HMGA1 promotes the progression of lung adenocarcinoma through the STAT1-mediated transcriptional activation of DDAH1

Abstract

Background

Advancements in precision oncology have generated increased interest in the prognostic and therapeutic capabilities of transcription factors, among which HMGA1 is significantly correlated with LUAD prognosis. However, our understanding of HMGA1 remains insufficient. This study seeks to elucidate the biological functions of HMGA1 and to investigate the underlying mechanisms.

Methods

The prognostic value of HMGA1 was validated across multiple independent patient cohorts with LUAD, and its impact on tumor proliferation was verified by both in vitro and in vivo models. A series of experiments were performed to investigate the underlying molecular mechanism, including RNA sequencing, co-immunoprecipitation and chromatin immunoprecipitation.

Results

HMGA1 plays a crucial role in promoting the proliferation of LUAD. The underlying mechanism involves the recruitment of STAT1 to the promoter region of DDAH1, which synergistically increases its transcription and subsequently activates the ADMA/NO signaling pathway. Notably, the STAT1 inhibitor fludarabine has been shown to effectively impede the progression of LUAD models characterized by high levels of HMGA1.

Conclusion

Our research reveals a previously unrecognized mechanism through which the HMGA1/STAT1 complex facilitates LUAD proliferation by transcriptionally activating DDAH1. Moreover, we propose that fludarabine could serve as a promising therapeutic option for LUAD patients exhibiting elevated levels of HMGA1.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13062-025-00688-x.

Introduction

Lung adenocarcinoma (LUAD) is the predominant histological subtype of lung cancer, with complex pathogenesis and significant clinical heterogeneity [1]. Accordingly, elucidating the molecular mechanisms of LUAD progression, as well as identifying reliable prognostic biomarkers and potential therapeutic targets, will contribute to the individualized treatment of LUAD patients.

Transcription factors (TFs) have long been considered ideal targets for tumor precision therapy [2]. High mobility group A1 (HMGA1) is a specific TF, which lacks intrinsic transcriptional activity [3]. It binds to dsDNA via its N-terminal AT-hook and then promotes DNA conformational changes, recruiting other TFs to particular regions and modulating transcriptional processes [4, 5]. Previous studies have demonstrated that HMGA1 is overexpressed in several tumors and correlates with poor prognosis [6, 7]. Our group’s earlier work also revealed that HMGA1 was upregulated by IL-17 signaling in LUAD and facilitated tumor progression via Cyclin D1 expression [8]. Considering the crucial roles of HMGA1 in transcriptional regulatory networks, it is hypothesized that HMGA1 may play a more intricate role in the malignant progression of LUAD, with the underlying mechanisms requiring further elucidation.

Dimethylarginine dimethylaminohydrolase-1 (DDAH1) is the major enzyme that mediates the metabolism of asymmetric dimethylarginine (ADMA) [9]. It modulates the homeostasis of AMDA by catalyzing its hydrolysis to citrulline and dimethylamine [10]. As a potent endogenous inhibitor of nitric oxide synthase (NOS), ADMA suppresses the production of nitric oxide (NO) and increases the generation of superoxide radical, making it an independent risk factor for several cardiovascular events [11, 12]. In terms of tumor progression, the DDAH1 has been shown to promote prostate cancer growth and angiogenesis through NO upregulation [13].

In this study, we have established that HMGA1 facilitates LUAD proliferation. The underlying mechanism involves HMGA1’s recruitment of signal transducer and activator of transcription 1 (STAT1) to the promoter region of DDAH1, which in turn significantly enhances DDAH1 transcription and activates the ADMA/NO signaling pathway. Notably, the STAT1 inhibitor fludarabine effectively impedes the progression of LUAD models characterized by high levels of HMGA1. Collectively, these findings elucidate a novel regulatory mechanism by which HMGA1 influences the progression of LUAD, thereby offering new insights for therapeutic strategies targeting LUAD.

Methods and materials

Data collection and processing

A series of microarray datasets (GSE30219, GSE31210, GSE50081, GSE68465, GSE68571, and GSE72094) were obtained from the Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) [14–19]. The RNA sequencing (RNA-seq) profiles and corresponding clinical metadata for LUAD were achieved from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/). Probe IDs were mapped to gene symbols based on annotation files. All microarray and RNA-seq data were log₂ transformed for subsequent analyses. Batch effects across datasets were corrected using the “Combat” algorithm of the R package sva. A total of 1,868 LUAD patients were included in our study. Detailed clinicopathological characteristics were concluded in Table S1.

The complete catalogue of TFs was downloaded from the AnimalTFDB website (http://guolab.wchscu.cn/AnimalTFDB4/#/) [20]. After quality control (removal of entries without official gene symbols), 1,637 high-confidence TFs were retained for analysis.

