CircCDYL promotes glycolysis to drive the progression of nasopharyngeal carcinoma

Graphical abstract

CircCDYL enhances LDHA mRNA stability in NPC by binding to its 3’-UTR, resulting in increased LDHA protein levels and enhanced glycolysis in NPC cells. This leads to lactate accumulation, which, under the catalysis of AARS1, induces lactylation of CFL1 at the K22 site, extending its protein half-life. The increased CFL1 levels promote cytoskeletal remodeling, decreasing cell stiffness and adhesion, thereby facilitating the proliferation and metastasis of NPC cells.

Highlights

• •CircCDYL facilitates NPC cells proliferation, invasion and migration.
• •CircCDYL promotes the proliferation and metastasis of NPC by enhancing glycolysis.
• •CircCDYL binds to the 3’-UTR of LDHA mRNA and stabilizes it.
• •The lactylation of CFL1 at lysine 22 reduces cell adhesion and stiffness.

Abstract

Introduction

CircRNAs play critical roles in the onset and progression of nasopharyngeal carcinoma (NPC), although their underlying mechanisms remain incompletely understood.

Objectives

In this study, we investigated the mechanism of circCDYL in promoting the proliferation, invasion and migration of NPC in vitro and in vivo.

Methods

We reanalyzed RNA sequencing data (GSE137543, PRJNA391554) and validated findings with RT-qPCR and in situ hybridization experiments. Functional assays were subsequently performed both in vitro and in vivo.

Results

We identified circCDYL as being highly expressed in NPC, with its expression levels positively correlated with clinical tumor staging. Functionally, circCDYL promotes tumor proliferation and metastasis both in vitro and in vivo. Mechanistically, circCDYL binds to the 3’-UTR of LDHA mRNA, stabilizing it and upregulating LDHA protein expression, thereby enhancing cellular glycolysis and increasing lactate production and accumulation. The elevated lactate, in turn, promotes lactylation of the actin-binding protein CFL1 at lysine 22. This modification reduces cell adhesion and stiffness, ultimately facilitating tumor cell proliferation, invasion, and migration.

Conclusion

These findings indicate that circCDYL and its downstream pathways may serve as potential diagnostic markers or therapeutic targets for NPC.

Introduction

Nasopharyngeal carcinoma (NPC) is a highly prevalent malignant tumor in the head and neck region, characterized by distinct racial and geographic distribution. In southern China, the age-standardized incidence rate for NPC ranges from 20 to 50 cases per 100, 000 individuals, a figure nearly 100 times higher than that observed in predominantly white populations [1,2]. Despite advancements in treatment, which have achieved a five-year survival rate of 80 %, approximately 20 % of patients still face recurrence or metastasis [3]. The processes of NPC proliferation and metastasis are complex, involving multifactorial and multistep mechanisms. Thus, further investigation into these underlying mechanisms is essential to identify potential targets for early diagnosis and therapeutic intervention.

Circular RNAs (circRNAs) are a class of non-coding RNAs generated through the back-splicing of pre-mRNA, which, lacking 5’ and 3’ ends, confer greater stability than linear RNAs [4]. Tens of thousands of circRNAs have been identified, many of which play critical roles in normal physiological processes and disease development [5,6]. Recent studies have shown that circRNAs—such as circRNF13, circARHGAP12, and circCAMSAP1—regulate cellular functions through mechanisms, including mRNA binding and stabilization [[7], [8], [9]], direct protein interaction [10,11], modulation of protein subcellular localization [12], scaffolding of protein complexes [13,14], and acting as microRNA sponges [15,16]. Additionally, a small subset of circRNAs can be translated into functional proteins [17]. In NPC, circRNAs have been implicated in regulating key processes such as proliferation, invasion, migration, apoptosis, angiogenesis, and immune evasion [18]. However, the mechanisms underlying the functions of many circRNAs remain poorly understood [6].

We previously identified several differentially expressed circRNAs in NPC through high-throughput RNA sequencing, revealing their significant impact on NPC progression [[9], [10], [11],19,20]. Upon revisiting the RNA sequencing data, we discovered that circCDYL (hsa_circ_0008285), derived from exon 2 of the chromodomain Y-like (CDYL) gene, was highly expressed in NPC. Notably, no detailed studies have explored its regulatory effects in NPC. To address this gap, this study aimed to elucidate the mechanisms of circCDYL and evaluate its role in NPC growth and metastasis.

Our findings demonstrated that circCDYL was overexpressed in NPC and significantly correlated with tumor clinical stages. In vitro and in vivo experiments revealed that circCDYL boosted glycolysis in NPC, thereby promoting its proliferation and metastasis. Mechanistically, circCDYL stabilizes lactate dehydrogenase A (LDHA) mRNA by binding to its 3’-UTR, thereby increasing LDHA protein levels, enhancing glycolysis, and promoting lactate accumulation. Elevated lactate facilitates the lactylation of cofilin 1 (CFL1), a process catalyzed by lactyltransferase alanyl-tRNA synthetase 1 (AARS1), which prolongs the half-life of CFL1 [21]. Stabilized CFL1 then remodels NPC cytoskeletons, reducing stiffness and adhesion, thereby promoting NPC proliferation and metastasis.

Collectively, our study identifies circCDYL as a critical regulator of glycolysis in NPC, driving tumor growth and metastasis, and underscores its potential as a biomarker for early diagnosis and a target for therapy.

