Abstract

Background

   Acute myeloid leukemia (AML) is a highly heterogeneous disease
   characterized by complex genetic and molecular features that contribute
   to poor prognosis and low cure rates. Therefore, identifying novel
   therapeutic targets is crucial for improving treatment efficacy and
   patient survival. This study investigated the potential role of
   cyclin-dependent kinase 9 (CDK9), a known regulator of gene expression,
   in AML pathogenesis and prognosis.

Methods

   This study employed multiple bioinformatics approaches, including
   analysis of CDK9 expression across various cancers using the Tumor
   Immune Estimation Resource (TIMER2.0) database and further
   investigation of its expression and prognostic significance in AML
   using data from the Cancer Genome Atlas (TCGA) and Genotype-Tissue
   Expression (GTEx) databases. Survival analysis and Cox regression
   analysis were used to assess the association between CDK9 expression
   and patient prognosis. Gene Ontology (GO), Kyoto Encyclopedia of Genes
   and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA) were
   performed to elucidate the pathways and biological processes influenced
   by CDK9. Furthermore, the relationship between CDK9 expression and
   tumor immune infiltration was evaluated, and a protein–protein
   interaction (PPI) network was constructed. In vitro experiments,
   including Western blotting, CCK-8 assays, and flow cytometry, were
   conducted to validate the bioinformatics findings.

Results

   Bioinformatics analysis revealed significantly elevated CDK9 expression
   in AML samples, which correlated with poor patient prognosis.
   Functional enrichment analysis indicated that CDK9 is involved in key
   pathways related to cell proliferation, differentiation, and the tumor
   microenvironment. Moreover, the study observed a strong correlation
   between CDK9 expression and altered immune cell infiltration,
   suggesting a potential role in immune evasion. In vitro experiments
   confirmed that CDK9 overexpression promoted AML cell proliferation and
   inhibited apoptosis. Additionally, CDK9 showed a strong correlation
   with epithelial-mesenchymal transition (EMT)-related proteins,
   suggesting a potential role in AML progression and the EMT process.

Conclusions

   This study demonstrates that CDK9 is a potential prognostic biomarker
   and therapeutic target in AML. Its involvement in multiple key pathways
   during AML development and its influence on the tumor immune
   microenvironment support further exploration of CDK9-targeted therapies
   to improve AML treatment outcomes.

Supplementary Information

   The online version contains supplementary material available at
   10.1007/s12672-025-02841-4.

   Keywords: CDK9, Acute myeloid leukemia, Biomarkers, Immune
   infiltration, Prognosis

Introduction

   Acute myeloid leukemia (AML) is a clonal hematopoietic malignancy
   originating from multipotent stem cells or early progenitor cells and
   is the most common acute leukemia in adults. It is characterized by the
   accumulation of leukemic blasts in the bone marrow and blood, with
   infiltration into organs and tissues [[40]1]. Prognosis, beyond age and
   comorbidities, is largely determined by AML's intrinsic biological
   characteristics [[41]2]. Recent advances in understanding AML biology
   and treatment have led to several therapeutic strategies that
   significantly benefit patients [[42]3]. However, AML prognosis remains
   poor, with a 5-year survival rate below 30% and a high relapse risk,
   presenting significant challenges [[43]4]. Therefore, discovering novel
   therapeutic targets and developing more effective treatments are
   crucial for improving patient outcomes.