Bioinformatic analyses

Survival data from LUAD cohorts were processed with a 5-year overall survival endpoint, as mortality events beyond this timepoint were unlikely to be tumor-related [16]. Harrell’s concordance index (C-index) was calculated for each TF using the R package survival across cohorts, and the TF with the highest average C-index was considered to be the most related with LUAD prognosis [21]. Kaplan-Meier curves were then depicted by the survival package, and the differences in survival curves were assessed using the log-rank test [22]. Multivariable Cox regression analysis was performed to estimate hazard ratios (HR) with 95% confidence intervals (CI), adjusting for clinical factors including age, sex, stage, smoking history and key driver gene mutations (e.g., KRAS, and EGFR). Additionally, a meta-analysis was conducted using the R package meta to evaluate the prognostic value of HMGA1 in multiple LUAD cohorts with the random effect model.

Differential expression of HMGA1 between LUAD and neighboring normal tissue was analyzed using ggplot2. Patients were stratified into HMGA1^high and HMGA1^low subgroups based on median expression levels. Differentially expressed genes (DEGs) were then identified by limma, with thresholds of |log₂ fold change| ≥ 1 and adjusted P-value < 0.05 [23]. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes database (KEGG) enrichment analysis of DEGs were performed by the clusterprofiler package [24]. The enrichment results were subsequently visualized as interaction networks using the Metascape platform (https://metascape.org) [25].

Patient tissue microarray (TMA) analysis

A TMA containing 30 pairs of LUAD and adjacent tissue were purchased from Outdo Biotech (HLugA060PG03; Shanghai, China) for immunohistochemistry (IHC) staining of HMGA1. IHC assay for TMA was implemented as previously described [26]. Details of antibodies and dilution ratios are provided in Table S2.

Cell lines and cell culture

Human normal bronchial epithelial cells BEAS-2B and human LUAD cell lines A549, PC9, NCI-H23 and NCI-H1299 were obtained from the Shanghai Institute of Biochemistry and Cell Biology (Shanghai, China), and cultured in RPMI-1640 medium (VivaCell, Shanghai, China) with 10% fetal bovine serum (FBS, VivaCell) at 37℃ in 5% CO₂.

Real-time quantitative PCR (RT-qPCR) and Western blot

RT-qPCR and Western blot experiments were performed according to previously published articles [27]. Details of antibodies and dilution ratios are provided in Table S2. Primer sequences are shown in Table S3.

Cell viability and proliferation assays

Cell counting kit-8 (CCK8), 5-ethynyl-2-deoxyuridine (EdU) incorporation, and colony formation assays were performed in accordance with our previously published articles [27, 28].

Cell transfection and lentivirus infection

The plasmids pCMV-HMGA1-3×FLAG and pCMV-STAT1-3×FLAG, along with their empty vectors, were constructed by GeneChem (Shanghai, China). Additionally, the pGL3-DDAH1-Promoter luciferase reporter plasmid, which contains the full-length DDAH1 promoter region (-2000 to + 0 nt), and the pRL-TK Renilla internal reference luciferase plasmid were also prepared by GeneChem. Small interfering RNAs (siRNAs) targeting HMGA1 were synthesized by Hanbio (Shanghai, China). Its sequences were listed in Table S4. Lipofectamine 3000 (Thermo Fisher, USA) was used for transfection, according to the manufacturer’s protocol.

The lentiviral plasmid encoding short hairpin RNA (shRNA) targeting HMGA1, DDAH1 and STAT1, as well as their negative control (shNC) were also obtained from Hanbio. Notably, the sequences of shHMGA1 were identical to si-HMGA1 #2. A549 and PC9 cell lines were transduced with lentiviruses at multiplicities of infection (MOI) of 5 and 10 for 24 h, respectively. Stable cell lines were established by selection with puromycin (1 µg/ml), and validated by qPCR and Western blot.

RNA sequencing (RNA-Seq) analysis

Total RNA was isolated from A549 cells infected with shNC or shHMGA1 using TRIzol (Invitrogen, USA), following the manufacturer’s protocol. After rRNA removal, mRNA was reverse-transcribed into cDNA for library construction and sequenced by the Alliance of Nanjing Biopharmaceutical Platform (Nanjing, China). High-quality reads were aligned to the GRCh38 human reference genome, and gene expression was normalized using transcripts per million (TPM). Differential analysis was conducted with the DESeq2 package to identify DEGs after HMGA1 knockdown [29].

ADMA and NO detection

ADMA concentrations in cell lysates were analyzed by the ADMA ELISA Kit (E-EL-0042; Elabscience, Hubei, China), according to the manufacturer’s recommended guideline. Absorbance at 450 nm was detected by Multiskan FC microplate photometer (Thermo Fisher) after adding stop solution.