Materials and methods

Clinical tissue samples

Clinical tissue samples for this study were collected from Xiangya Hospital of Central South University. All samples were pathologically confirmed by the hospital’s pathology department. The study protocols were approved by the Ethics Committee of Central South University, and informed consent was obtained from all patients. Detailed clinical sample information is provided in Tables S1–S3.

Cell lines, plasmids, siRNA, and transfection reagents

The NPC cell lines CNE2 and HNE2 were obtained from the Cancer Research Institute of Central South University and cultured in RPMI-1640 medium (Life Technologies, California, USA) supplemented with 10 % fetal bovine serum (Gibco, California, USA). The 2nd exon of circCDYL was amplified via PCR and inserted into the circular vector pcDNA3.1(+). siRNAs were purchased from GenePharma (GenePharma, Shanghai, China) (Table S4). For transfections, Neofect (Neofect Biotech, Beijing, China) and Hiperfect (Qiagen, Hilden, Germany) were used for plasmids and siRNA, respectively. The LDHA inhibitor GSK2837808A (MedChemExpress, New Jersey, USA) was used at a concentration of 10 μM for 24 h. 2-Deoxy-D-glucose (2-DG, MedChemExpress, New Jersey, USA), lactate (Coolaber, Beijing, China), and glucose (Coolaber, Beijing, China) were applied at concentrations of 1 mM, 10 mM, and 25 mM, respectively, for 24 h. Cycloheximide (CHX, AbMole, Texas, USA) was used at 150 μM for 0, 6, 12, 24, 36 h.

RNA extraction and RT-qPCR

Cells were lysed on ice using TRIzol reagent (Solarbio Life Sciences, Beijing, China). Reverse transcription was carried out with the HiScript cDNA Synthesis Kit (Vazyme, Nanjing, China), followed by RT-qPCR using 2 × SYBR Green qPCR Master Mix (Bimake, Texas, USA). The primer sequences used were provided in Table S4.

RNase R and actinomycin D assay

RNA was treated with RNase R (Thermo Fisher, California, USA) at a final concentration of 20 U/μL at 37 °C for 30 min, followed by inactivation at 70 °C for 10 min. NPC cells were treated with actinomycin D (Sigma, Missouri, USA) at a final concentration of 1 μg/mL for 0, 6, 12, and 24 h.

Glycolytic stress test

The extracellular acidification rate of NPC cells was assessed using the Agilent Seahorse XFp Glycolysis Stress Test Kit in conjunction with the Seahorse XF HS Mini Analyzer (Agilent Technologies, California, USA).

H&E staining, in situ hybridization (ISH), fluorescence in situ hybridization (FISH), and immunohistochemistry (IHC)

H&E staining, ISH, and FISH were procedured as previously described [9]. For IHC, the Elivision™ Plus Polymer HRP (Mouse/Rabbit) IHC Kit (Maxim, Fuzhou, China) was utilized.

Phalloidin staining

Cells on slides were fixed with pre-warmed 4 % paraformaldehyde at 37 °C for 30 min and permeabilized with 0.1 % Triton X-100. The slides were incubated with AbFluor™ 488-Phalloidin (Abbkine, Wuhan, China) at 37 °C in the dark for 1 h.

Wound healing and transwell invasion assays

For the wound healing assay, a 10 μL pipette tip was used to gently scratch the monolayer of NPC cells. The images of the scratch area were captured at 0, 12, and 24 h at the same location. For the transwell invasion assay, cells were seeded into the upper chamber of a transwell insert (Millipore, Massachusetts, USA) pre-coated with 10 % Matrigel (BD Biocoat, Shanghai, China). The invaded cells on the underside of the membrane were counted in three randomly selected microscopic fields.

MTT and colony formation assays

For the MTT assay, 1, 000 cells were seeded into each well of a 96-well plate. MTT (Beyotime, Shanghai, China) was added and incubated at 37 °C for 4 h. The resulting precipitate was dissolved in DMSO, and the absorbance was measured at 490 nm on days 0 to 5. For the colony formation assay, 500 cells were plated in each well of a 12-well plate. After two weeks, cells were fixed with 4 % paraformaldehyde, stained with 0.1 % crystal violet, and then scanned and counted.

Liquid chromatography coupled to tandem mass spectrometry

CNE2 cells transfected with circCDYL or empty vector for 48 h were used to extract total protein and perform digestion treatment. The protein identification and data processing protocols followed those described in previous reports [9]. The detailed data obtained were presented in Table S6.

Lactate concentration assay

The culture medium supernatant was collected 24 h after changing the medium, and the lactate level was measured using a Cobas 6000 Analyzer (Roche, Basel, Switzerland). The cell count was measured to calibrate the corresponding sample.

Circular RNA Immunoprecipitation assay

Biotin-labeled circCDYL probes and circCDYL overexpression plasmids were co-transfected into NPC cells. Then, the cells were co-incubated with Dynabeads M−280 Streptavidin magnetic beads (Invitrogen, California, USA) to isolate and extract the immunoprecipitated RNA. The extracted RNA was then subjected to reverse transcription and RT-qPCR. The sequences of the probes and primers used were provided in Table S4.