   Cyclin-dependent kinases (CDKs) are serine/threonine (Ser/Thr) protein
   kinases essential for cell growth, proliferation, and transcriptional
   regulation [[44]5–[45]7]. Cyclin-dependent kinase 9 (CDK9), a member of
   the CDK family, regulates cell cycle progression by forming complexes
   with cyclin T1. These complexes, known as positive transcriptional
   elongation factor b (P-TEFb), control transcriptional elongation and
   mRNA maturation [[46]8]. CDK9 is frequently overexpressed in various
   cancers and is closely associated with tumorigenesis, invasion,
   metastasis, and prognosis, making it a promising cancer therapeutic
   target [[47]9–[48]12]. The importance of CDK9 has driven the
   development of CDK9 inhibitors, which have shown promising efficacy in
   preclinical studies across multiple cancers, including some
   demonstrating potential therapeutic value in AML [[49]13–[50]15].
   However, previous studies on the role of CDK9 in AML have primarily
   focused on its involvement in pathways affecting anti-apoptotic factors
   such as myeloid cell leukemia 1 (MCL-1), B-cell lymphoma-2 (Bcl-2), and
   X-linked inhibitor of apoptosis protein (XIAP), which are crucial for
   cancer cell survival [[51]16, [52]17]. The prognostic potential of CDK9
   in AML and its potential immunomodulatory functions, among other
   molecular mechanisms, warrant further investigation.

   This study integrates bioinformatics analysis to assess the diagnostic
   and prognostic significance of CDK9, coupled with Western blotting to
   detect CDK9 expression in AML cell lines and its correlation with
   epithelial-mesenchymal transition (EMT)-related proteins. In vitro
   CCK-8 proliferation assays and flow cytometry were used to analyze the
   impact of CDK9 on cell proliferation and apoptosis. Furthermore, we
   constructed a CDK9 co-expression network and performed functional
   enrichment analysis to identify associated signaling pathways and
   explore the regulatory role of CDK9 in the AML immune microenvironment.
   Through this integrated analytical approach, we aim to provide valuable
   insights into the significant role and underlying mechanisms of CDK9 in
   AML.

Materials and methods

Data download

   AML datasets were retrieved from The Cancer Genome Atlas (TCGA)-LAML
   via the Cancer Genomics Hub (CGHub) and TCGA data portal
   ([53]https://portal.gdc.cancer.gov/), comprising 150 AML samples
   [[54]18]. Clinical data corresponding to TCGA-LAML samples were
   obtained from the UCSC Xena database ([55]http://genome.ucsc.edu)
   [[56]19]. Additionally, the TCGA_GTEx-LAML dataset, which included 70
   GTEx normal samples and 173 TCGA tumor samples, was sourced from the
   Xena database. RNA expression data were normalized using log2(value
   + 1).

CDK9 expression analysis in AML

   Pan-cancer analysis was first performed using the"Gene_DE module"of the
   Tumor Immune Estimation Resource (TIMER2.0,
   [57]https://cistrome.shinyapps.io/timer/) to assess CDK9 expression in
   various tumor and normal tissues [[58]20]. To further validate its
   expression in AML, we evaluated the differential expression between 173
   AML samples and 70 normal samples from the TCGA_GTEx-LAML dataset. Data
   visualization was performed using the R package “ggplot2” (v3.4.4).

Clinical correlation analysis of CDK9 expression in AML

   The TCGA-LAML dataset was used to analyze the differential expression
   of CDK9 across various clinical subgroups in AML to assess the
   correlation between CDK9 expression levels and clinicopathological
   characteristics. The Wilcoxon rank-sum test was used for clinical
   variables with only two groups (e.g., age, FL3 mutation), while for
   variables with more than two groups, a one-way ANOVA was performed,
   followed by Tukey's post-hoc test for multiple comparisons. Data
   visualization was performed using the R packages ggplot2, stats
   (v4.2.1), and car (3.1-0).

Independent prognostic and diagnostic value in AML patients

   Survival analyses involved 139 AML samples with complete survival data
   from the TCGA-LAML dataset. Kaplan–Meier (K-M) survival curves were
   plotted using the survminer R package (v0.4.9) to examine the
   association between CDK9 expression and survival. Diagnostic
   performance in distinguishing AML patients from healthy individuals was
   evaluated by plotting receiver operating characteristic (ROC) curves
   using “ggplot2.” Univariate and multivariate Cox regression analyses
   were performed with the “survival” package (v3.3.1), and forest plots
   were visualized using “ggplot2.” The proportional hazard assumption was
   tested using the “survival” package, and Cox regression analyses
   predicted overall survival. A nomogram was constructed and visualized
   using the “rms” package (v6.3-0). Risk scores were calculated based on
   the median expression of CDK9 and visualized with “ggplot2.” Predictive
   performance was further validated with calibration curves and
   time-dependent ROC (tROC) curves using the “rms,” “survival,”
   “ggplot2,” and “timeROC (v0.4)” packages.

