.tsv. The RNA expression data were processed by log₂ (count+1) standardization and the genes with >20% missing values (value=0) were removed. Additionally, the clinical information of the patients with AML was collected from TCGA.

Between December 2019 and December 2020, the peripheral blood of 20 healthy individuals and 20 patients with AML was obtained from Kunming Yan'An Hospital (Kunming, China). There were 10 males and 10 females, with a median age of 41 years (range, 8–79 years). The diagnostic criteria of AML patients followed FAB classification (29) and NCCN (National Comprehensive Cancer Network) AML diagnosis and treatment guidelines (30). The present study was approved by the ethics committee of Kunming Yan'An Hospital (Kunming, China; approval no. 2021-03-01). All participants provided written informed consent. The clinical information of the patients with AML was also collected (Table I).

The human acute promyelocytic leukemia cell line (HL-60) was purchased from Cobioer Biosciences Co., Ltd. HL-60 cells were cultured in Iscove's modified Dulbecco's medium (cat. no. 12440053; Thermo Fisher Scientific, Inc.) containing 20% fetal bovine serum (cat. no. 10099141; Thermo Fisher Scientific, Inc.) and 1% penicillin-streptomycin solution (cat. no. C0222; Beyotime Institute of Biotechnology). Cells were grown with 5% CO₂ at 37°C. The HL-60 cells were cultured in hypoxia under 95% N₂ and 5% CO₂. The HL-60 cells were treated with 0.5 µM rotenone (cat. no. R8875; MilliporeSigma) or 1 µM oligomycin (cat. no. SC0366; Beyotime Biotechnology Inc.) at 37°C for 24 h. The endoribonuclease-prepared small interfering RNAs (esiRNAs) targeting human mitochondrion (MT)-ND1, MT-ND4 or MT-ND5 (cat. nos. EHU100901, EHU101821 and EHU101561; MilliporeSigma) were used to silence ND1, ND4 or ND5 in HL-60 cells. esiRNA targeting enhanced green fluorescent protein (cat. no. EHUEGFP; MilliporeSigma) was used as a control. pCMV3-C-GFPSpark (cat. no. CV026; Sino Biological, Inc.) was used as an overexpression vector to construct the pCMV3-ND1/4/5 recombinant plasmid, and pCMV3-untagged (cat. no. CV011; Sino Biological, Inc.) was used as the negative control vector. siRNA/plasmids (20 pmol) were added to 50 µl opti-MEM (cat. no. 11058021; Thermo Fisher Scientific, Inc.) medium without serum. 1 µl Lipofectamine (cat. no. A12621; Thermo Fisher Scientific, Inc.) was also added to 50 µl opti-MEM serum-free medium. The two were mixed and placed at room temperature for 20 min to form the complex. The mixture was added to the cell suspension which was cultured at 37°C and 5% CO₂. After 48 h, other experimental steps were carried out. HL-60 cells stably transfected with siRNA negative control (siNC) or siRNA ND1/4/5 (siND1/4/5) were constructed. siRNA ND1: TGATCAGGGTGAGCATCAAACTCAAACTACGCCCTGATCGGCGCACTGCGAGCAGTAGCCCAAACAATCTCATATGAAGTCACCCTAGCCATCATTCTACTATCAACATTACTAATAAGTGGCTCCTTTAACCTCTCCACCCTTATCACAACACAAGAACACCTCTGATTACTCCTGCCATCATGACCCTTGGCCATAATATGATTTATCTCCACACTAGCAGAGACCAACCGAACC. siRNA ND4: CAGCCACATAGCCCTCGTAGTAACAGCCATTCTCATCCAAACCCCCTGAAGCTTCACCGGCGCAGTCATTCTCATAATCGCCCACGGACTCACATCCTCATTACTATTCTGCCTAGCAAACTCAAACTACGAACGCACTCACAGTCGCATCATAATCCTCTCTCAAGGACTTCAAACTCTACTCCCACTAATAGCTTTTTGATGACTTCTAGCAAGCCTCGCTAACCTCGCCTTACCCCCCACTATTAACCTACTGGGAGAACTCTCTGTGCTAGTAACCACGTTCTCCTGATCAAA siRNA ND5: ACATCTGTACCCACGCCTTCTTCAAAGCCATACTATTTATGTGCTCCGGGTCCATCATCCACAACCTTAACAATGAACAAGATATTCGAAAAATAGGAGGACTACTCAAAACCATACCTCTCACTTCAACCTCCCTCACCATTGGCAGCCTAGCATTAGCAGGAATACCTTTCCTCACAGGTTTCTACTCCAAAGACCACATCATCGAAACCGCAAACATATCATACACAAACGCCTGAGCCCTATCTATTACTCTCATCGCTACCTCCCTG.