RNA pulldown and Immunoprecipitation

The RNA pulldown kit (Thermo Scientific, California, USA) was used for RNA pulldown assay. For immunoprecipitation, 50 μL of protein A/G magnetic beads (Bimake) were incubated with the antibody on a DNA mixer for 2 h. The cell lysate was added to the beads and incubated on a DNA mixer at 4 °C overnight. The proteins bound to the beads were then collected and analyzed by western blotting.

Dual-luciferase reporter assay

In NPC cells, circCDYL overexpression plasmids or circCDYL siRNAs, along with pRL-TK plasmids and pMIR-REPORT plasmids containing specific fragments of the LDHA mRNA 3’-UTR, were co-transfected. After 48 h, firefly and renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega, Wisconsin, USA) on a microplate reader (Molecular Devices, California, USA).

Atomic force microscopy test

Cells in the culture dishes were fixed with 2.5 % glutaraldehyde for 45 s, followed by fixation with 4 % paraformaldehyde for 30 min. After washing with PBS, the cells were detected using PBS as the aqueous phase. The stiffness and adhesion of NPC cells were measured using Atomic Force Microscopy (AFM) (JPK NanoWizard 4, JPK Instruments, Berlin, Germany) with the HYDRA6V-100NG probe (AppNano, California, USA). The probe parameters were set as described in previous reports [20].

Animal experiments

The animal experiments were approved by the Institutional Animal Care and Use Committee of Central South University (NO. 2019sydw0090) and conducted in strict accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Female nude mice were randomly divided into three groups (n = 5 per group). CNE2 cells transfected with empty vector (EV), negative control (NC), circCDYL, or sicircCDYL-1 were injected either subcutaneously or via tail vein to establish subcutaneous tumor or lung metastasis xenograft models, respectively. In the subcutaneous tumor model, tumor volume was measured every 3 days, and after 27 days mice were sacrificed to collect tumors for volume and weight assessment. In the lung metastasis model, mice were monitored every 3 days, and after 8 weeks were sacrificed to collect lung tissues for nodule counting. Tumors were subsequently examined by H&E staining and IHC.

Statistical analysis

Data were presented as mean ± standard deviation (SD). GraphPad Prism 8.0 was used for statistical analysis and graph creation. Comparisons between two independent groups were performed using an unpaired t-test for independent samples. A p-value of < 0.05 was considered statistically significant.

Results

CircCDYL is highly expressed in NPC and drives its progression

To identify circRNAs with potential roles in NPC progression, we reanalyzed high-throughput RNA sequencing data from NPC cells (GSE137543, PRJNA391554) and identified circCDYL as highly expressed, with no previous studies reporting its role in NPC [10]. Sanger sequencing confirmed that circCDYL is generated by the back-splicing of exon 2 of the CDYL gene, resulting in a 667nt circular RNA (Fig. 1A). To further validate the differential expression of circCDYL in NPC, we analyzed RNA sequencing data from the GEO database (GSE68799). This analysis revealed that circCDYL was significantly overexpressed in 41 NPC tissues compared with 4 non-tumor nasopharyngeal epithelial (NPE) tissues (Fig. S1A). Subsequently, we collected 30 NPC tissues and 12 chronic inflammatory NPE tissues and performed RT-qPCR to measure circCDYL expression. The results showed that circCDYL was markedly overexpressed in NPC tissues, with even higher expression levels in middle and late clinical stages (Fig. 1B). ISH further demonstrated significantly elevated circCDYL expression in 82 NPC tissues compared to 29 adjacent NPE tissues (Fig. 1C). Additionally, circCDYL expression was strongly correlated with clinical staging and TNM staging of NPC (Fig. 1D). These findings suggest that circCDYL may play a key role in NPC progression.

Fig. 1.

CircCDYLishighlyexpressed in NPC anddrives NPCprogression.A. Sanger sequencing confirmed circCDYL was formed by the reverse splicing of CDYL mRNA exon B. RT-qPCR was conducted to measure the expression levels of circCDYL in 30 nasopharyngeal epithelial (NPC) and 12 non-tumor nasopharyngeal epithelial (NPE) tissues, with statistical analysis based on the clinical staging of NPC. β-actin was used as the internal control. NPE, n = 12; NPC, n = 30; I/II, n = 7; III, n = 11; IV, n = 12. C. ISH was used to detect circCDYL expression in 82 NPC tissues and 29 adjacent NPE tissues. Representative images are shown on the left. × 200, scale bar = 50 µm; ×400, scale bar = 20 µm. The right panel shows the statistical results of the scoring data. NPE, n = 29; NPC, n = 82. D. Statistical analysis was performed to examine the correlation between circCDYL expression and NPC clinical staging and TNM staging. NPE (N), n = 29; I/II, n = 15; III, n = 27; IV, n = 40; T1/2, n = 33; T3/4, n = 49; N0/1, n = 38; N2/3, n = 44; M0, n = 69; M1, n = 13. E. MTT assays were performed to assess the relative growth of CNE2 and HNE2 cells following circCDYL overexpression or knockdown on days 0 through 5. EV, empty vector; NC, siRNA negative control. n = 3. F. Transwell invasion assay was conducted to evaluate the invasive capacity of CNE2 and HNE2 after overexpression or knockdown of circCDYL. n = 3. Scale bar = 100 µm. G. Wound healing assay was conducted to measure the effect of circCDYL on cell migration at 0, 12, and 24 h. n = 7. Scale bar = 200 µm. Data were presented as mean ± standard deviation (SD). Differences between two independent groups were assessed using an unpaired t-test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

