Lymph node metastasis-related circHELQ promotes progression and acts as a nanotherapeutic target in colorectal cancer

Abstract

Introduction

Studies have highlighted the important roles of circRNAs in various cancers. However, their specific functions in the lymph node metastasis (LNM) of colorectal cancer (CRC) remain largely unclear.

Methods

The Arraystar human circRNA microarray was used to identify circular RNAs (circRNAs) associated with LNM in colorectal cancer (CRC). Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to measure circHELQ expression in CRC cell lines and tissue specimens. A series of in vitro and in vivo functional assays were subsequently conducted to investigate the role of circHELQ in LNM and CRC progression. Additionally, fluorescence in situ hybridization, dual-luciferase reporter assays, RNA pull-down, and Western blot experiments were carried out to elucidate the underlying regulatory mechanisms of circHELQ in CRC.

Results

Our results demonstrated that circHELQ expression was significantly upregulated in CRC and that its expression level was closely correlated with LNM. Gain- and loss-of-function experiments revealed that circHELQ promotes CRC progression and metastasis—including proliferation, migration, invasion, lymphatic vessel formation, and LNM—both in vitro and in vivo. Mechanistically, circHELQ acts as a competing endogenous RNA (ceRNA) that sponges miR-4793-3p, thereby alleviating its suppression of VEGFC. This circHELQ/miR-4793-3p/VEGFC axis drives LNM in CRC by upregulating VEGFC expression. Furthermore, we developed a cancer cell membrane–based siRNA delivery system (cMDS) targeting circHELQ, and these nanoparticles effectively suppressed CRC proliferation and LNM.

Conclusions

CircHELQ drives CRC progression and LNM and is a candidate prognostic biomarker. Furthermore, its molecular function renders it a viable target for nanoparticle-based therapeutic strategies against CRC.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12951-026-04693-8.

Background

Colorectal cancer (CRC) is among the most common malignancies and poses a major global health burden [1–4]. Metastasis, especially through the lymphatic system, is a major cause of mortality in patients with CRC [5, 6]. Additionally, it is an important indicator for assessing tumor invasiveness, informing clinical staging and surgical decision-making, guiding postoperative adjuvant chemotherapy, and predicting the likelihood of tumor recurrence [7–11]. Despite its widely acknowledged clinical relevance, the precise molecular mechanisms underlying lymph node metastasis (LNM) in CRC remain largely elusive. Therefore, investigating the molecular mechanisms of LNM in CRC is essential for developing effective therapeutic strategies.

As a novel class of RNAs, circRNAs have emerged as crucial regulators of various malignancies [12]. Unlike linear RNAs, circRNAs are formed by back-splicing, a process in which a downstream 5’ splice site is covalently joined to an upstream 3’ splice site [13, 14]. Consequently, circRNAs are evolutionarily conserved and characterized by high stability and resistance to exonuclease-mediated degradation [15]. CircRNAs function via a plethora of distinct molecular mechanisms, such as acting as miRNA sponges, binding to proteins, and participating in the translation of polypeptides [16, 17]. Our previous studies demonstrated that dysregulated circRNA expression plays important roles in various biological processes associated with CRC. For example, circTATDN3 acts as a sponge for miR-511-5p to promote the Warburg effect in CRC [18]. CircPDIA3 is upregulated and associated with pyroptosis and chemotherapy resistance in CRC [19]. Overexpression of circATG4B promotes autophagy in CRC cells [20]. CircFNDC3B suppresses CRC epithelial–mesenchymal transition through the translation of a novel protein [21]. While the importance of circRNAs in CRC is increasingly recognized, their specific roles and underlying mechanisms in lymphatic metastasis remain poorly understood. Thus, further studies are needed to elucidate these aspects.

Therefore, the purpose of this study was to investigate the role of circHELQ, a circular RNA that is upregulated and associated with LNM in CRC. Additionally, circHELQ enhanced LNM and cancer progression both in vitro and in vivo via the miR-4793-3p/VEGFC axis. In recent years, circRNA-targeted therapies have emerged as novel therapeutic strategies. The use of lipid nanoparticles (LNPs) for the targeted delivery of small interfering RNA (siRNA) represents a practical and promising strategy for cancer therapy, primarily by conferring stealth properties and reducing in vivo heterogeneity [22, 23]. Recently, cell membrane–coated nanocarriers have attracted increasing attention as bioinspired drug delivery platforms with significant functional versatility [24, 25]. The construction of cell-membrane nanocarriers involves coating nanoparticles with membrane extracts derived from specific cells [26–28]. This process directly transfers the natural targeting functions of the source cells onto the nanocarrier platform. In this study, we successfully engineered a cancer cell membrane-based siRNA delivery system (cMDS) targeting circHELQ to effectively suppress tumor progression.

Materials and methods

CRC tissue specimens

Paired tumor and adjacent normal tissue samples were obtained from patients who underwent surgery for CRC at Guangdong Provincial People’s Hospital between January 1, 2020, and December 31, 2022. Upon collection, all 108 paired tissue specimens were immediately snap-frozen and stored at -80 °C. This study was approved by the Ethics Committee of Guangdong Provincial People’s Hospital and conducted in accordance with its guidelines.

Cell culture

The human CRC cell lines (HCT116, SW480, DLD1, and HT29) and normal human colon epithelial cells (NCM460) used in this study were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). McCoy's 5A medium (Sigma, USA) was used for HCT116 and HT‑29 cells; RPMI‑1640 (Gibco, USA) for DLD1 and NCM460 cells; and Leibovitz's L‑15 (Gibco, USA) for SW480 cells. All cell lines were maintained in their respective media supplemented with 10% fetal bovine serum (Gibco, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin (Fudebio, China), at 37 °C in a humidified atmosphere containing 5% CO₂..

RNA Isolation and quantitative real-time polymerase chain reaction

Total RNA was extracted from cell lines using TRIzol reagent (Invitrogen, USA), followed by reverse transcription into complementary DNA (cDNA) with HiScript II RT SuperMix (Vazyme Biotech, China) according to the manufacturer’s instructions. Subsequent quantitative real-time polymerase chain reaction (qRT-PCR) analysis was conducted with SYBR qRT-PCR Master Mix (Vazyme Biotech, China). The primer sequences used are listed in Supplementary Table S1.

