Fetal bovine serum, cell culture medium 1640, DMEM were obtained from Gibco and purchased from Thermo Scientific (USA). The lentivirus was purchased from Miaoling Biology (China). Experimental animals were purchased from Beijing Vital River Laboratory Animal Technology (China). The primers used were purchased from Keyybio (China). A Cell Counting Kit-8 was purchased from Vazyme Biotech (China). TRIzol was purchased from Thermo Scientific (USA). Annexin V-FITC/PI double-staining cell apoptosis detection kit was purchased from Bestbio (China). RT-qPCR reagent was purchased from ACCURATE BIOTECHNOLOGY(HUNAN) CO., LTD (China). All the above reagents were used according to the manufacturer’s instructions. RNA-seq was performed by Shanghai Biotechnology Corporation.

Samples and data were collected from AML patients with newly diagnosis and healthy donors at Qilu Hospital of Shandong University. All relevant procedures for patient sample collection were approved by the Research and Ethics Committee of Qilu Hospital of Shandong University, and all patients who provided clinical samples signed informed consent forms. The patient’s information can be found in Supplementary Table S1 . All the procedures were performed in accordance with the 1964 Declaration of Helsinki principles and its later amendments or comparable ethical standards.

Human acute myeloid leukemia cells (THP-1 and Molm13) were obtained from China National Collection of Authenticated Cell Cultures, human umbilical vein endothelial cell HUVEC was obtained from Shandong University of Physiology Teaching and Research Office; human embryonic kidney cell HEK-293T was obtained from China National Collection of Authenticated Cell Cultures, KMT2A::MLLT3-GFP cells were donated by the Institute of Hematology, Chinese Academy of Medical Sciences. The cells were cultured in an incubator with a continuous supply of 5% CO2 under humidified conditions (75-80% relative humidity). THP-1 cells, Molm13 cells and HUVECs were grown in RPMI-1640 culture medium supplemented with 10% fetal bovine serum (FBS), and HEK293T cells were grown in DMEM supplemented with 10% FBS. These cell lines were identified by short tandem repeats (STR) and stored in the laboratory of Hematology, Qilu Hospital, Shandong University. All cell lines have been authenticated using STR within the last three years.

Total RNA was extracted using the TRIzol method according to the manufacturer’s instructions and reverse-transcribed into cDNA using SYBR^® Green Premix Pro Taq HS. Quantitative real-time PCR was performed using TB Green Premix Ex Taq™ II on a Light Cycler 480 II (Roche). GAPDH was used as an endogenous control. The data were analyzed using the 2^-△△CT method. The primers used for RT−qPCR were as follows: UNC5B for human: forward: 5′′-GTCGGACACTGCCAACTATAC-3′, reverse:5′-CCGCCATTCACGTAGACGAT-3′. GAPDH for human: forward: 5′- TGCACCACCAACTGCTTAGC-3′, reverse: 5′-GGCATGGACTGTGGTCATGAG-3′. MPZL1 for human: forward: 5′-GGCAGAGAATCCTCACCAGT-3′, reverse: 5′-GTGTGAGCAGCTTCCTTCAG-3′, CLDN23 for human: forward: 5′-AGTGGACGTGGAGTTGTACC-3′, reverse: 5′-AGCGAGGTGACCATGAGTG-3′, IGF2 for human: forward: 5′- ATGACACCTGGAAGCAGTCC -3′, reverse: 5′-TGGGTGGGTAGAGCAATCAG-3′, WNT7B for human: forward: 5′-GCAGTGCAACTGCAAATTCC-3′, reverse: 5′-CACTTGCAGGTGAAGACCTC-3′.

6-8 week-old male C57BL/6 mice were selected, with 5-6 mice in each group. All animals were kept in specific pathogen-free (SPF) level animal rooms at the Animal Center of Qilu Hospital, Shandong University. All animal experiments were approved by and conducted under the guidance of the Animal Ethics Committee of Shandong University Qilu Hospital.

