Identification of NSCLC-related DEGs in LUAD
The intersection analysis showed that 76 DEGs overlapped among the GSE118370, GSE63459, and GSE27262 datasets (Figure 1A). The top ten targets were VIPR1, ADARB1, PECAM1, CLDN18, NOTCH4, FHL1, TIMP3, TCF21, MACF1, and CD36, considered key genes in NSCLC. Notably, FHL1 has been considered as a tumor suppressor gene that exerts an inhibitory effect through various mechanisms underlying cancer growth, invasion, and metastasis[9,12-14]. A recent study found that FHL1 can also play a promoting role in tumors. Therefore, our study aimed to identify the function of FHL1 in NSCLC progression using bioinformatics.
Downregulated expression of FHL1 in LUAD and LUSC
Transcriptomic data were analyzed to systematically investigate the mRNA expression levels of FHL1 across diverse cancers based on TCGA and GTEx databases (Figure 1B). The results showed that the FHL1 expression level was significantly lower in 17 types of tumors than that in adjacent normal tissues, especially in LUAD and LUSC. GEPIA and GEO databases were used further to validate the findings of TCGA and GTEx web and displayed a low expression of FHL1 in LUAD and LUSC tissues compared with that of normal samples (Figure 1C-D). In addition, UALCAN was performed to examine the protein expression level of FHL1 based on the CPTAC database (Figure 1E), and the results showed that the FHL1 protein was remarkably decreased in tumor tissues compared to that of normal samples. Immunohistochemistry images from the HPA database showed results similar to those of CPTAC (Figure 1F). These findings suggest that both mRNA and protein expression levels of FHL1 are downregulated in LUAD and LUSC compared to those of normal tissues.
Correlation Between FHL1 and Clinical Characteristics in NSCLC
To clarify the association between the transcription of FHL1 and clinicopathological parameters in LUAD and LUSC patients, the information was analyzed using Wilcoxon and logistic regression analysis (Figure 2). The results of multiple subgroup analysis showed that FHL1 expression was significantly associated with age, sex, and smoking status and was downregulated in younger (age < 65 years), male, and smoking patients (p < 0.05, respectively). However, FHL1 expression was not correlated with other clinicopathological parameters, such as pathologic stage, TNM stage, and anatomic neoplasm subdivision (Figure 2A). Moreover, no statistical correlation was found between FHL1 expression and clinicopathological characteristics in LUSC, including age, sex, smoking status, TNM stage, pathologic stage, and anatomic neoplasm subdivision (Figure 2B).
Diagnosis and Prognosis value of FHL1 in LUAD and LUSC
ROC curve analysis was performed to identify the role of FHL1 in distinguishing LUAD and LUSC samples from normal samples. As shown in Figure 3A and B, the area under the curve (AUC) of FHL1 was 0.993 (95% CI, 0.989–0.998) in LUAD and 0.998 (95% CI: 0.995–1.000) in LUSC, indicating that FHL1 may be a strong identification biomarker for LUAD and LUSC. The curves illustrate the association between FHL1 expression and overall survival (OS) and disease-specific survival (DSS), which helps to investigate the prognostic value of FHL1 in LUAD and LUSC based on K-M analysis. Figure 3A shows that LUAD patients with low expression of FHL1 were associated with shorter OS (P = 0.025) and poor DSS (P = 0.054). However, the levels of FHL1 were not significantly correlated with OS and DSS in LUSC patients (P = 0.475, P = 0.533, respectively) (Figure 3B). These findings suggest that low expression of FHL1 could be a promising biomarker to diagnose LUAD and LUSC, as well as the poor prognosis of LUAD patients. Because FHL1 expression is not associated with the prognosis and clinical characteristics of LUSC, the following study focused on the role of FHL1 expression in LUAD.
PPI Networks and Functional Annotations
To explore FHL1-correlated genes and FHL1-binding proteins, GGI and PPI were generated using GeneMANIA and STRING databases (Figure 4A and B). The correlation analysis suggested that the proteins (AKT1, IGFBP5, INPP5A, KCNA5, RBPJ, STAT3, STAT5A, STAT5B, and TTN) in the PPI network had a significant relationship with FHL1 expression, except for RING1 (Figure 4D). Figure 4C shows that FHL1 is associated with the biological functions of DNA-binding transcriptional activator activity, RNA polymerase l-specific, and RNA polymerase II repressing transcription factor binding and participates in the JAK-STAT signaling pathway, interleukin 15 mediated signaling pathway, and response to interleukin-9.