Vector construction and cell transfection

For circHELQ knockdown, a small interfering RNA (siRNA) targeting its back‑splice junction and a non‑targeting siRNA control (siRNA‑NC) were designed and synthesized by Hanyin Biotechnology (Shanghai, China). An overexpression construct for circHELQ was generated by PCR amplification of the full-length cDNA and subsequent ligation into the pLCDH-ciR vector; the corresponding empty vector was used as a negative control. Stable upregulation of circHELQ in CRC cells was achieved by lentiviral transduction. Furthermore, miRNA mimics and inhibitors were obtained from GenePharma (Shanghai, China). All transfections were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions. The siRNA sequences are provided in Supplementary Table S2.

Cell proliferation and apoptosis assays

Cell proliferation was evaluated using the Cell Counting Kit-8 (CCK-8) assay. Cells were seeded into 96-well plates, and at the indicated time points, the absorbance at 450 nm was measured. Data from three independent experiments were normalized to obtain relative proliferation levels.

Colony formation, migration and invasion transwell assays

For the colony formation assay, 1 × 10³ cells were seeded into 6-well plates and incubated at 37 °C for two weeks. Following fixation with methanol, colonies were stained with 0.1% crystal violet for visualization. For transwell migration and invasion assays, 1 × 10⁵ cells were seeded into the upper chamber of inserts (BD Biosciences, USA) with or without Matrigel coating, while in both cases, the lower chamber was filled with medium containing 10% FBS. After 24 h, the migrated cells were fixed with methanol and stained with 0.1% crystal violet.

HLEC tube formation assays and transwell assay

Conditioned medium was collected and concentrated from CRC cells (HCT116 and SW480) following transfection with siRNA, plasmids, miRNA mimics, or miRNA inhibitors, using ultrafiltration tubes (Millipore, USA). For the tube formation assay, human lymphatic endothelial cells (HLECs) were plated in Matrigel-coated 96-well plates and cultured for 24 h. The concentrated conditioned medium was then added, and lymphatic tube formation was assessed after 12 h under an inverted microscope. In the migration assay, HLECs were first treated with the concentrated conditioned medium from CRC cells for 24 h. Subsequently, 1 × 10⁵ HLECs were seeded into the upper chamber of a transwell insert, with medium containing 10% FBS placed in the lower chamber. Following 24 h of incubation, cells that had migrated were subsequently fixed with methanol and visualized by staining with 0.1% crystal violet.

Immunohistochemistry (IHC)

Tissue specimens were fixed in 4% paraformaldehyde, embedded in paraffin, and subsequently sectioned at 5 μm thickness. After antigen retrieval, sections were incubated with primary antibody overnight at 4 °C. After washing, slides were incubated with a HRP‑conjugated secondary antibody for 1 h, then developed with diaminobenzidine (DAB) and counterstained with hematoxylin. Images were acquired using a light microscope.

Fluorescence in situ hybridization (FISH)

Fluorescence in situ hybridization (FISH) was conducted to detect the subcellular localization of circHELQ and miR-4793-3p in CRC cells. Cell slides were hybridized overnight at 37 °C with Cy3‑labeled probes targeting circHELQ and miR‑4793‑3p, followed by DAPI counterstaining. Images were acquired using a fluorescence microscope.

VEGFC quantification

The concentration of VEGFC in cell lysates was quantified using a Human VEGFC ELISA Kit (Elabscience, E-EL-H1600), according to the manufacturer’s instructions. Following antibody incubation as specified in the ELISA protocol, absorbance at 450 nm was measured, and VEGFC levels were determined based on a standard curve.

RNA Immunoprecipitation (RIP)

The interaction between the target protein and RNA was evaluated using an RNA immunoprecipitation (RIP) assay kit. Briefly, 1 × 10⁷ cells were harvested and lysed. The cell lysates were incubated with either anti-IgG (negative control) or specific antibodies against the target protein. RNA co‑immunoprecipitated with the protein complexes was subsequently extracted and quantified by qRT‑PCR using gene‑specific primers.

Dual-luciferase reporter assay

The full-length sequences of circHELQ and VEGFC, including either their wild-type or mutant binding sites, along with the corresponding linear gene sequences, were cloned into psiCHECK-2 reporter vectors. CRC cells were co-transfected with these constructs together with either negative control (NC) or miR-4793-3p mimics. At 48 h post-transfection, luciferase activity was measured using the dual-luciferase reporter assay system (Vazyme, Nanjing, China).

Western blot analysis

Cells were harvested and lysed in RIPA buffer (CWBIO, China) for protein extraction. Following SDS-PAGE, the proteins were transferred onto nitrocellulose membranes and probed with specific primary antibodies. The membranes were then incubated with corresponding secondary antibodies at room temperature for 1 h. Protein signals were detected with Immobilon ECL substrate (Millipore, Germany) and captured using an Optimax X-ray film processor (Protec, Germany).

Mouse xenograft model

To evaluate the effect of circHELQ on tumor growth in vivo, 4-week-old BALB/c nude mice were inoculated subcutaneously into the right flank with HCT116 cells (5 × 10⁶ cells in 0.2 mL PBS) following stable transfection with circHELQ-overexpressing or empty vector. Tumor growth was monitored every 7 days, and tumor volumes were calculated using the formula: (length × width²)/2. All mice were euthanized at 35 days post-injection.

PDX mouse model

Patient-derived xenograft (PDX) models of CRC were established in NOD/SCID and BALB/c nude mice. Briefly, primary tumor tissues were collected from CRC patients undergoing surgical resection and immediately preserved in ice-cold culture medium containing 50% Matrigel and 1% penicillin/streptomycin. These tissues were implanted into NOD/SCID mice and serially passaged for two generations. When the engrafted tumors reached 1–2 cm³ in size, they were harvested, fragmented, and serially transplanted into BALB/c nude mice. When the tumor volume in nude mice reached approximately 200 mm³, the mice were administered saline, si-NC@LNP, si-circHELQ@LNP, or si-circHELQ@cMDS via tail vein injection twice a week at a dose of 200 μg/kg. Tumor volumes and body weights were monitored regularly. At the endpoint, tumors were excised, weighed, and processed for subsequent analysis.