Co-expression analysis of CDK9 in AML

   The expression and transcriptional landscape of CDK9 in AML was
   analyzed using the LinkedOmics online platform
   ([59]http://www.linkedomics.org/login.php) [[60]21]. We identified 9120
   and 10,328 genes down-regulated and up-regulated in co-expression with
   CDK9, respectively. A heatmap visualized the top 50 most significantly
   downregulated and upregulated genes. Subsequently, Pearson correlation
   analysis assessed the strength of correlation among the top three
   up-regulated and down-regulated genes, with visualization
   using"ggplot2.".

Gene ontology (GO) and Kyoto rncyclopedia of genes and genomes (KEGG)
enrichment analyses

   AML patients in the TCGA-LAML dataset were divided into low and high
   CDK9 expression groups based on the median expression level. The DESeq2
   R package (v1.36.0) was then used to identify 126 differentially
   expressed genes (DEGs) with thresholds of |log2-fold-change (FC)| > 2
   and adjusted P-value < 0.05. Metascape ([61]http://www.metascape.org/)
   was used for functional annotation of CDK9 and its associated genes
   from the TCGA-LAML dataset [[62]22]. Gene Ontology (GO) enrichment
   analysis categorized biological processes (BP), cellular components
   (CC), and molecular functions (MF), identifying CDK9's roles in key
   metabolic pathways. Kyoto Encyclopedia of Genes and Genomes (KEGG)
   pathway enrichment analysis identified significant pathways associated
   with CDK9.

Gene set enrichment analysis (GSEA)

   Gene set enrichment analysis (GSEA) was performed using the
   “clusterProfiler” (v4.4.4) R package, with significance determined by
   an adjusted P-value < 0.05 and FDR q-value < 0.25, to identify
   functional and pathway differences between the high and low CDK9
   expression groups.

CDK9 expression, immune infiltration, and protein interaction network
analysis

   To assess the potential relationship between CDK9 expression and immune
   cell infiltration, single-sample gene set enrichment analysis (ssGSEA)
   was performed using the"GSVA"package (v1.46.0) for 24 immune cell types
   in the context of CDK9. The Wilcoxon rank-sum test compared enrichment
   scores for the 24 immune cell types between high and low CDK9
   expression groups. Spearman correlation analysis evaluated the
   correlation between CDK9 expression and immune cell infiltration. A
   protein–protein interaction network was constructed using the STRING
   database ([63]http://string-db.org/) by setting the minimum confidence
   to “medium confidence (0.400)” and using other default parameters
   [[64]23].

Cell culture

   Human AML cell lines (HL-60, THP-1, Kasumi-1) and normal HMSC-BM bone
   marrow cells were obtained from Anhui Baioujing Technology Co., LTD.,
   China. Cells were cultured in 1640 medium supplemented with 10% fetal
   bovine serum and 1% penicillin–streptomycin at 37 °C in a 5% CO2
   incubator. Cells were passaged when the growth density reached 80–90%.

Cell transfection

   THP-1 cells at the logarithmic growth stage were seeded into 6-well
   plates at a density of 1 × 10^5 cells per well. Transfections were
   performed for 24–48 h using the NanoTrans 60™ transfection kit (China
   Biomedical). Cells were grouped as NC, OE-CDK9, siCtrl, siCDK9-1,
   siCDK9-2, and siCDK9-3 based on transfection conditions, following the
   manufacturer’s instructions. After transfection, cells were incubated
   at 37 °C for 48 h. CDK9 protein expression levels were assessed using
   Western blot analysis. Compared to the siCtrl group, the siCDK9-1,
   siCDK9-2, and siCDK9-3 groups exhibited reduced CDK9 protein levels,
   while the OE-CDK9 group showed increased CDK9 expression relative to
   the NC group, confirming successful CDK9 plasmid transfection.

