CEACAM5 inhibits the lymphatic metastasis of head and neck squamous cell carcinoma by regulating epithelial–mesenchymal transition via inhibiting MDM2

Head and neck squamous cell carcinoma (HNSCC) originates from mucosal epithelial cells in the mouth, pharynx, and larynx [1] and affects approximately 600000 people per year worldwide [2,3]. Oral cavity tumors and laryngeal cancer are often associated with excessive tobacco and alcohol consumption, while there is growing evidence that pharyngeal cancer is linked to human papillomavirus (HPV) infection, mainly HPV-16 [4]. Moreover, HPV-negative and HPV-positive HNSCC differ both genetically and in clinicopathological appearance [5]. More than 50% of patients with HNSCC have locally advanced lesions at first diagnosis and between 60 and 80% of them exhibit lymph node (LN) metastasis when starting treatment [6–9]. Despite the availability of a variety of comprehensive therapeutic options, the prognosis of HNSCC patients remains poor, with a five-year survival of approximately 60% for patients with laryngeal cancer and 25% for those with hypopharyngeal cancer [10–12]. Accordingly, novel strategies are needed for the effective prevention and treatment of LN metastasis.

RNA sequencing is widely used in both cancer diagnosis and research into therapeutic strategies for this disease [13,14], including the exploration of biomarkers and the characterization of cancer heterogeneity, drug resistance, and the immune microenvironment [15–17]. Single-cell sequencing has shown that the epithelial–mesenchymal transition (EMT) signature can independently predict LN metastasis, cancer grade, and adverse pathological features [18,19]. The CEA family member 5 (CEACAM5) family of proteins includes both membrane junction and secretory glycoproteins. Membrane junction glycoproteins are either embedded within membranes via glycosylphosphatidylinositol (GPI) anchors or attached to the cell surface through transmembrane domains. GPI-anchored family members include CEACAM5, CEACAM6, CEACAM7, and CEACAM8 [20]. CEACAM5 is involved in the regulation of cell differentiation, apoptosis, and cell polarity, and can thus affect tumor development [21]. This protein can also serve as a biomarker of gastrointestinal cancer [22]. Murine double minute 2 (MDM2) is a RING-domain E3 ubiquitin ligase that, when phosphorylated, translocates to the nucleus, where it regulates the expression of a variety of genes in both tumor and normal cells [23]. Interaction between the NH2-terminal domain of MDM2 and the α-helix in the NH2-terminal domain of p53 induces the degradation of the latter, thereby promoting carcinogenesis [24].

EMT involves the acquisition of mesenchymal features by epithelial cells and is normally active during tissue development and repair, as well as stem cell functions [25]. EMT reactivation is associated with tumorigenesis, cancer cell invasion and metastasis, and drug resistance [26–28]. This transformation in function is largely regulated by the transcription factors Twist, Slug, Snail, and ZEB1. In addition, EMT can inhibit cell senescence and apoptosis and increase the tolerance to radiotherapy and chemotherapy [29,30].

In the present study, CEACAM5 expression was observed to be significantly lower in both HNSCC samples and cell lines. Importantly, we found that CEACAM5 inhibits HNSCC lymphatic metastasis by negatively regulating MDM2 expression and thereby inhibiting EMT both in vivo and in vitro. These findings provide insights into the function of CEACAM5 in HNSCC.

Fresh HNSCC and adjacent normal tissue specimens were collected from patients undergoing tumor resection in the First Affiliated Hospital of Chongqing Medical University between 2014 and 2022. The specimens included 57 paraffin-embedded tumor tissues that were used for histological analysis. Patients who had received preoperative antitumor therapy such as radiotherapy or chemotherapy were excluded. The study was carried out in accordance with the Declaration of Helsinki and was authorized by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University.

Total RNA was extracted from fresh frozen samples using an E.Z.N.A. Total RNA Kit I (Omega Bio-Tek, U.S.A.). The extracted RNA was reverse transcribed into cDNA (37°C for 15 min, 70°C for 5 min) using a PrimeScript RT Reagent Kit (Takara, Dalian, China). Quantitative-PCR (qPCR) was performed with the SYBR PrimeScript RT-PCR Kit (Takara). GAPDH served as the internal control and relative gene expression levels were calculated using the 2^−ΔΔCt method. The sequences of the primers used for qPCR are shown in Additional file 1: Supplementary Table S1.

Total protein was isolated using a protein extraction kit (KGP250, KeyGen, Jiangsu, China). The lysates were centrifuged and the protein concentration in the supernatants was measured using the BCA method (P0010S, Beyotime, Shanghai, China). Equal amounts of protein were separated via 10% SDS-PAGE and electroblotted onto polyvinylidene fluoride (PVDF) membranes. After blocking, the membranes were incubated overnight at 4°C with primary antibodies targeting CEACAM5 (Abcam, U.S.A.), MDM2 (Cell Signaling Technology, U.S.A.), E-cadherin (Cell Signaling Technology), N-cadherin (Cell Signaling Technology), vimentin (Cell Signaling Technology), LYVE-1 (Abcam), and GAPDH (Abcam). After washing, the membranes were incubated with a secondary antibody for 1 h at room temperature. An enhanced chemiluminescence (ECL) kit (Thermo, Shanghai, China) was used for visualization. Detailed antibody information is provided in Additional file 2: Supplementary Table S2.

An immunohistochemistry (IHC) kit (SP-9000, Beijing, Zhongshan Jinqiao, China) was used to stain 4-μm-thick paraffin-embedded tumor sample sections. Specifically, after dewaxing and rehydration in fresh xylene and a graded alcohol series, antigen retrieval was performed with citrate buffer at 95°C for 15 min, followed by incubation with an endogenous peroxidase blocker for 15 min at 37°C. The sections were then incubated first with primary antibodies overnight at 4°C and then with HRP-conjugated streptavidin (100 μL) for 15 min, followed by counterstaining with hematoxylin and visualization with diaminobenzidine (DAB). Different fields of view of each section were randomly imaged under a light microscope at ×200 magnification. Gray levels were estimated in ImageJ and graded as follows: 0, negative; 1–3, weak; 4–5, medium; and 6–7, strong.

FaDu and SCC15 cells were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells were grown in Dulbecco’s Modified Eagle’s Medium (DMEM; Gibco, MA, U.S.A.) containing 10% fetal bovine serum (FBS; Gibco, Australia) and 1% penicillin/streptomycin (Beyotime, Shanghai, China) at 37°C with 5% CO₂.

The siRNA used in the present study was obtained from GenePharma (Shanghai, China). Cells were seeded in 6-well plates (2.5 × 10⁵ cells per well) and allowed to grow to between 60 and 70% confluence. The cells were then transfected with siRNA using Lipofectamine iMAX (Invitrogen, U.S.A.) and incubated in serum-free DMEM for 6 h. The medium was then replaced with DMEM supplemented with 10% FBS and the cells were cultured for a further 48–72 h. RNA and protein were extracted for the assessment of transfection efficiencies as well as to select the best sequence for use in subsequent experiments.

To knock down CEACAM5, short hairpin RNA (shRNA)-targeting human CEACAM5 was cloned into the hU6-MCS-Ubiquitin-firefly_Luciferase-IRES-puromycin lentiviral vector (GV344, Genechem, Shanghai, China). For CEACAM5 overexpression, full-length CEACAM5 cDNA was cloned into the Ubi-MCS-firefly_Luciferase-IRES-Puromycin lentiviral vector (GV260, Genechem). Specifically, cells were grown in 6-well plates (1 × 10⁵ cells per well) to between 20 and 30% confluence. Lentivirus was transfected into the cells at a multiplicity of infection (MOI) of 10 using HiTransG A/P infection-enhancing solution. After 24 h, the cells were incubated in DMEM with 10% FBS and 2 μg/mL puromycin for 1 week, followed by selection.

Colony formation assays were used to assess the tumorigenic ability of the cells. Stably lentivirus-transduced FaDu and SCC15 cells were seeded in 6-well plates (1000 cells/well and 650 cells/well, respectively) in DMEM with 10% FBS and grown for approximately 2 weeks. The medium was replaced every 4 days. The cells were subsequently fixed in 4% paraformaldehyde and stained with 0.1% crystal violet for 20 min, and the number of colonies was counted.

The migratory and invasive potential of the cells was assessed using 8-μm-pore Transwell inserts (Corning, U.S.A.) coated or not with Matrigel (BD Biosciences, San Jose, CA, U.S.A.). Specifically, 1 × 10⁵ cells were suspended in 200 μL of serum-free medium and placed in the upper chamber of the inserts. The lower chamber contained 600 μL of DMEM supplemented with 15% FBS. After incubation for 24 h, cells remaining on the upper chamber of the membrane (unmigrated cells) were wiped off with a cotton swab. Cells that had crossed the membrane (migrated cells) were fixed in 4% paraformaldehyde and stained with 0.1% crystal violet for 20 min. The number of migrated cells was counted under a microscope.

For the analysis of apoptosis, an Annexin V-FITC/PI Detection Kit (SunGENE BioTEch, Tianjin, China) was used to label apoptotic cells. Cells were grown in 6-well plates (1 × 10⁵/well) to 90% confluence, harvested, and subjected to flow cytometry (FCM; BD Biosciences, U.S.A.). For the cell cycle assay, stably transfected cells were fixed in 70% ice-cold ethanol overnight at 4°C, incubated with RNase A and propidium iodide, and then analyzed in a flow cytometer.

An EdU detection kit (RiBoBio, Guangzhou, China) was used for the cell proliferation assay. Briefly, cells were grown in 6-well plates to 70–80% confluence and then incubated with EdU (50 μM) at 37°C for 2 h. The cells were subsequently fixed in 4% paraformaldehyde, stained with Apollo for 30 min, and imaged under an inverted microscope.

A total of 2.5 × 10⁵ cells were seeded per well of a 6-well plate, allowed to grow to 60–70% confluence, and then transfected using a liposome transfection kit (NEOFECTTMDNA transfection reagent, Beijing, China). The medium was replaced after 24 h and the cells were grown to approximately 95% confluence. A 200-μL pipette tip was used to make four scratches in the cell monolayer. After being allowed to grow for 24 h in serum-free DMEM, the cells were imaged under an inverted microscope.

Experiments involving mice were performed at a specific pathogen-free facility at Chongqing Medical University. Male BALB/c nude mice aged between 4 and 6 weeks were purchased from HFK Bioscience Limited Company (Beijing, China). SCC15 cells (5 × 10⁶ in 40 μL of PBS) transfected with the empty vector or the CEACAM5 overexpression plasmid were injected into the foot pads of the mice. After 35 days, the mice received intraperitoneal injections of 15 mg/mL D-Luciferin (Beyotime). Mice were monitored using in vivo bioluminescence imaging and, following euthanasia, the foot pad tumors and ipsilateral inguinal LNs were collected for further analysis.

Transcriptome sequencing was performed on 11 tumor tissues from HNSCC patients with or without LN metastasis [31]. Differentially expressed genes (DEGs) were identified using a threshold of P<0.05 and volcano plots were drawn using the Enhanced Volcano package in R. Protein–protein interaction (PPI) networks for genes related to CEACAM5 were compiled using the STRING database (https://cn.string-db.org/) [32] and visualized in Cytoscape [33]. GO enrichment analysis of the DEGs was performed through the Network Analyst database (https://www.networkanalyst.ca/) [34]. The HNSCC high-throughput sequencing datasets were downloaded from The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/) and Gene Expression Omnibus (GEO) databases (dataset GSE2379, platform GPL91).

GraphPad Prism version 8.0 (GraphPad Software, U.S.A.) and SPSS 20.0 (IBM Corp, NY, U.S.A.) were used for statistical analysis. T-tests were used for comparisons between two groups. One-way ANOVA was used for univariate comparisons among three or more groups. Multivariate between-group analysis was performed using two-way ANOVA. Pearson χ² tests were used to assess correlations between CEACAM5 expression and clinicopathological features. Kaplan–Meier and logarithmic rank analyses were used to evaluate prognostic factors. All data were expressed as means ± standard deviation (SD). A P-value <0.05 was considered significant.

To explore the role of CEACAM5 in human HNSCC, transcriptome sequencing was performed on six HNSCC tissues with LN metastases and five without. A total of 2294 DEGs between the two tissue types were identified (shown in the heatmap in Figure 1A and the volcano plot in Figure 1C). Expression profile microarray data for ten HNSCC cases with and four without LN metastases were obtained from the GEO database (depicted in the heatmap in Figure 1B). Analysis of the data indicated that CEACAM5 expression was lower in cases with LN metastases than in those without. We then conducted a pan-cancer analysis of CEACAM5 levels in 33 common human cancers together with an assessment of the disease risk ratio. We found that the levels of CEACAM5 were significantly higher in normal tissues (n=44) than in HNSCC tissues (n=520) and that the CEACAM5 risk ratio in HNSCC was less than 1 (P<0.05) (Figure 1D,E). In addition, CEACAM5 expression in HPV-positive HNSCC was higher than that in HPV-negative HNSCC (Figure 1D). These findings indicated that CEACAM5 may act as a tumor suppressor in HNSCC. The mRNA levels of CEACAM5 were markedly higher in normal tissues than in tumor tissues and were also higher at the N0 stage than at the N1 and N3 stages based on TCGA data (Figure 1F). Survival analysis based on the same dataset confirmed that high CEACAM5 levels were associated with better overall survival (OS), disease-free survival, and disease-specific survival (DSS) (Figure 1G–I). Combined, these findings suggested that CEACAM5 expression is significantly correlated with the N (LN) stage in HNSCC.

The sequencing datasets for 53 normal and 520 HNSCC tissues were downloaded from TCGA. Analysis of the data showed that CEACAM5 levels were significantly higher in normal tissue than in HNSCC tissue (Figure 2A). Next, we analyzed CEACAM5 mRNA levels in nine HNSCC specimens without and nine with LN metastases, as well as in control tissues. The results showed that CEACAM5 levels in the nonmetastasis group were higher than those in the metastasis group (Figure 2C). The CEACAM5 protein levels were consistent with the results of the mRNA analysis as determined by western blotting (WB) (Figure 2B,D) and further confirmed by IHC (Figure 2E–G). Subsequent univariate and multivariate Cox regression analyses indicated that CEACAM5 expression was a prognostic factor for HNSCC. The clinicopathological data are shown in Tables 1 and 2. Finally, survival analysis demonstrated that high CEACAM5 levels were linked with improved OS in patients with HNSCC (Figure 2H). Together, these findings indicated that CEACAM5 expression is down-regulated in HNSCC and that elevated CEACAM5 levels are associated with better clinical outcome.

Cell culture experiments were employed to assess the role of CEACAM5 in HNSCC. First, FaDu cells were transfected with three siRNAs targeting different sites in CEACAM5 (si-CEACAM5#1, si-CEACAM5#2, and si-CEACAM5#3), and the silencing efficiency of the siRNAs was assessed by qRT-PCR and WB (Figure 3A,B). Si-CEACAM5#1 displayed the best knockdown efficiency and was used for subsequent studies. Next, we established stable CEACAM5 overexpression and knockdown FaDu and SCC15 cell lines using lentiviral infection. Overexpression and knockdown efficiencies were confirmed at both the mRNA and protein levels (Figure 3C–F). EdU cell proliferation assays were performed on stably transfected HNSCC cells in the logarithmic growth phase. The results showed that the proliferative ability of CEACAM5-silenced cells (both the FaDu and SCC15 lines) was higher than that of the respective controls, whereas the opposite was observed in CEACAM5-overexpressing cells (Figure 3H,I). In addition, CEACAM5-silenced cells formed more and larger colonies compared with control cells (Figure 4A). CEACAM5-silenced cells also displayed increased migratory and invasive capacity relative to that of control cells; however, this trend was reversed in CEACAM5-overexpressing cells (Figure 4B,C). Additionally, compared with the respective controls, CEACAM5 overexpression increased the cell apoptosis rate, whereas CEACAM5 depletion exerted the opposite effect (Figure 4D). These data indicated that CEACAM5 inhibits the proliferation, survival, and motility of HNSCC cells.

To evaluate the role of CEACAM5 in lymphatic metastasis in vivo, we xenografted SCC15 cells transfected with the empty vector or the CEACAM5 overexpressing plasmid into the foot pads of nude mice. The sizes and weights of the primary tumors and inguinal metastatic LNs were smaller in the CEACAM5-overexpressing group relative to those of the empty vector group (Figure 5A–E,G–I). Additionally, CEACAM5 overexpression reduced the LN metastatic potential of SCC15 cells (Figure 5F). CEACAM5 overexpression in the tumors was confirmed by qRT-PCR and WB (Figure 5J,K). Lastly, immunohistochemical analysis of the foot pad tumors showed greater staining intensities and proportion of CEACAM5 in tumors formed of CEACAM5-overexpressing cells, in contrast with reduced staining of LYVE-1 (Figure 5L). These findings indicated that CEACAM5 inhibits lymphatic metastasis in vivo.

Numerous transcription factors have been implicated in tumorigenic processes. Among them, the EMT process is believed to be involved in tumor invasion and early metastasis. Accordingly, we used Spearman’s correlations to assess the relationships between CEACAM5 and key EMT-related transcription factors in HNSCC based on high-throughput data from TCGA. The results showed that CEACAM5 was negatively associated with these transcription factors, especially Snail, Slug, and Twist (Figure 6A). In addition, qRT-PCR analysis of the stably transduced FaDu and SCC15 cells showed that, compared with the controls, the levels of the epithelial marker E-cadherin were reduced in CEACAM5-silenced cells, whereas those of key EMT-related transcription factors and the interstitial markers N-cadherin and vimentin were increased. In contrast, in CEACAM5-overexpressing cells, the mRNA levels of E-cadherin were up-regulated, while those of Slug, N-cadherin, and vimentin were suppressed, compared with that seen in cells transfected with the empty vector (Figure 6B,C). These findings were confirmed at the protein level (Figure 6D,E). These observations implied that CEACAM5 inhibits EMT in HNSCC cells.

It has been reported that the phosphorylation of MDM2 allows its entry into the nucleus where it regulates the expression of numerous genes [35,36]. Here, using the STRING database, we constructed a PPI network based on the HNSCC transcriptome-sequencing data (Figure 7H). CEACAM5 and MDM2 were seen to share many similarities in GO term enrichment in the Biological Process category (Figure 7A,B). In addition, we analyzed the expression of CEACAM5 and MDM2 in HNSCC based on data from TCGA and found that CEACAM5 levels were significantly higher in normal tissues than in tumor tissues, whereas the MDM2 levels showed the opposite trend (Figure 7C,E). This suggested that a negative regulatory relationship may exist between CEACAM5 and MDM2 (Figure 7D). To confirm this possibility, we next assayed the stably transfected FaDu and SCC15 cell lines for MDM2 expression using qRT-PCR, with the results showing that MDM2 expression was higher in CEACAM5-knockdown cells than in control cells. In contrast, MDM2 levels were lower in CEACAM5-overexpressing cells than in cells transfected with the empty vector (Figure 7F,G). These findings were confirmed by WB (Figure 7J). Furthermore, the MDM2 mRNA levels were lower in CEACAM5-overexpressing mouse xenograft tumors compared with the empty vector group (Figure 7I). These findings confirmed the existence of a negative regulatory relationship between CEACAM5 and MDM2.

To further verify that a regulatory relationship indeed exists between CEACAM5 and MDM2, we silenced MDM2 in CEACAM5-knockdown cells. After puromycin selection for 1 week, the sh-CEACAM5-NC + sh-MDM2-NC, sh-CEACAM5-NC + sh-MDM2, sh-CEACAM5 + sh-MDM2-NC, and sh-CEACAM5 + sh-MDM2 SCC15 cell lines were constructed. We found that the mRNA level of the epithelial marker E-cadherin in sh-CEACAM5-NC + sh-MDM2 cells was increased in comparison with that in sh-CEACAM5-NC + sh-MDM2-NC cells. Moreover, the E-cadherin level was reduced in sh-CEACAM5 + sh-MDM2-NC cells compared with that in sh-CEACAM5-NC + sh-MDM2 cells and was higher in sh-CEACAM5 + sh-MDM2 cells than in sh-CEACAM5 + sh-MDM2-NC cells. Meanwhile, the mRNA expression levels of the EMT-related transcription factors as well as those of the interstitial proteins N-cadherin and vimentin displayed the opposite trend to that of E-cadherin (Figure 8A). These results were confirmed at the protein level (Figure 8B). Thus, the EMT process in cells in which both CEACAM5 and MDM2 were silenced was reversed in comparison with that in cells where only CEACAM5 was silenced.

Cell cycle assays subsequently showed that the proportion of sh-CEACAM5-NC + sh-MDM2 cells in the S phase was reduced in comparison with that of sh-CEACAM5-NC + sh-MDM2-NC cells. Additionally, the proportion of sh-CEACAM5 + sh-MDM2-NC cells in the S phase was higher than that for sh-CEACAM5-NC + sh-MDM2 cells and was lower in sh-CEACAM5 + sh-MDM2 cells than in sh-CEACAM5 + sh-MDM2-NC cells. Finally, wound-healing assays showed that the migratory ability of sh-CEACAM5-NC + sh-MDM2 cells was reduced compared with that of sh-CEACAM5-NC + sh-MDM2-NC cells. Meanwhile, the migratory potential of sh-CEACAM5 + sh-MDM2-NC cells was greater than that of sh-CEACAM5-NC + sh-MDM2 cells and was lower in sh-CEACAM5 + sh-MDM2 cells than in sh-CEACAM5 + sh-MDM2-NC cells. In conclusion, the depletion of both MDM2 and CEACAM5 counteracted the inhibitory effect of CEACAM5 silencing on the EMT, resulting in EMT inhibition in HNSCC cells.

Cancer-related morbidity and mortality are increasing rapidly worldwide [37,38]. While excessive smoking is associated with an increased incidence of HNSCC in developing countries, HPV infection is increasingly recognized as an important factor contributing to the increased incidence of oropharyngeal tumors in nonsmokers in developed countries [39]. The therapeutic modality of HPV-positive and HPV-negative HNSCC is essentially similar, including traditional surgical and chemoradiotherapy modalities, as well as new surgical techniques such as transoral robotic surgery and laser microsurgery [40]. The specific treatment plan depends on the site and stage of tumor development. OS for patients with HNSCC remains poor despite the advances in its diagnosis and treatment [41], while LN metastasis and extranodal extension are both important predictors of worse prognosis in HNSCC [42–44]. These observations underline the urgent need for further exploration of the mechanisms underlying HNSCC occurrence and progression, as well as identify novel targets and prognostic markers and develop effective strategies for the diagnosis and treatment of this cancer.

CEA was originally isolated from the blood of patients with colorectal cancer by Gold and Freedman [45]. The expression of this protein has since been observed to be up-regulated in many other human tumors, including pancreatic, lung, liver, gallbladder, breast, and bladder cancers, among others [46–48]. Accordingly, studies have mostly focused on its role as an oncogene in promoting cancer progression. In the present study, transcriptome sequencing in HNSCC tissues from patients with and without LN metastasis and analysis of high-throughput data downloaded from TCGA and GEO databases indicated that CEACAM5 expression was significantly reduced in HNSCC tumors compared with that in normal tissues. Generally, HPV-positive cases are more sensitive to chemoradiotherapy and, thus, have a better prognosis and a higher survival rate compared with HPV-negative cases [49]. However, in the present study, CEACAM5 expression was found to be higher in HPV-positive HNSCC than in HPV-negative HNSCC, and its expression level was also significantly lower in HNSCC tissues than in normal tissues, with a disease risk ratio of less than 1. This suggests that alterations in CEACAM5 expression are more likely to be a protective manifestation against this disease. Subsequently, survival analysis confirmed that elevated CEACAM5 expression was linked with improved OS, progression-free survival, DSS, and lower N-stage classification in patients with HNSCC.

p53 has been described as the guardian of the genome and controls multiple cellular processes, including DNA repair, cell-cycle arrest, apoptosis, autophagy, senescence, and metabolism through numerous mechanisms [50,51]. Ubiquitination is a reversible post-translational modification that modulates protein degradation and signal transduction through the binding of ubiquitin to protein substrates. This process mainly involves the E1-E2-E3 enzyme cascade and requires the activity of E1 ubiquitin-activating enzymes, E2 ubiquitin-conjugating enzymes, and E3 ubiquitin ligases. In addition, the regulation of NF-κB signaling by ubiquitination may be a potential therapeutic target for HNSCC [52]. MDM2 is an E3 ubiquitin ligase that can modulate p53 levels by promoting its degradation by the proteasome [53,54] or by binding to and stabilizing Slug mRNA independently of its ubiquitination activity, thereby stimulating cell invasion and metastasis [55–57]. Here, by analyzing transcriptome sequencing data to obtain DEGs and employing the STRING database to construct a PPI network, we found that CEACAM5 and MDM2 may have a negative regulatory relationship. Through the Network Analyst database, we assigned GO terms to genes associated with CEACAM5 and MDM2 expression and found similar enrichment patterns. In addition, analysis of TCGA dataset indicated that CEACAM5 and MDM2 expression exhibited opposing trends in HNSCC. Combined, these data suggested that CEACAM5 may regulate EMT processes by suppressing the expression of MDM2.

Malignant tumors induce lymphatic dilation (lymphangiogenesis) in the primary tumor and the drainage sentinel area by releasing lymphatic growth factors such as VEGF-C and LYVE1, thereby promoting LN metastasis [58–61]. Moreover, EMT is known to drive the metastasis of cervical cancer cells to LNs [62]. Here, we investigated the effect of CEACAM5 on LN metastasis in vivo using bioluminescence imaging and used IHC to assess the expression of LYVE1 in foot pad tumors derived from xenografted SCC15 cells in mice. The obtained results indicated that CEACAM5 overexpression abrogated the lymphatic metastasis of SCC15 cells. Despite our important findings, our study had several limitations. First, the number of samples used for transcriptomic analysis was small. Second, although our study indicated that CEACAM5 inhibits MDM2 expression in HNSCC, further investigation is required to clarify the specific underlying mechanisms.

In conclusion, this is the first study to show that CEACAM5 expression is down-regulated in HNSCC and that CEACAM5 expression is correlated with survival outcomes in HNSCC patients. CEACAM5 was also found to inhibit EMT in HNSCC cells by suppressing the expression of MDM2 and, ultimately, also LN metastasis. These findings contribute to the understanding of the molecular mechanism involved in HNSCC development.

• Background: OS in patients with HNSCC remains poor despite the advances in its diagnosis and treatment. Additionally, both LN metastasis and extranodal extension are important predictors of the prognosis of this cancer.

• Results: Using in vivo and in vitro experiments as well as bioinformatics analysis, we found that CEACAM5 suppresses LN metastasis by suppressing EMT via inhibiting the expression of its downstream target MDM2.

• Significance: This is the first study to suggest that CEACAM5 may play a role in inhibiting LN metastasis and also provides a potential therapeutic target and prognostic marker for HNSCC.

All data generated or analyzed during the present study are included in this published article.

The authors declare that there are no competing interests associated with the manuscript.

The present study was supported by the (National Natural Science Foundation of China) [grant number 82173303] and the (China Postdoctoral Science Foundation) [grant number 2021XM3029].

Xudong Wang: Resources, Data curation, Software, Investigation, Visualization, Methodology, Writing—original draft. Yanshi Li: Software, Supervision. Min Pan: Validation, Writing—review & editing. Tao Lu: Software, Methodology. Min Wang: Formal Analysis. Zhihai Wang: Software. Chuan Liu: Methodology. Guohua Hu: Resources, Supervision, Project administration, Writing—review & editing.

Not applicable.

Clinical specimens were obtained from patients undergoing hypophargiectomy at the First Affiliated Hospital of Chongqing Medical University. Informed consent was obtained from each patient. Meanwhile, all mice were suffocated with 1% sodium pentobarbital plus a high concentration of carbon dioxide. All experiments in the present study were conducted in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). All protocols for clinical specimens and in vivo experiments in mice were approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (2022-K184), and were conducted in accordance with the UK Animals (Scientific Procedures) Act and the National Institutes of Health guidelines.

Thanks to Molecular Medicine and Cancer Research Center of Chongqing Medical University for providing experimental equipment and experimental platform.

BP

biological process

ANOVA

analysis of variance

PBS

phosphate buffered saline

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

CEACAM5

carcinoembryonic antigen-related cell adhesion molecule 5

DEG

differentially expressed gene

EMT

epithelial–mesenchymal transition

FBS

fetal bovine serum

GEO

Gene Expression Omnibus

GPI

glycosylphosphatidylinositol

HNSCC

head and neck squamous cell carcinoma

HPV

human papillomavirus

IHC

immunohistochemistry

L

lymphatic metastasis

LM (+)

lymphatic metastasis

LM (−)

non-lymphatic metastasis

LN

lymph node

LYVE-1

lymphatic vessel endothelial hyaluronan receptor 1

MDM2

murine double minute 2

MOI

multiplicity of infection

NAT

normal adjacent tissue

NC

negative control

NL

no lymphatic metastasis

PPI

protein–protein interaction

qRT-PCR

quantitative reverse transcription-PCR

shRNA

short hairpin RNA

siRNA

small interfering RNA

T

tumor tissue

TCGA

The Cancer Genome Atlas

WB

western blotting