Popliteal lymphatic metastasis model

To investigate the role of circHELQ in lymph node metastasis-positive CRC, a popliteal lymphatic metastasis model was developed. 4–5-week-old BALB/c nude mice were randomly allocated into experimental groups. A third-party operator, blinded to the experimental design, injected 1 × 10⁶ transfected HCT116 cells suspended in 50 µL phosphate-buffered saline into the footpad of each mouse. When the primary footpad tumors reached about 200 mm³, metastatic progression was monitored non-invasively using an IVIS spectral imaging system. Primary tumors and popliteal lymph nodes were then harvested for IHC staining. Histopathological evaluation of metastatic lesions in hematoxylin and eosin (H&E)-stained sections, quantification of LYVE-1-positive lymphangiogenesis, and assessment of Ki-67-positive proliferative foci were performed independently by two board-certified pathologists who were blinded to all group assignments. To evaluate the efficacy of the developed cMDS, treatment was initiated once tumor volumes in nude mice had stabilized. Mice were administered saline, si‑NC@LNP, si‑circHELQ@LNP, or si‑circHELQ@cMDS via tail vein injection. Lymph node volume and the lymphatic metastasis ratios were subsequently assessed.

Statistical analysis

Normality was assessed using the Shapiro-Wilk test. Data with a P-value > 0.05 were considered to be normally distributed. Homogeneity of variances was evaluated using Levene’s test. Comparisons with P-value > 0.05 were considered to have equal variances across groups. A two-tailed Student’s t-test was applied only when both assumptions were met. The correlation between circHELQ expression and clinicopathological features was evaluated using Student’s t-test or the chi-square test, as appropriate. Overall survival (OS) was estimated using the Kaplan–Meier method, and survival differences were compared with the log-rank test. Univariate and multivariate Cox proportional hazards regression models were employed to identify independent prognostic factors. A two-sided P-value < 0.05 was considered statistically significant. All statistical analyses were performed using Stata (version 13.1; StataCorp, College Station, TX, USA) and GraphPad Prism (GraphPad Software Inc., San Diego, CA, USA).

Results

Identification of a novel lymph node metastasis-related circRNA circHELQ in CRC cells

Differential circRNA expression profiling between 3 lymph node metastatic and 3 non‑metastatic CRC tissue samples was conducted via circRNA microarray to investigate the underlying molecular mechanisms of CRC progression (Fig. 1A). The expression of the top 5 upregulated circRNAs among the differentially expressed candidates in primary CRC cells from CRC patients with or without LNM was evaluated by qRT-PCR. CircHELQ, which was clearly overexpressed in primary tumor cells from CRC patients with LNM, was selected for further exploration (Fig. 1B). Additionally, we also observed that circHELQ expression was upregulated in CRC cell lines compared with NCM460 cells (Fig. S1A). In accordance with the human reference genome, circHELQ is derived from exons 13–17 of the HELQ gene and spans 680 nucleotides (nts), and its back‑spliced junction was validated by Sanger sequencing (Fig. 1C-D). Furthermore, circHELQ was specifically amplified from cDNA but not from gDNA, confirming its circular rather than genomic origin (Fig. 1E). Minimal detection of circHELQ with oligo‑dT primers suggests a lack of a poly‑A tail, which is consistent with its circular structure (Fig. 1F). To verify that circRNAs were more stable than their corresponding linear RNAs, we performed stability assays for circHELQ. RNase R treatment revealed that compared with its linear HELQ mRNA counterpart, circHELQ was significantly more resistant (Fig. 1G). Following actinomycin D treatment to inhibit transcription, circHELQ exhibited a significantly longer half-life than HELQ mRNA (Fig. 1H). These findings indicated that circHELQ is highly stable as a circRNA. CircHELQ was predominantly localized in the cytoplasm of CRC cells, as demonstrated by subcellular fractionation experiments and FISH (Fig. 1I–J). Subsequent analysis revealed that circHELQ expression was significantly upregulated in CRC tissues compared with adjacent normal tissues (Fig. S1B), and further increases were observed in CRC tissues with LNM compared with those without LNM (Fig. 1K). Patients were stratified into high and low circHELQ expression groups according to the median expression level, and their associations with clinicopathological features were analyzed (Table S3). Kaplan‑Meier analysis demonstrated that high circHELQ expression correlated with significantly poorer outcomes in patients with CRC (P < 0.01) (Fig. S1C). Univariate and multivariate Cox regression analyses indicated that circHELQ expression was associated with independent prognostic value for CRC (HR = 3.42; 95% CI: 1.36–8.65; P = 0.009) (Table 1). These findings suggested a significant association between circHELQ expression and LNM, suggesting that circHELQ is a potential therapeutic target for CRC.

Fig. 1.

Characterization of circHELQ in CRC. (A) Differential expression volcano plot analysis of circRNAs between CRC tissues with/without LNM. (B) The top five circRNAs expression was evaluated by qRT-PCR in primary colorectal cancer cells from CRC patients with or without LNM. (C) Exon 13-17 form circHELQ by back-splicing. (D) Sanger sequencing shows the specific site of back-splicing connection. (E) CircHELQ was amplified only in cDNA and not in gDNA, and the negative control was GAPDH. (F) Detection of circHELQ expression by qRT-PCR using oligo-dT primers and random primers. (G) The expression of circHELQ and HELQ mRNA was detected by qRT-PCR after using RNase R in CRC cell lines. (H) Stability of circHELQ and HELQ mRNA in CRC cells at gradient time points was determined by actinomycin D assay. (I) Nuclear-cytoplasmic fractionation assay identified the localization of circHELQ. Actin and U6 were used as cytoplasmic and nuclear controls, respectively. (J) Fluorescence in situ hybridization assay was conducted to determine the subcellular localization of circHELQ in CRC cells. Scale bars, 10 μm. K. The expression of circHELQ in tumor tissues of CRC patients with/without LNM. P < 0.001***

Table 1.

Univariate and multivariate Cox-regression analysis in CRC patients

	HR	Univariate analysis	P	HR	Multivariate analysis	P
		95% CI			95% CI
T stage
T1/T2	1			1
T3/T4	11.00	(2.60, 45.57)	0.001**	8.69	(1.98, 38.08)	0.004**
N stage
N0	1			-
N1/N2	4.51	(1.36,15.05)	0.014*	-	-	-
Gender
Female	1			-
Male	1.44	(0. 64, 3.24)	0.374	-	-	-
Differentiation
Well/Moderate	1			-
Poor	1.23	(0.42, 3.58)	0.702	-	-	-
Location
Colon	1			-
Rectal	1.12	(0.49, 2.60)	0.785	-	-	-
Expression of circHELQ
Low	1			1
High	4.24	(1.70, 10.59)	0.002**	3.42	(1.36,8.65)	0.009**
Age	1.04	(1.01,1.07)	0.015*	1.01	(0.98,1.04)	0.489

P < 0.05*, P < 0.01**

CircHELQ promotes LNM and progression of CRC

To investigate the biological function of circHELQ in CRC cells, overexpression and RNA interference (RNAi) vectors (Fig. 2A) targeting circHELQ were constructed. CircHELQ was knocked down and overexpressed in CRC cells transfected with RNAi and overexpression vectors, respectively (Fig. 2B, Fig. S2A). The results of CCK8 assay demonstrated that downregulation of circHELQ markedly suppressed the proliferation of CRC cells, whereas its upregulation enhanced CRC cell growth (Fig. 2C, Fig. S2B).