For the AML mice model, KMT2A::MLLT3 cells with lentivirus transfection (Vector & Over-expression(OE)-UNC5B) were injected into C57BL/6 mice via tail vein (1 × 10⁷ cells per mouse for homing analysis and 5 × 10⁶ cells per mouse for before and after treatment analysis). Homing analysis was performed after 16 hours, before and after treatment analysis was performed after 7 days. We injected Ara-C and doxorubicin (DOX) intraperitoneally for three days after day 7. On day 11, we harvested the unilateral femur bone marrow cells and splenocytes. The percentage of tumor burden was analyzed by flow cytometry.

For the tool cell line HEK-293T, PMD2g and PspaX2 packaging plasmids, the vector plasmid pLV3-CMV-UNC5B-3×FLAG-mCherry-Puro, and Lipofectamine 2000 were used for virus packaging. The viruses were transduced, and the collected supernatant was centrifuged and filtered through a 0.45 µm filter. For lentivirus infection, UNC5B virus was added directly to the cells. Screening was performed with 1μg/mL of puromycin.

For the detection of cell adhesion, a total of 5×10⁵ AML cells were seeded in 6-well plates covered with HUVECs. After direct contact for 6 h, the culture supernatant was removed, 2 mL of PBS was slowly added along the edge of the plate, the nonadherent cells were washed off, and the cells were washed again. The PBS was removed, a fluorescence microscope was used to view the cells, and the number of mCherry-positive cells adhering to the HUVECs was counted.

UNC5B was overexpressed in AML cells, Molm13 and THP-1 cells were seeded on 96-well plates, and the proliferative ability of the control group was examined via a CCK-8 kit, after which the proliferation of the cells was measured at 0, 24, 48, 72, and 96 h. In the CCK-8 assay, WST-8 is the key chemical. It is reduced by NAD+ to a water-soluble yellow formazan product in the presence of viable cells and active metabolism. The more live cells present, the more formazan is produced, resulting in a darker color. The viability of AML cells treated with Ara-C and DOX were detected via CCK-8 assays. We have set the drug concentrations for Ara-C and DOX based on previous research (17). Molm13 and THP-1 cells were seeded on 96-well plates, Molm13 and THP-1 cells were treated with Ara-C and DOX doses of 0.01 μM, 0.05 μM, 0.25 μM, 1.25 μM, and 6.25 μM for each drug. The absorbance at 450/630nm was measured with a microplate reader.

The percentage of apoptotic cells was measured using an Annexin V-FITC/PI double-staining cell apoptosis detection kit. Q2: (AnnexinV+FITC)+/PI+; Q3: (AnnexinV-FITC)+/PI-; Q4: (AnnexinV-FITC)-/PI-. Molm13 and THP-1 cells were stained according to the kit manufacturer’s protocol and analyzed using flow cytometry. The apoptotic rate was defined as the percentage of early and late apoptotic cells among all cells in the sample (Q2 and Q3, respectively).

For RNA-seq, the VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina^® (Vazyme, #NR604-01) was used to establish the library. The cells in each group were sequenced with an Illumina HiSeq 2000 platform, and the length of the single-end sequences ranged from 200-300 bp. Sequence reads were compared with version 38 of the human genome (GRCH38) using a standard Illumina sequence analysis pipeline.

In brief, the gene barcode count matrix for the GSE178912 dataset was analyzed using the Seurat R software package (version 4.0.2). The GSE178912 dataset contains single-cell RNA-Seq data of residual AML cells purified from patient-derived xenograft bone marrow, with or without treatment using venetoclax plus cytarabine. Cells with > 200 genes and < 10% mitochondrial gene expression were screened via downstream analysis. After the samples were converted to Seurat, the combined Seurat data were normalized and scaled by regression of UMI count and mitochondrial gene percentage. In terms of dimension reduction, the Find Variable Genes function was used to identify the most variable genes. PCA was subsequently used for dimensionality reduction, and the UMAP graph was generated by the Run UMAP Seurat function (Seurat version 3.1.3). GEPIA (Gene Expression Profiling Interactive Analysis) is an online bioinformatics tool that uses RNA sequencing data from 9,736 tumor tissues and 8,587 normal tissues in the TCGA and GTEx databases. It retrieves gene expression values in various tumor samples, calculates gene expression levels in specific tumors, and analyzes the relationship between genes and tumor prognosis, as well as gene co-expression. (http://gepia.cancer-pku.cn/).