Quantitative RT-qPCR analysis

   cDNA synthesis was performed using AG RNAex Pro RNA extraction reagent
   (Ecorui Bio, China), followed by PCR amplification using SYBR Green Pro
   Taq HS premix (Ecorui Bio, China). Real-Time PCR conditions and
   reaction details are listed in Supplementary Table [65]S1. Each sample
   was analyzed in triplicate, with GAPDH serving as the internal
   reference gene for both experimental and control groups. Relative gene
   expression levels were calculated using the 2^−ΔΔCt method. Primer
   sequences are provided in Supplementary Table [66]S2.

Western blot assay

   Total protein was extracted from cells or tissues using RIPA lysis
   buffer (R0010, Solarbio). Protein concentration was determined with the
   BCA assay (P0010, Beyotime). A total of 30 μg of protein was separated
   by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked
   with 5% skim milk in PBS and incubated overnight at 4 °C with primary
   antibodies. They were then incubated with Goat anti-Rabbit IgG-HRP
   (abs20040, Ambition) or Goat anti-Mouse IgG-HRP (abs20039, Ambition) at
   37 °C for 2 h. Immunoreactive bands were visualized using an ECL kit
   (PK10002, Proteintech).

Cell proliferation test

   Cell Counting Kit-8 (CCK-8) assay was performed. Cells at the
   logarithmic growth phase were seeded into 96-well plates at a density
   of 1 × 10^4 cells per well. After incubation, MTT solution (5 g/L) was
   added and incubated for 4 h. Following centrifugation and removal of
   the supernatant, 150 μL of DMSO was added to each well and incubated in
   the dark for 15 min. Absorbance at 490 nm was measured with a
   microplate reader to evaluate cell proliferation activity.

Apoptosis test

   Flow cytometry was used to analyze apoptosis. Cells at the logarithmic
   growth phase were seeded into 24-well plates at a density of 1 × 10^5
   cells per well. After transfection and 48 h of incubation, cells were
   collected, washed with cold PBS, and resuspended in 100 μL of Binding
   Buffer. Cells were stained with 5 μL of Annexin V-FITC and 5 μL of PI
   and incubated in the dark at room temperature for 10 min. Apoptosis
   rates were measured using flow cytometry within 1 h of staining.

Statistical analysis

   Statistical analyses were conducted using GraphPad Prism (v9.5.1), R
   software (v4.2.1), and online tools. Kaplan–Meier survival curves were
   analyzed with the log-rank test. Correlation coefficients were
   calculated using Pearson or Spearman tests, as appropriate. Data are
   presented as mean ± standard deviation (SD) from at least three
   independent experiments. Statistical significance was defined as P <
   0.05 (*P < 0.05; **P < 0.01; ***P < 0.001; N.S., not significant).

Results

CDK9 expression analysis in pancarcinoma and AML

   Using the TIMER2.0 database, we observed differential CDK9 expression
   between tumor and adjacent normal tissues across 33 different cancers
   (Fig. [67]1A). Significantly different CDK9 expression (P < 0.05) was
   observed in ten cancer types: breast invasive carcinoma (BRCA),
   cholangiocarcinoma (CHOL), esophageal carcinoma (ESCA), head and neck
   squamous cell carcinoma (HNSC), clear cell renal cell carcinoma (KIRC),
   lung adenocarcinoma (LUAD), rectum adenocarcinoma (READ), stomach
   adenocarcinoma (STAD), thyroid carcinoma (THCA), and uterine corpus
   endometrial carcinoma (UCEC). However, due to the lack of normal AML
   samples in this database, we used the TCGA_GTEx-LAML dataset to compare
   CDK9 expression in 173 AML samples and 70 normal samples. Results
   (Fig. [68]1B) showed significantly higher CDK9 expression in AML
   samples compared to normal samples (P < 0.001). We further investigated
   the correlation between CDK9 expression and clinicopathological
   features (Fig. [69]1C–H). CDK9 expression was significantly lower in
   AML samples with t(8;21) chromosomal abnormalities compared to those
   with normal cytogenetics. Higher CDK9 expression was observed in the
   intermediate/normal and poor cytogenetic risk groups compared to the
   favorable group. The M1 FAB subtype showed higher CDK9 expression than
   the M4 subtype. No significant differences in CDK9 expression were
   observed among other clinicopathological subgroups.