Fig. 2.

Functions of circHELQ in CRC. (A) Construction of knockdown systems for circHELQ. (B) The efficiency of si-circHELQ was verified. (C) The proliferative capacity of CRC cells after down-regulation of circHELQ was verified by CCK8 assay. (D) The single-cell clonogenic capacity of CRC cells after down-regulation of circHELQ was verified by colony formation assay. (E) The migration ability of CRC cells after down-regulation of circHELQ was verified by transwell migration assay. (F) The invasion ability of CRC cells after down-regulation of circHELQ was verified by transwell invasion assay. (G) The effects of down-regulation of circHELQ on LNM and lymph node generation in CRC cells were verified by lymphatic endothelial cell migration assay and lymphatic vessel formation assay. P < 0.001***

Consistent with these findings, colony formation assays demonstrated that downregulation of circHELQ expression significantly suppressed the clonogenic capacity of CRC cells, whereas its upregulation enhanced this ability (Fig. 2D, Fig. S2C). Transwell migration assays demonstrated that circHELQ knockdown significantly inhibited CRC cell migration, whereas circHELQ overexpression promoted CRC cell migration (Fig. 2E, Fig. S2D). In cell invasion assays, circHELQ knockdown suppressed the invasive ability of CRC cells, while circHELQ overexpression enhanced it. (Fig. 2F, Fig. S2E). In vitro experiments demonstrated that the migration and tube formation of HLECs were affected in CRC cells transfected with knockdown or expression (Fig. 2G, Fig. S2F). These findings suggested that circHELQ contributes to tumor progression and lymphangiogenesis in CRC.

CircHELQ functions as a sponge for miR-4793-3p

Previous studies have shown that circRNAs can function as miRNA sponges in the cytoplasm across various tumor types. To explore the molecular mechanism of circHELQ, we predicted its potential targets using a miRNA target prediction database (TargetScan, miRanda, and circBank) (Fig. 3A). The analysis revealed that circHELQ contains conserved target sites for miR-4793-3p, miR-33a-5p, miR-548au-3p, miR-610 and miR-19b-2-5p, with high prediction scores.

Fig. 3.

CircHELQ sponges with miR-4793-3p. (A) Potential targets of circHELQ were predicted using miRNA target prediction software (TargetScan, miRanda and circBank). (B) Expression of five candidate miRNAs was quantified by qRT-PCR after the circHELQ probe pulldown assay in CRC cells. (C) Expression of circHELQ was quantified by qRT-PCR after the biotinylated-miR-4793-3p pulldown assay in CRC cells. (D–E) AGO2 RIP assay for circHELQ and miR-4793-3p in CRC cells. (F) FISH showing co-location of circHELQ and miR-4793-3p in CRC cells. Scale bars, 10 µm. (G) A schematic drawing showing the putative binding sites of circHELQ and miR-4793-3p. (H) Sequence alignment of circHELQ and miR-4793-3p. (I) Dual-luciferase validation illustrates the relationship between circHELQ and miR-4793-3p. P < 0.01**, P < 0.001***

In a pull-down assay using circHELQ probe, miR-4793-3p showed the highest enrichment, indicating its strongest binding affinity for circHELQ (Fig. 3B). Additionally, we found that miR-4793-3p biotin pulled down more circHELQ (Fig. 3C). We therefore hypothesized that circHELQ may act as a competing endogenous RNA (ceRNA) for miR-4793-3p in CRC. To elucidate this interaction, AGO2-RIP experiments were conducted. Our data demonstrated that AGO2 served as a binding platform, interacting with both circHELQ and miR-4793-3p (Fig. 3D). Furthermore, upregulation of circHELQ expression in CRC cells led to increased enrichment of miR-4793-3p by the AGO2 protein (Fig. 3E). FISH analysis in CRC cells revealed that circHELQ and miR-4793-3p are colocalized in the cytoplasm (Fig. 3F). Next, on the basis of the predicted molecular interaction between circHELQ and miR-4793-3p (Fig. 3G), the full-length wild-type (WT) circHELQ sequence and a mutant version lacking the miR-4793-3p binding site were cloned and inserted into a luciferase reporter vector (Fig. 3H). To investigate the direct binding between circHELQ and miR-4793-3p, a dual-luciferase reporter assay was employed. The results revealed that the miR-4793-3p mimics significantly reduced the luciferase activity of the WT reporter vector but had no effect on that of the MUT vector, confirming the specific interaction between circHELQ and miR-4793-3p (Fig. 3I). Collectively, these results supported the conclusion that circHELQ functions as a competing ceRNA for miR-4793-3p in CRC.

MiR-4793-3p can reverse the effects of circHELQ on CRC proliferation, migration, invasion and lymphangiogenesis

To validate the interaction between miR-4793-3p and circHELQ, we conducted the following rescue experiments using miR-4793-3p inhibitors and mimics (Fig. S3A-B). CCK8 experiments revealed that the upregulation of circHELQ promoted the proliferation of CRC cells, whereas the pro-proliferative effect of circHELQ was reversed by the miR-4793-3p mimics (Fig. 4A). Conversely, circHELQ downregulation suppressed proliferation, whereas the antiproliferative effect induced by circHELQ knockdown was attenuated by the miR-4793-3p inhibitors (Fig. S4A). Subsequent colony formation assays revealed that upregulating circHELQ expression significantly increased the cloning ability of CRC cells, whereas the miR-4793-3p mimics reversed this effect (Fig. 4B). Similarly, downregulation of circHELQ expression followed by treatment with the miR-4793-3p inhibitors restored the cloning ability (Fig. S4B). In the migration and invasion transwell assays, we found that upregulating circHELQ expression in CRC cells significantly promoted the migration and invasion of CRC cells. These effects were reversed by increasing the level of miR-4793-3p (Fig. 4C-D). When circHELQ knockdown was combined with subsequent miR-4793-3p inhibition, the resulting decrease in CRC cell migration and invasion was reversed (Fig. S4C-D). Through lymphatic endothelial cell migration and lymphangiogenesis experiments, we confirmed that upregulating circHELQ expression promoted lymphatic endothelial cell migration and lymphangiogenesis. A simultaneous increase in miR-4793-3p expression reversed these effects (Fig. 4E). In contrast, si-circHELQ inhibited lymphatic endothelial cell migration and lymphangiogenesis, whereas the miR-4793-3p inhibitors reversed these effects (Fig. S4E). Taken together, our data demonstrate that circHELQ functions as a molecular sponge for miR-4793-3p, alleviating its suppression of lymphangiogenesis and CRC progression.