To confirm the circRNA characteristics of circCDYL, we treated NPC cells with RNase R and found that circCDYL was significantly more resistant to RNase R treatment compared to CDYL mRNA (Fig. S1B), demonstrating that circCDYL exists in a stable circular configuration. Further validation using actinomycin D, which inhibits RNA transcription, revealed that the half-life of circCDYL was considerably longer than that of CDYL mRNA, highlighting the greater stability of circular circCDYL compared to its linear counterpart (Fig. S1C). Additionally, RNA FISH and nuclear-cytoplasmic fractionation assays demonstrated that circCDYL was distributed in both the cytoplasm and the nucleus (Fig. S1D and E), suggesting that it may have functional roles in multiple cellular compartments.

To investigate the role of circCDYL in NPC progression, we constructed an overexpression vector for circCDYL and designed two siRNAs (sicircCDYL-1 and sicircCDYL-2) targeting its back-splice junction (Fig. S2A). MTT and colony formation assays revealed that circCDYL overexpression significantly enhanced cell proliferation, while circCDYL knockdown inhibited it (Fig. 1E and Fig. S2B). Transwell and wound healing assays further demonstrated that overexpressed circCDYL promoted NPC cell invasion and migration, whereas its knockdown suppressed them (Fig. 1 F and G and Fig. S2C). These results collectively indicate that circCDYL plays a pivotal role in promoting the proliferation, invasion, and migration of NPC cells in vitro.

To determine whether circCDYL exerts similar functions in vivo, we established subcutaneous and lung metastasis tumor models by injecting CNE2 cells with either circCDYL overexpression or knockdown into the armpit or tail veins of nude mice, respectively. In the subcutaneous tumor model, mice injected with CNE2 cells overexpressing circCDYL developed significantly larger and heavier tumors compared to the control group, whereas the knocking down group showed the opposite effect (Fig. 2A–E). In the lung metastasis model, the number of metastatic lung nodules was quantified. The circCDYL overexpression group exhibited a significantly higher number of lung nodules compared to the control group, while the knockdown group demonstrated a notable reduction (Fig. 2F). H&E staining results of lung metastatic nodules showed that circCDYL knockdown reduced the tumor tissue area, whereas circCDYL overexpression increased it (Fig. 2G). ISH experiments confirmed the effective overexpression or knockdown of circCDYL (Fig. 2H). These findings demonstrated that circCDYL promotes the proliferation and metastasis of NPC in vivo.

Fig. 2.

CircCDYLpromotes theproliferation andmetastasis of NPCcellsinvivo.A. Subcutaneous tumor tissues from nude mice injected with CNE2 cells overexpressing or knocking down circCDYL. OE, overexpression vector. n = 5. B-D. The growth curve (B) of subcutaneous tumors from day 0 to 27, tumor volume (C), and tumor weight (D) after 27 days demonstrated the impact of circCDYL on tumor proliferation in nude mice. n = 5. E. IHC experiments were conducted to detect the expression of the proliferation marker Ki-67 in subcutaneous xenograft tumors in nude mice. × 200, scale bar = 50 µm; ×400, scale bar = 20 µm. F. Representative images of lung tissues from nude mice injected with CNE2 cells overexpressing or knocking down circCDYL (left). The number of nodules on the surface of the lung tissue (right). n = 5. G. Representative images of H&E staining of metastatic lung nodules. × 40, scale bar = 200 µm; ×100, scale bar = 100 µm; ×200, scale bar = 50 µm. H. Representative images of H&E and ISH staining of subcutaneous tumor and metastatic lung nodules. × 400, scale bar = 20 µm. Data were presented as mean ± SD. Differences between two independent groups were assessed using an unpaired t-test. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

CircCDYL promotes the proliferation and metastasis of NPC by enhancing glycolysis

Notably, we observed that the medium in the circCDYL overexpression group turned yellow more rapidly than in the control group (Fig. 3A), suggesting higher lactate production, which was confirmed by the lactate production assay (Fig. 3B). To further investigate whether circCDYL enhances lactate production through increased glycolysis, we measured the extracellular acidification rate (ECAR) in NPC cells as an indicator of glycolytic activity. The results showed that circCDYL overexpression markedly elevated both baseline glycolytic capacity and maximum glycolytic capacity, whereas circCDYL knockdown led to reductions in both metrics (Fig. 3C–E). Additionally, treatment with the glycolysis inhibitor 2-DG attenuated the promotional effects of circCDYL on NPC cell proliferation, invasion, and migration (Fig. 3F and G and Fig. S3). These findings demonstrate that circCDYL boosts glycolysis, thereby promoting the proliferation and metastasis of NPC cells.

Fig. 3.