Fig. 1.

   [70]Fig. 1
   [71]Open in a new tab

   Differential expression analysis of CDK9. A Differential expression of
   CDK9 in tumor and adjacent normal tissues across 33 different cancers
   in the TIMER2 database; B Difference in CDK9 expression levels between
   AML and normal tissues based on the TCGA_GTEx-LAML dataset; C-H
   correlation between CDK9 expression levels and clinicopathological
   features of AML, including age, FLT3 mutation, NPM1 mutation,
   prognostic stratification, and FAB classification. (*p < 0.05; **p <
   0.01; ***p < 0.001)

Prognostic analysis of CDK9 in AML

   AML patients from the TCGA database were categorized into high- and
   low-CDK9 expression groups based on average CDK9 expression levels.
   Kaplan–Meier (K-M) survival analysis revealed that patients in the
   low-CDK9 expression group had better survival rates (P < 0.05)
   (Fig. [72]2A), indicating that low CDK9 expression is associated with a
   favorable prognosis. ROC curve analysis demonstrated high accuracy of
   CDK9 expression in assessing tumorigenesis (Fig. [73]2B). Univariate
   Cox regression analysis showed significant correlations between AML
   prognosis and CDK9 expression, age, and prognostic stratification
   (Fig. [74]2C, Supplementary Table [75]S3).

Fig. 2.

   [76]Fig. 2
   [77]Open in a new tab

   The prognostic and diagnostic value of CDK9 in AML. A CDK9 expression
   and survival analysis for patient prognosis; B ROC curves illustrating
   the diagnostic value of CDK9; C Screening of clinically relevant
   prognostic features associated with AML prognosis; D Nomograph
   prediction model based on CDK9 expression levels and associated
   clinical features; E Calibration curves with one, two, and three years
   for the Cox regression prediction model; F Risk scores and maps of
   survival status for prognostic models that are clinically relevant; G
   Time-dependent ROC curve results of risk scores versus patient
   prognostic OS for clinically relevant prognostic models

   To develop a prognostic model, CDK9 expression was combined with
   clinical characteristics. The model demonstrated higher predictive
   value for age (Fig. [78]2D). Multivariate analysis confirmed the
   model's predictive performance. Prognostic calibration curves indicated
   high predictive accuracy (Fig. [79]2E), and risk factor plots visually
   displayed prediction results (Fig. [80]2F). Time-dependent ROC analysis
   further validated the model's predictive accuracy for 1-, 2-, and
   3-year survival rates (Fig. [81]2G).

CDK9 coexpression network

   To investigate the potential mechanisms of CDK9 in AML, a co-expression
   network analysis was performed using the LinkedOmics database. This
   identified 10,328 genes positively correlated and 9,120 genes
   negatively correlated with CDK9 expression (Fig. [82]3A). A heatmap
   highlighted the top 50 positively and negatively correlated genes
   (Fig. [83]3B, C), emphasizing CDK9's impact on the transcriptome. The
   top three positively correlated genes were WBP1 (R = 0.660, P < 0.001),
   FPGS (R = 0.819, P < 0.001), and RASGRP2 (R = 0.737, P < 0.001), while
   the top three negatively correlated genes were CPSF2 (R = − 0.201, P =
   0.013), ZMPSTE24 (R = − 0.225, P = 0.006), and PGM2 (R = − 0.103, P =
   0.211) (F[84]ig. [85]3D–I).

Fig. 3.

   [86]Fig. 3
   [87]Open in a new tab

   Coexpression network of CDK9 in LAML. A Volcano plot showing CDK9
   coexpressed genes; B, C Heat maps illustrating the top 50 genes
   positively and negatively correlated with CDK9 expression in AML; D–I
   Pearson correlation scatter plots showing the relationship between CDK9
   expression and the expression levels of WBP1, FPGS, RASGRP2, CPSF2,
   ZMPSTE24, and PGM2

CDK9 differential gene screening, functional enrichment, and pathway analysis

   The TCGA-LAML dataset was stratified into high and low CDK9 expression
   groups. R software was used to identify 126 differentially expressed
   genes (DEGs) using a cutoff of |log2 FoldChange|> 2 and P < 0.05. GO
   functional annotation and KEGG pathway enrichment analyses of these
   DEGs revealed enrichment in biological processes such as cell
   differentiation and signal receptor binding (Fig. [88]4A–C). KEGG
   analysis highlighted the involvement of cancer pathways, TGF-β
   signaling, and cytokine-cytokine receptor interaction (Fig. [89]4D).
   Gene Set Enrichment Analysis (GSEA) comparing the high and low CDK9
   expression groups identified key AML-related pathways, including
   mitochondrial complex III assembly (Fig. [90]4E), downstream CD8
   signaling (Fig. [91]4F), BCR signaling (Fig. [92]4G), and
   corticotropin-releasing hormone signaling (Fig. [93]4H). These findings
   suggest a role for CDK9 in immune infiltration and highlight its
   importance in AML immunooncology.

Fig. 4.

   [94]Fig. 4
   [95]Open in a new tab

   Identification of CDK9 DEGs and enrichment analysis of biological
   functions. A-C GO bubble plots; D KEGG enrichment analysis bubble plot;
   E Mitochondrial Complex III assembly pathway; F CD8 downstream
   signaling pathway; G BCR signaling pathway. H corticotropin-releasing
   hormone signaling pathway

Relationship between CDK9, immune cell infiltration, and the protein
interaction network

   To further explore the role of CDK9 expression in AML immune
   infiltration, the ssGSEA algorithm was used to estimate the
   infiltration proportions of 24 immune cells in each sample. Seven
   immune cells showed differential infiltration between high and low CDK9
   expression groups (Fig. [96]5A). High CDK9 expression was associated
   with increased infiltration of NK CD56bright cells, TFH cells, Treg
   cells, CD8 T cells, pDCs, and Th17 cells and decreased infiltration of
   Th2 cells compared to the low expression group. Spearman correlation
   analysis was then performed to investigate the correlation between CDK9
   and immune cells. Setting a threshold of |R|≥ 0.3 and p < 0.05 (as |R|<
   0.3 indicates weak correlation), NK CD56bright cells (R = 0.439, P <
   0.001) and follicular helper T cells (TFH) (R = 0.300, P < 0.001) were
   found to be significantly correlated with CDK9 (Fig. [97]5B, C). A CDK9
   protein–protein interaction (PPI) network was constructed using the
   STRING database to assess its functional impact on AML progression. The
   top 10 proteins identified in the PPI analysis were MLLT1, CCNT1, BRD4,
   CCNT2, AFF4, HEXIM1, CCNK, AFF1, MEPCE, and MED12 (details in
   Supplementary Table [98]S4, Fig. [99]5D).

Fig. 5.

   [100]Fig. 5
   [101]Open in a new tab

   Relationship between CDK9 and immune cell infiltration and protein
   interaction network. A Assessment of immune infiltration levels in high
   and low CDK9 expression groups; B correlation of CDK9 gene expression
   with immune cell infiltration; C Scatter plots showing the correlation
   between CDK9 and NK CD56bright cells and TFH cells; D Protein–protein
   interaction (PPI) network of CDK9

Effect of CDK9 on proliferation and apoptosis in AML cells

   Western blot analysis showed elevated CDK9 protein levels in HL-60,
   THP-1, and Kasumi-1 cells compared to HMSC-BM cells (Fig. [102]6A).
   CDK9 knockdown (siCDK9-1, siCDK9-2, siCDK9-3) and overexpression
   (OE-CDK9) plasmids were introduced in THP-1 cells. CDK9 protein
   expression decreased in the siCDK9 groups compared to siCtrl, while it
   increased in the OE-CDK9 group compared to NC (Fig. [103]6B).

Fig. 6.