Intracellular levels of NO were quantified by the nitrate reductase assay using the NO Colorimetric Assay Kit (E-BC-K135-M, Elabscience). Absorbance at 530 nm was measured with the same photometer (Thermo Fisher).

Immunoprecipitation (IP) and mass spectrometric (MS) analysis

Total proteins were extracted using RIPA lysis buffer (Beyotime, Shanghai, China) supplemented with 1% protease and phosphatase inhibitor cocktails. A portion of lysate was reserved as Input, while the remaining samples were equally divided and incubated overnight at 4 °C with either IgG or target-specific IP antibody. Following the protocol of the Immunoprecipitation Kit (P2179S; Beyotime), antigen-antibody complexes were captured with magnetic beads and then eluted with SDS loading buffer. To identify the HMGA1-interacting proteins, tryptic digests of A549 cells transfected with HMGA1 plasmids were analyzed by MS (BGI, Shenzhen, China). Antibody specifications and dilution ratios are detailed in Table S2.

Immunofluorescence (IF) staining and confocal microscopy

LUAD cells with overexpressed HMGA1 and STAT1 were seeded onto coverslips in 24-well plates and allowed to adhere for 24 h. After three PBS washes, the cells were fixed with 4% paraformaldehyde (PFA) for 30 min and permeabilized with 0.5% Triton X-100 for 30 min. After blocking with 1% BSA for 2 h at room temperature, cells were incubated with primary antibodies overnight at 4 °C. Following PBS washes, the cells were incubated with fluorescence-conjugated antibodies for 2 h at room temperature under light protection. Nuclei were counterstained with DAPI for 20 min. Intracellular colocalization images were captured with a Stellaris STED confocal microscope (Leica, Germany). Details of antibodies and dilution ratios are provided in Table S2.

Chromatin Immunoprecipitation (ChIP)

DNA-protein crosslinks in LUAD cells were fixed with 4% PFA. Cells were lysed with SDS buffer (Beyotime), and the chromatin was sheared by sonication to obtain DNA fragments of 200–1000 bp. Approximately 1% of the total chromatin was retained as Input, while the remaining samples were equally divided and incubated overnight at 4 °C with either IgG or target-specific ChIP antibody. According to the guidelines of the ChIP Assay Kit (P2080S; Beyotime), magnetic beads were utilized to capture antigen-antibody complexes, and DNA fragments were subsequently eluted using Elution buffer and analyzed by qPCR and agarose gel electrophoresis. Antibody details and dilution ratios are provided in Table S2, while primer sequences are listed in Table S3.

Luciferase reporter assay

After seeding in 24-well plates, cells were co-transfected with the pGL3-DDAH1-Promoter luciferase reporter plasmid and the pRL-TK Renilla luciferase plasmid. 48 h later, cell lysates were prepared based on the standard procedure of the Dual Luciferase Reporter Gene Assay Kit (RG027; Beyotime). Luciferase activity was quantified by a multifunctional microplate reader (Biotek, USA), with Renilla luciferase serving as the internal control for luciferase activity normalization.

Animal studies

All animal experiments were conducted in accordance with protocols approved by the Ethics Committee of Nanjing Medical University (IACUC-2109036). 4-week-old female BALB/c nude mice were obtained from the Animal Center of Nanjing Medical University. A549 cells (4 × 10⁶ cells/mouse) that stably silenced or expressed HMGA1, DDAH1 and STAT1, along with their corresponding control cells, were subcutaneously injected into the left axilla of each mouse. For fludarabine treatment experiments, 16 mice were randomly allocated into 4 groups. Treatment groups received intraperitoneal injections of fludarabine (40 mg/kg) every two days, while control groups were administered an equivalent volume of saline [30, 31]. Tumor dimensions were recorded weekly, and tumor volume was calculated using the standard formula (V = 0.5 × length × width²). At the end of experiments, all mice were euthanized. Excised tumors were immediately weighed, digitally documented, and fixed in 4% FPA for subsequent histopathological analysis, including hematoxylin-eosin (HE) staining and Ki-67 immunohistochemical evaluation.

Statistical analysis

All statistical analysis and representations were performed using R (version 4.2.3), SPSS (version 21.0) and GraphPad Prism (version 8.0.1) software. Student’s t-test was applied to analyze intergroup differences for variables with normal distribution. All in vitro experiments were repeated at least three times. A two-tailed P value less than 0.05 was used to assess statistical significance.