CircCDYLpromotes theproliferation andmetastasis of NPC byenhancingglycolysis. A. Representative images showing the color of the culture medium in CNE2 and HNE2 cells after overexpression or knockdown of circCDYL for 24 h. B. Lactate production assay measured the relative extracellular lactate levels in CNE2 and HNE2 following circCDYL overexpression or knockdown, normalized to cell count. n = 3. C-E. ECAR was assessed to evaluate the effect of circCDYL overexpression or knockdown on glycolysis in CNE2 and HNE2 cells (C). Statistical analysis was performed on basal glycolytic capacity (D) and maximum glycolytic capacity (E). Baseline ECAR represents basal glycolytic capacity, and stressed ECAR represents maximum glycolytic capacity. n = 3. F-G. MTT (F) and transwell invasion assays (G) assessed the impact of 2-DG (1 mM) treatment on the proliferation and invasion of CNE2 and HNE2 cells after overexpression or knockdown of circCDYL. n = 3. Transwell invasion (G) scale bar = 100 µm. Data were presented as mean ± SD. Differences between two independent groups were assessed using an unpaired t-test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

CircCDYL stabilizes LDHA mRNA to Drive NPC progression

To investigate how circCDYL promotes the proliferation, invasion, and migration of NPC cells via glycolysis, we performed mass spectrometry analysis on CNE2 cells transfected with circCDYL overexpression or empty vectors. A total of 2024 proteins were identified, and after applying data filtering criteria (fold change > 1.5 or < 0.67), 435 proteins with significant differences were identified (Table S6). KEGG pathway enrichment analysis of these 435 proteins revealed that circCDYL likely interacts with metabolic pathways (Fig. S4A, Table S7). To further explore circCDYL’s role, we analyzed the expression of glycolysis-associated genes in the clinical NPC datasets (GSE12452 and GSE53819), focusing on LDHA, GAPDH, ADH5, ALDH1B1, ALDOB, ENO2, ENO3, HK2, HK1, and ALDH9A1. Among these, LDHA, GAPDH, and ADH5 were significantly overexpressed (Fig. S4B).

RT-qPCR analysis revealed that circCDYL upregulated LDHA mRNA (Fig. 4A) and exerted a modest inhibitory effect on ADH5 mRNA, while showing no significant impact on GAPDH mRNA (Fig. S4C). The regulation of LDHA mRNA by circCDYL was particularly pronounced, suggesting a potential direct interaction. Consistently, RT-qPCR analysis of 27 NPC tissues and 9 chronic inflammatory NPE tissues confirmed that LDHA expression was elevated in NPC tissues and positively correlated with clinical stage (Fig. 4B). Furthermore, linear CDYL mRNA knockdown did not alter the expression of LDHA mRNA or circCDYL (Fig. S4D), indicating that circCDYL functions independently of linear CDYL. Western blotting demonstrated that circCDYL also increased the protein levels of LDHA expression (Fig. 4C). Together, these findings indicate that circCDYL enhances glycolysis in NPC cells primarily by upregulating LDHA.

Fig. 4.

CircCDYLdrives NPCprogression bystabilizing LDHA mRNA.A. RT-qPCR experiments assessed the regulatory effect of circCDYL on mRNA. n = 3. B. RT-qPCR experiments measured the expression of LDHA mRNA in 27 NPC tissues and 9 chronic inflammatory NPE tissues. β-actin was used as the internal control. C. The expression of LDHA protein was evaluated following overexpression or knockdown of circCDYL. D. CircRIP experiments detected the binding of circCDYL to 3’-UTR and CDS of LDHA mRNA. n = 3. E. The relative luciferase activity was measured after overexpression or knockdown of circCDYL. WT, wild-type; MT, mutant type. n = 3. F-G. After treating NPC cells with actinomycin D for 0, 1, and 2 h, RT-qPCR experiments measured the half-life of LDHA mRNA following overexpression or knockdown of circCDYL. 18S was used as the internal control. n = 3. Data were presented as mean ± SD. Differences between two independent groups were assessed using an unpaired t-test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

Given that circRNAs often function by sponging microRNAs, interacting with proteins, or binding directly to mRNAs, we explored these potential mechanisms. Using prediction tools (TargetScanHuman, miRTarBase, and circInteractome), we identified candidate microRNAs predicted to bind both circCDYL and LDHA mRNA. However, no microRNAs were identified as intermediaries, suggesting that circCDYL did not regulate LDHA mRNA through the competing endogenous RNA (ceRNA) mechanism. RNA pulldown assays confirmed that circCDYL did not directly bind to the LDHA protein (Fig. S4E).

Bioinformatics analysis using RNA22 predicted that the 162-188nt region (5’-GCAGAAGGAGAGCACATTGACCAG-3’) of circCDYL might interact with the 3’-UTR of LDHA mRNA. CircRIP experiments demonstrated that circCDYL significantly enriched the 3’-UTR of LDHA mRNA compared to its coding sequence (CDS) (Fig. 4D and Fig. S4F). To validate these findings, we constructed a wild-type (WT) plasmid containing the LDHA mRNA 3’-UTR binding site and a mutant type (MT) with deletions in the predicted binding region. Dual-luciferase reporter assays showed that luciferase activity was significantly increased in the circCDYL overexpression group and decreased in the knockdown group, and these effects were abolished in the MT group (Fig. 4E). We constructed a circCDYL mutant that lacks the 162-188nt segment. RT-qPCR experiments showed that the circCDYL mutant no longer upregulates LDHA mRNA expression. Instead, it even shows a certain downward trend (Fig. S4G). Actinomycin D treatment revealed that circCDYL overexpression enhanced LDHA mRNA stability, whereas knockdown of circCDYL reduced it (Fig. 4F and G). These results suggest that circCDYL binds to the 3’-UTR of LDHA mRNA, enhancing its stability and upregulating LDHA expression.