The sample size was chosen to be n ≥ 3. The data were analyzed and are expressed as the medium ± range, mean ± standard deviation. Unpaired t-test were used to analyzed the two groups as indicated; two-way ANOVA were used to compare the inter- and intragroup differences in two factors between two groups as indicated. All the above data were analyzed and plotted using GraphPad Prism 9.0 (La Jolla, CA), and P < 0.05 indicated statistical significance. Flow analysis was performed using FlowJo software (TreeStar, USA) and R software (version 4.2.3). The workflow for animal experiments and Graphical Abstract were illustrated by Figdraw.

UNC5B expression was found to be lower in AML patients than in healthy donors according to analysis of the GEPIA database ( Figure 1A ). Moreover, to verify this conclusion, we collected information from AML patients ( Supplementary Table S1 ) and healthy donors to analyzed the relationship between UNC5B expression levels in both groups. The expression levels of UNC5B were significantly lower in the AML group than in the healthy control group, which corresponds to the GEPIA results. An analysis of patient clinical data revealed an association between UNC5B expression and relapse or CR rate: low UNC5B expression indicated a higher relapse rate in AML patients ( Figure 1B ).

First, we searched the Gene Expression Omnibus (GEO) database and selected a dataset of AML patients, including those who underwent different treatment methods. Such as single-agent application of Cytarabine and Venetoclax; or combination therapy of Cytarabine and Venetoclax ( Supplementary Table S2 ). The following results below are based on the GEO dataset. Based on this, we used R language to analyze AML patients included in the dataset before and after treatment to examine UNC5B expression and its relationship with drug resistance in AML cells. The growth protocol of the dataset we have selected from GEO is the process of isolating and purifying human AML cells, xenotransplanting them into 8-week-old NSG mice for amplification, the human AML cells from the mice’s bone marrow were isolated and purified to conduct single-cell sequencing. The lack of immune cells in NSG mice made this data more accurately to demonstrate the effects of various drugs on leukemia cells. We conducted single-cell analysis on the sorted leukemia cells from the dataset, we divide AML cells which have received different treatments and who have not received treatment into 15 AML clusters ( Figure 1C ). Meanwhile, to verify that these clusters are all AML cells, we used the same marker, CD33, as the source data and verified that these clusters were all indeed leukemia cells ( Supplementary Figure S1 ). The treatment types in this dataset include single-agent application of Cytarabine and Venetoclax; or combination therapy of Cytarabine and Venetoclax ( Figure 1D ). By observed the changes in the cell proportion in each AML clusters before and after chemotherapy drugs, we found that the proportion of two clusters, clusters 4 and 6, was significantly decreased after treatment ( Figures 1E, F ). Therefore, we consider that these two groups of cells, which are significantly reduced after chemotherapy, will be called chemotherapy-sensitive cells, while the other cells were designated as drug-resistant cell subpopulations ( Figure 1G ). To further investigate the role of UNC5B in AML, we analyzed the expression levels of UNC5B in six groups of AML cells and found that the expression of UNC5B in the chemotherapy group were lower than untreated group ( Figure 1H ). Moreover, we extracted drug-resistant cells in the untreated group and compared the expression of UNC5B in the drug-resistant cells in other chemotherapy groups. For the drug-resistant cells, the expression of UNC5B was significantly reduced after treatment ( Figure 1I ). These findings suggested that decreased expression of UNC5B may enhance the drug resistance of resistant cells in AML cells. Taken together, these results support our conclusion that UNC5B expression is low in AML and is associated with drug sensitivity. So, overexpression of UNC5B may increase chemotherapy sensitivity.

In order to prove whether UNC5B is related to chemotherapy sensitivity of AML, we used in vitro and in vivo experiments to verify. In vivo experiment, we constructed a mouse model of KMT2A::MLLT3 leukemia. We isolated spleens from KMT2A::MLLT3 mice and obtained leukemia cells. These cells were then transfected with UNC5B and control vectors in vitro. After KMT2A::MLLT3 leukemia cells were transfected with UNC5B lentivirus, they were injected into C57BL/6 mice via the tail vein. Bone marrow from the homing model was obtained at 16 h, and a before and after treatment model with DOX + Ara-C injection was used to assess whether UNC5B influenced chemosensitivity in AML cells ( Figure 2A ). We performed in vivo data analysis of the homing model and found that UNC5B overexpression decreased the percentage of GFP-positive AML cells in the bone marrow and spleen of mice, indicating that UNC5B overexpression suppressed the homing of AML cells to the bone marrow and spleen ( Figures 2B, C ).