   [104]Fig. 6
   [105]Open in a new tab

   AML cell proliferation is stimulated, apoptosis is inhibited, and the
   expression of proteins related to EMT is regulated by CDK9. A The
   abundance of CDK9 protein cells was detected by WB; B WB identified
   CDK9's interference and overexpression efficiency; C CCK8 cell
   viability was detected (Each experimental group's OD450 multiplicity
   from day 1 to day 5 in relation to day 1 is expressed as OD450/fold,
   which shows the multiplicity of proliferation on each day.); D
   Apoptosis was detected by flow cytometry. E qPCR and WB were used to
   detect the expression of EMT-related proteins. (*p < 0.05; **p < 0.01;
   ***p < 0.001)

   CCK-8 assays demonstrated that OE-CDK9 significantly enhanced THP-1
   proliferation (**P < 0.01), while siCDK9 significantly reduced it
   (**P < 0.01) (Fig. [106]6C). Flow cytometry analysis indicated that
   CDK9 overexpression inhibited apoptosis, whereas CDK9 knockdown
   promoted apoptosis (Fig. [107]6D).

   Subsequently, epithelial-mesenchymal transition (EMT)-related protein
   expression was assessed. In the OE-CDK9 group, mRNA and protein levels
   of E-cadherin decreased significantly, while N-cadherin, Vimentin, and
   Snail1 levels increased compared to NC. Conversely, the siCDK9 group
   showed the opposite trends, with increased E-cadherin and decreased
   N-cadherin, Vimentin, and Snail1 levels (Fig. [108]6E).

Discussion

   Recent years have witnessed extensive research dedicated to elucidating
   the pathogenesis and treatment of AML, aiming to identify novel
   clinical targets and therapeutic agents with improved specificity and
   reduced side effects [[109]24, [110]25]. CDK9, the catalytic subunit of
   P-TEFb, has been shown to be widely expressed in various cancers and
   plays a crucial role in their development and progression, including
   some hematological malignancies and solid tumors [[111]13].

   Given CDK9's established role in various malignancies, we first
   investigated its differential expression between normal and tumor
   samples across different cancer types using public databases. We
   observed differential CDK9 expression in 10 cancer types with
   comparable sample availability. However, due to the lack of normal AML
   samples, we further evaluated CDK9 expression using the TCGA_GTEx_LAML
   database, revealing increased CDK9 expression in AML patients.
   Interestingly, a recent study reported no significant difference in
   CDK9 expression between AML and normal samples [[112]26]. This
   discrepancy may be attributed to the different data filtering methods
   used; they employed the GEPIA database. Subsequently, Western blot
   analysis confirmed our findings, demonstrating increased CDK9
   expression in AML, thus highlighting its potential significance. We
   further investigated the association between CDK9 expression and AML
   clinicopathological features and prognosis. Prognostic stratification
   and French-American-British (FAB) subtypes were significantly
   associated with CDK9 expression in AML patients. Survival analysis and
   both univariate and multivariate Cox regression analyses revealed that
   high CDK9 expression was correlated with poor prognosis. A nomogram
   prediction model further underscored the importance of CDK9 expression
   in predicting AML prognosis. In summary, these results highlight the
   significance of CDK9 as a prognostic biomarker and potential
   therapeutic target in AML.

   To explore the potential functional mechanisms of CDK9 in AML, we
   analyzed CDK9 co-expressed genes in the TCGA-LAML dataset using the
   LinkedOmics database. The results showed that a large number of genes
   were significantly positively or negatively correlated with CDK9
   expression, suggesting a broad impact of CDK9 on the AML transcriptome.
   Specifically, FPGs showed a significant positive correlation with CDK9,
   consistent with previous reports associating FPGs with poor prognosis
   in acute leukemia [[113]27]. Subsequently, we stratified the TCGA-LAML
   dataset into high and low CDK9 expression groups and identified 126
   differentially expressed genes (DEGs). GO functional enrichment
   analysis revealed that these DEGs were significantly enriched in
   biological processes such as cell differentiation and signal receptor
   binding. KEGG pathway enrichment analysis showed that CDK9 is involved
   in the regulation of cancer-related pathways, the TGF-β signaling
   pathway, and cytokine-cytokine receptor interactions. Furthermore, GSEA
   revealed significant associations between CDK9 and key pathways,
   including mitochondrial complex III assembly, downstream CD8 signaling,
   BCR signaling, and corticotropin-releasing hormone signaling. These
   results suggest that CDK9 may play a crucial role in AML development
   and progression by regulating multiple key signaling pathways,
   influencing AML cell proliferation, differentiation, apoptosis, and the
   immune microenvironment.