To determine whether LDHA is a critical pathway through which circCDYL promotes NPC proliferation, invasion, and migration, we constructed an LDHA overexpression vector and designed siRNA targeting LDHA (Fig. S5A). In NPC cells with circCDYL knockdown, subsequent overexpression of LDHA rescued the inhibitory effects of circCDYL knockdown on proliferation, invasion, and migration (Fig. 5A–D and Fig. S5B). Similarly, in NPC cells overexpressing circCDYL, knockdown of LDHA reversed the promotional effects of circCDYL on these cellular processes (Fig. S5C–F). Treatment with the LDHA inhibitor (LDHAi) GSK2837808A also mimicked the effects of LDHA knockdown, further supporting the dependence of circCDYL’s function on LDHA activity (Fig. S5G–J). These findings confirm that circCDYL promotes NPC proliferation, invasion, and migration by binding to the LDHA mRNA 3’-UTR, enhancing its stability, and driving cellular glycolysis through LDHA.

Fig. 5.

CircCDYLpromotes theproliferation,invasion, andmigration of NPC via LDHA. A-D. MTT (A), colony formation (B), transwell invasion (C), and wound healing assays (D) were conducted to assess how circCDYL promoted the proliferation, invasion, and migration of NPC cells via LDHA. n = 3. Transwell invasion (C) scale bar = 100 µm. Wound healing assays (D) scale bar = 200 µm. Data were presented as mean ± SD. Differences between two independent groups were assessed using an unpaired t-test. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

CircCDYL promotes CFL1 lactylation

Post-translational modifications (PTMs) play a crucial role in disease progression by modulating protein expression and function. Lactylation, an important PTM, is closely associated with lactate accumulation. To investigate whether circCDYL induces protein lactylation, we overexpressed or knocked down circCDYL. The results showed that circCDYL regulated total protein lactylation (pan-Kla) levels in NPC cells (Fig. 6A). Nuclear-cytoplasmic fractionation experiments revealed that circCDYL predominantly modulated lactylation of non-histone proteins in NPC cells (Fig. 6B).

Fig. 6.

CircCDYLenhancesglycolysis andpromotes thelactylation of CFL1.A. The effect of overexpression or knockdown of circCDYL on protein lactylation in CNE2 and HNE2 cells. pan-Kla, total protein lactylation. B. Nuclear-cytoplasmic fractionation experiments assessed the localization of lactylation in CNE2 and HNE2 cells. β-tubulin served as a cytoplasmic reference, and histone 3 (H3) used as a nuclear reference. C-D. Immunoprecipitation experiments detected CFL1 lactylation (C), and the binding of CFL1 to AARS1 (D) following overexpression of flag-CFL1 and treatment with lactate (15 mM) for 24 h in CNE2 and HNE2 cells. E. Immunoprecipitation experiments detected CFL1 lactylation at K22R and K45R sites after knocking down endogenous CFL1, overexpressing mutant type plasmids, and treating with lactate for 24 h. F. CFL1 protein levels were measured after lactate (15 mM) or glucose (25 mM) treatment for 24 h in CNE2 and HNE2 cells. G. CFL1 protein levels were measured after treatment with 0, 5, 10, and 15 mM lactate for 24 h in CNE2 and HNE2 cells. H. The effect of circCDYL overexpression or knockdown on CFL1 protein levels in CNE2 and HNE2 cells. Data were presented as mean ± SD. Differences between two independent groups were assessed using an unpaired t-test.

To identify proteins undergoing lactylation regulated by circCDYL, we analyzed the lactylome data from Gao et al. and integrated it with the circCDYL proteome dataset [22]. This analysis revealed 252 differentially expressed molecules encompassing 542 lactylation sites (Fig. S6A). GO enrichment analysis highlighted the regulation of cytoskeleton-related proteins (Fig. S6B, Table S8). Among these, CFL1, CFL2, and destrin (DSTN) were enriched, with potential lactylation sites identified at CFL1 K22, CFL1 K45, CFL2 K19, and DSTN K22. These proteins are closely linked to tumor cell proliferation, invasion and migration [23,24].

Recent studies suggest that AARS1 functions as a lactate transferase, facilitating the transfer of lactyl groups to proteins it binds [25]. Using the HDOCK protein docking prediction tool, we predicted AARS1 binding interactions with CFL1, CFL2, and DSTN. The results showed high confidence for AARS1 binding to CFL1 and DSTN, with confidence scores around 0.7 and docking scores around −200, indicating a strong likelihood of interaction (Fig. S6C). To validate these findings, we constructed Flag-tagged overexpression vectors for CFL1 and DSTN. Immunoprecipitation assays demonstrated that under lactate treatment, CFL1 lactylation levels increased (Fig. 6C), and its binding to AARS1 was enhanced as well (Fig. 6D). However, DSTN did not exhibit increased lactylation or changes in AARS1 interaction despite lactate treatment (Fig. S6D). These results indicate that CFL1, but not DSTN, undergoes lactylation.

Given the critical role of AARS1 as a lactyltransferase, we further examined whether circCDYL or lactate treatment influences its expression. Both treatments significantly increased AARS1 expression (Fig. S6E and S6F), suggesting that AARS1 not only functions as a lactate sensor but is also upregulated in response to lactate.

