Abstract

Objective

   This study aimed to identify potential biomarkers and elucidate
   molecular mechanisms in oral squamous cell carcinoma (OSCC) by
   leveraging an integrated bioinformatics approach.

Methods

   We conducted high-throughput RNA sequencing on OSCC and adjacent normal
   tissues to profile mRNA and lncRNA expression. Differentially expressed
   genes were identified and subjected to Gene Ontology (GO) and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. A
   competing endogenous RNA (ceRNA) network was constructed, and
   protein–protein interaction (PPI) networks were analyzed using STRING
   and Cytoscape software. Hub genes were identified using the Cytohubba
   plug-in in Cytoscape.

Results

   Our analysis identified 5362 differentially expressed mRNAs and 2801
   differentially expressed lncRNAs. GO analysis revealed that
   dysregulated mRNAs were associated with system development and
   responses to organic substances. At the same time, lncRNAs were
   enriched in muscle system processes and cell–cell signaling. KEGG
   pathway analysis highlighted cancer-related pathways, including
   cytokine–cytokine receptor interactions and the NF-kappa B signaling
   pathway. The constructed ceRNA network highlighted key regulatory
   nodes, including hub genes IGF2BP1, CLDN6, and HLA-G, which may play
   pivotal roles in OSCC.

Conclusions

   This study provides a comprehensive lncRNA-mRNA regulatory network and
   identifies biomarkers that could advance OSCC therapeutic strategies.
   The findings offer new insights into OSCC pathogenesis and potential
   targets for clinical application.

Supplementary Information

   The online version contains supplementary material available at
   10.1007/s12672-025-02922-4.

   Keywords: Oral squamous cell carcinoma, LncRNA, mRNA, Bioinformatics,
   CeRNA network

Introduction

   Oral cancer incorporates all malignant tumours arising in the oral
   cavity. The GLOBOCAN 2020 database cites that in the year 2020 alone,
   377,700 new cases and 177,800 deaths were recorded globally, with
   increasing incidence and mortality rates in many countries [[34]1].
   Oral squamous cell carcinoma (OSCC) is the most common form of oral
   cancer. Though advances in diagnostics and therapeutics have occurred,
   the most recently reported 5-year survival rate for OSCC stands at
   40–50%, depending on macro-regions and the stage of diagnosis
   [[35]2–[36]4]. Early OSCC is much more amenable to cure, yet early
   detection is hard to achieve. Once OSCC progresses to advanced stages,
   survival averages sink dramatically, and distant metastasis is
   frequently seen [[37]2, [38]3, [39]5]. For that reason, understanding
   the molecular processes that drive OSCC proliferation and ensuing
   metastasis is critical for developing new strategies to curb the
   progression of the disease [[40]6, [41]7].

   Long non-coding RNAs (lncRNAs) are RNA molecules longer than 200
   nucleotides that do not code for proteins but play a crucial role in
   regulating gene expression. More than 5000 lncRNAs were identified
   based on transcriptome sequencing and microarray technologies
   [[42]8–[43]11]. Recent evidence suggests the existence of complex
   regulatory interactions among lncRNAs, microRNAs (miRNAs), and
   messenger RNAs (mRNAs) in cancer pathogenesis, which form competing
   endogenous RNA (ceRNA) networks. In addition, such networks
   significantly contribute to tumorigenesis by controlling gene
   expression at transcriptional and post-transcriptional levels
   [[44]12–[45]17]. In particular, the regulatory interactions within the
   lncRNA-miRNA-mRNA axis have been pointed out as important mechanisms in
   developing OSCC: tumor cell proliferation, invasion, metastasis, and
   resistance to therapy [[46]18–[47]20].

   Some investigations have reported on the predictive and progressive
   association of lncRNAs with OSCC. For example, high levels of
   expression of CASC9 lncRNA correlate with larger sizes of the tumor,
   more advanced clinical stages, and poorer overall survival [[48]21].
   Similarly, knockdown of lncRNA LSINCT5 will inhibit the proliferation
   and migration of OSCC, and its overexpression leads to poor prognosis
   [[49]19]. In addition, other important lncRNAs, such as SNHG16, CCAT1,
   and PART1, have been shown to play a role in the proliferation,
   migration, and invasion of OSCC [[50]18–[51]20]. Despite these studies,
   the very complicated regulatory mechanisms and interactions of
   lncRNA-miRNA-mRNA in the context of OSCC are still scarcely explored.

   This study aims to bridge those gaps in knowledge by utilizing
   high-throughput RNA sequencing (RNA-seq) analysis on 10 pairs of OSCC
   tumors and corresponding adjacent normal tissues. We constructed a
   complete lncRNA–mRNA interactome network using this information to
   illuminate key regulatory interactions involved in OSCC pathogenesis.
   Future studies should include in vitro and in vivo validations to
   complement the findings of the current performative analysis.

Materials and methods

Sample collection

   Histopathological confirmation of a diagnosis of OSCC was secured by
   exploding the ten OSCC patients contributed with tissue samples. Both
   samples of the tumor, as well as non-tumorous adjacent tissues, were
   carried out with standardized clinical protocols. Inclusion criteria
   for this study included patients with histologically confirmed OSCC and
   staging according to the tumor-node-metastasis (TNM) staging system.
   Only patients with Stage I-III OSCC, excluding those with distant
   metastasis, were included. Tumor grading was conducted according to the
   World Health Organization (WHO) histopathological classification.
   Immediately after excising, the tissues were flash-frozen in liquid
   nitrogen and sectioned into 2–3 mm thick portions to preserve RNA
   integrity. Samples were stored in 1.5 mL RNase-free screw-cap tubes,
   tightly sealed with parafilm, and kept at − 80 °C until further
   analysis. All patients were recruited from Hospital AA and given
   written consent before participation. The study was approved by the
   Ethics Committee of The First People's Hospital of Foshan (Approval No.
   Lunshenyan (2025) No. 202) and was carried out in accordance with the
   Declaration of Helsinki.

Isolation of total RNA and sequencing

   The TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was utilized in
   accordance with the manufacturer’s protocol to isolate total RNA from
   tumor tissue samples and normal tissues of OSCC patients. Gel
   electrophoresis in agarose was conducted to analyze the RNA purity and
   integrity of the samples. The initial quantification and detection of
   RNA concentration and purity were performed utilizing the NanoDrop-1000
   instrument (Thermo Fisher Scientific Inc., Waltham, MA, USA). The
   Agilent 2100 Bioanalyzer was used to check the insert size of the
   library, with an anticipated insert size of around 250–300 bp. Illumina
   PE150 sequencing was conducted in accordance with the manufacturer’s
   guidelines Hisat2 ([52]http://ccb.jhu.edu/software/hisat2). A total of
   20 paired samples (10 tumor and 10 adjacent normal tissues) were
   analyzed, obtaining 1,750,496,736 raw reads; after quality control,
   1,707,758,234 clean reads were left to be analyzed.

PPI network construction and hub gene identification

   A PPI network has been constructed using the STRING database version
   11.5 with a confidence score threshold of 0.7 to evidence molecular
   interactions. The top 3096 DEGs were selected for network analysis
   based on adjusted p value (< 0.01) and fold change criteria (|log2FC| ≥
   2). The outputted PPI network was visualized in Cytoscape version
   3.9.1, and hub genes were determined using the CytoHubba plugin,
   ranking the nodes based on the MCC algorithm.

Isolation of LncRNA

   Following standard protocols, total RNA, including long non-coding RNA
   (lncRNA), was extracted from the OSCC tumor and adjacent normal tissues
   using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). RNA quality and
   concentration were assessed with a NanoDrop spectrophotometer, and
   integrity was verified using an Agilent 2100 Bioanalyzer; only samples
   with an RNA integrity number (RIN) > 7 were used. Lithium chloride
   precipitation was performed to enrich lncRNA and remove small RNAs,
   selectively isolating lncRNAs and discarding small RNA fractions. The
   final lncRNA samples were stored at – 80 °C for further analyses.

Screening for differential expression genes

   To quantify transcript fold changes and identify differential
   expression mRNA and lncRNAs, we employed the edgR package in R (R
   version 3.38.5) with a threshold of |log[2]FC| ≥ 2.0 and FDR < 0.01.
   Volcano plots and heat maps were generated in R software using the
   heatmap and ggplot2 packages, respectively.

Principal component analysis (PCA)

   PCA was performed to analyze the variation in gene expression patterns
   between the tumor and adjacent normal tissues of OSCC. Three PCs
   accounting for more than 75% of the variance in gene expression
   profiles were chosen for visualization, providing robust separation
   between tumor and normal groups.

Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway
analysis

   The enrichment analyses using GO and KEGG pathways were performed in R
   with the clusterProfiler package (version 4.2.2). The hypergeometric
   test was used to determine the significance of enrichment, and p value
   adjustments for multiple comparisons were made using the
   Benjamini–Hochberg method, using an adjusted p value threshold < 0.05.
   Significantly enriched biological processes, cellular components, and
   molecular functions were identified and visualized using ggplot2.

Co-expression of differential LncRNAs and mRNAs

   To create the co-expression network for both mRNA and lncRNA, we
   integrated the differential expression mRNAs and lncRNAs, then
   calculated the Pearson correlation coefficient for all pairs. We
   considered pairs of mRNAs and lncRNAs with coefficients greater than
   0.9 or less than − 0.9 to be correlated. The resulting mRNA-lncRNA
   network was visualized using Cytoscape (version 3.9.1). We then
   utilized the Molecular Complex Detection (MCODE) plug-in in Cytoscape
   to identify hub genes within the network, providing insights into
   potential key regulators in the system.

Co-expression network construction

   The interplays of differentially expressed lncRNAs and mRNAs were
   further explored by constructing a co-expression network that involved
   Pearson correlation analysis. A count of pairs of correlation
   coefficients > 0.9 or < − 0.9 was considered sufficiently big. The
   network was depicted in Cytoscape (3.9.1), while the Molecular Complex
   Detection (MCODE) plugin identified hub genes, thus allowing a glimpse
   into potential key regulators in OSCC.

Results

RNA sequencing and data quality control

   RNA sequencing was performed on 20 paired OSCC samples of tumor and
   adjacent normal tissues, giving rise to 1,750,496,736 raw reads (Table
   S1). Stringent quality control removed all but 1,707,758,234
   high-quality clean reads (Fig. [53]1A). The Phred quality scores (Q30)
   were above 95%, indicating that the sequencing performed was highly
   accurate. Read alignment to the human reference genome (GRCh38) using
   Hisat2 showed a mapping efficiency of over 98%, establishing the
   reliability of sequence data for all samples. Through PCA, tumor tissue
   can be distinctly separated from adjacent tissue (Fig. [54]1B).

Fig. 1.

   [55]Fig. 1
   [56]Open in a new tab

   Identification of differential expression mRNAs and lncRNAs inRNA-seq.
   A Box plot of gene expression FPKM distribution in samples. B PCA
   results for the RNA-seq data. C Volcano plot of differential expression
   mRNAs. D Volcano plot of differential expression lncRNAs. E Heatmap of
   differential expression mRNAs. F Heatmap of differential expression
   lncRNAs. N normal, T tumor

Identification of differentially expressed genes (DEGs) and LncRNAs

   A total of 5362 DEGs and 2801 differentially expressed lncRNAs were
   identified (|log2FC| ≥ 2.0, FDR < 0.01) (Table S2). The most
   significantly upregulated DEGs included IGF2BP1 (log2FC = 4.67,
   p < 0.001), CLDN6 (log2FC = 4.21, p < 0.001), and HLA-G (log2FC = 3.98,
   p < 0.001). The most significantly downregulated genes included
   LINC00472 (log2FC = − 4.51, p < 0.001), MUC19 (log2FC = − 4.33,
   p < 0.001), and ZG16B (log2FC = − 4.12, p < 0.001). Volcano plots
   (Fig. [57]1C, D) and hierarchical clustering heatmaps (Fig. [58]1E, F)
   visualize these significant differences between OSCC tumors and normal
   tissues.

Gene ontology (GO) and KEGG pathway enrichment analysis

   As shown in Fig. [59]2A, the top 20 biological processes (BP) terms in
   tumor tissues were associated with single-organism processes, system
   development, and responses to organic substances; the top 20 CC terms
   were those associated with integrated components of the plasma
   membrane, extracellular regions, and cytoplasm; top 20 MF terms were
   involved in molecular functions, protein binding, and protein
   dimerization activities (Table S3, Fig. [60]2A). The KEGG pathway
   analysis identified the top significantly enriched pathways for DEGs,
   which included metabolic pathways, pathways in cancer,
   cytokine–cytokine receptor interaction, viral carcinogenesis, and
   systemic lupus erythematosus (Fig. [61]2B). These pathways are
   essential during OSCC pathogenesis, with the PI3K-Akt signaling
   pathways known to regulate tumor proliferation, survival, and
   resistance to apoptosis (Fig. [62]2B).

Fig. 2.

   [63]Fig. 2
   [64]Open in a new tab

   Functional analysis of differential expression mRNAs. A GO analysis of
   differential expression mRNAs. B KEGG pathway analysis of differential
   expression mRNAs

   The GO and KEGG analysis of differential expression lncRNAs revealed
   important pathways linked to OSCC (Fig. [65]3). The top 20 terms for
   BP, CC, and MF are presented in Fig. [66]3A. In KEGG analysis, the top
   five significantly enriched pathways for DEGs were metabolic pathways,
   pathways in cancer, regulation of actin cytoskeleton, neuroactive
   ligand–receptor interaction, and cytokine–cytokine receptor interaction
   (Table S4, Fig. [67]3B).

Fig. 3.

   [68]Fig. 3
   [69]Open in a new tab

   Functional analysis of differential expression lncRNAs. A GO analysis
   of differential expression lncRNAs. B KEGG pathway analysis of
   differential expression lncRNAs

PPI network construction and hub gene identification

   A network of protein–protein interactions (PPIs) was determined by
   using the STRING database, including 3096 DEGs, and allowing
   intermedium-high confidence scores for integration (> 0.7)
   (Fig. [70]4). Selected hub genes are recognized and ranked by the
   Maximal Clique Centrality (MCC) algorithm via the plug-in CytoHubba.
   The list of the top 10 hub genes includes IGF2BP1, CLDN6, C1orf94,
   HLA-G, MUC19, LINC00472, LINC02582, [71]AC073365.Y, HTN1, and ZG16B
   (Fig. [72]4). These genes are key to cancer progression, immune
   evasion, and tumor microenvironment regulation.

Fig. 4.

   [73]Fig. 4
   [74]Open in a new tab

   Construction of deferential expression lncRNA–mRNA co-expression
   network. Green ovals depict the representation of key genes; blue ovals
   represent differential expression mRNAs; differential expression
   lncRNAs are indicated by red diamonds

lncRNA-mRNA co-expression network

   A co-expression network was constructed to investigate the lncRNA–mRNA
   interactive network. Spearman correlation analysis identified
   significant associations (|r| >0.9, p < 0.01). Cytoscape visualization
   showed that IGF2BP1, CLDN6, and LINC00472 are the major regulatory
   hubs, indicating their possible roles in the modulation of
   OSCC-associated pathways (Fig. [75]4). These results provide
   significant insights into the OSCC pathogenesis, highlighting specific
   key genes and pathways that could be candidates for either therapeutic
   intervention or as potential prognostic biomarkers.

Discussion

   OSCC is an extremely heterogeneous malignancy; various factors,
   including ethnicity, tumor stage, and molecular subtypes, may lead to
   certain variations in gene expression patterns and impact patient
   prognosis [[76]22, [77]23]. We identified differentially expressed
   genes (DEGs) and long non-coding RNAs (lncRNAs) associated with OSCC,
   revealing several key regulatory pathways. Incorporating TCGA RNA-seq
   data into the analysis could interactively reinforce these findings by
   correlating biomarker gene expression with survival signatures in
   various patient cohorts [[78]24].

   The results of our work strongly proposed the importance of
   lncRNA-mediated regulatory networks in OSCC development. The competing
   endogenous RNA (ceRNA) hypothesis states that lncRNAs act as molecular
   sponges that modulate the stability and translation of target mRNAs
   through the sequestration of certain miRNAs [[79]25, [80]26]. Although
   key lncRNA–mRNA interactions were identified, miRNA profiling was not
   fully integrated into our study, which limited the extent to which
   ceRNA interactions could be explored. This enhances the potential for
   further insights into the interconnected regulatory networks directing
   OSCC pathogenesis, using miRNA expression profiling complemented by
   target prediction tools such as miRDB or TargetScan [[81]27].

   The protein–protein interaction (PPI) network analysis revealed several
   key hub genes, including IGF2BP1, CLDN6, and HLA-G. IGF2BP1 is an
   oncogenic RNA-binding protein associated with enhanced tumor cell
   proliferation and metastasis, with high expression levels linked to
   poor prognosis in multiple cancers [[82]28]. CLDN6, a tight junction
   protein, is involved in epithelial-mesenchymal transition (EMT),
   contributing to OSCC invasiveness and metastatic potential [[83]29].
   HLA-G, an immune checkpoint regulator, has been implicated in immune
   evasion mechanisms, suggesting a potential role in OSCC immune escape
   [[84]30].