   The tumor microenvironment (TME) in AML is characterized by
   immunosuppression, promoting immune tolerance and immune evasion of
   malignant cells [[114]28, [115]29]. Therefore, studying immune cell
   infiltration within the tumor may provide valuable insights into the
   efficacy and mechanisms of immunotherapy in AML. Given the potential
   impact of CDK9 on the immune microenvironment, we further assessed
   immune cell changes at different CDK9 expression levels using the
   ssGSEA algorithm. NK CD56bright cells and TFH cells showed a
   significant positive correlation with high CDK9 expression. Notably,
   CD56bright CD16- NK cells are a major component of NK cells in various
   immune-tolerant organs, and their function is known to be suppressed in
   the tumor immune environment [[116]30]. Therefore, we hypothesize that
   high CDK9 expression may lead to increased infiltration of NK
   CD56bright cells in the TME of AML patients, resulting in tumor immune
   evasion and ultimately poor prognosis. Furthermore, protein–protein
   interaction (PPI) analysis identified interactions between CDK9 and
   proteins such as MLLT1, CCNT1, BRD4, and CCNT2, which also play
   important roles in AML development and progression, providing new
   avenues for further investigation into CDK9's mechanism of action.

   In vitro experiments confirmed CDK9's role in promoting proliferation
   and inhibiting apoptosis in AML cells. qPCR and western blot analyses
   of epithelial-mesenchymal transition (EMT)-related proteins in CDK9
   knockdown and overexpression groups revealed reduced E-cadherin mRNA
   and protein levels in the OE-CDK9 group, while N-cadherin, Vimentin,
   and Snail1 levels were significantly elevated. These findings suggest
   that CDK9 is closely associated with EMT-related proteins, potentially
   contributing to AML progression.

   This study has several limitations. First, our findings are based on
   analyses of public databases with limited sample sizes. Current
   datasets lack information on specific epigenetic mutations for all
   patients, including those in key genes highlighted by the recently
   updated WHO classification [[117]31], which is crucial for a more
   detailed understanding of CDK9's mechanism in AML. Future studies
   should include larger and more comprehensive cohorts for validation.
   Second, we only confirmed the effects of CDK9 on AML cell
   proliferation, apoptosis, and EMT marker expression through in vitro
   experiments. Further validation of CDK9 co-expressed genes, related
   signaling pathways, and interacting proteins associated with CDK9
   expression is lacking. Future studies should include more functional
   experiments and in vivo studies. Finally, this study did not evaluate
   the efficacy and sensitivity of immune-targeted drugs and CDK9
   inhibitors. Given that studies have shown that selective CDK9
   inhibitors and degraders such as Dinaciclib, NVP-2, LDC000067, FIT-039,
   Atuveciclib, PROTAC 2, and Compound 45 [[118]26, [119]32–[120]36] can
   inhibit tumor growth in hematological and solid tumors in vitro and in
   vivo, and that combination therapy may be a promising strategy to
   enhance the efficacy of selective CDK9 inhibitors and degraders in
   tumors, future research should explore the efficacy of these drugs in
   AML and combination therapy strategies to promote the clinical
   translation of CDK9 inhibitors.

Conclusions

   In conclusion, we have identified a correlation between high CDK9
   expression in AML and poor prognosis, suggesting that CDK9 may promote
   AML development and progression by influencing immune cell infiltration
   and the EMT process. This study enhances our understanding of CDK9's
   role in AML.

Supplementary Information

   [121]Additional file 1.^ (26.3KB, docx)

Acknowledgements