Furthermore, we knocked down the endogenous expression of wild-type CFL1 (Fig. S6G) and then overexpressed two CFL1 mutants, K22R and K45R, respectively (Fig. S6H). Immunoprecipitation assays revealed that lactylation was abolished in the K22R mutant, whereas it remained inducible by lactate in the K45R mutant, identifying K22 as the key lactylation site of CFL1 (Fig. 6E). Considering that our previous lactylation analyses were derived from liver cancer, we used a lactylation prediction tool covering 14 species (HybridKla) to investigate whether lactylation at the CFL1 K22 site is universally present [26]. The analysis indicated that CFL1 has high lactylation potential across multiple species, with K22 being highly conserved and predicted as the site with the highest lactylation probability (Fig. S6I and J).

Lactylation has been shown to reshape protein function, enhance stability, promote nuclear localization, and influence gene expression [27,28]. In our study, treatment with lactate or high glucose upregulated CFL1 expression (Fig. 6F), with a dose-dependent increase in CFL1 levels observed as lactate concentration rose within a certain range (Fig. 6G). Cycloheximide (CHX) assays further revealed that lactate treatment extended the half-life of CFL1 (Fig. S6K), indicating increased protein stability under lactylation-inducing conditions. Western blotting analysis demonstrated that overexpression of circCDYL elevated CFL1 protein levels, while circCDYL knockdown reduced them (Fig. 6H). However, circCDYL did not alter CFL1 mRNA expression (Fig. S6L), suggesting that its effect on CFL1 was post-transcriptional. All together, these findings indicate that circCDYL regulates CFL1 expression by promoting its lactylation, thereby increasing its protein level.

CircCDYL promotes the proliferation, invasion, and migration of NPC by enhancing CFL1 lactylation

To further validate the role of CFL1 lactylation in promoting NPC proliferation and metastasis, atomic force microscopy (AFM) was employed to assess the stiffness and adhesion force of NPC cells. CircCDYL overexpression reduced cell stiffness and adhesion force, enhancing metastatic potential, whereas glycolysis inhibition with 2-DG increased both parameters (Fig. 7A). Exogenous lactate or CFL1 overexpression reduced the stiffness and adhesion force of NPC cells (Fig. 7B). We then observed that lactate restored the increased stiffness and adhesion caused by CFL1 knockdown (Fig. 7C). Notably, mutation of the CFL1 K22 site abolished this rescue effect, whereas mutation of the K45 site had no impact, indicating that K22 is essential for lactate-mediated restoration (Fig. 7C). Phalloidin staining experiments showed that CFL1 overexpression or lactate treatment caused F-actin to predominantly bundle and localize at the cell periphery (Fig. S7A). We found that lactate treatment could rescue the effects of CFL1 knockdown. This rescue was abolished when the CFL1 K22 site was mutated, whereas mutation of K45 did not impair lactate’s restorative effect (Fig. S7A).

Fig. 7.

CircCDYLdrives theproliferation,invasion, andmigration of NPC byenhancing CFL1lactylation.A-C. AFM was used to measure the stiffness and adhesion force of NPC cells. Lac, lactate. n = 3. Scale bar = 20 µm. D. IHC analysis of clinical tissue samples confirmed the expression levels of LDHA, CFL1, and pan-Kla. NPE, n = 29; NPC, n = 82. E. Pearson correlation analysis was conducted between circCDYL IHC scores and LDHA, pan-Kla, and CFL1 IHC scores, respectively. n = 82. F. Schematic diagram illustrating how circCDYL promotes the proliferation and metastasis of NPC. The image was created using BioRender.com. Data were presented as mean ± SD. Differences between two independent groups were assessed using an unpaired t-test. **, p < 0.01; ****, p < 0.0001.

IHC staining of clinical tissues demonstrated elevated levels of LDHA, CFL1, and pan-Kla in NPC compared to NPE tissues (Fig. 7D). A positive correlation trend was observed between these markers and both clinical stage and TNM classification (Fig. S7B–D). For analysis, clinical stages of NPC were assigned numerical values (Stage I = 1, Stage II = 2, Stage III = 3, Stage IV = 4). Correlation analysis indicated that the expression levels of circCDYL, LDHA, pan-Kla, and CFL1 are positively associated, to a certain extent, with NPC clinical stage (Fig. S7E). Additionally, circCDYL expression positively correlated with LDHA, CFL1, and pan-Kla levels in clinical samples (Fig. 7E). Survival analysis using TCGA Head and Neck Squamous Cell Carcinoma (HNSCC) data showed that patients with higher LDHA and CFL1 expression had poorer prognosis (Fig. S7F). These findings suggest that, in clinical samples, circCDYL is positively associated with CFL1, LDHA, and pan-Kla, and that these markers are linked to tumor malignancy and progression.

Discussion

In this study, we found that circCDYL is highly expressed in NPC tissues, and its expression is significantly correlated with both TNM staging and clinical staging. Further investigations revealed that circCDYL enhanced glycolysis by directly binding to and stabilizing LDHA mRNA, leading to excessive lactate production. These findings highlight the pivotal role of circCDYL in driving the malignant progression of NPC and demonstrate its potential as a novel target for early diagnosis and therapeutic intervention.