   Our pathway enrichment analysis highlighted PI3K-Akt and NF-kappa B
   signaling as major pathways involved in the biology of OSCC. Those
   pathways are vital for tumor cell survival, proliferation, and
   resistance to therapy [[85]31, [86]32]. The PI3K-Akt pathway, often
   activated in OSCC, has been linked with poor prognosis, while NF-kappa
   B signaling contributes mainly to inflammation-driven carcinogenesis
   and chemoresistance [[87]33]. Small-molecule inhibitors specifically
   targeting these pathways or applying immunotherapeutics might provide
   novel therapeutic options for OSCC patients.

   The relative differential coding potential of mRNAs, as opposed to
   non-coding RNAs, raises important considerations in directing their
   contributions to OSCC progression. mRNAs would code for active
   proteins, leading to oncogenic processes, while on the other hand,
   lncRNAs would primarily function transcribing and post-transcribing to
   regulate expression [[88]34]. Literature reviews hint at specific
   lncRNAs, such as LINC00472, playing tumor-suppressing roles in oncogene
   expression regulation. In contrast, lncRNAs such as LINC02582 are
   posited to play oncogenic protein roles by promoting tumor growth and
   invasion [[89]35, [90]36].

   All the same, it must be said that the authors are cognizant of
   limitations in their study. These small samples hinder generalizability
   and thus call for larger multi-center studies. The results mentioned
   here, while interesting, should be treated with caution since a lot
   still needs to be learned, particularly in terms of experimental
   validation through in vitro and in vivo models. Further investigation
   would benefit greatly from integrating clinical data, including patient
   survivorship and treatment response, which would enhance the
   translational potential of those biomarkers identified. Future efforts
   should focus on functional assays and patient-derived OSCC samples to
   establish clinical correlations. Therefore, from this analysis of
   lncRNA–mRNA interactions in OSCC, key pathways and intricate regulatory
   networks implicated in tumor progression emerged. By integrating
   bioinformatics analyses with experimental verification and clinical
   datasets, it may be possible to follow these findings with multi-center
   studies in a broader scope that could ultimately lead to novel
   diagnostic and therapeutic strategies for OSCC. How the interactions
   between these molecular players impact OSCC prognosis and response to
   therapy remains an important area of inquiry for the future.

Conclusion

   Our study identified ten RNAs as potential biomarkers for OSCC,
   providing valuable insights into their regulatory mechanisms in the
   lncRNA-mRNA ceRNA network. From building a PPI network, we identified
   key hub genes that might contribute to OSCC progression and therapeutic
   resistance. While our findings identified potential molecular targets,
   further experimental validation and clinical studies are required to
   examine their diagnostic and prognostic significance. Future studies
   could refine these biomarkers into novel OSCC diagnosis and treatment
   strategies when integrated with multi-omics data and functional assays.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [91]Supplementary Material 1^ (636.6KB, xls)
   [92]Supplementary Material 2^ (369.4KB, xls)
   [93]Supplementary Material 3^ (166KB, xls)
   [94]Supplementary Material 4^ (2.1MB, xls)

Author contributions

   ShiWei Liu, Jin Li and Qing Shao carried out study concepts and design,
   clinical and experimental studies, data acquisition and analysis,
   statistical analysis and manuscript preparation and editing; JuFeng
   Chen and Chen Zou contributed to literature research; YiLong Ai was the
   guarantor of integrity of the entire study and helped to mnuscript
   review. All authors have read and approved this article.

Funding

   None.

Data availability

   The datasets used or analysed during the current study are available
   from the corresponding author on reasonable request.

Declarations

Consent for publication

   Informed consent was obtained from all individual participants included
   in the study.

Ethical approval

   All procedures performed in studies involving human participants were
   in accordance with the 1964 Helsinki Declaration and its later
   amendments or comparable ethical standards. This study is approved by
   the Ethics Committee of Foshan First People’s Hospital. Written
   informed consent was obtained.

Competing interests

   The authors declare no competing interests.

Footnotes

   Publisher’s note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

   ShiWei Liu, Jin Li and Qing Shao are co-first authors.

References