Human AML cell lines THP-1, OCI-AML3, KG-1α, MV4-11, and Molm13 were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The cells were cultured in RPMI 1640 medium (Gibco, USA) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin/streptomycin (Gibco, USA). Guide RNA sequences for CRISPR-Cas9 were designed using GenScript CRISPR Design Tool. Two oligonucleotides targeting K-Env were then inserted. The first gRNA (#1) sequence was 5'-tattccagtcacattgtaactgg-3,’ and the second gRNA (#2) was 5'-tgacagggttaagatttgcgagg-3.’ Lentiviruses were prepared by packaging plasmids with PMD2G and PAX2 along with the target plasmid containing gRNA, and then transfected into THP-1 and MV4-11 cells. The knockout efficiency was verified by qPCR and western blot at 3–4 days after transfection. In vitro experiments, including flow cytometry, LDH assays, and MTT assays, were conducted 5–7 days after the knockout results were verified.

Flow cytometry was performed using a NovoCyte flow cytometer (ACEA Biosciences Inc.). Commercial antibodies were purchased and the following information was obtained: caspase-1 (FAM-FLICA Caspase 1 YVAD Assay Kit, Cat ICT97#), Annexin V (Annexin V, Cat 640,920), and 7-AAD (Biogems, Cat 61,410–00).

Cells from KO and control groups were collected and seeded evenly in 96-well plates. Cell supernatants were collected and LDH release was measured using an LDH assay kit (MCE, Cat HY-K1090). The absorbance at 490 nm was recorded using a microplate reader.

Total RNA was extracted from the cells using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. Total RNA (1 μg) was reverse transcribed using a reverse transcription kit (Takara RR036A), and the resulting cDNA served as the template. Quantitative real-time PCR (qRT-PCR) was performed on a LightCycler 480 system using premix (Takara RR820A) following the manufacturer’s instructions. Relative gene expression levels were calculated using the 2^−△△Ct method. The primers used in this study are listed in Table 1.

Total protein was extracted from cultured cells using the M-PER Mammalian Protein Extraction Reagent (Invitrogen) supplemented with 1% protease and a phosphatase inhibitor cocktail (Invitrogen). After centrifugation at 12,000 rpm for 15 min at 4 °C, protein concentrations were determined using the BCA Protein Detection Kit (Beyotime). Equal amounts of proteins were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride (PVDF) membranes (Bio-Rad Laboratories, Hercules, CA, USA). The membranes were incubated with commercial primary antibodies and visualized using the Tano 5200 Chemiluminescent Imaging System.

The commercial antibodies used in the western blotting assay were as follows: GAPDH (60,004–1-Ig, Proteintech), IL-1β (16,806–1-AP, Proteintech), S100A9 (26,992–1-AP, Proteintech), Caspase-1/P20 (22,915–1-AP, Proteintech), NLRP3 (27,458–1-AP, Proteintech), Cas9 (26,758–1-AP, Proteintech), and GSDMD (AF4012, Affinity). HERV-K102 Env protein was detected as previously described [22].

Raw sequencing data were filtered using Fastp (v0.23.2), low-quality reads were discarded, and the adaptor sequences were trimmed. Clean reads were further processed using in-house scripts to eliminate the duplication bias introduced during library preparation and sequencing. Briefly, clean reads were clustered according to their UMI sequences, with reads sharing the same UMI grouped into one cluster. Reads within each cluster were compared using pairwise alignment, and those with over 95% sequence identity were extracted into subclusters. Consensus sequences for each subcluster were generated through multiple sequence alignments, thereby eliminating errors and biases introduced by PCR amplification or sequencing. Sequences were then used for standard mRNA-seq analysis and mapped to the reference genome using the STAR software (version 2.7.6a). Gene- and transcript-level counts were obtained using featureCounts (Subread-1.5.1; Bioconductor) and expression levels were calculated. Differentially expressed genes and transcripts were identified using the edgeR package (version 3.40.2) with a p-value cutoff of 0.05 and a fold-change threshold of 2. Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) enrichment analyses for differentially expressed genes were performed using R (version 4.0) with a p-value cutoff of 0.05.