Results

HMGA1 is the main TF correlated with poor prognosis in LUAD

To identify the most critical TF associated with poor prognosis in LUAD, the C-indices of 1,637 TFs were calculated across 7 independent cohorts (TCGA, GSE30219, GSE31210, GSE50081, GSE68465, GSE68571 and GSE72094) (Fig. 1A, Table S5). Noteworthy, HMGA1 exhibited the highest normalized C-index (0.743) among all TFs, indicating its pivotal role in predicting unfavorable outcomes in LUAD. Moreover, the mRNA expression of HMGA1 was elevated in LUAD tissues in comparison to unpaired or paired adjacent normal tissue (Fig. 1B C). Meanwhile, this level underwent a gradual increase in proportion to tumor stage (Fig. 1D). IHC analysis of TMA further confirmed that HMGA1 protein levels was significantly overexpressed in LUAD tissue at both low-grade (I-II) and high-grade (III-IV) stages than corresponding neighboring tissue (Fig. 1E, Figure S1A-S1C).

Fig. 1.

HMGA1 correlates with poor prognosis in LUAD. A Top 20 TFs with the highest average C-index across LUAD cohorts. B-C HMGA1 expression in LUAD tissue versus unpaired or paired adjacent neighboring tissue. D Differential HMGA1 expression between high-grade and low-grade LUAD tissue. E Representative IHC results of HMGA1 expression in LUAD TMA scale bars = 100 μm. F Kaplan-Meier survival curves between HMGA1^high and HMGA1^low groups in the TCGA cohort. G Multivariate Cox regression analysis in the TCGA dataset. H Meta-analysis of HMGA1 expression across LUAD cohorts. I-J GO and KEGG enrichment analyses between HMGA1^high and HMGA1^low subgroups. **p < 0.01, ***p < 0.001

Subsequently, the 5-year survival curves in diverse LUAD cohorts were depicted (Fig. 1F, Figure S1D-I). Patients with higher HMGA1 expression had poorer prognoses across all cohorts. Multivariate Cox analysis identified that HMGA1 as an independent risk factor in LUAD (hazard ratio (HR) [95% confidence interval (CI)] = 1.36 [1.14–1.63], P = 0.001; Fig. 1G). In addition, the meta-analysis of above LUAD cohorts confirmed that LUAD patients with high HMGA1 levels exhibited unfavorable prognosis (HR [95% CI] = 1.44 [1.25–1.66], P = 0.03; Fig. 1H).

GO/KEGG enrichment analysis revealed that HMGA1 was closely correlated with cellular proliferation (Fig. 1I and J). Therefore, we selected the correlation between HMGA1 and tumor proliferation for further investigation.

Elevated HMGA1 promotes LUAD proliferation in vitro and in vivo

The expression levels of HMGA1 were evaluated across various cell lines, revealing that all LUAD cell lines exhibited elevated HMGA1 expression compared to the normal bronchial epithelial cell line BEAS-2B, with the A549 and PC9 cell lines showing particularly high levels (Fig. 2A). Thus, these two cell lines were selected in the follow-up experiments.

Fig. 2.

HMGA1 promotes LUAD proliferation in vitro and in vivo. A Endogenous HMGA1 expression profiles in various LUAD cell lines. B-C Validation of HMGA1 overexpression plasmid and siRNA transfection efficiency by qPCR and Western blot. D-E CCK8 assays were used to assess the viability of A549 and PC9 cells transfected with si-HMGA1 and HMGA1 overexpression plasmids. F-G EdU assays were used to analyze the proliferation in transfected A549 and PC9 cells scale bars = 100 μm. H-I Colony formation assays were performed to detect the clonogenicity of A549 and PC9 cells after HMGA1 manipulation. J Tumor volumes were calculated after injection per week. K Image of LUAD xenografts in groups injected with HMGA1-slienced or control A549 cells. L Tumor weight was assessed from excised tumors. M Representative images of tumors subjected to HE and Ki-67 staining (scale bars = 50 μm). Data represent mean ± SD for n = 3 independent replicates. *p < 0.05, **p < 0.01, ***p < 0.001

To elucidate the function of HMGA1, we performed plasmid-mediated HMGA1 overexpression and siRNA-mediated HMGA1 knockdown in LUAD cells, and the transfection efficiency was evaluated by qPCR and Western blot (Fig. 2B and C). Cell proliferation assays using the CCK8 method showed that HMGA1 overexpression significantly promoted cell proliferation, while the knockdown of HMGA1 markedly inhibited this effect (Fig. 2D and E). Additionally, EdU assays indicated a positive correlation between HMGA1 levels and the percentage of EdU positive cells (Fig. 2F and G). Moreover, colony formation assays revealed that HMGA1 silencing attenuated the clonogenic capacity of LUAD cells (Fig. 2H and I).