Fig. 4.

MiR-4793-3p can reverse the effect of circHELQ in CRC. (A) CCK-8 assays showed that miR-4793-3p mimics could reverse the promotive effect of circHELQ overexpression on CRC cell proliferation. (B) Colony formation assays showed that miR-4793-3p mimics could reverse the promotive effect of circHELQ overexpression on the CRC cell clonogenic capacity. (C) Transwell migration assays showed that miR-4793-3p mimics could reverse the promotive effect of circHELQ overexpression on CRC cell migration. (D) Transwell invasion assays showed that miR-4793-3p mimics could reverse the promotive effect of circHELQ overexpression on CRC cell invasion. (E) Lymphatic endothelial cell migration assays and lymphatic vessel formation assays showed that miR-4793-3p mimics could reverse the promotive effect of circHELQ overexpression on HLEC migration and tube formation. P < 0.01**, P < 0.001***

VEGFC is directly targeted by miR-4793-3p

By cross-referencing the TargetScan, miRDB, miRTarBase, and miRWalk databases, we identified five LNM‑related genes—VEGFA, VEGFC, ITGA9, APLN, and AGRN—as potential targets of miR‑4793‑3p. We found that the overexpression of miR-4793-3p reduced VEGFC mRNA expression, whereas the miR-4793-3p inhibitors increased VEGFC expression in CRC cells (Fig. 5A). On the basis of these findings, VEGFC was selected as the downstream target. Next, on the basis of the bioinformatically predicted interaction between miR-4793-3p and VEGFC, both the full-length WT VEGFC sequence and a mutant version lacking the miR-4793-3p binding sites were cloned and inserted into a luciferase reporter vector (Fig. 5B). The luciferase activity of the reporter vector containing the wild-type VEGFC 3’UTR was significantly reduced by miR-4793-3p mimics compared with the control (Fig. 5C). To investigate whether circHELQ regulates VEGFC through miR-4793-3p, we constructed a dual luciferase reporter system. As shown in Fig. 5D, the suppression of luciferase activity of the wild-type VEGFC 3’UTR reporter induced by miR-4793-3p could be reversed by cotransfection with circHELQ. In contrast, cotransfection with a plasmid carrying the mutant VEGFC 3’UTR had no significant effect on luciferase activity (Fig. 5D). Furthermore, downregulation of circHELQ expression in CRC cells decreased the expression of VEGFC mRNA and its protein, whereas downregulation of miR-4793-3p expression increased VEGFC expression (Fig. 5E-F, Fig. S5A-B). Additionally, compared with NC transfection, cotransfection of circHELQ and miR-4793-3p did not affect VEGFC expression (Fig. 5G-H, Fig. S5C-D). Consistent with these findings, similar changes in VEGFC levels were observed in the conditioned medium of CRC cells subjected to the same treatments (Fig. 5I-J, Fig. S5E-F). Furthermore, analysis of CRC patient samples revealed a positive correlation between the VEGFC levels and circHELQ in serum and tissue of CRC patients (Fig. 5K-L). Collectively, these data indicated that circHELQ regulates VEGFC expression in CRC by functioning as a ceRNA that sequesters miR-4793-3p.

Fig. 5.

VEGFC is directly targeted by miR-4793-3p. (A) Following overexpression or down-regulation of miR-4793-3p in CRC cell lines, the expressions of the previously predicted LNM-associated genes were assessed. (B) Bioinformatic analysis screened for miR-4793-3p interaction with wild-type VEGFC 3’UTR. (C) Dual-luciferase validation illustrated the relationship between VEGFC and miR-4793-3p. (D) The luciferase assay of VEGFC 3’UTR in CRC cells co-transfected with miR-4793-3p and circHELQ. (E–F) The VEGFC mRNA and protein expression was evaluated in HCT116 following circHELQ downregulation and treatment with miR-4793-3p inhibitors. (G–H)The VEGFC mRNA and protein expression was evaluated in SW480 following circHELQ upregulation and treatment with miR-4793-3p mimics. (I–J) CircHELQ regulated the secretion of VEGFC through miR-4793-3p. (K–L) Correlation analysis between circHELQ and VEGFC levels in serum and tissue of CRC patients. P < 0.05*, P < 0.01**, P < 0.001***

VEGFC knockdown abolishes the proliferative, migratory, invasive, and lymphangiogenic effects induced by circHELQ overexpression

Furthermore, we investigated the role of the circHELQ/miR-4793-3p/VEGFC axis in regulating the malignant progression of CRC. CCK-8 assays demonstrated that knockdown of VEGFC attenuated the proliferative advantage conferred by circHELQ overexpression (Fig. 6A). Similarly, VEGFC deletion diminished the enhanced colony-forming ability induced by circHELQ (Fig. 6B). Moreover, VEGFC silencing suppressed the migratory capacity of CRC cells overexpressing circHELQ (Fig. 6C). Consistent with these findings, the VEGFC knockdown also impaired the increase in invasive potential induced by circHELQ upregulation (Fig. 6D). Furthermore, the deletion of VEGFC reduced HLEC migration and lymphangiogenesis promoted by conditioned medium from circHELQ-overexpressing CRC cells (Fig. 6E). In addition, given that circHELQ can sponge miR-4793-3p, we performed rescue experiments using miR-4793-3p inhibitors and anti-VEGFC antibodies. The results also indicated that VEGFC serves as a key downstream functional target of miR-4793-3p (Fig. S6A-E). Taken together, these results indicated that the ability of circHELQ to promote CRC proliferation, migration, invasion, and lymphangiogenesis can be reversed by inhibiting VEGFC.

