Two tiers of data, denoted as Level 1 and Level 4, were retrieved from the TCGA repository. Level 4 primarily encompasses fundamental patient attributes, including sex, age, race, and the Tumor Lymph Node Metastasis (TNM) staging; whereas Level 1 provides more detailed information, including specifics such as medication dosage, treatment efficacy, radiotherapy records, and other relevant information for each follow-up day. In total, 504 LUSC tumor samples were procured for analysis. RNAseq expression values, which had been corrected utilizing Recursive Structural Equation Model (RSEM), were obtained from the Level 3 dataset of the LUSC project within the TCGA database hosted at Firehost (https://gdac.broadinstitute.org/). Subsequently, known lncRNA expression values were discerned from this dataset, and a total of 51 normal samples and 501 tumor samples were included in the ensuing analysis.

DESeq2 software was employed to analyze the differential expression of lncRNAs between normal and cancer samples. The lncRNA expression values were subjected to statistical analysis with a predefined significance threshold (p < 0.05, |log2FoldChange| > 1). The software was configured accordingly to derive the differentially expressed lncRNA profiles.

We first compiled a list of lncRNAs associated with survival outcomes through single-factor Cox proportional hazard regression analysis. Subsequently, we conducted multivariable Cox proportional hazard regression analysis to identify independent prognostic factors influencing survival. Finally, a stepwise selection procedure was employed in the context of multivariable Cox proportional hazard regression analysis to identify lncRNAs with significant prognostic relevance based on their associated P- values.

The identification of potential lncRNA (in cis acting) target genes was performed by screening for differentially expressed genes within 10 kb upstream and downstream of the lncRNAs. This approach is grounded in the recognition that interactions between lncRNAs and their target genes typically occur in close genomic proximity. The selection of a 10-kilobase range serves to focus the research, mitigate computational complexity, and enhance prediction accuracy. In contrast, the identification of potential lncRNA target genes acting in a trans manner commenced with an initial screening using the blast algorithm (e-value < 1e-5) and was subsequently refined using RNAplex software (G<-20). It is worth noting that lncRNAs exert their influence on mRNA expression through various mechanisms, including: 1) direct cis and trans interactions with mRNAs; 2) binding to microRNAs (miRNAs), thereby impeding miRNA-mediated mRNA regulation; and 3) binding to RNA-binding proteins (RBPs), influencing RBP-mRNA interactions. The scope of this analysis was limited to the examination of RNA-seq data from the TCGA database, which exclusively provides expression profiles of lncRNAs and mRNAs. Therefore, the initial phase of analysis focused on discerning potential interactions between these molecules at the expression level through correlation analysis.

For KEGG analysis, the p-value was determined utilizing Fisher’s exact test. Signal transduction and disease pathways exhibiting statistical significance with a threshold of p < 0.05 were selected for further analysis. For GO analysis, the p-value was computed employing the hypergeometric distribution method. Annotations characterized by a high frequency and possessing a p-value less than 0.05 were retained for subsequent investigation.

RNA extraction was performed employing the Jinbaite RNA Extraction Kit. Subsequently, DEPC water was introduced, and the concentration of RNA was quantified. Quantitative polymerase chain reaction (qPCR) experiments were executed utilizing the Zhongshi Tongchuang Reverse Transcription and Amplification Kit. The primer sequences employed are detailed in Table 1 .

The isolation of nuclear and cytoplasmic RNA was accomplished employing Norgen’s RNA Nucleocytoplasmic Separation Kit in strict accordance with the manufacturer’s provided instructions. H226 and H1703 cells were collected into enzyme-free EP tubes, and 30 μL of solution buffer was added to each. The procedure began with an initial centrifugation step at 2000 rpm for two minutes, followed by a subsequent centrifugation at 14000 rpm for one minute. Upon removal of the supernatant, lysis buffer was introduced, and centrifugation was repeated. The supernatant, denoted as “tube 2”, was carefully transferred to a new tube, while the original EP tube’s precipitate retained the nuclear RNA and was labeled as “tube 1”. To both tube 1 and tube 2, 400μL of Buffer SK was added, followed by 200 μL of Buffer SK. Vortex mixing for 10 seconds was carried out prior to the addition of 200μL of anhydrous ethanol, followed by further vortex mixing for 10 seconds. The resultant mixture was subsequently loaded into centrifuge columns, with one column allocated for the nuclear fraction and another for the cytoplasmic fraction. Centrifugation was performed at 12000 rpm for one minute, facilitating the separation of nucleic material. Subsequently, the elution process involved the removal of the flow-through liquid from the collection tube, followed by the addition of 400uL of Wash Solution A, accompanied by centrifugation at 12000 rpm for one minute. This washing step was repeated thrice. Finally, a centrifugation at 12000 rpm for two minutes was conducted, after which the centrifuge columns were transferred to new tubes. Here, 50μL of Elution Buffer E was added, and further centrifugation ensued. The resultant eluate was collected, and the RNA concentration was measured.

Preparation for transfection commenced with the addition of 500μL of Opti-MEM serum-free medium approximately 30 minutes prior to the transfection process. Transfection mixture A, intended for the introduction of one plasmid into each well of a 6-well plate, was created as follows: plasmid DNA (3 µg) was combined with Opti-MEM serum-free medium to achieve a final volume of 50 µL, and the resultant mixture was gently homogenized. Transfection mixture B was prepared by adding 3μL of Lipofectamine 2000 (Invitrogen, USA) to Opti-MEM serum-free medium until the final volume reached 50 µL. This mixture was subsequently thoroughly mixed and allowed to incubate for a duration of five minutes. The two distinct solutions, transfection mixture A and transfection mixture B, were then merged and incubated for an additional 20 minutes prior to their introduction into the target cells. Subsequently, following a 6-hour incubation period, 1.5 mL of complete medium was introduced.

100 μL of cells at a concentration of 1×10⁴ were dispensed into each well of a 96-well plate, with three replicates per group, ensuring that the periphery of the plate was fully surrounded by sterile phosphate-buffered saline (PBS). Following a 48-hour incubation period, 10 μL of the CCK-8 assay solution were added to each well while taking care to prevent bubble formation. Subsequently, the 96-well plate was incubated at room temperature for one hour before being subjected to machine-based testing.

Following a 48-hour period post-transfection, 5×10³ cells from each experimental group were inoculated onto individual wells of a 6-well plate, with three wells allocated per sample. Subsequently, crystal violet staining was performed once cellular clusters had formed.

To conduct the invasion experiment, an extracellular matrix coating solution was meticulously prepared, and precisely 50 µL of this solution was aseptically dispensed into each well. Subsequently, a serum-free cell suspension, comprising 2×10⁵ cells per well, was introduced into the upper compartment of the transwell system. The lower compartment was filled with a culture medium supplemented with 20% Fetal Bovine Serum (FBS). Following a 48-hour incubation period, the transwell chambers were carefully extracted. The cells were subsequently subjected to fixation using methanol for a duration of seven minutes, followed by a 7-minute staining procedure with crystal violet. Following the staining protocol, the cells underwent a thorough washing process. The membrane, bearing the adherent cells, was affixed onto a microscope slide and subsequently subjected to microscopic examination, with three random fields captured for further analysis. The migration experiment, in contrast, did not entail the utilization of matrix adhesive. However, all other procedural steps remain consistent with those employed in the invasion experiment.

In this study, statistical analyses were conducted utilizing the Student’s t-test and the χ² test. The difference between two groups was assessed via a two-tailed t-test analysis, with a significance threshold set at p < 0.05. Survival analysis was conducted employing the Kaplan−Meier method, and intergroup differences among patients were compared utilizing the log-rank test.

We conducted a comprehensive statistical analysis of key clinical attributes pertaining to LUSC patients, as delineated in Table 2 . Principal component analysis (PCA) indicated that the selected samples could be effectively categorized into a single cohesive group for subsequent analyses, as presented in Figure 1A . Subsequently, a rigorous differential expression analysis of lncRNAs was undertaken on the samples, yielding a volcano plot that depicted the expression variations across all lncRNAs. Within this analysis, a total of 160 lncRNAs exhibited a significant upregulation, while 110 lncRNAs displayed a significant downregulation, as illustrated in Figure 1B . In order to screen for lncRNAs with potential synergistic effects or similar regulation, and to help explore the functions of lncRNAs, clustering analysis was conducted. The clustering analysis results showed the expression trend of all differential lncRNAs in all samples, as shown in Figure 1C .

By conducting an in-depth analysis of the correlation between overall survival (OS) and lncRNAs through univariate Cox proportional hazard regression, we successfully identified 271 lncRNAs with significant prognostic implications. Subsequently, a refined selection process involving multifactorial Cox proportional hazard regression analysis led to the identification of twelve lncRNAs. To provide insight into the expression patterns of these selected lncRNAs in LUSC and adjacent tissues, we performed t-tests and generated corresponding box plots. In the graphical representation, the designation “N” pertains to the adjacent cancer sample group, while “T” signifies the cancer sample group. Our analytical findings unveiled noteworthy disparities in the expression levels of these 12 lncRNAs between the cancer and adjacent cancer samples. Among the identified lncRNAs, seven (BRE-AS1, CCL15-CCL14, DNMBP-AS1, LINC00482, LOC100129034, MIR22HG, PRR26) exhibited significant downregulation, while five (FAM83A-AS1, LINC00628, LINC00923, LINC01341, LOC100130691) demonstrated notable upregulation, as depicted in Figure 2 .

Table 3 presents the predicted outcomes concerning the target genes associated with the 12 identified lncRNAs. It is noteworthy that the target genes of the differentially downregulated lncRNAs exhibited a pronounced enrichment in pivotal biological pathways, specifically those linked to metabolism, cancer, MAPK signaling, and PI3K-Akt signaling pathways ( Figure 3 ). Conversely, the differentially upregulated lncRNAs demonstrated enriched target genes predominantly within metabolic and PI3K-Akt signaling pathways ( Figure 4 ). Additionally, an overarching analysis of both downregulated and upregulated lncRNA target genes revealed a predominant enrichment in functions related to protein binding, cellular components, and cellular transformation, as depicted in Figures 5A, B . This extensive array of functions encompasses vital processes such as lung epithelial phosphorylation and cellular hypermetabolism, both of which play significant roles in the initiation and progression of LUSC.

In our assessment of the expression profiles of BRE-AS1, CCL15-CCL14, DNMBP-AS1, LINC00482, LOC100129034, MIR22HG, PRR26, FAM83A-AS1, LINC00628, LINC00923, LINC01341, and LOC100130691 within H226 and H1703 cells, we observed that relative to normal lung epithelial cells, BRE-AS1 and CCL15-CCL14 exhibited downregulation in both H226 and H1703 cells, while LINC00923 and LINC01341 exhibited upregulation in H226 and H1703 cells, respectively. These findings are graphically represented in Figure 6 . The expression levels of the remaining lncRNAs in H226 and H1703 cells exhibited slight deviations from the predicted outcomes, which could be attributed to the inherent heterogeneity among tumor cells. Given the detectability of upregulated lncRNAs in patient fluids and various samples, their study vis-à-vis their impact on tumorigenesis is often more feasible. Consequently, LINC00923 and LINC01341 were prioritized for further investigation. Initially, we selected these upregulated lncRNAs, LINC00923 and LINC01341, to assess their nucleocytoplasmic expression in H226 and H1703 cells. Our analyses indicated that LINC00923 and LINC01341 were predominantly localized within the nuclei of H226 and H1703 cells, as depicted in Figure 7 . Given the conspicuous nuclear presence of LINC00923 and LINC01341, we conducted a comprehensive evaluation of their roles in the proliferation, invasion and migration ability of LINC00923 and LINC01341 in LUSC H226 and H1703 cells. Employing Antisense Oligonucleotide (ASO) primers, we embarked on a series of experiments encompassing CCK-8 assays, colony formation assays, and transwell assays. These assays were conducted to assess the proliferation, invasion, and migration capabilities of H226 and H1703 cells in response to the knockdown of LINC00923 and LINC01341. The transfection efficiency was confirmed through quantitative reverse transcription PCR (qRT-PCR), as illustrated in Figures 8A–D . Our results from the CCK-8 assay ( Figures 8E–H ) and the colony formation assay ( Figures 8I–L ) demonstrated that the downregulation of LINC00923 and LINC01341 exerted inhibitory effects on the proliferation of H226 and H1703 cells. Furthermore, our transwell experiments corroborated that the suppression of LINC00923 and LINC01341 led to attenuated invasive and migratory capabilities of H226 and H1703 cells, as illustrated in Figure 9 .