Given the pronounced inhibitory effect of si-HMGA1 #2 on LUAD proliferation, this specific sequence was chosen for shRNA construction to create stable HMGA1 knockdown cell lines (Figure S2A-B). Subsequently, HMGA1-silenced A549 cells, along with negative control cells, were subcutaneously implanted into nude mice. The results exhibited significant reductions in tumor volume, weight, and Ki-67 positivity, indicating that HMGA1 knockdown effectively suppressed tumorigenesis in LUAD xenograft mouse models (Fig. 2J and M, Figure S2C).

HMGA1 activates DDAH1 transcription by binding with STAT1

To investigate the mechanism by which HMGA1 accelerates LUAD proliferation, a transcriptomic analysis was conducted following the knockdown of HMGA1 (Fig. 3A and B). Notably, DDAH1 emerged as one of the significantly downregulated genes in response to HMGA1 depletion, in contrast to traditional cell cycle regulators. It implied that HMGA1 might facilitate its pro-proliferative effects, at least in part, by influencing DDAH1 expression. These observations were further validated by qPCR and Western blot (Fig. 3C, Figure S3A). Moreover, the knockdown of HMGA1 in LUAD cells was associated with a decreased degradation rate of ADMA, leading to an accumulation of ADMA and a concomitant reduction in NO production (Fig. 3D and E).

Fig. 3.

HMGA1 activates the ADMA/NO pathway by elevating DDAH1 expression. A Transcriptomic profiling of A549 cells after HMGA1 knockdown. B-C DDAH1 decreased its expression in the shHMGA1 subgroup. D-E Concentrations of ADMA and NO between HMGA1 knockdown or normal LUAD cell lines. F Validation of DDAH1 exogenous plasmid and shRNA lentivirus efficiency. G CCK8 assays were used to assess the viability of A549 and PC9 cells infected with DDAH1 plasmid and shDDAH1 lentivirus. H EdU assays were uesd to analyze the proliferation in A549 and PC9 cells after DDAH1 manipulation scale bars = 100 μm. I Colony formation assays were performed to detect the clonogenicity of infected A549 and PC9 cells. J Tumor volumes were calculated after injection per week. K Image of xenografts in groups injected with DDAH1-slienced or control A549 cells. L Tumor weight was assessed from excised tumors. M Representative images of tumors subjected to HE and Ki-67 staining (scale bars = 50 μm). Data represent mean ± SD for n = 3 independent replicates. *p < 0.05, **p < 0.01, ***p < 0.001

Following this, the phenotypes of DDAH1 regarding cell proliferation in LUAD cell lines were further validated. The expression of DDAH1 in A549 and PC9 cell lines was regulated by exogenous plasmid and shRNA lentivirus (Fig. 3F, Figure S3B-C). Overexpression of DDAH1 significantly enhanced proliferative capacity, as evidenced by CCK8, EdU, and colony formation assays, whereas DDAH1 knockdown attenuated this effect (Fig. 3G and I). Xenograft experiments revealed that DDAH1 knockdown resulted in a decrease in tumor volume, weight and ki-67 positivity (Fig. 3J and M, Figure S3D). These findings indicated that DDAH1 exerted a pro-proliferative effect in LUAD cell lines, and HMGA1 might promote tumor proliferation through modulating DDAH1 expression.

Due to the structural limitations, HMGA1 does not possess intrinsic transcriptional activity; instead, it exerts its regulatory functions through interactions with other TFs [5]. Consequently, MS analysis was implemented. Among HMGA1-associated proteins, STAT1 was identified as the sole TF, exhibiting high abundance in both IP assays and intracellular contexts (Fig. 4A). IF staining demonstrated that HMGA1 is localized within the nucleus, where it co-localizes with both STAT1 and its phosphorylated form at Tyr701 (Fig. 4B). Co-IP assays further validated the physical interaction between HMGA1 and both STAT1 and phosphorylated STAT1 (Tyr701) (Fig. 4C).

Fig. 4.

HMGA1 activates DDAH1 transcription through STAT1 binding. A MS analysis of HMGA1-interacting proteins. B IF staining indicated that HMGA1 was co-localized with STAT1 and p-STAT1 Tyr701 in the nucleus scale bars = 20 μm. C Co-IP experiments confirmed endogenous interaction between HMGA1 and STAT1/p-STAT1 Tyr701. D JASPAR database prediction of potential binding sites in the DDAH1 promoter region. E ChIP-qPCR analysis demonstrated that HMGA1 and STAT1 bind to the DDAH1 promoter region. F Luciferase reporter assays exhibited transcriptional activity of the DDAH1 promoter regulated by HMGA1 and STAT1. G Representative IHC results of STAT1 expression in LUAD TMA (scale bars = 100 μm). Data represent mean ± SD for n = 3 independent replicates. *p < 0.05, **p < 0.01, ***p < 0.001