Fig. 6.

VEGFC knockdown abolishes the effects induced by circHELQ overexpression. (A) The proliferation ability of CRC cells following circHELQ upregulation and VEGFC downregulation was assessed using CCK8 assays. (B) The single-cell clonogenic capacity of CRC cells following circHELQ upregulation and VEGFC downregulation was assessed using colony formation assays. (C) The migration of CRC cells following circHELQ upregulation and VEGFC downregulation was assessed using transwell migration assays. (D) The invasion of CRC cells following circHELQ upregulation and VEGFC downregulation was assessed using transwell invasion assays. (E) The migration and tube formation of HLECs treated with conditioned medium from circHELQ-upregulated and VEGFC-downregulated CRC cells were assessed using lymphatic endothelial cell migration and lymphatic vessel formation assays. P < 0.01**, P < 0.001***

CircHELQ facilitate s CRC progression and LNM in vivo

To further investigate the role of circHELQ in CRC proliferation in vivo, we established subcutaneous xenograft models. BALB/c nude mice were injected subcutaneously with HCT116 cells stably transfected with either an empty vector or a circHELQ‑overexpressing lentivirus. Compared with control tumors, tumors derived from circHELQ‑overexpressing cells were significantly larger in volume and greater in weight (Fig. 7A-C). IHC staining for Ki‑67 further confirmed the enhanced proliferative activity of circHELQ‑overexpressing tumors (Fig. 7D). We also employed a popliteal LNM model by inoculating luciferase‑labeled CRC cells transfected with either control or circHELQ overexpression vectors into the right footpad of nude mice (Fig. 7E). When the primary footpad tumor reached approximately 200 mm³, lymph node metastasis was assessed. In vivo imaging system (IVIS) analysis revealed significantly stronger luciferase signals in the circHELQ‑overexpression group, which corresponded to increased popliteal lymph node enlargement (Fig. 7F-H). Additionally, the levels of VEGFC in both the footpad primary tumor and the serum of the circHELQ-expressing group were significantly higher than those in the control group (Fig I-J). IHC detection of GFP‑positive cells confirmed the presence of metastatic CRC cells in the lymph nodes (Fig. 7K). Moreover, the incidence of popliteal LNM was markedly greater in the circHELQ‑overexpression group (Fig. 7L). IHC analysis further revealed that circHELQ overexpression increased the microlymphatic vessel density (MLD) within the primary tumors (Fig. 7M). In summary, these in vivo findings demonstrated that circHELQ promotes tumor proliferation and lymph node metastasis in CRC.

Fig. 7.

CircHELQ facilitates CRC progression and LNM in vivo. (A) Representative picture of subcutaneous xenograft tumors. (B–C) Curves of tumor volumes and weights showed promotive effects of circHELQ on the formation of subcutaneous xenograft tumors. (D) The effect of circHELQ on tumor proliferation was verified by detecting Ki67 expression by IHC assay. Scale bar, 50 μm. (E) Schematic diagram of constructing a popliteal LNM model in nude mice. (F) Validation of the effect of circHELQ upregulation on popliteal LNM in popliteal LNM model mice by detecting popliteal Luminescence values by IVIS. (G–H) The effect of circHELQ upregulation on popliteal LNM was verified by calculating popliteal lymph node volume. (I) VEGFC expressions in footpad primary tumors were assessed by qRT-PCR. (J) VEGFC expressions in serum were assessed by ELISA. (K) IHC detection of anti-GFP expression in popliteal lymph nodes of nude mice overexpressing circHELQ popliteal LNM model. Scale bar, 50 μm. (L) The metastasis ratio of popliteal LNM was calculated to verify the effect of circHELQ. (M) Validation of the effect of circHELQ upregulation on MLD by IHC detection of LYVE1 expression in primary tumors. P < 0.01**, P < 0.001***

Synthesis and characterization of cMDS (si-circHELQ@cMDS) in vitro

This study aimed to develop an efficient siRNA delivery vehicle targeting circHELQ as a potential therapeutic strategy for CRC. The results presented above demonstrated that siRNA-mediated knockdown effectively reduced circHELQ expression and suppressed its ability to promote cell proliferation, migration, invasion, and lymphangiogenesis in vitro. Given that naked siRNA cannot readily cross cell membranes, it is essential to establish robust methods for in vivo siRNA delivery. Owing to the clearance of lipid nanoparticles (LNPs) in the complex bloodstream and their nonspecific distribution throughout the body, the therapeutic potential of LNPs is significantly limited. Therefore, the biomimetic cancer cell membrane engineered lipid nanoparticle method was used to prepare a cancer membrane delivery system (cMDS) loaded with si-circHELQ and LNPs. Transmission electron microscopy (TEM) revealed that the cMDS particles (si-circHELQ@cMDS) exhibited a spherical morphology with an additional membrane-like layer coating their outer surface (Fig. 8A). With a diameter of 143 nm, cMDS particles (si-circHELQ@cMDS) have a negative zeta potential of -4.3 mV (Fig. S7A-B). The polydispersity index (PDI) of si-circHELQ@cMDS was approximately 0.125 (Fig. S7C). To evaluate cellular uptake in vitro, cells were treated with coumarin-6-labeled LNPs and analyzed via confocal microscopy. Compared with the LNP group, the cMDS group exhibited stronger intracellular fluorescence from coumarin-6‑labeled LNPs under fluorescence microscopy (Fig. 8B). These results suggested that the delivery vehicle we developed enhances cellular drug uptake.

Fig. 8.