CircCDYL exhibits diverse roles in cancer progression that vary by tumor type. It promotes proliferation in liver and breast cancers [[29], [30], [31]], enhances metastasis in lung cancer [32], but suppresses both proliferation and metastasis in colorectal cancer [33]. The underlying also differ: in breast cancer it functions as a microRNA sponge [31], in liver cancer it induces autophagy and promotes oncogene expression [30], while in colorectal cancer it inhibits the AKT pathway [33]. In lung cancer, circCDYL functions as a molecular scaffold to facilitate transcription factor binding and enhance target gene transcription [32]. Additionally, exosome-derived circCDYL drives liver cancer proliferation, with m⁶A modifications promoting its synthesis and exosome formation, forming a positive feedback loop [29]. Our study revealed that circCDYL directly binds to LDHA mRNA 3’-UTR, thereby enhancing glycolysis in NPC cells. This leads to CFL1 lactylation, which promotes tumor proliferation and metastasis. As the first report of circCDYL’s role in NPC, our findings expand its functional repertoire and provide novel insights into its contribution to cancer biology.

During our study, we noticed that the culture medium acidified more rapidly in NPC cells overexpressing circCDYL, suggesting elevated production of acidic metabolic byproducts, which was later confirmed through further experiments. Lactate, a key product of glycolysis, not only acts as an important intracellular signaling molecule but also serves as a precursor for macromolecule synthesis [[34], [35], [36]]. In tumor cells, lactate drives the development of an acidic tumor microenvironment (TME), which is critical for sustaining the malignant phenotype [37]. For instance, in melanoma, lactate promotes the conversion of normal dendritic cells into mature regulatory dendritic cells, thereby suppressing anti-tumor immunity [38]. Lactate modulates RNA splicing-related enzyme levels in Treg cells within the TME, enhancing CTLA-4 expression [39]. In glioma, lactate activates GPR on tumor-associated macrophages, promoting tumor progression [40]. In our study, we found that circCDYL regulated lactate concentration in NPC cells, thereby facilitating the malignant progression of NPC.

Lactate also plays a significant role in tumorigenesis and progression by inducing protein lactylation. Zhao et al. discovered that histone lactylation occurs at lysine residues, with lactylation levels dynamically regulated by lactate concentrations. This process influences the expression of inflammation-related genes in macrophages [41]. In cancer cells, lactylation exerts significant regulatory effects. Gao et al. team applied a label-free quantification approach to analyze global lactylome data from liver cancer and adjacent normal tissues, revealing extensive protein lactylation in cancer cells beyond histones [22].

Protein lactylation modulates protein function and expression, thereby influencing tumor progression [27,28,[42], [43], [44]]. For example, lactylation of AK2 reduces its enzyme activity, promoting liver cancer cell proliferation [22]; MRE11 lactylation enhances its function, contributing to tumor chemoresistance [28]; lactylation of DCBLD1 extends its half-life, promoting the proliferation and metastasis of cervical cancer [27]. In this study, we identified lactylation of CFL1 at the K22 site for the first time.

CFL1 is a key regulatory molecule of the actin cytoskeleton, and its levels and stability are crucial for the formation of invasive pseudopodia, which regulate essential processes of cell movement [45]. CFL1 also plays a role in spindle positioning during mitosis and cytoplasmic division in mammals, thereby influencing cell proliferation [46,47]. In liver cancer, bladder cancer, and thyroid cancer, high expression of CFL1 promotes tumor proliferation and metastasis [[48], [49], [50]]. In endometrial cancer, CFL1 knockdown inhibits cell proliferation and invasion [51]. CFL1 can undergo various PTMs, such as phosphorylation at the Ser3 site and SUMOylation on the N-terminal α-amino group [52,53]. Our findings demonstrate that lactylation stabilizes CFL1 by extending its half-life and elevating its protein level, which in turn drives NPC proliferation and metastasis.

By reshaping cellular physical properties, cytoskeletal remodeling plays a critical role in controlling cell proliferation, invasion, and migration. AFM is a powerful tool used to analyze surface morphology and adhesion forces at the nanoscale, providing insights into cell proliferation and metastatic potential [54]. Our AFM analysis of NPC cells revealed that CFL1 lactylation reduces stiffness and adhesion force, thereby promoting proliferation and metastasis.

Conclusion

In summary, our study identified the overexpression of circRNA circCDYL in NPC, which enhances LDHA mRNA stability in NPC by binding to its 3’-UTR, resulting in increased LDHA protein levels and enhanced glycolysis in NPC cells. This leads to lactate accumulation, which, under the catalysis of AARS1, induces lactylation of CFL1 at the K22 site, extending its protein half-life. The increased CFL1 levels promote cytoskeletal remodeling, decreasing cell stiffness and adhesion, thereby facilitating the proliferation and metastasis of NPC cells (Fig. 7F). When glycolysis was inhibited by 2-DG or an LDHA inhibitor, both proliferation and metastasis were reduced, suggesting that targeting circCDYL or its downstream pathways could serve as potential therapeutic strategies for NPC.

Compliance with ethics requirements

All Institutional and National Guidelines for the care and use of animals (fisheries) were followed (NO. 2019sydw0090).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Author contributions.

YJY and YAW contributed equally to this work, designed the study and conducted most of the experiments. HKQ, DW, JSG, PW, QJY, and PC performed some of the experiments. QJY and LS collected tissue samples. PC, BX, MZ, CG, WX and LS analyzed the data. YJY wrote the manuscript, while YAW, ZYZ, and WX revised it. WX, LS and ZYZ supervised the research and acquired funding. All authors read and approved the final manuscript.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: