Cuproptosis-related LINC02454 as a biomarker for laryngeal squamous cell carcinoma based on a novel risk model and in vitro and in vivo analyses

Laryngeal squamous cell carcinomas (LSCCs) are aggressive tumors with the second-highest morbidity rate in patients with head and neck squamous cell carcinoma. Cuproptosis is a type of programmed cell death that impacts tumor malignancy and progression. The purpose of this study was to investigate the relationship between cuproptosis-related long non-coding RNAs (crlncRNAs) and the tumor immune microenvironment and chemotherapeutic drug sensitivity in LSCC, and crlncRNA impact on LSCC malignancy. Clinical and RNA-sequencing data from patients with LSCC were retrieved from the Cancer Genome Atlas. Differentially expressed prognosis-related crlncRNAs were identified based on univariate Cox regression analysis, a crlncRNA signature for LSCC was developed and validated using LASSO Cox regression. Finally, the effect of LINC02454, the core signature crlncRNA, on LSCC malignancy progression was evaluated in vitro and in vivo. We identified a four-crlncRNA signature (LINC02454, AC026310.1, AC090517.2, and AC000123.1), according to which we divided the patients into high- and low-risk groups. The crlncRNA signature risk score was an independent prognostic indicator for overall and progression-free survival, and displayed high predictive accuracy. Patients with a higher abundance of infiltrating dendritic cells, M0 macrophages, and neutrophils had worse prognoses and those in the high-risk group were highly sensitive to multiple chemotherapeutic drugs. Knockdown of LINC02454 caused tumor suppression, via cuproptosis induction. A novel signature of four crlncRNAs was found to be highly accurate as a risk prediction model for patients with LSCC and to have potential for improving the diagnosis, prognosis, and treatment of LSCC.


Introduction
Laryngeal carcinomas are among the most common head and neck carcinomas, and approximately 95% are squamous cell carcinomas (Leemans et al. 2011).Given that many laryngeal squamous cell carcinomas (LSCCs) have highly heterogeneous pathogeneses, most patients diagnosed with laryngeal cancer are already at an advanced locoregional stage (Steuer et al. 2017).The survival rate of patients with advanced laryngeal cancer is 29%-56%, compared to 56%-93% of patients with early-stage laryngeal cancer (Haws and Haws 2014).Although treatment and diagnosis have greatly improved, patients with advanced laryngeal cancer continue to have a poor prognosis due to lymph node metastases (Wu et al. 2020).Increasing the early diagnosis rate of laryngeal cancer, with subsequent access to prevention and intervention strategies, is crucial to improving clinical outcomes.Thus, there is a great need for the identification of sensitive biomarkers for early diagnosis and efficient treatment of LSCC.
Copper (Cu) is an indispensable trace element in the body with important roles on myriad biological processes, including cell proliferation, angiogenesis, and metastasis of malignant tumors (Bui et al. 2020).It also plays a central role in cuproptosis, a novel type of cell death that has distinct morphology, biochemistry, and regulatory mechanisms, compared to other cell death pathways, including necroptosis, pyroptosis, apoptosis, autophagy, and ferroptosis (Tsvetkov et al. 2022).Cuproptosis is inherently reliant on mitochondrial metabolism and the tricarboxylic acid cycle (Das et al. 2022;Tang et al. 2022).Upon mitochondrial dysfunction, the sensitivity of cells to Cu 2+ increases.These ions combine with the lipoylated components of the tricarboxylic acid cycle, leading to abnormal accumulation of lipoylated proteins and depletion of iron-sulfur cluster proteins (Tsvetkov et al. 2022).The resulting disturbances in protein expression induce proteotoxic stress and, ultimately, cell death.Recent research suggests that Cu 2+ levels in patients with cancer differ markedly from those in healthy individuals, and it is known that abnormal Cu accumulation may activate malignant processes in cancer cells that lead to tumorigenesis and tumor development (Ge et al. 2022).Thus, the exploration of cuproptosis-related long non-coding RNAs (crlncRNAs) may provide novel targets for the clinical diagnosis and prognosis of LSCC.
LncRNAs are a type of RNA that exceeds 200 nucleotides in size (Lee et al. 2016).These RNA molecules are non-protein-coding and exert epigenetic, transcriptional, and post-translational regulation functions that greatly impact various cellular processes (Liu et al. 2017), including cell differentiation, proliferation, metabolism, and apoptosis, as well as tumorigenesis and metastasis.Recently, lncRNAs have been recognized as diagnostic biomarkers for multiple tumors; thus, their association with cancer progression may help improve the understanding of tumor pathophysiology and advance the development of novel cancer therapeutics (Slack and Chinnaiyan 2019;Eptaminitaki et al. 2021).Currently, there is limited understanding of the role of crlncRNAs in LSCC, their prognostic value, and their relationship with the tumor microenvironment (TME).Therefore, the objectives of the present study were to identify crlncRNAs associated with LSCC, to use these to construct a risk signature, and to investigate the relationship of the signature with the clinical outcome, clinical features, TME, and antitumor therapy.Collectively, our findings establish a reliable crl-ncRNA predictive signature that provides new perspectives for the clinical treatment of LSCC.

Collection and processing of clinical and transcriptomic data
Clinicopathological and RNA-sequencing data for patients with LSCC (n = 111) were obtained from the Cancer Genome Atlas (TCGA).Clinical features included age, sex, TNM stage, survival period, and survival status.Perl was used to convert Ensembl IDs to gene symbols in the expression matrix profile.Using the R package "caret" in a 1:1 ratio, 111 patients with LSCC were randomized into training (n = 56) and test (n = 55) groups.The lncRNA expression matrices of cancer cell lines were acquired from the CCLE dataset (https:// porta ls.broad insti tute.org/ ccle).The R package (v4.2.1) "ggplot2" was used for analysis.

Establishment of a prognostic crlncRNA signature
Prognosis-related crlncRNAs (n = 30) were identified in differentially expressed crlncRNAs (n = 519) using univariate Cox regression analysis.The prognostic crlncRNA signature was then obtained via LASSO regression analysis in the R "glmnet" package.The risk score was calculated using the following equation: where n is the number of prognosis-related lncRNAs in the risk model, coef (crlncRNA) is the regression coefficient of the risk model in multivariate Cox regression, and expr (crlncRNA) is the expression level of risk-related crlncRNA.

Validation of the crlncRNA prognostic signature
Kaplan-Meier (KM) plots and receiver operating characteristic (ROC) curves were used to evaluate the value of the risk prediction model by using the R packages "survival," "survminer," and "ggplot2."To confirm the reliability of the crlncRNA signature, a principal component analysis (PCA) was performed through the "scatterplot3D" R package.The resulting C-index was used to compare the predictive ability of the crlncRNAs with that of other clinical characteristics through the R packages "rms," "dplyr," "survival," and "pec."The signature was then validated using test data.

Establishment of a risk score-based nomogram
Based on clinical information and risk scores, a nomogram was developed for forecasting 1-, 3-, and 5-year overall survival (OS) in patients with LSCC using the R packages "survival" and "rms."Subsequently, with the help of calibration curves, the predicted survival rates were compared against observed survival rates to determine the precision of the model.Furthermore, decision curve analysis with the Cox regression results was performed using the "DecisionCurve" function.

Enrichment analysis
For Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of differentially expressed genes (DEGs), we utilized the R package "clusterProfiler."GO analysis focused on biological process, molecular function, and cellular component, while the KEGG analysis focused on chemical, genomic, and system functional information.An enrichment analysis of gene sets was implemented using gene set enrichment analysis (GSEA) software (version 4.2.3), and graphs were produced using the R packages "org.Hs.eg.db," "clusterProfiler," "DOSE," and "enrichplot."

Assessment of the tumor immune landscape by the prognostic crlncRNA signature
We assessed immune infiltration levels in LSCC using eight methods, namely ssGSEA xCell, Tumor Immune Estimation Resource, quanTIseq, Microenvironment Cell Population counter, Estimating the Proportions of Immune and Cancer cells, CIBERSORT, and CIBERSORT-ABS.The "pheatmap" and "ggcorrplot" packages were used to generate expression and correlation heatmaps, respectively.Subsequently, a comparison was made between the high-and low-risk groups regarding immune checkpoint-and m6Arelated genes using the "limma" and "ggplot2" packages.

Antitumor drug sensitivity analysis
RNA-sequencing expression profiles and corresponding clinical information for patients with LSCC were downloaded from the TCGA dataset (https:// portal.gdc.com).We predicted the chemotherapeutic response for each sample based on the largest publicly available pharmacogenomics database [the Genomics of Drug Sensitivity in Cancer (GDSC), https:// www.cance rrxge ne.org/].The prediction process was implemented using the R package "pRRophetic," including "car," "ridge," "preprocessCore," "genefifilter," and "sva."The samples' half-maximal inhibitory concentration (IC50) was estimated by ridge regression.Differences in the values between the two groups were evaluated using Wilcoxon signed rank test.Pearson correlation analysis was used to calculate the correlation between the risk scores and IC50 values of common antineoplastic agents.All parameters were set as the default values.The batch effect of combat and tissue type of all tissues and the duplicate gene expression were summarized as mean values.All the analyses were performed using R, version 4.2.1.

Consensus clustering analysis
To explore the features of crlncRNAs, the "ConsensusClus-terPlus" R package was utilized to classify patients with LSCC into two clusters.The KM plots were used to analyze progression-free survival (PFS) and OS in the clusters.Differences in immune infiltration between the two clusters were evaluated using the ssGSEA algorithm and Sankey analysis used to visualize the proportional changes in the high-risk patients.

Clone formation tests
TU686 and TU212 cells were transfected in a sixwell plate (Corning, NY, USA) with negative control and targeted siRNA before performing the colony formation assay (2 × 10 5 cells/well).Transfected cells (1 × 10 3 ) were implanted in a six-well plate and cultured for 2 weeks.Subsequently, 0.01% crystal violet was added for 10 min to stain the colonies.

EdU cell proliferation assay
TU686 and TU212 cells (5 × 10 4 ) were plated in glassbottomed 15-mm dishes (NEST, San Diego, CA, USA, #801002) and transfected with NC or LINC02454 siRNA, respectively.The EdU staining test was performed using an Edu staining kit (Beyotime, Shanghai, China, #C0078S).The EdU solution was added to the cells and incubated for 2 h and then fixed with 4% paraformaldehyde for 20 min.The staining process was performed according to the manufacturer's instructions.Images were captured using a fluorescence microscope (Olympus, Tokyo, Japan).

Spheroid morphology and cell viability assay
TU686 cells (1 × 10 4 ) were seeded into Corning 96-well spheroid microplates (Corning, #4515).The cells were observed for spheroid morphology at 2-day intervals and, after 7 days, were stained using the LIVE/DEAD® Viability/ Cytotoxicity Kit (KeyGEN, Nanjing, China, #KGAF001) following the manufacturer's instructions, after which the cells were gently washed with PBS (Gibco).A working solution containing calcein-acetoxymethyl ester (2 µM) and propidium iodide (8 µM) was prepared in PBS and added to the microplates.The spheroid cells were imaged with a fluorescence microscope (Olympus, Tokyo, Japan).

Cell counting kit-8 (CCK8) assay
Cell viability was monitored using a CCK8 kit (Yeasen, #40203ES60), according to the manufacturer's instructions.TU686 and TU212 cells were transfected for 48 h with NC or LINC02454 siRNAs in a six-well plate.Next, 3 × 10 3 cells were seeded in triplicate in 96-well plates and cultured in complete DMEM/high glucose medium.After 24, 48, and 72 h, cells were incubated with 10% CCK-8 solution at 37 °C for 1 h.Absorbance at 450 nm was measured using a microplate reader (BioTek /Agilent, Winooski, VT, USA, H1).Cells were co-cultured with elesclomol (ES; GLPBIO, Montclair, CA, USA #GC13885) and CuCl 2 (Sigma-Aldrich, St. Louis, MO, USA, #751944) for 24 h at a concentration of 20 nM each.The CCK-8 assay was conducted at 6 h intervals to measure cell viability.

Migration and invasion assays
To assess the migration abilities of TU686 and TU212 cells, Transwell chambers with 8-mm pores (Corning, #3422) and 24-well plates were used.Cells (2 × 10 4 cells) in 200 µL of serum-free medium were added to the Transwell upper chamber.The lower chambers were filled with complete medium containing 10% fetal bovine serum.After incubation at 37 °C for 24 h, cells from the upper side of the membrane were washed and removed, and those that had migrated through the membrane and attached to its underside were treated with 4% paraformaldehyde for 20 min and stained with 10% crystal violet (Zhu et al. 2021).For invasion assays, Matrigel (Corning, #354234) was implanted in the upper chambers 24 h before the assay.Digital images of the membranes were obtained, and counting was performed in three randomly chosen fields.

In vivo xenografts
Zebrafish xenograft models were established as previously described (Zhu et al. 2021).Zebrafish were maintained and bred at the ENT Research Center, Air Force Military Medical University, Shangxi, China (Permission code IACUC-20220908).TU686 cells were washed twice with PBS and labeled with DiI (Beyotime, #C1036) at a concentration of 5 g/mL.After 48 h of fertilization, zebrafish were anesthetized with 1.2-mM tricaine and placed in a modified agarose gel mold.Tumor cells in DMEM medium (300 cells/5 nL) were then injected into the perivitelline cavity of zebrafish embryos.
Male BABL/c nude mice (Bui et al. 2020) were purchased from Beijing Weitong Lihua Experimental Animal Technology Co., Ltd.TU686 cells (1 × 10 7 ) that were stably transfected with Lv-NC and Lv-sh-LNC02454 with cellular luciferase expression were re-suspended in 1 mL of basic medium, and 2 × 10 6 cells were subcutaneously injected into the right axilla of nude mice.d-Luciferin was injected intraperitoneally into mice to examine tumor growth over time (XenoLight D-Luciferin-NA + Salt, GLPBIO, #GC43497, 150 mg Luciferin/kg).The tumor volume (mm 3 ) was measured using an in vivo imaging system (IVIS, Biolight, Guangzhou, China) and vernier caliper.The tumor volumes were calculated using the following formula: tumor volume (mm 3 ) = (tumor width) 2 × length/2.We euthanized the mice 18 days after tumor implantation and removed, fixed, weighed, photographed, and preserved the xenografts.

Hematoxylin-eosin (HE) staining
At day 5 post-injection, the zebrafish xenograft specimens were fixed with neutral formalin, embedded in paraffin, and sectioned to a thickness of 10 μm.Sections were stained with HE via standard methods and visualized under an optical microscope.

Detection of mitochondria and reactive oxygen species (ROS)
Mito-Tracker Green FM (Invitrogen, #M7514) is a green fluorescent dye that specifically stains stable mitochondria, independent of the mitochondrial membrane potential.Mito-Tracker Red CMXRos probe (Invitrogen, #7512) is a red fluorescent dye that passively diffuses through the cell membrane and assembles directly in active mitochondria.The accumulation of this dye depends on the mitochondrial membrane potential.In the logarithmic growth phase, TU686 and TU212 cells (1 × 10 5 ) were seeded overnight in confocal culture dishes.The cells were washed three times with serum-free medium, and then treated with a mixture of 200 nM preheated Mito-Tracker Green FM and 100 nM preheated Mito-Tracker Red CMXRos.The cells were then incubated in the dark, at 37°C, for 20 min and then stained with Hoechst 33342 (1 μg/mL), for nuclear counterstaining, for 10 min.Finally, the cells were washed three times with PBS and fresh medium was added.The mitochondrial staining was imaged in a fluorescence microscope.
Another group of logarithmic growth phase cells was seeded at a density of 1 × 10 5 cells per well in confocal culture dishes and incubated overnight.After treatment, the culture medium was removed, and the cells were washed three times with Hanks' Balanced Salt solution (HBSS).Then, 1 mL of 200 nM MitoSOX (Invitrogen, #M36008) solution was added, and the cells were incubated at 37 °C for 10 min.After washing three times with HBSS, the cells were stained with 1 μg/mL Hoechst 33342 for 10 min and then washed three times with HBSS.A confocal fluorescence microscope was used to image the cells.

Statistical analysis
Statistical analysis was performed using Student's t-test and one-way ANOVA with GraphPad Prism software (version 9.4.1;San Diego, CA, USA).In this study, p-values less than 0.05 were considered statistically significant.

Identification of crlncRNAs with prognostic value in LSCC
We first obtained the expression profiles of 19 CRGs and 16,876 lncRNAs from the TCGA database.According to Pearson's correlation analysis, we obtained 900 crl-ncRNAs (R 2 > 0.4 and p < 0.05; Fig. 1a and Table S3).Subsequently, crlncRNA differential expression analysis between normal and tumor samples identified 519 training sets (e), testing sets (f).The ROC curves for survival prediction in the entire sets (g), training sets (h), testing sets (i).PCA results show an important distribution difference between high-and low-risk groups in all datasets (j), training datasets (k), testing datasets (l).Risk score of each patient in all datasets (m), training datasets (n), testing datasets (o).Survival status of each patient in the whole datasets (p), training datasets (q), testing datasets.(r).The expression patterns of four risk crlncRNAs in all datasets (s), training datasets (t), testing datasets (u)   S4).We then identified 30 prognosis-related lncRNAs via univariate Cox analysis (Fig. 1c, d); the correlation analysis of which is shown in Fig. 1e.

Construction and assessment of the crlncRNA prognostic signature
The 30 crlncRNAs were further investigated for their prognostic significance: first, LASSO regression analysis reduced the number of crlncRNAs to seven (Fig. 2a, b); then, multivariate Cox analysis produced a prognostic signature comprising four lncRNAs (Fig. 2c).Signature risk scores for each patient with LSCC were calculated using the following equation: risk score = (0.7808 × LINC02454) + (0.579 5 × AC026310.1)+ (0.9217 × AC090517.2) + (− 0.9285 × AC000123.1).Patients with LSCC were divided into highand low-risk groups based on their median risk scores.To validate the predictive accuracy of the prognostic signature, 111 patients were randomly divided into training and test groups (clinical information is presented in Table S5).Next, to assess the OS for each patient group, KM survival curves were generated and revealed that the high-risk group displayed an increased rate of mortality compared with that in the low-risk group (Fig. 2d-f).To further evaluate the reliability of the crlncRNA prognostic signature, area under curve (AUC) graphs for the entire, the training, and the test groups were calculated and found to be 0.794, 0.831, and 0.734, respectively, at 3 years (Fig. 2g-i).Additionally, PCA analysis of the LSCC cohort confirmed that high-and lowrisk patients are unequally distributed in the TCGA dataset (Fig. 2j-l).The risk score (Fig. 2m-o) and survival status (Fig. 4p-r) of the high-and low-risk patients showed equal distribution in the training and test datasets.At higher risk scores, patient survival times decreased, and the number of deaths increased.Figure 2s-u shows the expression patterns of four risk-related crlncRNAs in the high-and low-risk groups.Hence, we concluded that patients with LSCC could be discriminated based on risk level using the crlncRNA signature.

Independence of the crlncRNA signature in predicting OS and PFS
To further confirm the prognostic ability of the crlncRNA signature, we performed univariate and multivariate Cox regression analysis and found that the signature exhibited independent prognostic significance for both OS and PFS in patients with LSCC (Fig. 3a-d).The AUC of the signature for the 3-year OS was 0.794 (Fig. 3e), which was significantly higher than those of other clinical parameters.Furthermore, the C-index indicated that crlncRNA prognostic signature had a higher predictive power than the other clinicopathological characteristics (Fig. 3f).Meanwhile, the AUCs of the signature for 1-, 2-, and 3-year PFS were 0.724, 0.775, and 0.731, respectively (Fig. 3g), whereas that for 3-year PFS patients with T-stage tumors was only 0.567 (Fig. 3h).In addition, there was a statistically significant difference between the two groups of patients with LSCC in terms of PFS (p = 0.013; Fig. 3k).
To determine the contributions of individual prognostic factors to the total risk score, a nomogram was created (Fig. 3i).An analysis of calibration was performed to identify the discriminating power and clinical practicability of the nomogram, and the plots showed that the nomogrampredicted OS was highly proportional to the observed OS (Fig. 3j).In terms of prediction accuracy, the ROC and decision curves showed that the nomogram had a high predictive ability (Fig. 3l, m) and that it was more accurate than any single clinical feature, which may be beneficial for clinical decision-making and treatment options.
To examine the potential clinical utility of the crlncRNA signature, patients were stratified based on different clinical characteristics (Fig. 3n).The resulting histogram and boxplot suggested that the proportion of patients at the T4 stage was higher in the high-risk group (Fig. 3o), and the risk score was considerably higher in the T4 stage than in the T1-3 stages (Fig. 3p).

Functional analysis
We conducted GO and KEGG enrichment analyses on 627 selected DEGs between the high-and low-risk groups (Table S6).The GO enrichment analysis revealed that the DEGs were predominantly enriched in immunoglobulin receptor binding, immunoglobulin complexes, antigen binding, and extracellular matrix structural constituents (Fig. 4a,  c, and Table S7).The KEGG pathway analysis suggested that these genes were mainly enriched in proteoglycans in tumors, Wnt signaling pathway, relaxin signaling pathway, focal adhesion, protein digestion and absorption, and neutrophil extracellular trap formation (Fig. 4b, d, and Table S8).We then performed GSEA to assess the differences in potential pathways between the two groups based on the prognostic signature.Changes in the Hedgehog signaling pathway, extracellular matrix-receptor interaction, basal cell carcinoma, and focal adhesion were enriched in high-risk patients; conversely, in low-risk patients the enriched genes were associated with allograft rejection and primary immunodeficiency (Fig. 4e, f).

Differences in the immune landscapes and targeted drug sensitivity
To better understand the impact of tumor immune regulation in LSCC, we used CIBERSORT and ssGSEA algorithms to probe for different patterns of immune cell infiltration in the two risk groups.The immune infiltration for each patient was computed using the ssGSEA algorithm and presented as a heatmap (Fig. 5a).Antigen-presenting cell co-stimulation, inflammation promotion, T-cell co-stimulation, and follicular T helper (Tfh) and type-2T helper (Th2) cells greatly differed between groups (Fig. 5b); the correlation between these and the risk scores is shown in Fig. 5c.Using the CIBERSORT algorithm, we found comprehensive differences in immune cell abundance between the different subgroups (Fig. 5d) and determined the proportion of different cellular components during the immune response (Fig. 5e).Furthermore, we investigated correlations between risk scores and immune cell types using several platforms (xCell, Tumor Immune Estimation Resource, quanTIseq, Microenvironment Cell Population counter, Estimating the Proportions of Immune and Cancer cells, CIBERSORT, and CIBERSORT − ABS).As shown in the bubble chart in Fig. 5f, B cells, CD8 + T cells, Tfh cells, and neutrophils were negatively correlated with the risk score, whereas cancer-associated fibroblasts, naive CD4 + T cells, and M0 macrophages were positively correlated.Moreover, certain effective checkpoint immunotherapy targets, such as CD276 and CD70, showed increased expression in the highrisk group (Fig. 5g); the expression of specific m6A-related genes (FTO and HNRNPC) greatly differed between the two groups (Fig. 5h).
Next, we examined the relationship between the immune TME and LSCC prognosis using KM analysis.Patients with a higher expression of CD8 + T cells, checkpoint, cytolytic activity, immature dendritic cells (DCs), pro-inflammatory responses, natural killer (NK) cells, T cell co-stimulation, Tfh and Th1 cells, and tumor-infiltrating lymphocytes had a better prognosis (p < 0.01; Figure S1 A-J).In addition, patients with a higher abundance of naive B cells, resting memory CD4 + T cells, resting dendritic cells, and plasma cells had a better prognosis (p < 0.01; Figure S2 A-E), whereas those with a higher infiltration of activated DCs, M0 macrophages, neutrophils, and resting NK cells had a worse prognosis (p < 0.01; Figure S2 F-H).These data show that immune responses differ between the patient groups and may provide guidance for LSCC immunotherapy.
To forecast the responses to immune checkpoint blockade therapy in patients with LSCC, the TIDE algorithm was applied.A higher exclusion (TIDE) score was detected among patients in the high-risk group (Figure S3), which is indicative of a weaker response to immunotherapy (Dumbrava et al. 2018).The IC50 levels of several antitumor drugs were then investigated, with 20 chemotherapy drugs showing obvious differences between the high-and low-risk groups in LSCC (p < 0.05; Figure S4 A-T).In particular, ponatinib, midostaurin, and thapsigargin were identified as potential therapies for treating high-risk patients with LSCC.We investigated the correlation coefficient between risk score and the IC50 levels of 20 common chemotherapy drugs (Figure S5 A-T); based on these findings, we concluded that the crlncRNA signature may predict therapeutic drug responses.

Consensus clustering of crlncRNAs in molecular subtypes of LSCC
Based on the expression of the OS-associated crlncR-NAs, consensus clustering was applied to investigate the molecular subtypes of LSCC.The patients with LSCC were divided into two subtypes (cluster 1, n = 74; cluster 2, n = 37) via consensus clustering, with K = 2 as the optimal value (Fig. 6a-c and Table S9).To determine the stability and reliability of the molecular subtype, KM curves were generated to confirm differences in survival between the two subtypes and revealed that cluster 2 had a worse survival (OS and PFS) outcome compared with cluster 1 (p < 0.05; Fig. 6d,  f).A heatmap, calculated using the ssGSEA algorithms, was generated to show the distribution of the immune cell infiltration in the two clusters (Fig. 6e).We found that B cells, CCR (Chemokine Receptor), immature DCs, Th1 cells, neutrophils, NK cells, and Th2 cells greatly differed in the two clusters (Fig. 6g), as did the risk scores (Fig. 6h), indicating that cluster 2 samples primarily belong to the high-risk group of patients (Fig. 6i).

Evaluation of the relationship between the four crlncRNAs and LSCC prognosis
To elucidate the underlying mechanisms of LSCC malignant progression, we explored the expression of the four risk-related crlncRNAs and their relationship with LSCC prognosis.The expression correlation between the four ).e Heatmap showed the distribution of the immune cell infiltration landscape in the two clusters.f The KM curves showed the difference in PFS between the two clusters (p = 0.006).g The abundance of immune infiltrates differed between the two clusters.h The risk scores were notably different between clusters 1 and 2. i The Sankey plot showed a higher proportion of high-risk patients in clusters 2 ◂ crlncRNAs and 19 CRGs is shown in Fig. 7a.LINC02454, AC026310.1,and AC090517.2were highly expressed in the high-risk group (Fig. 7b) and positively correlated with the crlncRNA risk score (Fig. 7e and Figure S6).Paired and unpaired differential expression analyses were performed by comparing the normal and LSCC tumor groups, revealing that the four risk-related crlncRNAs were highly expressed in the patients with tumors (Fig. 7c, d).After addition of other clinical features, univariate and multivariate Cox analyses identified only LINC02454 as being related to the prognosis of patients with LSCC (p < 0.05 Figure S7A-D and Fig. 7g-j).KM analysis revealed that the OS of patients with high expression of LINC02454, AC026310.1,and AC090517.2was significantly reduced, and that prognosis of patients with LSCC who had high expression of AC000123.1 was improved (Figure S7E-H).Subsequently, we evaluated the expression of LINC02454 in the CELL dataset; we found that this crlncRNA was detected in most head and neck cancer cell lines and was especially highly expressed in squamous cell lines (Figure S8).Moreover, ROC analysis showed that LINC02454 was a more accurate prognostic indicator than other risk-related crlncRNAs at 3 years (Fig. 7f).Therefore, LINC02454 is a core gene of the crlncRNA prognostic signature and may play a considerable role in the malignant progression of LSCC.

Knockdown of LINC02454 inhibited LSCC cell proliferation and migration
To explore the potential biological functions of LINC02454 in LSCC, we first validated the expression levels of LINC02454 in LSCC cell lines (TU686, TU212) and HaCaT cells (Fig. 8a).The results showed that LINC02454 was highly expressed in LSCC cell lines.Subsequently, we established a LINC02454-knockdown protocol in LSCC cell lines (TU686 and TU212) and verified knockdown efficiency using qRT-PCR (Fig. 8b).Subsequent bioinformatic analyses revealed that LINC02454 was significantly overexpressed in patients with T4-stage LSCC (Fig. 8c), indicating that it may play a critical role in the malignant progression of LSCC.LINC02454 knockdown decreased EdU incorporation into the LSCC cell lines, indicating reduced cell proliferation (Fig. 8d, h).Moreover, Colony formation assays showed that LINC02454 downregulation significantly decreased the proliferative ability of TU686 and TU212 cells (Fig. 8e, g), which was confirmed via CCK8 assays (Fig. 8i,  j).Three-dimensional LSCC cell culture models, similar to the in vivo TME, were used to evaluate cell viability and proliferation; we observed that LINC02454 increased the viability of TU686 cells (Fig. 8f).
In vitro migration and invasion experiments suggested that LINC02454 knockdown decreased the migratory and invasive abilities of TU686 and TU212 cells (Fig. 8k-n).Meanwhile, cell immunofluorescence results suggested that epithelial mesenchymal transition of TU686 and TU212 cells was promoted by LINC02454 (Fig. 8o-r), which is essential for cancer metastasis (Yilmaz and Christofori 2009).
Based on the above findings, we verified the effect of LINC02454 on the malignant progression of LSCC in vivo.We subcutaneously injected TU686 cells transfected with Lv-NC and Lv-sh-LINC02454 cells into 5-week-old nude mice.After 18 days, we used a live imaging system to measure the luciferase signal; the relative photon flux of LSCC tumors was measured in nude mice from the Lv-NC or Lv-sh-LINC02454 groups (Fig. 9a), and the tumors were dissected (Fig. 9b).Compared with the Lv-NC group, tumor growth in nude mice in the Lv-sh-LINC02454 group was slower and the tumor was lighter in weight (Fig. 9c, d).We also employed a zebrafish tumor model to examine the influence of LINC02454 on metastasis in LSCC.DiI-labeled TU686 cells with or without LINC02454-knockdown were injected into the perivitelline space of 48-h zebrafish embryos.Fluorescence microscopy was used to count migrated cells, and bioluminescence measurement was used to monitor tumor cell growth and survival.Three days after injection, TU686 cell migration decreased following LINC02454 knockdown (Fig. 9e, f).At day 5 postinjection, the tumor masses in zebrafish were visualized using HE staining (Fig. 9g).Taken together, our findings suggested that LINC02454 participates in the occurrence and development of LSCC tumors.

Knockdown of LINC02454 enhanced the sensitivity of LSCC cells to cuproptosis
Our findings showed that LINC02454, as a cuproptosisrelated core lncRNA, promoted the proliferation and Fig. 7 Evaluation of the relationship between the four risk crlncR-NAs and prognosis in LSCC. a The expression correlation between the four crlncRNAs and 19 CRGs.b Expression levels of the four risk crlncRNAs (LINC02454, AC026310.1,AC090517.2and AC000123.1) in the high and low risk groups.c, d The four riskrelated crlncRNAs were highly expressed in patients with tumors using unpaired and paired differential expression analyses.e This plot shows the associations of the four risk-related crlncRNAs with the risk score of the prognostic signature.f ROC analysis also showed that LINC02454 was a more accurate prognostic indicator than other risk-related crlncRNAs at 3 years.g-j The independent prognostic ability of the four crlncRNAs was analyzed by multivariate cox analysis metastasis of LSCC cells, but whether this lncRNA was involved in regulating cuproptosis in LSCC was unclear.Therefore, we performed CCK-8 assays to observe the effect of LINC02454 downregulation on LSCC cell viability.We found that downregulation of LINC02454 significantly reduced TU686 and TU212 cell viability following copper and elesclomol treatment (Fig. 10a-d).This indicated that LSCC cell lines with downregulated LINC02454 were more sensitive to cuproptosis than control cells.In addition, the spheroid formation assay and live/dead cell staining confirmed the induction of cuproptosis upon LINC02454 knockdown.We observed that knockdown of LINC02454 significantly reduced the size of spheroids formed by LSCC cells after copper treatment.Live/dead cell staining showed that downregulation of LINC02454 increased the number of dead cells in the LSCC cell population after copper treatment and reduced the number of live cells (Fig. 10e-g).Mitochondria are critical for cellular energy production, and damage to these can lead to a decrease in energy production and the accumulation of ROS (Vyas et al. 2016).The accumulation of ROS in the cell can further exacerbate mitochondrial damage, leading to increased oxidative stress and, ultimately, cell death.Cuproptosis has been shown to cause significant mitochondrial damage in various cell types including LSCC cells.Several studies have demonstrated the link between cuproptosis and mitochondrial damage, highlighting the crucial role of mitochondria in copper toxicity (Chen et al. 2022).As part of this study, MitoTracker Red CMXRos and MitoTracker Green FM were used to further examine the effects of LINC02454 on cuproptosis-induced mitochondrial damage.The results showed that decreased expression of LINC02454 significantly decreased MitoTracker Red staining intensity (Fig. 10h, j).ROS production, evaluated using Mito-Sox Red staining, was found to be increased in LSCC cell lines with LINC02454 knockdown (Fig. 10i, k).Overall, our data provide evidence that LINC02454 plays a crucial role in protecting against copper death-related mitochondrial damage in LSCC cells.

Discussion
Laryngeal squamous cell carcinoma is one of the most aggressive head and neck squamous cell carcinomas, and it has a poor treatment efficacy and high recurrence rate (Steuer et al. 2017).Lack of definitive methods for initial diagnosis and identification of recurrence partly account for the poor prognoses of most patients with LSCCs.Therefore, the development of novel biomarker tools for prognosis is necessary to improve patient outcomes.A growing body of evidence has shown that lncRNAs function as tumor suppressors or oncogenes in myriad tumor types and are potential therapeutic targets (Mercer et al. 2009).Meanwhile, as a new type of cell death, cuproptosis has been implicated in the malignant progression of multiple cancers.However, the specific molecular mechanism underlying LSCC malignant progression remains unclear, particularly in terms of the roles played by lncRNAs.In fact, the functions of crl-ncRNAs in LSCC have not been previously reported.Our study explored the potential biological mechanism between lncRNAs and cuproptosis and defined a crlncRNA signature that may facilitate the prediction of LSCC prognosis and improve its treatment efficacy.
We identified 19 CRGs and 519 differentially expressed crlncRNAs from TCGA-LSCC data and obtained 30 prognosis-related lncRNAs using univariate Cox analysis.Subsequently, by Lasso Cox regression and univariate and multivariate Cox proportional hazard regression analyses, we developed a four-crlncRNA risk signature to predict the OS and PFS of patients with LSCC.Through KM analysis, we confirmed that patients in the high-risk group had worse prognoses, while those in the low-risk group had better life expectancy, indicating that the signature established in this study has strong prognostic potential.The crlncRNA signature showed a high level of accuracy based on the ROC curves, demonstrating its considerable discriminative ability for patients with LSCC.
We then evaluated how risk scores of the signature relate to clinical variables, immune cell infiltration, and antitumor drug sensitivity in LSCC, while also attempting to define molecular biological functions of the core crlncRNA.Our and Hoechst (blue).e The colony-forming ability of TU686 and TU212 cells was assessed by colony-forming assay.g The number of colonies was analyzed by Student's t test.**p < 0.01, ***p < 0.001.h EdU (%) was computed as EdU-positive cells/total Hoechst-positive cells.***p < 0.001, Student's t test.i, j Cell proliferation was detected using CCK8 assay in the TU686 and TU212 cell lines, respectively.*p < 0.05, **p < 0.01, ***p < 0.001, Student's t test.f Three-dimensional spheroid growth experiment was used to detect the vitality of TU686 cells.Dead cells were labeled with red fluorescence, live cells were labeled with green fluorescence.k The migration ability of TU686 and TU212 cells was detected by transwell migration assay.l Transwell invasion assay assessed the invasion capacity of TU686 and TU212 cells.The numbers of migratory (m) and invasive (n) cells were compared to the negative control group, respectively.**p < 0.01, ***p < 0.001, Student's t test.Immunofluorescence (IF) assay was performed to reveal the E-cadherin (o, q) and N-cadherin (p, r) protein expression in TU686 and TU212 cells transfected with either NC or LINC02454 siRNA ◂ findings suggest that the risk model, comprising LINC02454, AC026310.1,AC090517.2,and AC000123.1,can be used as a novel prognostic signature for LSCC.In papillary thyroid cancer, recent studies have shown that as a potential oncogene, LINC02454 can inhibit cell apoptosis, promote cell proliferation, and act as a significant biological marker for diagnosis and prognosis.However, to our knowledge, no data has been reported regarding the four risk-related crl-ncRNAs in LSCC; thus, further investigation is required to further elucidate the functions of these lncRNAs.Combined with clinical factors, the multivariate analysis suggested that LINC02454 represents the only risk gene associated with LSCC OS.Moreover, the ROC curve of the four LSCC risk genes revealed that LINC02454 exhibited the highest accuracy; thus, we focused our study on LINC02454.
Using in vivo and in vitro approaches, we found that downregulation of LINC02454 inhibited the proliferation, migration, and invasion of LSCC cells, suggesting that LINC02454 may have tumor-promoting properties.Subsequent analyses suggested that knockdown of LINC02454 can enhance the sensitivity of LSCC cells to cuproptosis, which demonstrated the potential role of LINC02454 in regulating cell death pathways and highlight its significance in LSCC.Previous studies have shown that ROS can cause significant mitochondrial damage and exacerbate oxidative stress, which promotes the activation of cell death mechanisms (Chen et al. 2022).Therefore, strategies that target cuproptosis have emerged as potential therapeutic options in cancer treatment by inducing tumor cell death.In this study, knockdown of LINC02454 was found to enhance the sensitivity of LSCC cells to cuproptosis, which may aid in the development of novel strategies to combat cancer.In addition, LINC02454 has been implicated in various cellular processes, including proliferation, migration, and invasion, suggesting its possible involvement in tumorigenesis.The current study sheds light on the role of LINC02454 in cuproptosis-mediated cell death regulation in LSCC and provides insight into the potential mechanisms underlying its involvement in tumorigenesis.
By conducting biological function and pathway enrichment analyses, we investigated the potential biological mechanisms involved in LSCC, allowing for more precise biological inferences.To identify the functions of DEGs, GO and KEGG enrichment analyses were carried out and found to be primarily enriched in phagosomes, immunoglobulin receptor binding, immunoglobulin complexes, proteoglycans in cancer, and Wnt signaling pathway.The GSEA further indicated that allograft rejection, autoimmune thyroid diseases, and primary immunodeficiency were significantly enriched in the low-risk group.Most of the DEGs focused on immune function, which may provide a new direction in LSCC immunotherapy.
Although immunoglobins are produced by B cells, they can be expressed by tumor cells.In fact, immunoglobulins expressed by epithelial cancer cells have been shown to contribute to the malignant progression of tumors.Recent studies have reported that the clinical application of programmed cell death protein 1/programmed death-ligand 1 monoclonal antibodies provides promising efficacy for tumor immunotherapy (Kishton et al. 2017).Park et al. proposed a model and strategy for anti-human epidermal growth factor receptor 2/neu antibody-mediated tumor clearance that further enhances subsequent antibody-induced immunity through the increased influx of innate and adaptive immune cells into the TME (Park et al. 2010).Given that immunoglobulins are critical for the effective treatment and diagnosis of numerous diseases, including many cancers, they may represent novel therapeutic targets for the improvement of LSCC treatment efficacy.
The TME, particularly the immune microenvironment, significantly contributes to tumor pathophysiology (Wang et al. 2018).Indeed, the TME is widely recognized as an important factor in cancer progression, enabling primary, invasive, and metastatic growth (Tekpli et al. 2019).We investigated the correlation between immunologically active cells and the four-crlncRNA signature.Naive B cells, M0 and M2 macrophages, Th2 cells, and inflammation-promoting tumors significantly differed between the groups.M2 macrophages in the TME are essential for immunological tolerance and tumor progression (Pan et al. 2020).Recent studies have shown that M2 macrophage depletion may inhibit tumor relapse after cerebrotendinous xanthomatosis treatment.Moreover, tumor-associated inflammation increases angiogenesis, promotes metastatic spread, inhibits local immunosuppression, and affects genomic stability (Hughes et al. 2015;Liu et al. 2019).Mitochondria function as the regulatory centers of cell-signaling processes associated with inflammation.Additionally, mitochondrial dysfunction may increase the sensitivity of cells to Cu 2+ , promoting cuproptosis induction (Barrett et al. 2004).These findings suggest a possible link between the immune TME and cuproptosis in LSCC.Moreover, our data reveal that the expression levels of immune checkpoint genes (CD276 and CD70) were upregulated in the high-risk group, while CD27 was downregulated.CD276 helps cancer stem cells escape immune surveillance, and its blockade effectively enhances T cell-mediated antitumor immunity, while CD70 regulates CD44 and SRY-box 2 expression, affecting tumor migration and growth, as well as macrophage infiltration (Sanchez-Correa et al. 2019).In contrast, the CD27 signaling pathway inhibits tumor growth.CD27 co-stimulation can enhance the expansion, effector function, and survival of human CAR-T cells in vitro, and enhance the persistence and antitumor activity of human T cells in vivo (Song et al. 2012).In the clinical and research settings, tumor immunotherapy is rapidly advancing.Compared with conventional therapies, immunotherapy provides increased tumor suppression coupled with decreased toxicity (Jiang et al. 2018).The findings from the current study may provide novel insights into the development of immunotherapies for LSCC.
Currently, multidrug resistance is the main therapeutic obstacle for human laryngeal cancer chemotherapy.Owing to late metastases and resistance to chemotherapy and radiotherapy, patients with laryngeal cancer have a low survival rate.Consequently, it is necessary to enhance the sensitivity of tumor cells to chemotherapeutic agents and to identify potential markers of chemotherapy efficacy.Our findings indicate that patients with LSCC in the high-risk group were susceptible to thapsigargin, midostaurin, and ponatinib, but may have been resistant to belinostat, GW 441756, Akt inhibitor, and THZ-2-102-1.These findings may improve the prognoses of patients undergoing chemotherapy and reduce the financial burden on patients and hospitals caused by administration of ineffective therapies.Moreover, the LSCC samples were divided into clusters 1 and 2 via consensus clustering analysis based on the crlncRNAs associated with prognosis.Immune infiltration analysis showed that neutrophils, NK cells, Th1 cells, and Th2 cells were expressed at low levels in cluster 1.Compared with cluster 1, cluster 2 showed higher risk scores and worse survival outcomes, which is consistent with cluster 2 primarily being concentrated in the high-risk group.
Despite these findings, there are several limitations to this study.First, only a limited number of patients were included, which might have caused certain deviations in the results.Inclusion of more patients may enhance the reliability and stability of our model; thus, further preclinical studies are warranted.Second, the precise molecular mechanisms by which these lncRNAs affect cuproptosis remain unclear and require further exploration.This study only provided a preliminarily investigation of the malignant biological functions of crlncRNAs in LSCC and, thus, their specific molecular mechanisms warrant further analysis.
In conclusion, our study established a four-crlncRNA signature that may serve as a reliable predictive tool for LSCC prognosis and treatment decision-making, thereby providing new insights into LSCC tumorigenesis and progression.Our in vitro and in vivo experiments demonstrated that LINC02454 can accelerate the growth and metastasis of LSCC cells.In addition, our findings suggest that targeting LINC02454 may be a promising approach to enhance the efficacy of cuproptosis-induced cancer cell death and improve the outcomes of LSCC treatment.cells that were co-cultured with ES (20 nm) and CuCl 2 (20 nm) for 24 h.b, d The cell (TU686, TU212) viability was determined by CCK8 assay.e-g Cells were co-cultured with CuCl 2 and ES for 12 h, and then the cellular viability was tested by spheroid growth assay.Live cells (green), dead cells (red), cell nucleus (blue), respectively.***p < 0.001, ****p < 0.0001, Student's t test.h, j Mitochondria were stained with Mito-Tracker red CMRos (red) and Mito-Tracker Green FM (green) in LSCC cells that were co-cultured with ES (20 nm) and CuCl 2 (20 nm) for 12 h, respectively.**p < 0.01, ***p < 0.001, one-way ANOVA.i, k After 12 h of co-culture, cells were stained by MitoSOX Red (red) mitochondrial superoxide indicator and the mean fluorescence intensity (MFI) of MitoSOX was quantified, respectively.*p < 0.05, ****p < 0.0001, one-way ANOVA

Fig. 1
Fig. 1 Identification of the differential crlncRNAs with prognostic value in LSCC. a Sankey plot displayed the network of CRGs and crlncRNAs co-expression (R 2 > 0.4 and p < 0.05).b The volcano plot showed the differential crlncRNAs among normal and tumor samples, with red representing upregulated crlncRNAs, green representing downregulated crlncRNAs, and black represents non-differentiated

Fig. 2
Fig. 2 Construction and validation of the crlncRNA prognostic signature.a Seven candidate crlncRNA were screened by LASSO Cox regression.b The trace of each individual crlncRNA.c A forest plot listed the four risk crlncRNAs of the prognostic signature based on multivariate analysis.Survival curves for the low and high risk groups in the entire sets (d),training sets (e), testing sets (f).The ROC curves for survival prediction in the entire sets (g), training sets (h), testing sets (i).PCA results show an important distribution difference between high-and low-risk groups in all datasets (j), training datasets (k), testing datasets (l).Risk score of each patient in all datasets (m), training datasets (n), testing datasets (o).Survival status of each patient in the whole datasets (p), training datasets (q), testing datasets.(r).The expression patterns of four risk crlncRNAs in all datasets (s), training datasets (t), testing datasets (u)

Fig. 3
Fig. 3 Independence of the crlncRNA signature in predicting OS and PFS.a-d Univariate and multivariate Cox regression analyses of the crlncRNA signature for OS and PFS in LSCC.e The AUC of the signature for the 3-year OS was 0.794, which was greater than those of other clinical parameters.f The C-index of the risk score was superior to that of the other clinical characteristics.g The AUC of the crl-ncRNA signature for 1-, 2-, and 3-year PFS.h The AUC of the clinical parameters for PFS.k KM analysis in PFS between two groups

Fig. 4
Fig. 4 Functional enrichment analysis of DEGs in LSCC. a GO enrichment analysis regarding the DEGs was shown in the bar graph.b KEGG enrichment analysis of the DEGs was displayed in the bubble plot.c A Circos plot demonstrating the relationship between

Fig. 6
Fig. 6 Consensus clustering of crlncRNAs in molecular subtypes of LSCC. a The consensus clustering matrix for the 111 TCGA samples with k = 2. b The consensus clustering cumulative distribution function (CDF) is shown for k = 2-10.Relative change in area under CDF curve for k = 2 to 10. c The area under CDF curve for k = 2-10 shows a relative change.d There was a significant difference between the KM curves of the two clusters in OS (p = 0.002).e Heatmap showed the distribution of the immune cell infiltration landscape in the two clusters.f The KM curves showed the difference in PFS between the two clusters (p = 0.006).g The abundance of immune infiltrates differed between the two clusters.h The risk scores were notably different between clusters 1 and 2. i The Sankey plot showed a higher proportion of high-risk patients in clusters 2

Fig. 8
Fig. 8 Knockdown of LINC02454 inhibits the proliferation and migration of LSCC cells in vitro.a Relative expression of LINC02454 determined by qRT-PCR in LSCC cells and HaCaT cells.b Relative expression of LINC02454 in TU686 and TU212 cells treated with LINC02454 siRNA.***p < 0.001, one-way ANOVA.c Box plot analysis of LINC02454 expression in T1-3 stage and T4 stage LSCC patients according to TCGA database.*p < 0.05, Student's t test.d The cells were fixed and stained with EdU (red)and Hoechst (blue).e The colony-forming ability of TU686 and TU212 cells was assessed by colony-forming assay.g The number of colonies was analyzed by Student's t test.**p < 0.01, ***p < 0.001.h EdU (%) was computed as EdU-positive cells/total Hoechst-positive cells.***p < 0.001, Student's t test.i, j Cell proliferation was detected using CCK8 assay in the TU686 and TU212 cell lines, respectively.*p < 0.05, **p < 0.01, ***p < 0.001, Student's t test.f Three-dimensional spheroid growth experiment was used to detect the vitality of TU686 cells.Dead cells were labeled with red fluorescence, live cells were labeled with green fluorescence.k The migration ability of TU686 and TU212 cells was detected by transwell migration assay.l Transwell invasion assay assessed the invasion capacity of TU686 and TU212 cells.The numbers of migratory (m) and invasive (n) cells were compared to the negative control group, respectively.**p < 0.01, ***p < 0.001, Student's t test.Immunofluorescence (IF) assay was performed to reveal the E-cadherin (o, q) and N-cadherin (p, r) protein expression in TU686 and TU212 cells transfected with either NC or LINC02454 siRNA

Fig. 9
Fig. 9 LINC02454 knockdown inhibits LSCC proliferation and metastasis in vivo.a Imaging of nude mice with xenograft tumors.b Nude mice xenografted tumors.c The tumor weight.d Tumor growth curve.e TU686 cells transfected with NC or LINC02454 siRNA and labeled with DiI were injected into the zebrafish at 48 h.On the