Next, we used DOX in combination with Ara-C as the chemotherapeutic regimen and assessed how UNC5B expression affected chemosensitivity in vivo. We found that the overexpression of UNC5B significantly increased the sensitivity of AML model mice to chemotherapeutic agents, and the size and weight of the spleen significantly decreased ( Figures 2D, E ). The tumor burden was reduced in the bone marrow and spleens of AML mice which were treated with chemotherapeutic drugs ( Figures 2F, G ). These results might indicate that UNC5B improved chemosensitivity in AML cells in vivo.

Cell adhesion refers to the use of specific surface adhesion molecules (AMs) attached to adjacent cells or the cell matrix. It is a key link in the occurrence and development of AML. At present, an increasing number of studies have focused on leukemia-targeted adhesion (16). An essential characteristic of tumor cells is that they continually proliferate and affect cancer progression (17). Strategies for targeting and inhibiting the proliferation of tumor cells are of great significance in inhibiting the development of AML. To further clarify the mechanism of drug sensitivity, we performed further experiments. By coculturing with HUVECs, we found that the adhesion of AML cells in the UNC5B-overexpressing group was reduced compared to the control group, which is consistent with the results of our in vivo experiments ( Figures 3A, B ). Next, we performed a proliferation assay to examine the proliferative capacity of 0-96 h Molm13 cells and THP-1 cells by a CCK-8 assay. Moreover, the ability of cells to proliferate was significantly lower in the UNC5B-overexpressing group than in the control group, supporting our view that UNC5B overexpression suppressed the adhesion and proliferation of cells ( Figures 3C, D ). To further verify the role of UNC5B in chemosensitivity, we used Ara-C and DOX to treat Molm13 and THP-1 cells. The viability rate of Ara-C and DOX-treated cells in the overexpressing group was lower than that of PBS-treated controls under Ara-C and DOX drug treatment, thus, UNC5B overexpression improved the chemosensitivity of AML cells ( Figures 3E, F ). Afterwards, we tested the apoptotic ability of AML cells under low-dose drug treatment and found that after treatment with low-dose chemotherapy drugs, the apoptosis of AML cells overexpressing UNC5B increased ( Figures 3G, H ). Finally, we conducted an apoptosis assay to measure the apoptotic rate of Molm13 and THP-1 cells without drugs using flow cytometry. The results showed that there was no significant difference in the apoptosis rate between the UNC5B overexpressing group and the control group ( Supplementary Figure S3 ). Overexpression of UNC5B can inhibit cell adhesion and proliferation, and reduce cell activity under the pressure of chemotherapy, which making leukemia cells more sensitive to the effects of chemotherapy.

To verify whether UNC5B affects adhesion and proliferation-related genes, we performed an RNA-seq analysis. After overexpression of UNC5B, the expression levels of many genes were altered ( Figure 4A ). Based on these differentially expressed genes, we conducted GO pathway enrichment analysis and found that the main altered pathways were related to adhesion and proliferation related ( Figure 4B ). Moreover, we found that the adhesion and proliferation molecules were significantly altered in these pathways. We evaluated molecules in adhesion- and proliferation-related pathways, such as myelin Protein Zero-Like Protein (MPZL1), Claudin-23 (CLDN23), Wnt Family Member 7B (WNT7B), and Insulin-like growth factor 2 (IGF2), and found that they significantly decreased after overexpressing UNC5B ( Figure 4C ). We then analyzed AML and normal samples from the TCGA and GTEx databases, focusing on the expression levels of four genes. Our analysis revealed that these genes were expressed at higher levels in AML patients compared to normal samples ( Supplementary Figure S4 ). Subsequently, we validated the expression of these molecules using RT−qPCR experiments. The results confirmed that UNC5B could affect the development of AML by inhibited the expression of adhesion and proliferation genes of AML cells ( Figures 4D, E ).