ChIP and luciferase reporter assays were then utilized to illustrate the mechanism by which the HMGA1/STAT1 complex activates DDAH1 transcription. Two potential binding sites within the DDAH1 promoter region were identified based on predictions from the JASPAR database (Fig. 4D). ChIP-qPCR analysis demonstrated that both HMGA1 and STAT1 could bind to the DDAH1 promoter, enhancing its transcriptional activity (Fig. 4E). Additionally, luciferase reporter assays showed that DDAH1 promoter activity was diminished following the knockdown of HMGA1 and STAT1, and enhanced with STAT1 overexpression (Fig. 4F, Figure S4A). IHC analysis of TMA further validated the critical roles of STAT1 in the LUAD cohort. Compared to adjacent normal tissue, STAT1 was substantially elevated in both low-grade (I-II) and high-grade (III-IV) stages, indicating that STAT1 may play a role in the pathogenesis of LUAD (Fig. 4G, Figure S4B-S4D).

HMGA1 promotes LUAD proliferation via STAT1-mediated upregulation of DDAH1

Subsequent rescue experiments were conducted to determine whether the pro-proliferative effect of HMGA1 is contingent upon the STAT1-mediated upregulation of DDAH1. The efficiency of transfection and the corresponding expression levels were validated by qPCR and Western blot analysis (Fig. 5A and B, Figure S5A). HMGA1 knockdown reduced DDAH1 expression, whereas STAT1 overexpression effectively reversed this suppression. The activation status of the ADMA/NO pathway was also assessed (Fig. 5C and D). It was found that STAT1 upregulation rescued the blockage of ADMA degradation caused by HMGA1 knockdown and concurrently promoted the synthesis of NO.

Fig. 5.

HMGA1 promotes LUAD proliferation via STAT1-mediated upregulation of DDAH1. A-B Transfection efficiency and downstream DDAH1 expression level after rescue experiments were validated by qPCR and Western blot. C-D ADMA and NO concentrations of LUAD cell lines in rescue experiments. E-G STAT1 overexpression reversed the anti-proliferative effects of HMGA1 silencing, as demonstrated by CCK8, EdU and colony formation assays. H Tumor volumes were calculated after injection per week. I Image of LUAD xenografts in rescue experiments. J Tumor weight was assessed from excised tumors. K Representative images of tumors subjected to HE and Ki-67 staining (scale bars = 50 μm). Data represent mean ± SD for n = 3 independent replicates. *p < 0.05, **p < 0.01, ***p < 0.001

Following this, a series of functional experiments were conducted to confirm the restorative effect of STAT1 on the anti-proliferative characteristics associated with HMGA1 knockdown. Cell viability assay revealed that the overexpression of STAT1 counteracted the proliferation inhibition resulting from HMGA1 knockdown (Fig. 5E). Additionally, EdU assays indicated that STAT1 reinstated the elevated proportion of EdU-positive cells that had been diminished by HMGA1 silencing (Fig. 5F). Furthermore, colony formation assays demonstrated that the upregulation of STAT1 partially restored the clonogenic potential of cells with HMGA1 silencing, as evidenced by an increased number of colonies (Fig. 5G). In alignment with these observations, xenograft tumor experiments indicated that the presence of STAT1 markedly mitigated the reductions in tumor volume, weight, and the proportion of Ki-67 positive cells upon HMGA1 knockdown, suggesting a successful reversal of the tumor growth inhibition associated with HMGA1 silencing (Fig. 5H and K, Figure S5B).

Furthermore, loss-of-function assays were conducted for HMGA1 and STAT1 to elucidate the regulatory roles of HMGA1 and STAT1 in DDAH1-mediated proliferation. Knockdown of either HMGA1 or STAT1 differentially suppressed DDAH1 expression and the activation of the ADMA/NO signaling pathway, while simultaneous depletion of both HMGA1 and STAT1 resulted in more pronounced inhibition (Fig. 6A and D, Figure S5C). Additional CCK8, EdU incorporation, and colony formation experiments also confirmed these observations (Fig. 6E and G). Notably, HMGA1 exhibited more potent anti-proliferative effects than STAT1. In vivo experiments further revealed that knockdown of either HMGA1 or STAT1 significantly suppressed tumor progression, as reflected by obvious reductions in tumor volume, weight, and ki-67 positivity, and the double knockdown of HMGA1/STAT1 exacerbated these effects (Fig. 6H and K, Figure S5D). Our findings suggested that HMGA1 and STAT1 cooperatively regulate tumor proliferation through the modulation of DDAH1 expression and the subsequent activation of the ADMA/NO signaling cascade.

Fig. 6.