Characterization and efffcacy of si-circHELQ@cMDS by systemic injection animial model. (A) The TME image of si-circHELQ@cMDS. (B) Coumarin-6 assay was used to detect the effect of NPs on drug absorption in HCT116 cells. (C) The VEGFC levels in conditioned medium were detected by ELISA. (D–G) Targeting of nanomaterials was tested by mouse imaging in vivo and tissue imaging ex vivo. (H–J) The PDX model of CRC was treated by injecting drugs into the mouse tail vein. Tumor growth was significantly inhibited following treatment with si-circHELQ@cMDS, compared to the control group, si-NC@LNP group, and si-circHELQ@LNP group. (K) HE staining showed that intravenous NPs had no significant toxicity to major organs. (L–M) The BALB/c model of CRC was treated by injecting drugs into the mouse tail vein. Popliteal LNM was significantly inhibited following treatment with si-circHELQ@cMDS, compared to the control group, si-NC@LNP group, and si-circHELQ@LNP group. (N) The metastasis ratio of popliteal LNM was calculated to verify the effect of NPs. (O) Schematic diagram demonstrating the molecular mechanisms underlying circHELQ in CRC. P < 0.001***

The in vitro release profiles of free si‑circHELQ, si‑circHELQ@LNP, and si‑circHELQ@cMDS were evaluated. Overall, compared with si‑circHELQ@LNPs, si‑circHELQ@cMDS resulted in a higher rate of siRNA release. Moreover, the release rate of siRNA from si‑circHELQ@cMDS was faster at pH 6.5 than at pH 7.4. After 48 h, the cumulative siRNA release from si‑circHELQ@cMDS reached 78.5% at pH 6.5, significantly exceeding the 23.5% observed at pH 7.4 (Fig. S7D-E). Overall, the comparison between cMDS and conventional LNPs was shown in Table S4.

We further evaluated the knockdown efficiency of circHELQ in CRC cells (Fig. S7F) and the level of VEGFC in the conditioned medium (Fig. 8C) following treatment with si‑circHELQ@LNP or si‑circHELQ@cMDS. Following tail‑vein injection of DIR‑labeled LNPs, fluorescence signals were predominantly localized in the liver. In contrast, DIR‑labeled si‑circHELQ@cMDS also accumulated in mouse tumor sites, demonstrating the tumor‑targeting ability of the cMDS formulation (Fig. 8D-G).

Efficacy of si‑circHELQ@cMDS by systemic injection in an animal model

We investigated the effect of si‑circHELQ@cMDS on tumor growth in vivo using CRC patient-derived xenograft (PDX) mouse model (Fig. S8A). Mice were treated via tail vein injection with saline, si-NC@LNP, si-circHELQ@LNP, or si‑circHELQ@cMDS. Both si-circHELQ@LNP and si‑circHELQ@cMDS significantly suppressed tumor growth, with si‑circHELQ@cMDS demonstrating a more pronounced inhibitory effect (Fig. 8H). Consistent results were observed for the volume and weight of excised xenograft tumors from the corresponding treatment groups (Fig. 8I-J). Potential systemic toxicity was assessed by histopathological examination of major organs using H&E staining. The findings indicated that intravenous administration of si‑circHELQ@cMDS did not induce significant toxicity in major organs including the liver, kidney, lung, spleen, or heart (Fig. 8K). Safety and long-term toxicity were assessed by monitoring mouse body weights and performing serum biochemical analysis, which included markers such as hemoglobin (Hb), white blood cell count (WBC), platelet count (PLT), alanine transaminase (ALT), aspartate transaminase (AST), creatinine (Cr), and blood urea nitrogen (BUN), thereby demonstrating the safety of cMDS (Fig. S8B-I).

Furthermore, we established a popliteal lymph node metastasis (LNM) model to evaluate the effect of si‑circHELQ@cMDS on LNM in vivo. Treatment with si-circHELQ@LNP or si‑circHELQ@cMDS reduced popliteal lymph node size, with si‑circHELQ@cMDS having a more potent inhibitory effect (Fig. 8J–M). Correspondingly, a decrease in the volume of excised lymph nodes was observed in the same treatment groups. The incidence of popliteal LNM was also significantly reduced following si‑circHELQ@cMDS administration (Fig. 8N). Collectively, these findings indicated that si‑circHELQ@cMDS can suppress both CRC proliferation and lymph node metastasis. A graphical summary of the proposed mechanism is shown in Fig. 8O.

Discussion

LNM is a robust indicator of poor clinical outcomes in CRC patients. Numerous studies have highlighted the crucial role of tumor lymphangiogenesis in promoting LNM across various cancers, including CRC [29–31]. Consequently, there is a clear need to identify effective therapeutic targets associated with LNM. In this study, we revealed that LNM-associated circHELQ is significantly upregulated in CRC patients and that its expression is positively correlated with LNM status. Furthermore, our experimental results demonstrated that circHELQ promotes both LNM and CRC progression, indicating its potential as a novel biomarker and therapeutic target for LNM in CRC.

Previously, miR-4793-3p was shown to be differentially expressed in a range of human diseases, with its clinical relevance further validated in the context of cancer. In particular, elevated levels of miR-4793-3p have been associated with the initiation and progression of hepatitis B virus-associated liver cirrhosis, suggesting a potential link to bacterial infection and tumor modulation [32]. In gastric cancer, miR-4793-3p functions as a molecular sponge for circEVI5, thereby contributing to the suppression of tumorigenesis [33]. Furthermore, in breast cancer, miR-4793-3p exerts tumor-suppressive effects through the regulation of RNF169, which in turn modulates the JAK2/STAT3 signaling pathway [34]. In this study, we initially screened five miRNAs predicted by TargetScan, miRanda, and circBank, among which miR-4793-3p was validated to inhibit tumor lymphangiogenesis. Mechanistically, miR-4793-3p was found to directly bind to circHELQ. Given that circRNAs can function as miRNA sponges to modulate tumor progression, we further investigated whether circHELQ acts as a molecular sponge for specific miRNAs. Through a series of experiments, we demonstrated that circHELQ sponges endogenous miR-4793-3p, thereby suppressing its activity and subsequently promoting tumor lymphangiogenesis. These findings suggested that the circHELQ/miR-4793-3p axis plays a critical role in regulating LNM in CRC.

Through bioinformatics analysis and expression validation, we selected VEGFC from five downstream targets of miR-4793-3p for further investigation. VEGFC is closely associated with tumor cell proliferation and LNM. By promoting VEGFR3-dependent lymphatic vessel growth, VEGFC serves as a key driver of lymphatic network expansion within the tumor microenvironment, thereby facilitating LNM [35]. In this study, we found that circHELQ overexpression significantly upregulated VEGFC expression in CRC cells. This finding is consistent with the positive correlation observed between serum VEGFC levels and tissue circHELQ expression in CRC patients. Mechanistically, circHELQ alleviates the inhibitory effect of miR-4793-3p on VEGFC, thereby promoting VEGFC expression and secretion in CRC cells. Previous studies have reported that miR-4793-3p can regulate other downstream targets in the biological behaviors of various tumors. In our future studies, we will further investigate whether miR-4793-3p also regulates additional targets to exert its functions in CRC.