A total of 20 BALB/c-nu mice (male, 6 weeks old, 18–20 g) were purchased from SipeiFu Biotechnology Co., Ltd. Nude mice were adaptively fed in a specific pathogen-free environment for 7 days. The study protocol was ethically approved by the Kunming Yan'An Hospital Experimental Animal Ethics Committee (Kunming, China; approval no. 2021016). The whole animal experiment process also followed the Animal Research: Reporting In Vivo Experiments guidelines (31). Mice were randomly divided into a control group (siNC) and an experimental group (siND1/4/5) with 10 mice in each group. The cell suspension (HL-60-siNC or HL-60-siND1/4/5) with 2×10⁶ cells per mouse was injected into the right axillary skin of mice after inhalation anesthesia with 2% isoflurane (cat. no. 26675-46-7; MilliporeSigma). The induction concentration of isoflurane in mice was 4% and the maintenance concentration was 2%. The whole experimental period was 23 days starting with the receipt of nude mice. Mice were raised in a specific pathogen-free level environment with suitable temperature (26–28°C), humidity (40–60%), ventilation (10–15 times per hour) and light (10 h of light, 14 h of no light). Each mouse had an independent sterile cage to ensure sufficient activity space. The feed and drinking water of mice were also pathogen-free and was used for ad libitum feeding. The health and behavior of the mice were observed and recorded every day. Weight loss, loss of appetite, weakness, infection of body organs and excessive tumor volume in mice were regarded as the humane end points. A total of 20 mice were euthanized at the terminal point of the experiment and no mice died during the experiment. All mice were sacrificed using intraperitoneal injection of 200 mg/kg pelltobarbitalum solution (cat. no. P-010; MilliporeSigma) on the 18th day after injection. Before euthanasia, the mice were given orally administrated ibuprofen (40 mg/kg; cat. no. 14883; MilliporeSigma) with water to relieve pain. The criteria for euthanasia were respiratory and cardiac arrest and disappearance of nerve reflex. The long diameter (A) and short diameter (B) of the tumor were measured every 2 days and tumor volume was calculated according to the following formula: V = AB²/2. The tumors were removed surgically and images were captured using a camera (Alpha 7S III; Sony Corporation).

Isolation of protein-coding genes from the RNA expression matrix from the TCGA database was performed using the dplyr and tidyr packages (32,33) in the R programming language (version 3.6.2; http://www.r-project.org/). DEmRNAs were screened using the limma (version 3.50.1) package (34) in R3.6.2 (35) according to the threshold defined as P<0.01 and |log₂^foldchange|>4 (AML group vs. normal group). DEmRNAs were visualized by a heatmap using the pheatmap (version 1.0.12) package (36) in R3.6.2.

The differential expression of ND1, ND4 and ND5 across cancers was analyzed using the Gene Expression Profiling Interactive Analysis (http://gepia2.cancer-pku.cn/#index) online website (37).

Total RNA was extracted from the cell lines or tumor tissues using a Micro Scale RNA Isolation Kit (cat. no. AM1931; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The total RNA concentration was measured by a micro ultraviolet-visible spectrophotometer (NanoDrop™ One; Thermo Fisher Scientific, Inc.). Equal amounts of RNA (2 µg) in each group were reverse transcribed into cDNA using the High Capacity cDNA Reverse Transcription Kit (cat. no. 4374967; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The reverse transcription conditions were as follows: 25°C for 10 min; 37°C for 120 min; and 85°C for 4 min. qPCR assays were performed using a DyNAmo ColorFlash SYBR Green qPCR Kit (cat. no. F416S; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The qPCR conditions were as follows: Pre-denaturation (95°C for 7 min, 1 cycle); denaturation and annealing (95°C for 10 sec, 60°C for 20 sec, 40 cycles); extension (95°C for 15 sec, 60°C for 15 sec, 1 cycle). The following primer sequences were used: ND1 forward, 5′-CAACATCGAATACGCCGCAG-3′ and reverse, 5′-AATCGGGGGTATGCTGTTCG-3′; ND4 forward, 5′-ACAAGCTCCATCTGCCTACG-3′ and reverse, 5′-GAAGCTTCAGGGGGTTTGGA-3′; ND5 forward, 5′-CACATCTGTACCCACGCCTT-3′ and reverse, 5′-TGCTATAGGCGCTTGTCAGG-3′; and GAPDH forward, 5′-GGAGCGAGATCCCTCCAAAAT-3′ and reverse, 5′-GGCTGTTGTCATACTTCTCATGG-3′. GAPDH was the internal reference for the RT-qPCR assay. The relative mRNA expression levels were calculated using the 2^−ΔΔCq method (38).

ROS accumulation was detected using a Reactive Oxygen Species Assay Kit (cat. no. CA1410; Beijing Solarbio Science & Technology Co., Ltd.). Briefly, dichlorofluorescin diacetate (DCFH-DA) was diluted with serum-free medium at a ratio of 1:1,000 to make the final concentration 10 µmol/l. The cells were fixed at room temperature with 95% ethanol for 30 min. After fixation, the cells (10⁶ cells/ml) were suspended in diluted DCFH-DA and incubated in a cell incubator at 37°C for 20 min. The cells were washed three times with serum-free cell culture medium. The staining state of cells was directly observed under a fluorescence microscope (BX53; Olympus Corporation). Quantitative analysis was performed by flow cytometry (Model: BD Accuri C6; BD Biosciences Inc.) with an excitation wavelength of 525 nm and an emission wavelength of 488 nm. The result of the flow cytometry was analyzed using BD Accuri C6 software (version: 1.0.264.21).

ELISA was used to detect the contents of ND1, ND4 and ND5 in peripheral blood of AML patients and healthy people according to the manufacturer's instructions of Human NADH Dehydrogenase Subunit 1, 4 and 5 ELISA kit (cat. nos. abx576646, abx536145 and abx152448; Abbexa, Ltd.). The optical density was determined at OD₄₅₀ nm using an microplate photometer (Multiskan FC; Thermo Fisher Scientific, Inc.).

ATP in cells and tissues was detected using an ATP content detection kit (cat. no. BC0300; Beijing Solarbio Science & Technology Co., Ltd.) according to the manufacturer's instructions. ATP content was detected at a wavelength of 340 nm by an enzyme-labeled instrument (Multiskan FC; Thermo Fisher Scientific, Inc.).

Extracellular oxygen consumption in cells was detected using an Extracellular Oxygen Consumption Assay kit (cat. no. ab197243; Abcam) according to the manufacturer's instructions. The extracellular O₂ consumption signal was measured at 1.5 min intervals for 90 min by an enzyme-labeled instrument (Multiskan FC; Thermo Fisher Scientific, Inc.). The excitation wavelength was 380 nm and the emission wavelength was 650 nm.

Human NADH dehydrogenase (complex I) content in cells and mouse tumor tissues was detected using a Human NADH Dehydrogenase SimpleStep ELISA^® Kit (cat. no. ab178011; Abcam) according to the manufacturer's instructions. The absorbance value was recorded at 450 nm and substituted into the standard curve to calculate the content of complex I.

Cell suspension was cultured in a 96-well plate (100 µl; 5,000 cells per well). The cells were cultured at 37°C with 5% CO₂, and 10 µl CCK-8 solution (cat. no. CA1210; Beijing Solarbio Science & Technology Co., Ltd.) was added to each well. The cells were cultured in the incubator for 4 h. The absorbance at 450 nm was measured by an enzyme-labeled instrument (Multiskan FC; Thermo Fisher Scientific, Inc.).

The R programming language (version 3.6.2; http://www.r-project.org/) and SPSS 15.0 software (SPSS, Inc.) were used for statistical analysis. All experiments were repeated at least three times and data are presented as the mean ± SD. A two-tailed unpaired Student's t-test was used to determine the statistical significance of the experimental results. One-way ANOVA with the Least Significant Difference post hoc test was used for the comparison of multiple groups. P<0.05 was considered to indicate a statistically significant difference.

RNA sequencing data from a total of 132 patients with AML were included in the present study. The average age of the patients with AML was 53 years and only 55 patients were over 60 years old. There were 61 female patients and 71 male patients. Among them, 80 patients succumbed to AML disease and the average survival time was 418 days. The French-American-British classification of patients was as follows: M0, 12 (9.09%); M1, 32 (24.24%); M2, 32 (24.24%); M3, 14 (10.61%); M4, 27 (20.45%); M5, 12 (9.09%); M6, 2 (1.52%); and M7, 1 (0.76%). Detailed clinical information is presented in Table II.