Knockdown of HMGA1 and STAT1 inhibited the DDAH1-mediated LUAD proliferation. A-B Transfection efficiency and downstream DDAH1 expression level after rescue experiments were validated by qPCR and Western blot. C-D ADMA and NO concentrations of LUAD cell lines in rescue experiments. E-G Knockdown of HMGA1 and STAT1 inhibited the proliferative activity of A549 and PC9, as reflected by CCK8, EdU and colony formation assays. H Tumor volumes were calculated after injection per week. I Image of LUAD xenografts in rescue experiments. J Tumor weight was assessed from excised tumors. K Representative images of tumors subjected to HE and Ki-67 staining scale bars = 50 μm). Data represent mean ± SD for n = 3 independent replicates. *p < 0.05, **p < 0.01, ***p < 0.001

Fludarabine inhibits the progression of HMGA1^high LUAD

Considering that HMGA1 plays a critical role in LUAD, targeting HMGA1 may provide clinical benefits for LUAD patients by improving survival outcomes. However, drugs precisely targeting HMGA1 are still lacking. Since HMGA1 exerts its biological functions through cooperative interactions with other TFs, we postulated that disrupting the HMGA1/STAT1 complex might achieve similar therapeutic benefits. The STAT1 inhibitor fludarabine exhibited the half-maximal inhibitory concentration (IC₅₀) values of 24.13µM and 36.38µM in A549 and PC9 cells, respectively (Fig. 7A). Treatment with fludarabine effectively attenuated proliferative advantage conferred by HMGA1 upregulation (Fig. 7B and C).

Fig. 7.

Fludarabine inhibits the progression of HMGA1^high LUAD. A The IC₅₀ value was determined after treating LUAD cells with fludarabine at concentration gradients ranging from 0 to 40 µM for 48 h. B-C Relative proliferation and clone formation capacity of LUAD cells with empty vector or with overexpression of HMGA1 under treatment with fludarabine at the corresponding IC₅₀ values. D Experimental design of fludarabine treatment in BALB/c nude mice harboring HMGA1^high or control LUAD xenograft models. E Tumor volumes were calculated after injection per week. F Image of LUAD xenografts. G Tumor weight was assessed from excised tumors. H-I Representative images of tumors subjected to HE and Ki-67 staining scale bars = 50 μm). J Schematic model illustrating the mechanism of HMGA1-mediated cellular proliferation in LUAD. Data represent mean ± SD for n = 3 independent replicates. *p < 0.05, **p < 0.01, ***p < 0.001

Accordingly, we established an in vivo treatment model (Fig. 7D). Nude mice were inoculated subcutaneously with either HMGA1-overexpressing or normal A549 cells, and subsequently treated with fludarabine. Our results elucidated that HMGA1 upregulation promoted tumor growth, whereas fludarabine effectively hindered tumor progression, particularly in HMGA1^high LUAD xenograft models (Fig. 7E and I). These observations indicate that targeting STAT1 with fludarabine may represent a potential approach to counteract the pro-tumorigenic mechanism mediated by HMGA1 and could serve as a novel therapeutic strategy for HMGA1^high LUAD patients.

Discussion

As illustrated in Fig. 7J, our study identified HMGA1 as the predominant TF contributing to the unfavorable prognosis of LUAD. In detail, HMGA1 collaborated with STAT1 to transcriptionally upregulate DDAH1 expression, thereby accelerating LUAD proliferation. Importantly, fludarabine significantly inhibited the progression of HMGA1^high LUAD models. These findings uncover a novel molecular mechanism by which HMGA1 drives LUAD progression and highlight its potential as a dual-purpose biomarker and therapeutic target for LUAD patients.

The pro-proliferative phenotype of HMGA1 has been reported in several tumors. In gastric cancer (GC), HMGA1 facilitated cell proliferation by activating the Wnt/β-catenin pathway [32]. Similarly, in pancreatic cancer, HMGA1 drove tumorigenesis and stroma formation through the secretion of FGF19 [6]. Beyond its pro-proliferative effects, HMGA1 has also been implicated in remodeling the tumor immune microenvironment (TIME). For instance, HMGA1B/2 transcriptionally upregulates POU1F1 in GC, which in turn induced macrophage polarization through the CXCL12/CXCR4 axis [33]. Likewise, in colorectal cancer (CRC), the long non-coding RNA RP11-417E7.1 cooperated with HMGA1 to facilitate the chromatin loop formation, thereby enhancing the transcription of THBS2. Subsequently, CRC-derived exosomes containing THBS2 further promoted M2 macrophage polarization and contributed to an immunosuppressive TIME [34]. In this study, we unveiled a novel link between HMGA1 and the activation of the DDAH1-ADMA/NO pathway. Notably, NO exhibits a concentration-dependent biphasic effect on tumor proliferation [35]. High concentrations of NO induce cytotoxicity and tissue damage through oxidative stress-mediated generation of reactive oxygen species (ROS), whereas low concentrations of NO function as transmembrane signaling molecules that promote angiogenesis and tumor proliferation [36]. Interestingly, HMGA1 has been reported to directly bind to the inducible nitric oxide synthase (iNOS, NOS2) promoter region, cooperating with NF-κB to enhance NOS2 transactivation in endotoxemia [37, 38].