Cells were seeded in 96-well plates at 5,000 cells per well. Twenty microliters of thiazolyl blue tetrazolium bromide (MTT; Sangon Biotech, China) were added to each well at designated time points. Cells were incubated with MTT for 4 h and then dissolved overnight in a buffer containing 10% sodium dodecyl sulfate (SDS), 5% isobutanol, and 0.012 M HCl. After the formazan crystals dissolved, the absorbance was measured at 562 nm using a microplate reader, and cell growth curves were generated using GraphPad Prism 10.1.

All experimental data were analyzed using GraphPad Prism 10.1. If the data followed a normal distribution with a homogeneity of variance, a t-test was performed; otherwise, a two-sided nonparametric Wilcoxon rank-sum test was performed. Statistical significance was defined as p < 0.05.

An optimized in-house MV4-11 model was used. Female NOD.Cg-Prkdc^scid IL2rgtm^1Wjl/SzJ (NSG) mice aged 6–8 weeks were intravenously inoculated with 1 × 10⁶ MV4-11-Cas9 or MV4-11-Cas9-K-Env KO cells. One week later, mice were administered doxycycline (Dox) (2 mg/kg, day 0–5) to induce K-Env knockout. The tumor burden was monitored using an IVIS Lumina LT Series III in vivo imaging system (Perkin Elmer, USA).

To investigate the expression of HERV-K102 in AML, we compared datasets from the pan-tissue TCGA and Genotype-Tissue Expression (GTEx) projects. The expression profile of HERV-K102 was downloaded from an online database (https://xena.ucsc.edu/) based on its chromosomal location (chr1:155,596,457–155,605,636) [2]. HERV-K102 was found to be highly expressed in AML cells (Fig. 1A). The prognostic significance of HERV-K102 was assessed using Kaplan–Meier analysis. As shown in Figure S1, HERV-K102 expression was not significantly associated with overall survival. We then examined K-Env expression in different AML cell lines and observed heterogeneous expression levels. K-Env was highly expressed in OCI-AML3, THP-1, and MV4-11 cells, but showed extremely low expression in KG-1α cells (Fig. 1B).

To further investigate the role of K-Env in AML, it was knocked out using the CRISPR-Cas9 gene editing system. Two guide RNAs (gRNAs) targeting K-Env were designed; their specific sequences are described in the Methods section. Here, we selected two cell lines, THP-1 and MV4-11, with high K-Env expression for subsequent experiments. KO efficiency was evaluated by qRT-PCR, and comparative analysis revealed that gRNA2 achieved higher KO efficiency than gRNA1 (Fig. 2A). Western blot analysis confirmed the reduction in HERV-K-Env protein levels (Fig. 2B). Next, we examined whether K-Env knockout affects the survival of leukemia cells. The knockout of K-Env led to a significant decrease in Ki67 expression levels, indicating that cell proliferation was inhibited (Fig. 2C,D). The MTT assay showed that the knockout of K-Env by gRNA2 significantly suppressed the proliferation of leukemia cells (Fig. 2E). Apoptosis was further evaluated using Annexin V and 7-AAD staining, followed by flow cytometry. K-Env knockout by gRNA2 resulted in 59.63% and 60.03% of Annexin V-positive MV4-11 and THP-1 cells, respectively, suggesting the induction of early apoptosis (Fig. 2F,G).

