Transcription factor ZNF266 suppresses cancer progression by modulating CA9-mediated intracellular pH alteration in lung adenocarcinoma

Abstract

Background

Lung cancer remains the leading cause of cancer-related mortality globally, with lung adenocarcinoma (LUAD) being the most prevalent subtype. Despite extensive research efforts, the role of transcription factors in LUAD progression remains largely uncharacterized. In this study, we focused on ZNF266, a transcription factor whose impacts on LUAD have not been investigated.

Methods

Using high-throughput sequencing data, we observed a significant downregulation of ZNF266 expression in LUAD tissues. To validate this finding, we conducted a retrospective analysis of nearly three thousand LUAD patients’ data from public databases and our institution. Functional studies were performed using cell lines, organoids, and xenograft models to assess the role of ZNF266 in LUAD progression. RNA sequencing, chromatin immunoprecipitation, DNA pull-down assays, and dual-luciferase reporter assays were employed to elucidate the underlying mechanism. Additionally, adeno-associated virus (AAV)-mediated overexpression of ZNF266 was used to evaluate its therapeutic potential.

Results

Patients with low ZNF266 expression had poorer prognosis compared to those with high expression. ZNF266 inhibits the malignant phenotypes of LUAD, including proliferation, migration, and invasion. Mechanistically, ZNF266 binds to the promoter region of CA9, suppressing its transcription. This leads to a reduction in intracellular pH and subsequent inhibition of the mTOR signaling pathway, which is crucial for cancer cell growth and survival. Furthermore, AAV-mediated overexpression of ZNF266 significantly inhibited tumor growth in patient-derived xenograft models.

Conclusions

Our study demonstrated that ZNF266 inhibits LUAD progression in a pH-dependent manner via modulating CA9 expression, uncovering its therapeutic significance for LUAD treatment.

Background

Lung cancer remains one of the most frequently diagnosed malignancies and is the leading cause of cancer-related death globally [1]. Among its types, non-small cell lung cancer (NSCLC) accounts for approximately 85% of cases, with lung adenocarcinoma (LUAD) as the predominant subtype [2]. The primary treatment modalities for patients with LUAD encompass surgery, chemotherapy, radiation therapy, molecular targeted therapies, and immunotherapy [3]. Although great progress has been made in cancer treatment recently, the prognosis for LUAD patients remains poor, with a five-year survival rate under 26% [4].

Transcription factors (TFs) are proteins that recognize specific DNA sequences to control chromatin and transcription [5]. Aberrant expression of transcription factors leads to abnormal cell activities, including sustaining proliferative signaling, evading growth suppressors, and activating invasion and metastasis, which are also hallmarks of cancer [6]. Multiple agents targeting transcription factors, such as the MYC regulator RGT-61,159 [7] and the STAT3 inhibitor TTI-101 [8], have been assessed in clinical trials and exhibited promising efficacy in cancer treatment. Although thousands of transcription factors exist, the roles and therapeutic potential of many remain inadequately understood.

In this study, we demonstrated that Zinc Finger Protein 266 (ZNF266) potently inhibits the proliferation, migration, and invasion of LUAD both in vitro and in vivo. ZNF266 expression is significantly downregulated in LUAD, where it transcriptionally represses carbonic anhydrase IX (CA9), a pH regulator critical for maintaining cancer cell intracellular pH (pHi). This CA9–pHi axis activates the mammalian target of rapamycin (mTOR) signaling pathway, driving the malignant phenotypes of cancer. Most importantly, our findings from organoids and murine xenograft models confirmed the inhibitory role of ZNF266 in carcinogenesis. Notably, gene therapy utilizing peritumoral injection of adeno-associated virus (AAV) to deliver ZNF266 expression to the tumor significantly reduced the size of the xenograft tumors. Taken together, our study provides fundamental mechanistic insights into the role of ZNF266 and highlights its potential in the treatment of LUAD.

Methods

Bioinformatic analysis

Transcriptomic profiles alongside clinical and survival metadata for LUAD cases were retrieved from TCGA and GEO repositories. Employing the limma package (R/Bioconductor), differentially expressed genes (DEGs) were detected via an empirical Bayesian framework to model variance-stabilized expression shifts using moderated t-tests, with significance thresholds set at an adjusted p < 0.001 and absolute log2 fold change > 1.0. Survival outcomes were analyzed using Kaplan-Meier methodology through the survminer and survival R packages, supplemented by the Kaplan-Meier Plotter online resource (http://kmplot.com). The human transcription factor repertoire was obtained from the hTFtarget database. Publicly available ChIP-seq datasets were acquired from ENCODE and visualized using the Integrative Genomics Viewer (IGV). For functional annotation, DEGs derived from ZNF266-OE and negative control groups underwent pathway enrichment analysis via clusterProfiler, with Kyoto Encyclopedia of Genes and Genomes (KEGG) terms deemed significant at an adjusted p < 0.05 and false discovery rate (FDR) < 0.05.

Patients selection

Between February 2016 and October 2023, we gathered resected tumor tissue and corresponding clinical information from 173 patients with LUAD who underwent surgical resection at the Thoracic Surgery Department of Zhongshan Hospital, Fudan University. These procedures were performed by highly skilled thoracic surgeons, and the resected specimens, including tumors and lymph nodes, underwent thorough examination and confirmation by at least two competent pathologists. The staging of the tumors was determined according to the TNM classification system, as outlined in the eighth edition of the Cancer Staging Manual published by the American Joint Committee on Cancer (AJCC). This study strictly adhered to the ethical guidelines established in the Declaration of Helsinki and received approval from the Research Ethics Committee of Zhongshan Hospital, Fudan University (approval number: B2022-180R).

Cell lines and compounds

LUAD cell lines (A549 and PC-9) were sourced from the Chinese Academy of Sciences cell bank. Short Tandem Repeat (STR) profiling confirmed the genetic identity and purity of these lines, with cells under 30 passages used for experimental work. Cells were cultured in DMEM high-glucose medium (KeyGEN BioTECH, Nanjing, China), enriched with either 10% fetal bovine serum (Lonsera, Shuangru Biotechnology Co., Ltd, Suzhou, China) or dialyzed fetal bovine serum (VivaCell, Shanghai, China), along with 0.1 mg/ml streptomycin and 100 U/ml penicillin (Beyotime, Shanghai, China). Culturing was conducted in a 37 °C incubator under a 5% CO₂, humidified environment. Sodium hydroxide (NaOH in PBS) and hydrochloric acid (HCl, 37 wt % in water) were purchased from Aladdin (Shanghai, China).