A total of 284 mRNAs were differentially expressed between the normal and AML groups. Among them, 245 (86.27%) mRNAs were upregulated in the AML samples, while 39 (13.73%) were downregulated in the AML samples from TCGA database (Table SI). A heatmap was used to visualize differentially expressed genes (Fig. 1A). The mRNA expression levels of ND1, ND4 and ND5 were upregulated in the AML samples compared with the normal samples (Fig. 1B; P<0.001). Peripheral blood samples were collected from 20 healthy individuals and 20 patients with AML to verify the differential expression of ND1, ND4 and ND5. RT-qPCR results demonstrated that the mRNA expression levels of ND1, ND4 and ND5 were upregulated in the AML samples compared with in the normal samples (Fig. 1C; P<0.05). ELISA results revealed that the secretion levels of ND1, ND4 and ND5 in the serum of patients with AML were higher than those in the normal group (Fig. 1D; ***P<0.001, *P<0.05).

The differential expression of ND1, ND4 and ND5 across cancers was analyzed using the Gene Expression Profiling Interactive Analysis online website. Compared with in normal samples, ND1 was significantly upregulated only in AML samples. ND4 was significantly upregulated in AML and thymoma samples. ND5 was significantly upregulated in AML and kidney chromophobe samples (Fig. 2; P<0.001). Thus, the upregulation of ND1, ND4 and ND5 expression is almost a unique phenomenon in AML.

siRNA-mediated silencing of ND1, ND4, ND5 or ND1/4/5 in HL-60 cells resulted in decreased mRNA expression levels of these genes (Fig. 3A). Additionally, ND1, ND4 and ND5-silenced HL-60 cells showed decreased cell viability, with the triple-KD (siND1/4/5) group showing the largest decrease to 24.22±7.12% compared with the control (siNC) group (Fig. 3B). A xenograft model was used to identify the effect of ND1/4/5 on AML growth in vivo. As shown in Fig. 3C, tumors in the siND1/4/5 group grew slower than those in the negative control group. RT-qPCR was performed on all tumor tissues. RT-qPCR showed that the mRNA expression levels of ND1, ND4 and ND5 were significantly downregulated in the siND1/4/5 group compared with the siNC group (P<0.001; Fig. 3D). The volume of tumors in the siNC was larger than that of siND1/4/5 groups over time (Fig. 3E). The average volume of tumors in the siNC or siND1/4/5 group at day 18 was 1,339.33±247.53 mm³ or 1,004.91±257.25 mm³. Compared with siNC group, the average volume in siND1/4/5 group was decreased (P<0.001; Fig. 3F). The average weight of tumors in the siND1/4/5 group at 18th days was 77.21±16.72% of that in the control (siNC) group (P<0.001; Fig. 3G).

Most tumors can survive and proliferate rapidly under hypoxia (39). The present study demonstrated that the viability of HL-60 cells was decreased under hypoxia; this was 81.89±3.80% of that under normoxia (Fig. 4A). Furthermore, when HL-60 cells were treated with 0.5 µM rotenone or 1 µM oligomycin (respiratory chain inhibitors) for 24 h, cell viability was decreased to 63.56±7.38% and 75.12±2.13%, respectively, compared with that in the control group (Fig. 4B). The present study explored whether ND1/4/5 overexpression could alleviate the negative effects of rotenone. After verification of the overexpression of ND1/4/5 in HL-60 cells transfected with the PCMV3-ND1/4/5 recombinant plasmid, it was found that it could not alleviate the inhibition of cell viability in AML cells caused by rotenone, an inhibitor of respiratory chain complex I (Fig. 4C). However, silencing of ND1/4/5 downregulated the expression levels of complex I in HL-60 cells and mouse tumor tissues (Fig. 4D). Next, the effect of ND1/4/5 on the energy metabolism of AML cells was assessed in terms of ATP production in HL-60 cells or tumor tissues. After silencing of ND1/4/5, ATP production in HL-60 cells or tumor tissue was decreased to 50.48±7.56% or 40.48±21.68% of the control (siNC), respectively (Fig. 4E). After silencing of ND1/4/5, the oxygen consumption of HL-60 cells was decreased significantly to 78.56±8.34% compared with that in the control (siNC) group (Fig. 4F). Similarly, after silencing of ND 1/4/5, the ROS production of HL-60 cells was decreased significantly to 38.26±2.13% of that of the control (siNC) group (Fig. 4G). In conclusion, these data suggest that the silencing of ND1/4/5 can inhibit the proliferation of AML cells and reduce the oxidative phosphorylation.