To elucidate the molecular mechanisms by which HERV-K102 Env contributes to AML pathogenesis, we performed transcriptome profiling of K-Env KO MV4-11 and THP-1 cells alongside mock controls. Comparative analyses revealed widespread transcriptomic alterations following K-Env ablation. In MV4-11 cells, K-Env KO resulted in 1762 differentially expressed genes (DEGs), with 1,351 upregulated and 411 downregulated genes (Fig. 3A). Among the most significantly upregulated genes, S100A9, C3, CHI3L1, NCF1, and NCF1C are key mediators of inflammation and immune responses. GO enrichment analysis showed that these DEGs were strongly associated with the inflammatory response, neutrophil degranulation, and cytokine-mediated signaling pathways (Fig. 3B). KEGG pathway analysis further indicated robust activation of the NOD-like receptor, PI3K-Akt, and NF-κB signaling pathways (Fig. 3C), suggesting strong engagement of innate immune and survival-related molecular networks. Similarly, in THP-1 cells, K-Env depletion significantly altered 2,421 genes, with 1,706 and 715 upregulated and downregulated genes, respectively (Fig. 3D). Notably, S100A9, C3, HMOX1, CXCL8, and LACC1 were the most upregulated genes, reinforcing the pattern of innate immune activation. GO analysis revealed enrichment in the inflammatory response, immune activation, and neutrophil degranulation (Fig. 3E). KEGG analysis also confirmed significant involvement of the NOD-like receptor, PI3K-Akt, NF-κB, and cytokine–cytokine receptor interaction pathways (Fig. 3F). Collectively, these findings indicate that HERV-K102 Env ablation triggers a strong inflammatory response and activates key immunomodulatory and survival signaling pathways, suggesting that K-Env may suppress immune recognition and help maintain cellular homeostasis in AML.

Given that K-Env knockout upregulated the genes involved in inflammatory regulation, we investigated whether K-Env knockout induces pyroptosis in leukemia cells. Caspase-1 is a key protease that initiates pyroptosis. The proportion of caspase1⁺PI⁺ cells was significantly higher in the K-Env KO mice (Fig. 4A). On day 6 after K-Env knockout, we measured LDH levels. The results showed that cells in the knockout group released more LDH (Fig. 4B). To further elucidate potential molecular mechanisms, we conducted additional tests on molecules related to pyroptosis. The mRNA levels of S100A9, caspase-1, IL-1β, NLRP3, and GSDMD were significantly elevated in the MV4-11 K-Env KO group (Fig. 4C). Consistent with these findings, western blot analysis confirmed increased protein expression of S100A9, caspase-1 P20, IL-1β, GSDMD-N, and NLRP3 in the knockout group (Fig. 4D). Similar results were observed in THP-1 cells (Fig. 4E,F). Here, we hypothesized that knockout of K-Env may cause pyroptosis by upregulating S100A9 and the NOD-like receptor signaling pathway. During this process, the expressions of caspase-1, IL-1β, GSDMD, and NLRP3 also increase.

To further evaluate the in vivo efficacy of K-Env knockout, which was shown to inhibit cell proliferation, we established an inducible knockout model using the Tet-On system. First, the MV4-11 cell line was transduced with the pcw-Cas9 lentivirus. Upon induction with 2 μg/mL Dox, Cas9 protein expression was significantly and stably upregulated (Fig. 5A). Subsequently, plx-sgRNA lentivirus was transduced into MV4-11-Cas9 cells. Western blot analysis confirmed that K-Env protein levels were markedly reduced following Dox treatment (Fig. 5B). Consistent with this, the MTT assay performed on day 5 after Dox induction showed inhibited proliferation in cells in the Dox-induced knockout group only (Figure S2). For in vivo validation, MV4-1-Cas9-mock and MV4-11-Cas9-K-Env KO cells were injected into mice via the tail vein on day 7 (Fig. 5D). Tumor implantation was observed through fluorescence imaging on day 0, and the results showed no significant difference in tumor burden between the two groups. Dox was used to induce Cas9 expression in knockout K-Env. Subsequently, the growth of tumor cells in the mice was monitored using fluorescence imaging. The results showed that tumor signals in the KO group were significantly lower than those in the control group (Fig. 5E,F). These results demonstrate that induced silencing of K-Env effectively suppressed AML cell proliferation in vivo.