Lentiviral infections and CRISPR/Cas9-mediated gene knockout

Lentiviral vectors for ZNF266, CA9, and negative control were developed by Genechem Technology (Shanghai, China). Cells were transfected with these lentiviruses using HiTrans A reagent (Genechem Technology) as per the manufacturer’s instructions and then selected with puromycin (Beyotime) for 48 h. To achieve precise gene knockout, CRISPR/Cas9 technology was employed in the cell lines. Specific sgRNAs targeting ZNF266 were designed and incorporated into GV392 plasmids, which included both puromycin resistance and the hSpCas9 gene. To achieve the overexpression of ZNF266 after knockout, the vectors carrying modified ZNF266 cDNA sequence refractory to the inhibition of sgRNAs were generated. These modified vectors were subsequently packaged into lentiviruses within HEK-293T cells. Genechem Technology conducted the entire process of vector design, assembly, and lentivirus packaging. Table S3 provides further details on the sgRNA used.

RNA sequencing

To investigate the transcriptional response of ZNF266 overexpression in A549, total RNA was extracted using the TRIzol reagent (TIANGEN, Beijing, China). Sequencing was conducted by OE Technology (Shanghai, China), and the raw data were normalized to Fragments Per Kilobase of exon model per Million mapped fragments (FPKM) format for subsequent analysis.

RNA isolation and quantitative real-time PCR (qRT-PCR)

RNA extraction was carried out using TRIzol reagent sourced from TIANGEN. Following this, the isolated RNA was converted into cDNA employing the Hifair II First-Strand cDNA Synthesis Kit, manufactured by YEASEN (Shanghai, China). The qRT-PCR reactions were conducted using Hieff qPCR SYBR Green Master Mix (YEASEN) on an ABI QuantStudio 5 real-time PCR platform provided by Thermo Fisher Scientific. Each experimental sample was assayed in triplicate, and the relative mRNA expression levels were calculated using the 2-ΔΔCT method. For normalization, β-actin was used as an endogenous reference gene. The primers used in this study are listed in Table S4.

Western blot

Cell lysates were generated by homogenizing samples in ice-cold RIPA buffer (Beyotime) containing a protease inhibitor cocktail and two phosphatase inhibitors (TargetMol). Protein levels were quantified via the BCA assay (YEASEN) according to the manufacturer’s protocol. For immunoblotting, 20 µg of protein per sample was resolved on SDS-PAGE gels and electrophoretically transferred to PVDF membranes (Millipore). Membranes were blocked for 1 h at room temperature with 5% non-fat milk dissolved in Tris-buffered saline containing 0.1% Tween-20 (TBST). Primary antibody incubations were performed overnight at 4 °C, followed by three 10-min TBST washes to remove unbound antibodies. Horseradish peroxidase (HRP)-linked secondary antibodies (Beyotime; 1:2,500 dilution) were applied for 1 h at room temperature. Chemiluminescent signals were captured using BeyoECL Moon substrate (Beyotime), with exposure times optimized for band clarity. β-actin was uniformly employed as the normalization control across all experimental conditions. A full list of primary antibodies, including dilutions and vendors, is provided in Table S5.

Cell viability assay

To assess cell proliferation, cells were plated in quintuplicate into 96-well plates at a density of 1,500 cells per well during their logarithmic growth phase. Then the cell viability was measured as previously described [9]. Following incubation at 37 °C for various durations (0, 24, 48, 72, 96, and 120 h), cell proliferation was evaluated using the Enhanced Cell Counting Kit-8 Viability Assay Kit (Topscience, Shanghai, China), adhering strictly to the manufacturer’s instructions.

High-content imaging system analysis

For the analysis utilizing the high-content imaging system, cells transfected with lentivirus carrying the Green Fluorescent Protein (GFP) sequence were seeded in quintuplicate into 96-well plates at a density of 1,800 cells per well during the logarithmic growth phase. Subsequently, the plates were positioned within the Operetta CLS High-Content Analysis System (PerkinElmer, Waltham, USA), which was set to a 37 °C incubator with 5% CO₂ and maintained under these conditions for 120 h. Meanwhile, the fluorescence images of the cells were captured and analyzed at 24-hour intervals using the Harmony 4.9 software.

Wound healing assay

On average, various cell groups were seeded into a 6-well plate. These cells were allowed to proliferate into a monolayer and subsequently scratched manually using a 200 µL pipette tip. The cells were then rinsed with phosphate-buffered saline (PBS) and incubated at 37 °C in media devoid of serum. Phase-contrast microscopy was used to capture images of the wounded areas at 24-hour intervals. Each experiment was independently replicated three times to ensure reproducibility.

Transwell migration and invasion assay

The abilities of cell migration and invasion were assessed using Transwell assays as previously described [10]. Specifically, LUAD cells (3 × 10⁴ cells/well) were plated in the upper chamber (BIOFIL, Guangzhou, China), either without (for the Transwell migration assay) or with (for the Transwell invasion assay) Matrigel coating (YEASEN), following the manufacturer’s instructions. Subsequently, serum-free medium was added to the upper chambers, while the lower chambers contained serum-rich medium. The cells were then incubated for 24 h at 37 °C, in an atmosphere of 5% CO₂. After incubation, the cells that had migrated or invaded into the lower chambers were fixed with 4% paraformaldehyde for 10 min and stained with 0.5% crystal violet. Images of each assay were captured using an inverted microscope (Olympus, Tokyo, Japan), and the cell counts within each chamber were determined using ImageJ software. The experiments were conducted in triplicate to ensure reproducibility.