Since HMGA1 lacks intrinsic transcriptional activity, it exerts its transcriptional regulatory function by interacting with other TFs [3]. For example, the HMGA1/FOXM1 complex promoted VEGFA transcription, thereby enhancing tumor angiogenesis and facilitating distant metastasis in breast cancer [7]. Additionally, HMGA1 cooperated with Sp1 and C/EBPβ to stimulate transcription of the human insulin receptor (IR) gene, consequently alleviating insulin resistance [39]. Our study demonstrated that HMGA1 co-localized with STAT1 in the nucleus. As previously reported, HMGA1 exhibits strict nuclear localization, whereas STAT1 is distributed in both the nucleus and cytoplasm, suggesting that HMGA1 selectively interacts with STAT1 in the nucleus. Noteworthy, phosphorylation at Tyr701 serves as a critical switch for STAT1 nuclear translocation and transcriptional activation [40]. Accordingly, HMGA1 might specifically interact with p-STAT1 (Tyr701) within the DDAH1 promoter region, synergistically promoting the transcription of DDAH1, thereby facilitating LUAD proliferation and progression.

HMGA1/STAT1 complex showed significant pro-proliferative impacts in LUAD, suggesting that targeting them could effectively block downstream pathway activation and thus suppress tumor progression. Netropsin, a minor groove DNA-binding agent, specifically interferes with protein binding at AT-rich regions, leading to its proposed role as a potential HMGA1 inhibitor in several studies [34, 41]. However, this inhibitory effect appears not exclusive to HMGA1, as it may also impair the catalytic activity of other proteins like topoisomerases [42]. Guided by the principle of precision medicine in oncology, we established an in vivo therapeutic model to validate the feasibility of our targeting strategy. Fludarabine is principally used in clinical practice for the treatment of chronic lymphocytic leukemia (CLL), and it also serves as a lymphodepleting agent prior to CAR-T therapy for solid tumors [43, 44]. Besides, several studies have revealed its radiosensitizing effects in solid tumor radiotherapy [30, 45]. In the present study, fludarabine significantly impeded progression in HMGA1^high LUAD models. This finding suggests that biomarkers identified through bioinformatic approaches may serve as stratification criteria to distinguish potential beneficiaries of existing therapeutic agents, facilitating the advancement of individualized treatment strategies.

However, this study still has several limitations need to be improved. Firstly, more LUAD samples should be collected to assess the best cutoff between HMGA1^high and HMGA1^low groups. Moreover, the STAT1 inhibitor fludarabine might exhibit off-target effects by suppressing DNA synthesis, which contributes to its antitumor activity. More precise STAT1-targeting strategies, such as proteolysis-targeting chimeras (PROTACs) and nanoparticle-based delivery systems, needs further investigation in future studies. Furthermore, the inhibitory effect of fludarabine on HMGA1^high LUAD requires further validation in patient-derived organoid models and preclinical mouse models.

Conclusion

In conclusion, our research indicates that the HMGA1/STAT1 complex facilitates LUAD proliferation and progression through the regulation of the DDAH1-ADMA/NO signaling pathway. Furthermore, experiments utilizing xenograft mouse models have demonstrated that LUAD characterized by high levels of HMGA1 exhibits increased sensitivity to fludarabine treatment. This underscores the potential of HMGA1 as both a biomarker and a therapeutic target in the management of LUAD.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Y-Q. Shu and D. Xu contributed to the conception and design of the study. T. Hu and R. Shi performed the bioinformatic analysis. T. Hu and S-Y. Yin implemented the experiments. T. Hu and R. Shi wrote the first draft of the manuscript. S-Y. Yin, T-T. Xu and D. Xu assisted with language improvement. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (#82404066 to D. Xu, #82172889 to Y-Q. Shu), Natural Science Foundation of Jiangsu Province (#BK20241139 to D. Xu) and Jiangsu Province Capability Improvement Project through Science, Technology and Education (#CXZX202204 to Y-Q. Shu).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

All procedures related to the LUAD TMA were approved by the Ethics Committee of Shanghai Outdo Biotech with an approval number (ZQTYHM-19-01). Participants gave informed consent to participate in the study before taking part.

Competing interests

The authors declare no competing interests.