Additionally, volcano plots and heatmaps were generated to identify differentially expressed genes (DEGs, | log (fold change) | > 1.5, and adjusted P-value < 0.05), based on the TCGA database (Figure 5A and B). Metascape was used to explore the role of FHL1-related DEGs in LUAD patients (Figure 5C-F). The results showed that the FHL1-related DEGs were involved in ERKl and ERK2 cascades, cell differentiation, and blood circulation. Biological process analysis further revealed that FHL1-related DEGs of LUAD participated in humoral immune response, vascular processes in the circulatory system, and neutrophil-mediated cytotoxicity. Moreover, KEGG analysis showed that these genes were closely related to the metabolism of xenobiotics by cytochrome P450. A recent study confirmed that cytochrome P450 is a crucial mediator of ferroptosis, considered a novel therapeutic strategy to treat NSCLC. Altogether, FHL1 is strongly linked to the immune response. Therefore, the correlation between FHL1 and the anti-cancer immune response was investigated.
Correlation of FHL1 expression with infiltrating immune cells
To better understand the expression status of FHL1 in different cell types, the violin plots showed that FHL1 expression was the most frequent in immune cells, while only parts of stromal cells, malignant cells, and other cells were expressed (Figure 6A). Furthermore, establishing scatterplots to evaluate the correlation of FHL1 expression with the TIICs (Figure 6B) illustrates that FHL1 expression was positively associated with the level of B cells (r = 0.157, P = 5.21e-04), CD8+T cells (r = 0.207, P = 4.05e-06), CD4+T cells (r = 0.279, P = 4.41e-10), macrophages (r = 0.433, P = 1.45e-23), neutrophils (r = 0.275, P = 7.44e-10), and dendritic cells (r = 0.31, P = 2.40e-12), while being negatively related to tumor purity (r = −0.313, P = 1.03e-12). The enrichment score boxplot further validated that all six types of immune cells had a higher degree of immune infiltration in the high FHL1 expression group than that in the low FHL1 expression group (Figure 6C). Further, the correlation analysis based on the TISIDB database was used to confirm that the level of FHL1 was clearly correlated with TILs in diverse cancers (Figure 7A). As shown in Figure 7B, 26 TILs were closely related to FHL1 expression in LUAD. Specifically, FHL1 was significantly positively associated with 24 types of TILs but negatively associated with active CD4+T cells and CD56 dim cells in LUAD.
Correlation of FHL1 expression with immune molecules
TILs are important constituents of the tumor immune microenvironment and play a crucial role in antitumor efficacy and prognostic ability[18,19]. Immune checkpoints inhibit the anti-tumor immune response of TILs, contributing to tumor cell immune escape. To identify whether FHL1 impacts TIL infiltration via immune checkpoints, a correlation analysis between FHL1 expression and 47 immune checkpoint genes was performed (Figure 8A). The results showed that FHL1 was associated with most immune checkpoint genes, including CD274, CD48, CD80, VTCN, and PVR.
Then, correlation analysis was performed to explore the relationship between FHL1 expression and chemokines (Figure 8B). The results showed that FHL1 expression levels were markedly associated with CCL5 (r = 0.097, P = 0.0267), CCL17 (r = -0.138, P = 0.00166), CCL20 (r = -0.172, P = 8.31e−05), and CXCL8 (r = -0.104, P = 0.018). These results revealed that FHL1 participated widely in regulating immune molecules, thereby affecting immune cell infiltration.
Correlation of the genomic alteration between FHL1 and immune checkpoint
Further, investigation of the prognosis of immune checkpoints based on TISIDB showed that PD-L1, PD-L2, CD80, CD86, VSIR, PVR, LGALS9, and CD48 were downregulated, while VTCIN, CD112, TNFSF4, CD70, and TNFSF18 were upregulated in LUAD (Figure 9). Notably, not all the different expressions of immune checkpoints have a prognostic role in LUAD. Briefly, low expression of CD80 and CD48 and high expression of VTCN1 indicated high OS and/or DFS, but downregulated PVR and was significantly related to poor OS and DFS (P = 0.012, 0.034 respectively).
Moreover, mutation analysis revealed that FHL1 was altered in 2.3% of all study subjects, including missense mutations, splice mutations, truncating mutations, structural variants, amplifications, and deep deletions (Figure 10A). Figure 10B shows the correlation between FHL1 and immune checkpoints. In addition, alterations in PVR, NECTIN2, and HHLA2 have a co-occurrence tendency with FHL1 alterations. These results indicate that FHL1 may participate in regulating immune checkpoints in LUAD.