Intracellular pH (pHi) detection in vitro

There are three methods for detecting pHi in vitro using BCECF (20,70-bis-(2-carboxyethyl)-5(and-6)-carboxyfluorescein)-AM probes (Dojindo, Kumamoto, Japan). In the first method, cells from various groups were seeded in 24-well plates. Following 24 h of incubation, cells were treated in PBS solution containing 2 µM BCECF-AM at 37 °C for 20 min. Fluorescence was then measured using an inverted microscope (Olympus) with excitation at 488 nm. In the second method, cells from various groups were seeded in 24-well plates and incubated for 24 h. They were subsequently treated in PBS solution containing 2 µM BCECF-AM at 37 °C for 20 min, then trypsinized, washed, and resuspended in 0.5 mL of PBS for flow cytometry analysis. At least 20,000 cells were analyzed per condition using FlowJo software (TreeStar, Woodburn, OR, USA). In the third method, as previously described [11, 12], cells were seeded in 96-well plates. Upon reaching 60–70% confluency, cells were treated with 2 µM BCECF-AM in phosphate-buffered saline (PBS) for 20 min at 37 °C. Fluorescent signals were quantified using a microplate reader, with dual excitation wavelengths (440 nm and 490 nm) and emission at 530 nm. For pH calibration, cells labeled with BCECF-AM were equilibrated for 10 min in high-potassium buffers adjusted to predefined pH ranges (6.2–6.6, 7.0–7.4, 7.8) containing 10 µM nigericin, a K+/H + ionophore, prior to signal acquisition.

Chromatin Immunoprecipitation (ChIP) assay

The ChIP assays were conducted using the SimpleChIP^® Plus Enzymatic Chromatin IP Kit (Magnetic Beads) (CST, Danvers, USA) as previously described [13]. Chromatin fragmentation was initiated by formaldehyde cross-linking of intact cells, followed by membrane disruption and digestion with micrococcal nuclease (MNase). Following fragmentation, previously cross-linked chromatin was subjected to immunoprecipitation using monoclonal antibodies specific to the target proteins, with incubations performed overnight at 4 °C. For positive and negative controls, a Histone H3 antibody (CST) and human IgG (CST) were used, respectively. The specific antibodies used are listed in Table S5. Following extensive washing, DNA was eluted from the beads and subjected to qPCR analysis.

DNA pull-down assay

For DNA pull-down assays, 5’-biotinylated primers were employed to generate oligonucleotide probes by PCR amplification. Probe integrity was confirmed by 1.5% agarose gel electrophoresis, followed by purification with a commercial gel extraction kit (TaKaRa). Streptavidin-coated magnetic beads (CST) were conjugated with 1 mg biotinylated DNA probes in 20 mL binding buffer. After magnetic capture and three washes with ice-cold BS/THES buffer, beads were incubated with 250 µg nuclear extracts. Negative controls utilized unmodified beads, while competition assays included 20 µg unlabeled probe. Rotational incubation at 4 °C for 12 h included 100 µg/mL salmon sperm DNA (Solarbio) and 10 µg/mL BSA to block nonspecific interactions. Protein-DNA complexes were eluted with SDS sample buffer prior to immunoblotting.

Dual-luciferase reporter assay

The predicted target sites’ CA9 promoter sequence and its corresponding mutated variant were cloned into firefly luciferase reporter plasmid vectors by Genechem. Subsequently, these plasmids were transfected into either control or ZNF266-overexpressing 293T cells utilizing the lipo8000 transfection reagent supplied by Beyotime. Additionally, renilla luciferase reporter plasmids were co-transfected during this process. 48 h post-transfection, dual-luciferase reporter assays were conducted using the Beyotime Luciferase Reporter Gene Assay Kit.

Establishment of patient-derived organoids (PDOs) and viability assay

LUAD tumor tissues were utilized to establish PDOs. The experimental procedures were performed as previously described [11, 14]. Initially, the tissues were meticulously dissected into smaller fragments and thoroughly rinsed three times with ice-cold Hank’s Balanced Salt Solution (HBSS), enriched with 0.3 mg/mL streptomycin and 300 units/mL penicillin. Enzymatic digestion was then performed using a specialized organoid digestive solution sourced from D1Med (Shanghai, China). Following filtration and centrifugation steps, the resultant cell precipitate was resuspended in a 25-fold excess of Matrigel (YEASEN) and plated into a 24-well dish. After allowing the Matrigel to solidify at 37 °C for 25 min, each well received 500 µL of human LUAD-PDO culture medium (D1Med) to support PDO growth.

For lentiviral transfection, PDOs were dissociated into individual cells by a specialized organoid digestive solution (Absin, Shanghai, China) and resuspended in 500 µL of human LUAD-PDO culture medium. Transfection was achieved using HiTrans A reagent, adhering to the manufacturer’s guidelines, for 24 h. Subsequently, the cells were harvested, resuspended in a 25-fold volume of Matrigel, and subjected to puromycin selection for 72 h.

For histological examination, PDOs were collected and fixed in 4% paraformaldehyde. They were then embedded in a combination of agarose and paraffin, sectioned, and processed through standard hematoxylin and eosin staining protocols, which included deparaffinization, dehydration, and staining steps.

To evaluate the viability of PDOs across different groups, they were plated in 96-well plates. Cell viability assessments were conducted using the CellTiter-Lumi Luminescent 3D Cell Viability Assay Kit (Beyotime).

Xenograft studies

The study adhered to the ethical standards set by the Research Ethics Committee at Zhongshan Hospital, affiliated with Fudan University. To establish cell-derived tumor xenograft models (CDXs), 1 × 10⁶ A549 cells from various groups were injected subcutaneously into the right flanks of 4-week-old male BALB/c nude mice, sourced from GemPharmatech (Nanjing, China), using 100 µL of PBS as the carrier. Tumor growth was assessed every 3 days with vernier calipers, starting 10 days post-tumor formation. After 4 weeks, the mice were euthanized, and tumor volumes were calculated employing the formula: (length × width²) / 2. To ensure objectivity, the researchers were blinded to the group allocations during the outcome evaluations.