Unlike conventional CRC targets such as KRAS, EGFR, or PD-L1, which are primarily involved in proliferation or immune evasion, circHELQ is specifically associated with LNM—a critical determinant of poor prognosis and treatment resistance. Our data demonstrated that circHELQ drives metastatic progression, positioning it as a potential target for patients with advanced or metastatic CRC, in whom current targeted therapies frequently fail. Furthermore, as a circRNA, circHELQ possesses inherent stability, offering a distinct advantage over linear RNAs or short-lived protein targets. Given that LNM is a major cause of chemoresistance and recurrence, combining circHELQ-targeted siRNA delivery with standard regimens or immune checkpoint inhibitors may yield enhanced anti-tumor effects. Specifically, reducing circHELQ expression may resensitize metastatic CRC cells to chemotherapy and improve anti-tumor immune responses by modulating the metastatic niche.

Therefore, we aimed to develop an efficient siRNA delivery vehicle targeting circHELQ. At present, the targeted delivery of siRNA using LNPs represents a practical and promising frontier in the development of cancer therapeutics. Nevertheless, these LNPs face several limitations, including poor biological stability, insufficient tumor targeting, potential toxicity, and rapid clearance by the immune system, in clinical applications [36, 37]. Among these, cell-derived biomimetic nanotechnology has been widely adopted in targeted tumor therapy because of its inherent advantages, such as high selectivity, favorable biocompatibility, low immunogenicity, and minimal side effects [38–41]. To enhance tumor-targeting efficacy, this study developed biomimetic tumor cell membrane-coated nanoparticles for the efficient delivery of si-circHELQ LNPs. We created a cell membrane delivery system based on the CRC cell membrane (si-circHELQ@cMDS). At a pH of 6.5, which simulates the tumor microenvironment, the release rate of siRNA from si-circHELQ@cMDS was notably faster than that at a pH of 7.4, indicating physiological conditions. After 48 h, the cumulative siRNA release from si‑circHELQ@cMDS reached 78.5% at pH 6.5, significantly exceeding the 23.5% release observed at pH 7.4. These observations suggested that si-circHELQ@cMDS enables rapid siRNA release in the weakly acidic tumor microenvironment. Furthermore, as cell membranes predominantly consist of lipids, proteins, carbohydrates, and other minor constituents—with lipids comprising more than 50% of the composition, including phospholipids, glycolipids, cholesterol, and sphingomyelins—their inherent surfactant-like properties facilitate drug dissolution and release [42]. Consequently, compared with si-circHELQ@LNPs, si-circHELQ@cMDS resulted in a higher rate of siRNA release. Such favorable siRNA release kinetics provided a robust foundation for the in vivo therapeutic efficacy of si-circHELQ@cMDS. Cancer cell membrane engineered LNPs enable industrial translation. Biomimetic coating enhances biocompatibility and targeting. Combined with FDA-approved LNP platforms, this offers a scalable, clinical strategy for circHELQ targeting, supporting long-term storage and batch reproducibility.

The present study has several limitations. First, although circHELQ represents a potential prognostic biomarker for CRC, a definitive cut-off value for distinguishing high versus low expression levels has yet to be established. In the present study, the median expression level, a cut-off commonly used in previous studies, was employed to distinguish high from low expression, which remains a limitation. In future studies, we will calculate the predictive cut-off for circHELQ using our institutional cohort and validate it in multicenter datasets to demonstrate the generalizability of circHELQ as a prognostic predictor. Second, whether circHELQ promotes CRC progression through interacting with RNA-binding proteins or by encoding novel proteins remains unclear. Subsequently, we plan to investigate both mechanistic possibilities to further elucidate the specific biological mechanisms by which circHELQ induces CRC progression. Additionally, the therapeutic efficacy of the biomimetic LNP-encapsulated circHELQ has only been validated in animal models, and its safety and effectiveness in patients remain unknown. Therefore, we will expand the sample size and may further validate its efficacy and safety in non-human primates.

Conclusions

CircHELQ expression is markedly elevated in CRC and is positively correlated with LNM. Mechanistically, circHELQ promotes VEGFC expression by acting as a sponge for miR-4793-3p through the miR-4793-3p/VEGFC axis, thereby accelerating LNM and CRC progression. These findings suggested that circHELQ is a promising biomarker and a potential therapeutic target for LNM in CRC. To explore this potential, we developed a biomimetic nanoparticle system coated with cell membranes and loaded with si-circHELQ for targeted CRC therapy. The results demonstrated that si-circHELQ@cMDS effectively targeted tumors and had antitumor activity. Collectively, the results of this study elucidated the functional role of circHELQ in CRC progression and offered a novel nanotherapeutic strategy for circRNAs in cancer treatment.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CRC

Colorectal cancer

OS

Overall survival

LNM

Lymph node metastasis

CircRNAs

Circular RNAs

miRNAs

microRNAs

ceRNA

Competing endogenous RNA

LNP

Lipid nanoparticle

qRT-PCR

Quantitative real-time polymerase chain reaction

CCK-8

Cell counting kit-8

FISH

Fluorescence in situ hybridization

IHC

Immunohistochemical

PDI

Polydispersity index

PDX

Patient-derived xenograft

ALT

Alanine transaminase

AST

Aspartate transaminase

BUN

Blood urea nitrogen

Author contributions

J.P: Data curation, investigation, methodology. W.S: Conceptualization, formal analysis, writing–original draft, writing–review and editing. M.L: Investigation, methodology. M.Z: Investigation, methodology. R.C: Investigation, methodology, funding acquisition. Y. L: Data curation, investigation. H.Z: Data curation, investigation. Y.L: Data curation, investigation. Z.L: Data curation, investigation. J.D: Data curation, investigation. J.L: Conceptualization, resources, formal analysis, writing–review and editing. Q.Z: Conceptualization, resources, formal analysis, funding acquisition, visualization. Z.P: Conceptualization, resources, formal analysis, supervision, funding acquisition, visualization, writing–original draft, writing–review and editing.

Funding

This work was supported by grants from the National Natural Science Foundation of China (82373350, 82573082), Science and Technology Program Guangzhou (2025A04J5520, 2024A04J4977), National key Clinical Specialty Construction Project (2022YW030009), the Excellent Young Talent Program of Guangdong Provincial People’s Hospital (KY012021186), the Talent Introduction Supporting Funds of Guangdong Provincial People’s Hospital (KY0120230009).

Data availability

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

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of our hospital and performed according to institutional guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.