For the development of patient-derived tumor xenograft models (PDXs), freshly resected tumor specimens from two patients with LUAD were collected and utilized. These samples were transplanted onto the backs of 4-week-old male NSG mice to generate two PDX models (PDX-1 and PDX-2). Before transplantation, the surgical tumor tissues were washed with a 10× solution of penicillin-streptomycin and subsequently cut into 2-mm³ fragments in PBS containing penicillin-streptomycin. The tumor fragments were mixed with Ceturegel Matrix (YEASEN) and PBS in a 1:1 ratio before implantation (P0). Once the graft tumors reached a volume of 200 mm³, they were excised, dissected into smaller pieces, and passaged to another set of NSG mice (P1). Subsequently, P2 PDX models were established. Tumor volumes were measured biweekly by determining the length and width using calipers and applying the formula: (length × width²) / 2. The safety profile of AAV-ZNF266 was rigorously evaluated through monitoring of hematological indices, serum biochemistry, and body weight in mice over a 28-day period. These assessments were conducted relative to peritumorally administered PBS controls to isolate vector-specific effects. In the treatment studies, AAV-Vector (1 × 10¹⁰ vg/mL) and AAV-ZNF266 (1 × 10¹⁰ vg/mL) were injected peritumorally into the mice bearing PDXs. AAV expressing human ZNF266 was produced and purchased from Genechem.

Immunohistochemical (IHC) staining

Tissue samples from patients with LUAD were formalin-fixed and paraffin-embedded. The slides underwent a baking process at 65 °C for a duration of overnight, followed by three consecutive dewaxing procedures in xylene, each lasting 15 min. Rehydration was accomplished through a series of graded ethanol immersions. Antigen retrieval was carried out using citrate buffer sourced from Sangon Biotech (Shanghai, China), and the slides were then treated with peroxidase blocking buffer at 37 °C for 30 min to reduce non-specific binding interactions. Following a 10-minute incubation in blocking buffer, the slides were incubated with primary antibodies at 4 °C overnight, with a subsequent incubation with a secondary antibody for 30 min at room temperature. The staining process was executed utilizing the GTVisionTM III Detection System/Mo&Rb (GeneTech, Shanghai, China), adhering to the manufacturer’s guidelines. The primary antibodies employed for IHC staining are detailed in Table S5.

For the quantitative assessment of IHC staining, the evaluation criteria encompassed both staining intensity and the proportion of positively stained cells. The immunoreactivity score (IRS) was calculated as follows: IRS = Staining intensity × Proportion of positive-stained cells. The staining intensity of the colored cells was graded as follows: 0 (absent/negative), 1 (light yellow/weak), 2 (brownish-yellow/moderate), and 3 (brown/strong).

Multiplex immunohistochemistry (mIHC)

Each tissue section underwent the IHC processing steps outlined previously, including paraffin removal, dewaxing with xylene, rehydration, antigen retrieval, temperature reduction, permeabilization, and blocking procedures. Following this, the sections were incubated with primary antibodies at 4 °C overnight, with subsequent washing in PBS to eliminate unbound antibodies. Thereafter, the sections were incubated with their respective secondary antibodies for one hour and then treated with Tyramide signal amplification (TSA) reagents sourced from Servicebio (Wuhan, China) for 10 min at room temperature. To strip the tissue-bound primary and secondary antibodies, the sections were immersed in a citrate repair solution (pH 6.0) and heated in a microwave oven to maintain boiling for 10 min. This blocking step was repeated, followed by a second application of primary antibodies, subsequent incubation with secondary antibodies, and another treatment with TSA reagents. For nuclear staining, the slides were counterstained with DAPI (Servicebio). The TSA reagents used were as follows: iF440-Tyramide (dilution 1:500), iF488-Tyramide (dilution 1:500), iF546-Tyramide (dilution 1:500), and iF647-Tyramide (dilution 1:500). The primary antibodies utilized in each staining protocol are detailed in Table S5.

Data presentation and statistical analysis

Each experiment was independently replicated a minimum of three times, and the outcomes were presented as the mean with its corresponding standard deviation. Statistical computations and graphical representations were performed using R software, version 4.4.0 (R Foundation for Statistical Computing, Vienna, Austria), ImageJ (National Institutes of Health, USA), IBM SPSS Statistics 20 (USA) and GraphPad Prism 10 (GraphPad Software, La Jolla, California, USA). Depending on the nature of the continuous variables, either an unpaired Student’s t-test or one-way or two-way ANOVA was employed to ascertain statistical differences. All tests were two-tailed, and statistical significance was established at a p-value threshold of less than 0.05.

Results

ZNF266 is decreased in LUAD and associated with poor prognosis

To identify TFs that may play a crucial role in the progression of LUAD, we analyzed data from The Cancer Genome Atlas Program (TCGA) and The Genotype-Tissue Expression (GTEx) databases. By overlapping the sets of human genes that were both differentially expressed in LUAD and associated with prognosis, we uncovered three significant genes: ZNF266, ZEB2, and ZNF493 (Fig. 1A). The effects and mechanisms of ZEB2 have been extensively studied in previous research [15, 16]. Meanwhile, the fold change of ZNF266, as well as its hazard ratio (HR) concerning survival, was more significant than ZNF493. Given the significance of the observed differences and the gaps in current research, we chose ZNF266 as the focus of our study.

We first investigated the clinical implications of ZNF266 in LUAD patients. RNA sequencing (RNA-seq) data from the TCGA and the GTEx databases, including both unpaired samples (normal group = 347, tumor group = 513) and paired samples (n = 58), revealed that the expression of ZNF266 was significantly downregulated in LUAD (Fig. 1B and C). Moreover, low expression of ZNF266 in LUAD patients indicated a poorer prognosis, as demonstrated by the Kaplan–Meier plotter database analysis, the TCGA database analysis and the GEO database analysis (Fig. 1D and E, S1A and S1B). Then, we obtained samples from LUAD patients who underwent radical surgery at Zhongshan Hospital Fudan University between February 2016 and October 2023. The quantitative real-time PCR (qRT-PCR), Western blot, and immunohistochemistry (IHC) analysis confirmed the downregulation of ZNF266 in LUAD tissue (Fig. 1F-I). According to the staining intensity of ZNF266 in IHC, 142 patients (clinicopathological characteristics summarized in Table S1) were divided into low-ZNF266 (n = 80) and high-ZNF266 (n = 62) groups. Notably, the low-ZNF266 group exhibited a higher T stage and N stage compared to the high-ZNF266 group. Kaplan-Meier survival analysis also demonstrated that the low-ZNF266 group had significantly shorter overall survival (OS) compared to the high-ZNF266 group (Fig. 1J). Furthermore, univariate and multivariate Cox regression analyses identified high ZNF266 expression as a significant and independent predictor of better OS for LUAD patients (Table S2). Based on the Cox regression analysis, a nomogram was constructed that incorporated the predictors of OS, including age, pathologic T stage, pathologic N stage, pathologic M stage, and ZNF266 (Fig. 1K). The calibration curves demonstrated good reproducibility and predictability of the nomogram model during internal validation (Fig. 1L). Taken together, the above results indicate that the transcription factor ZNF266 is downregulated and serves as a prognostic biomarker for LUAD.

Fig. 1

ZNF266 is decreased in lung adenocarcinoma (LUAD) and associated with poor prognosis. (A) Venn diagram demonstrates the sets of genes that were differentially expressed in LUAD, those associated with prognosis, and human transcription factors. (B-C) ZNF266 expression levels in normal and tumor samples from the TCGA-LUAD and GTEx databases, including both unpaired samples (normal group = 347, tumor group = 513) (B) and paired samples (n = 58) (C). (D-E) Kaplan–Meier analysis of overall survival comparing the high and low expression of ZNF266 in LUAD patients in the Kaplan–Meier plotter database (D) and the TCGA database (E). (F) RT-qPCR analysis of ZNF266 mRNA expression in 31 LUAD tumor tissues and paired normal tissues from Fudan University-affiliated Zhongshan Hospital cohort (FDU-ZSH cohort). (G) Western blot analysis of ZNF266 protein expression in paired LUAD tissues. (H-I) Representative images (H) and quantification (I) of immunohistochemistry (IHC) staining of ZNF266 in LUAD tissues and paired normal tissues. Scale bars, 200–100 μm. (J) Kaplan–Meier analysis of overall survival according to ZNF266 expression in LUAD patient samples from the FDU-ZSH cohort. (K) Nomogram plot predicting the overall survival probability. (L) Calibration curve of the nomogram using the bootstrap method in internal validation. Error bars validated the standard error of the mean within the current data set

ZNF266 inhibits the proliferation, migration, and invasion of LUAD

To explore the function of ZNF266 in LUAD, we first generated ZNF266-knockout (KO) and ZNF266-overexpressing-after-knockout (OE) LUAD cell lines (A549 and PC-9) using lentiviral vectors (Fig. 2A and S2A). CCK8 assays and high-content imaging analysis demonstrated that in comparison to the negative control (NC) cells, ZNF266-KO resulted in a significant increase in proliferation (Fig. 2B and C). Conversely, ZNF266-OE cells exhibited a decreased proliferation rate (Fig. 2B and C). Wound healing and transwell assays revealed that knockout of ZNF266 enhanced the migration and invasion of LUAD cells, whereas overexpression of ZNF266 exhibited opposite results (Fig. 2D and E).

Furthermore, we successfully established patient-derived organoids (PDOs) and created ZNF266-KO, OE, and NC PDO models (Fig. 2F). The histological and molecular characteristics of the primary LUAD tissues were accurately recapitulated in the LUAD PDOs (Fig. 2G and H). We examined the morphology and measured the ATPase activity of PDOs (Fig. 2I and J), and discovered that the level of ZNF266 expression impacts the vitality and growth of organoids, aligning with what we’ve observed in cellular assays.

To verify the findings in vivo, we conducted cell line-derived xenograft (CDX) tumor formation assays in nude mice (n = 6 for each group) (Fig. 2K). A549 cells from different groups were injected subcutaneously into nude mice. It was observed that ZNF266 overexpression significantly inhibited tumor formation, whereas ZNF266 deletion promoted tumor progression (Fig. 2L and M). Our results confirmed that ZNF266 expression level impacts the hallmarks of LUAD both in vitro and in vivo.

Fig. 2

ZNF266 inhibits the proliferation, migration, and invasion ability of LUAD. (A) Western blot analysis verifying the ZNF266 knockout and overexpression efficiency in A549 and PC-9 cells. (B-C) CCK8 assays (n = 6 per group) (B) and high-content imaging analysis (n = 3 per group) (C) show the effects of ZNF266 knockout and overexpression on cell proliferation in A549 and PC-9. Scale bars, 200 μm. (D) Wound-healing assays (n = 3 per group) show the effects of ZNF266 knockout and overexpression on cell migration in A549 and PC-9. Scale bars, 200 μm. (E) Transwell assays (n = 3 per group) show the effects of ZNF266 knockout and overexpression on cell migration and invasion in A549 and PC-9. Scale bars, 200 μm. (F-H) Establishment (F), HE staining images (G), and morphology (H) of LUAD patient-derived organoids (PDOs). Scale bars, 100–20 μm. (I-J) Representative brightfield images (I) and luminescence measurement (n = 3 per group) (J) of PDOs in different groups. PDOs were transfected with lentiviruses to knockout or overexpress ZNF266. P values by one-way ANOVA. Partial η² = 0.92. KO vs. NC 95% CI of diff. = (0.22 to 0.73), OE vs. NC 95%CI of diff. = (-0.52 to -0.01). Scale bars, 100 μm. (L-M) Tumor images (L) and growth curves (M) of each group (n = 6). P values by two-way ANOVA. Partial η² = 0.51. KO vs. NC 95% CI of diff. = (20.84 to 34.95), OE vs. NC 95%CI of diff. = (-39.95 to -25.85)

ZNF266 represses CA9’s transcription by binding to its promotor

To determine potential downstream targets of ZNF266, RNA-seq was conducted in ZNF266-NC and ZNF266-OE cells respectively (Fig. 3A) and differentially expressed genes (DEGs) were displayed (Fig. 3B). The results revealed that CA9 stood out as the most significant gene among the five with notably altered expression levels in both A549 and PC-9 cells (Fig. 3C), suggesting that CA9 might be a target molecule of ZNF266. Additional verification of the effect of ZNF266 on CA9 expression was conducted using qRT-PCR and Western blot analysis. The results demonstrated that overexpression of ZNF266 led to a decrease in CA9 levels, whereas ZNF266 knockout resulted in the upregulation of CA9 (Fig. S2B and 3D).

We speculate that ZNF266 transcriptionally suppressed CA9 by binding to its promotor. Utilizing the genome-wide chromatin immunoprecipitation sequencing (ChIP-seq) data published by the ENCODE project, we found three very strong ZNF266-binding peaks in the CA9 promotor region (between − 1,000 and + 500 bp to the transcription start site [TSS]) (Fig. 3E-G). To confirm the binding of ZNF266 to the CA9 promoter in LUAD cell lines, ChIP-qPCR assays were conducted, revealing significant recruitment of ZNF266 to the CA9 promoter region (Fig. 3H). Consistently, DNA pull-down assays demonstrated that oligonucleotides spanning the wild-type CA9 promotor region, but not the mutant, exhibited significant affinity for ZNF266 (Fig. 3I). These results confirm that ZNF266 can directly bind to the CA9 promotor.

Subsequently, the CA9 promoter sequences were cloned into the pGL4-luciferase reporter vector to assess the transcriptional functionality of ZNF266 binding. As a negative control, we predicted and modified all the sites where ZNF266 most likely binds (threshold ≥ 0.8) using the consensus motifs on the JASPAR website. It was observed that the expression of ZNF266 significantly reduced the activity of the luciferase reporter containing the wild-type CA9 promoter sequence (Fig. 3J). In contrast, the introduction of mutations to the CA9 promoter region attenuated this change. These observations suggest that ZNF266 directly regulates CA9 transcription by binding to its promotor region.

In addition, a negative correlation was observed between ZNF266 and CA9 expression in resected tissue from LUAD patients (Fig. 3K and L) and in immunofluorescence analysis of CDX tumors (Fig. 3M), also indicating their negative regulation.

Fig. 3

ZNF266 represses CA9’s transcription by binding to its promotor. (A) Schematic diagram of RNA sequencing in A549 and PC-9 cells stably overexpressing ZNF266. (B-C) Volcano plot (B) and heatmap (C) exhibit differentially expressed genes between ZNF266-OE and NC groups in A549 and PC-9. CA9 stands out. (D) Western blot analysis confirms regulated CA9 expression after ZNF266 knockout and overexpression. (E) Characteristic sequences predicted by JASPAR for transcriptional binding sites of ZNF266. (F-G) Genome-wide data and ChIP-seq results from the ENCODE project show the solid ZNF266-binding peak in the promoter region to the TSS of CA9. (H) ChIP-qPCR assays (n = 3 per group) depicting the enrichments of ZNF266 binding in A549 and PC-9. (I) Western blot analysis of exogenous ZNF266 and Lamin B1 (control) protein pulled down by biotin-labeled oligonucleotides containing the wild-type or mutant CA9 promoter. (J) Dual luciferase activity assays (n = 3 per group) to analyze the fluorescence intensity of ZNF266-OE and NC with or without mutations in the CA9 promoter region. (K-L) Pearson correlation analysis between ZNF266 and CA9 expression in LUAD patient samples (K) and representative IHC images (L). Scale bars, 200–100 μm. (M) Representative multiplex IHC (mIHC) images of CDX tumors. Scale bars, 20 μm

ZNF266 restrains the mTOR signaling pathway by decreasing intracellular pH

Data from the TCGA and GTEx databases showed that CA9 expression was elevated in LUAD (Fig. 4A), which indicated a poor clinical prognosis for patients (Fig. 4B). CA9 catalyzes the reversible hydration of CO₂ at the exofacial site of the plasma membrane, making CA9 centrally in the regulation of intracellular pH (pHi) and extracellular pH (pHe). CA9 maintains a steep outward-directed CO₂ gradient across the plasma membrane, facilitating CO₂ excretion, which subsequently leads to extracellular acidification and a more alkaline pHi [17,18,19]. Morphology and flow cytometry with BCECF-AM probes demonstrated a decreased pHi in the ZNF266-OE LUAD cells and an increased pHi in the ZNF266-KO LUAD cells. This finding is consistent with ZNF266’s CA9-suppressing role. (Figure 4C and D).

It is noteworthy that Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses indicated that the DEGs between ZNF266-NC and ZNF266-OE were predominantly enriched in the mTOR signaling pathway in both A549 and PC-9 (Fig. 4E and F), suggesting a regulatory role for ZNF266 in the mTOR signaling pathway in LUAD. Therefore, we detected the phosphorylation level of mTOR and S6K, the classic downstream of mTORC1, and found that ZNF266 overexpression reduced the phosphorylation of both mTOR and S6K and vice versa, supportively (Fig. 4G).

Furthermore, prestigious research has reported the impact of pHi on the mTOR signaling pathway, revealing that a reduction in pHi inhibits the activation of mTORC1, a key complex within the mTOR signaling pathway [20, 21], while dysregulated mTORC1 contributes causally to cancer development and progression [22,23,24]. We adjusted the pH of the LUAD cell culture medium for 48 h by adding HCl or NaOH (pH = 6.5, 7.0, and 7.5) and noticed the changed pHi (Fig. 4H), and the subsequent Western blot analysis confirmed that the mTOR signaling pathway was regulated by pHi in LUAD cells (Fig. 4I).

Based on these findings, we hypothesized that ZNF266 might regulate the activation of the mTOR pathway through CA9-mediated pHi alteration, thereby influencing the malignant phenotypes of cancer.

Fig. 4

ZNF266 restrains the mTOR signaling pathway via intracellular pH. (A) CA9 expression levels in normal and tumor samples from the TCGA-LUAD and GTEx databases (normal group = 347, tumor group = 513). (B) Kaplan–Meier analysis of overall survival comparing the high and low expression of CA9 in LUAD patients in the Kaplan–Meier plotter database. (C-D) Morphology (C) and flow cytometry (D) with BCECF-AM probes exhibit the effects of ZNF266 knockout or overexpression on pHi in A549 and PC-9. Scale bars, 100 μm. (E-F) KEGG enrichment analysis of differentially expressed genes in our RNA-seq data. (G) Western blot analysis of the effects of ZNF266 knockout and overexpression on phosphorylation of the mTOR signaling pathway in A549 and PC-9 cells. (H) Measurement of pHi (n = 4 per group) in A549 and PC-9 cells incubated with different pH for 48 h. (I) Western blot analysis confirms the effects of different pHi on phosphorylation of the mTOR signaling pathway in A549 and PC-9 cells incubated in medium with indicating pH for 48 h

CA9 expression abrogates the effect of ZNF266 on malignant phenotypes of LUAD

To validate our hypothesis, CA9 expression was restored in ZNF266-OE and NC LUAD cells (Fig. S2C, S2D, and 5C). We observed that CA9 restoration eliminated the difference between ZNF266-OE and NC in terms of the pHi and the phosphorylation level of the mTOR pathway (Fig. 5A and B). Subsequently, CCK8 assays and high-content imaging analysis showed that cells with restored CA9 expression exhibited similar proliferation capacities, regardless of the expression level of ZNF266 (Fig. 5C and D). Wound healing and transwell assays indicated that the overexpression of CA9 significantly mitigated the inhibitory effects of ZNF266 on migration and invasion in LUAD cells (Fig. 5E and F). Similar findings were observed in PDOs, where both morphology and the ATPase activity were consistent with these results (Fig. 6A and B). Additionally, immunohistochemistry analysis showed that the differences in Ki-67 levels were eliminated by CA9 restoration (Fig. 6C).

Subcutaneous tumor formation assays were performed to verify the results obtained above in vivo. In alignment with the in vitro results, tumor formation was significantly boosted in nude mice injected with CA9-overexpressing cells compared to the ZNF266-OE and NC groups. More importantly, irrespective of ZNF266 expression levels, tumorigenesis following subcutaneous injection of CA9-overexpressing cells in nude mice yielded similar results (Fig. 6D and E).

Collectively, the results above suggested that ZNF266 regulates the activation of the mTOR pathway via CA9-mediated pHi alteration, ultimately affecting the malignant phenotypes of LUAD.

Therapeutic potential of AAV-ZNF266 in LUAD

To assess the therapeutic potential of ZNF266 in LUAD, we investigated its efficacy in preclinical PDX models. Following confirmation of AAV-ZNF266 biosafety (Fig. S3A-3Q), two PDX models (PDX-1 and PDX-2) were generated from treatment naïve LUAD specimens. Third-generation xenografts received peritumoral AVV-ZNF266 delivery (Fig. 6F), which significantly suppressed tumor growth over an 18-day observation window (Fig. 6G and H), recapitulating the inhibitory effects observed in CDX models. Following euthanasia, tumors from NSG mice were dissected for gravimetric analysis and immunohistochemistry staining. ZNF266-overexpressing tumors exhibited marked downregulation of CA9 protein levels compared to controls, alongside diminished Ki-67 positivity (Fig. 6I), indicative of impaired proliferation capacity. These findings positioned ZNF266 restoration as a promising translational strategy for LUAD (Fig. 6J).

Fig. 5

CA9 expression abrogates the effect of ZNF266 on malignant phenotypes of LUAD cells. (A) Measurement of pHi (n = 6 per group) in A549 and PC-9 transfected with ZNF266-NC, ZNF266-OE, CA9-OE, and ZNF266-OE + CA9-OE lentiviruses. (B) Western blot analysis of ZNF266 expression level, CA9 expression level, and phosphorylation of the mTOR signaling pathway in A549 and PC-9 cells of different groups. (C-D) CCK8 assays (n = 6 per group) (C) and high-content imaging analysis (n = 3 per group) (D) show the effects of CA9 restoration on cell proliferation inhibited by ZNF266 in A549 and PC-9 cells. Scale bars, 200 μm. (E) Wound-healing assays (n = 3 per group) show the effects of CA9 restoration on cell migration inhibited by ZNF266 in A549 and PC-9 cells. Scale bars, 200 μm. (F) Transwell assays (n = 3 per group) show the effects of CA9 restoration on cell migration and invasion inhibited by ZNF266 in A549 and PC-9. Scale bars, 200 μm

Fig. 6

ZNF266 inhibits LUAD progression via regulating CA9 expression in PDOs and xenograft models. (A-C) Representative brightfield images (A), mIHC images (B), and luminescence measurement (n = 3 per group) (C) of PDOs in different groups. PDOs were transfected with lentiviruses to overexpress ZNF266 or CA9. Scale bars, 100 μm–20 μm. P values by one-way ANOVA. Partial η² = 0.97. ZNF266-OE vs. Vector 95% CI of diff. = (-0.51 to -0.01), CA9-OE vs. Vector 95%CI of diff. = (0.37 to 0.88), ZNF266-OE + CA9-OE vs. Vector 95% CI of diff. = (0.45 to 0.96), CA9-OE vs. ZNF266-OE + CA9-OE 95% CI of diff. = (-0.33 to 0.17). (D-E) Tumor images (D) and growth curves (E) of CDX models (n = 6 per group). P values by two-way ANOVA. Partial η² = 0.31. ZNF266-OE vs. Vector 95% CI of diff. = (-56.31 to -3.48), CA9-OE vs. Vector 95%CI of diff. = (53.73 to 113.53), ZNF266-OE + CA9-OE vs. Vector 95% CI of diff. = (66.80 to 126.60), CA9-OE vs. ZNF266-OE + CA9-OE 95% CI of diff. = (-42.97 to 16.83). (F) Schematic representation of establishment of two patient-derived xenograft (PDX) models and following adeno-associated virus (AAV) treatment. (G-H) Tumor images and volumes (n = 3 per group) of PDX-1 (G) and PDX-2 (H) injecting AAV-ZNF266 or AAV-Vector. (G) P values by Student’s t-test. Cohen’s d = 4.75. AAV-ZNF266 vs. ZNF266-Vector 95% CI of diff. = (54.77 to 154.60). (H) P values by Student’s t-test. Cohen’s d = 4.76. AAV-ZNF266 vs. ZNF266-Vector 95% CI of diff. = (55.89 to 157.67). (I) Representative mIHC images of PDX tumors. Scale bars, 20 μm. (J) Model for ZNF266 inhibiting LUAD progression by regulating CA9-mediated intracellular pH

Discussion

LUAD is characterized by prevalent and highly aggressive malignant phenotypes with a poor prognosis. It displays the fundamental traits of malignancies, including the ability to sustain chronic proliferation, bypass potent inhibitory mechanisms that regulate cellular growth, and facilitate invasion and metastasis. There is currently an urgent need to identify new therapeutic targets for LUAD and to elucidate the mechanisms underlying them [25].

TFs serve as pivotal regulators of gene expression across all cancer hallmarks, influencing processes such as cellular proliferation and metastatic dissemination. Historically regarded as undruggable due to structural disorganization and the lack of defined small-molecule binding sites, recent advancements in structural elucidation, fundamental biological understanding, and ligand design methodologies have facilitated the discovery of several promising therapeutic agents targeting TFs, such as RGT-61,159 and TTI-101 [7, 8]. Approximately 1,600 TFs are estimated to exist in the human genome, yet the functions of the majority of these factors and their therapeutic potential in cancer remain largely unexplored.

This study delved into the pivotal role of ZNF266 in LUAD. By leveraging a large cohort of nearly three thousand samples from public databases alongside our specimens, we present the initial comprehensive analysis of the clinical relevance of ZNF266 in LUAD. Our study provides the first evidence that ZNF266 mediates an inhibitory effect via a mechanism that encompasses the suppression of CA9 transcription and consequent inhibition of the mTOR pathway. Manipulation of ZNF266 expression levels in organoid models and overexpression of ZNF266 using adeno-associated virus in PDX models has led to significant findings.

In mammalian cells, zinc finger proteins (ZNFs) constitute the most extensive family of transcription factors. Dysregulation of ZNF expression is observed in various tumor types, implying potential functions as oncogenes or tumor suppressors [26, 27]. ZNF266 is a member of the zinc finger protein family and comprises many zinc-finger motifs arranged in tandem [28]. These zinc fingers serve as binding domains for proteins or nucleic acids and may participate in diverse functions, notably transcription regulation [5, 29, 30]. It has been reported by Jinhui Zhang et al. that ZNF266 may serve as a promising candidate gene with a causal association with the survival risk of cervical cancer [31]. Nonetheless, the mechanisms and regulatory functions of ZNF266 remain elusive and may vary across cancer types, necessitating further exploration.

Our research suggests that ZNF266-CA9-pHi-mTOR axis modulates the progression of LUAD. Solid tumors frequently exhibit acidic regions due to hypoxia. The acidic microenvironment poses a selective pressure on cancer cells, and CA9 plays a pivotal role in enabling cancer cell adaptation by converting high CO₂ levels into H⁺ and HCO₃⁻ [32]. These adaptations facilitate the maintenance of a slightly alkaline pHi in cancer cells [19, 33]. The mammalian target of rapamycin, discovered 30 years ago, is a highly conversed serine/threonine protein kinase that plays a central role in regulating cell growth and metabolism [34, 35]. Dysregulation of mTOR has been implicated in several major diseases [36], including many types of cancers [22, 37, 38]. Zandra E. Walton et al. have reported that acidic intracellular conditions promote the redistribution of lysosomes to the cellular periphery, thereby separating mTORC1 from its perinuclear activator RHEB and disrupting the circadian clock [20, 39]. This pHi-mTOR axis has not been previously reported in the context of tumors, and we validated the existence of this axis in LUAD for the first time and extrapolated the conclusion to the malignant phenotype of tumors. Cellular signaling architectures function as highly integrated, pleiotropic systems transcending simplistic linear cascades. Although our dissection of ZNF266 interactome dynamics may not exhaustively map all tertiary effectors within this multidimensional network, we establish through mechanistic and functional validation a novel tumor-suppressive axis (ZNF266-CA9-pHi-mTOR) essential for maintaining LUAD oncogene addiction.

While elucidating this downstream circuitry, upstream modulators governing ZNF266’s oncogenic subversion in LUAD remain unresolved, which is a limitation tied to the systems biology challenge of revealing transcription factor regulation in vivo. This limitation presents opportunities for leveraging multi-omics approaches to delineate mechanisms that may drive ZNF266 dysregulation through chromatin topology remodeling or non-coding RNA-mediated control [40], which we hope future studies will address.

Modulation of ZNF266 expression in organoid models, along with increased ZNF266 expression using AAV vectors in PDX models derived from patients, has yielded notable results. During long-term in vitro culture, the PDOs maintained the genetic alterations present in the original tumors, allowing for the investigation of the mechanisms underlying tumor evolution [41]. AAV-based gene delivery vectors have been employed in numerous clinical trials for gene therapy and have demonstrated a robust safety profile as well as efficacious treatment of monogenic disorders [42]. While the off-target effects and immune responses linked to AAV-mediated gene therapy remain incompletely elucidated [43], accumulating preclinical and clinical evidence highlights its promising therapeutic potential as a targeted intervention in oncology [44,45,46]. PDX models, in which tumor tissues obtained from patients are engrafted into immunocompromised or humanized mice, demonstrate enhanced accuracy in recapitulating cancer characteristics, such as the spatial organization and intratumoral heterogeneity of the tumor [47]. Application of AAV to PDX models enables more precise simulation of the clinical outcomes of ZNF266 therapy. However, PDX models utilized NSG mice, which inherently lack T cells, B cells, and NK cells, thereby precluding direct evaluation of AAV-ZNF266’s effects on immune cell infiltration. To resolve this limitation and rigorously dissect the interplay between AAV-ZNF266 and adaptive immunity, future studies will assess its therapeutic efficacy and potential synergies with immunotherapeutic agents in syngeneic or immunocompetent murine models.

Conclusion

Our study identified ZNF266 as a prognostic biomarker and demonstrated that ZNF266 inhibits LUAD progression in a pH-dependent manner via modulating CA9 expression. These findings revealed that ZNF266 might be a prospective target for LUAD treatment.