CBioPortal (http://www.cbioportal.org/) is widely used to investigate and visualize multidimensional cancer genomics data[15]. The types of comprehensive gene data in CBioPortal include those pertaining to somatic mutations, DNA copy number changes, mRNA and microRNA expression, among others. A dataset comprising information derived from 307 patients with ovarian serous cystadenocarcinoma contained mRNA data (RNA Seq V2) (TCGA, Firehose Legacy) and was selected to analyze genetic variation. In the present study, mutations, putative copy number alterations from GISTIC, and mRNA expression z-scores (RNA Seq V2 RSEM) were selected for genomic profiles. The data on the relationships between CNA (relative linear copy number) and the mRNA expression of each laminin member were plotted. A Kaplan-Meier plot provided insights into laminin gene mutations and their relationships with OS and DFS in OC patients. A log-rank test was used to determine the significance of the divergence in the survival curve. When p < 0.05, divergence was considered significant.

Oncomine (http://www.oncomine.org) is an microarray database of carcinoma and comprehensive platform intended for mining information about cancer genes [16]. Oncomine contains cancer mutation profiles, gene expression data, and relevant clinical information that can be used to identify new biomarkers or therapeutic targets. It combines RNA and DNA-seq data from GEO, TCGA, and reported literature sources. In our study, we obtained the transcriptional expression of laminins between cancer and relevant ovarian tissues. The differences in transcriptional expression were contrasted with the Student’s t-test, and the following parameters were set: p-value: 0.01, fold change: 1.5, gene rank: all, data type: mRNA, analysis type: Cancer vs. Normal analysis.

The GEPIA2 (http://gepia2.cancer-pku.cn/) web server is a reference source used for gene expression analyses. It is based on the Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) databases [17]. In the present study, the threshold of |log2FC| was 1, the p-value cutoff was 0.05. Spearman’s correlation coefficients were calculated. The function Similar Genes Detection was used to search for genes similar to those encoding laminins.

The Human Protein Atlas (https://www.proteinatlas.org) contains data on nearly 20 types of cancer based on immunohistochemical (IHC) expression[18]. It furnishes tissue and cell distribution information on 24,000 human proteins and is available in the public domain. In the present study, laminin expression in OC and normal tissues was compared via IHC. Antibody names and stain intensities are indicated in the figures.

The Kaplan-Meier plotter database (http://kmplot.com/analysis/) is an analytical prognostic database used for the analysis of malignant tumors. It is used to assess the impact of over 54,000 genes in the prognosis of 21 different types of cancer[19]. In medical training and scientific research, this database is used to explore the relationships between gene expression and tumor prognosis. The Kaplan-Meier plotter bioinformatics analysis platform can be used to inquire into the prognostic value of 12 laminin genes. Based on their median mRNA expression levels, tumor patients were separated into two groups using the Kaplan-Meier plotter and were verified based on K-M survival curves. The risk ratio (HR) and the log-rank p-values of the 95% confidence interval (CI) were used to determine OC patient OS and PFS. p < 0.05 was considered statistically significant.

The algorithm in TIMER (https://cistrome.shinyapps.io/timer/) is used to systematically analyze immune cell infiltration in cancer [20]. The associations between each laminin member and massive immune infiltration in OC were investigated. The immunocytes involved included B cells, CD4⁺ T cells, CD8+ T cells, neutrophils, macrophages, and dendritic cells. Spearman’s correlation coefficients were calculated.

GEO is a communal functional genomic database, which includes a large number of array and sequence data [21]. The two gene chip profiles GSE131978[22] and GSE58470 datasets [23] were obtained from NCBI-GEO. Users can use GEO2R to compare sets of samples in order to classify disparate genes expressed. In this study, the GEO2R online tools were used to distinguish DEGs (Differential expressed genes) between ovarian tumors and adjacent normal tissues, by the cut-off criteria of adjusted p < 0.05 and |log2FC| > 1.

Metascape is a web-based portal that provides inclusive gene list commentary and analysis resources [24]. We selected 20 genes similar to each laminin-family gene from GEPIA2. To identify the biological functions and pathways of these genes, we used Metascape to analyze Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment. Gene Ontology (GO) term enrichment analysis was performed to determine the functions of the target host genes in terms of biological process, cell composition, and molecular function.

We used the GeneMANIA (http://www.genemania.org/) and STRING (https://string-db.org) [25] databases to explore the gene–gene and protein–protein interaction networks of the laminins. GeneMANIA can use large functionally linked data to identify other genes related to target genes. STRING is used to search for known protein interactions. Interactions revealed by STRING are primarily based on the confidence score (reliability index) as well as other collateral information such as protein domains and three-dimensional protein structures.

We evaluated genetic alterations in laminins and their relationships with the OS and DFS of OC patients registered in the cBioPortal database. Figure 1A depicts high laminin mutation rates in the OC samples. Laminin gene expression levels were altered in 191 samples obtained from 303 ovarian cancer patients (63%), and LAMA5 was observed to be the most frequently mutated gene (30%). Other genes with high mutation rates were LAMC1(16%), LAMC2 (12%), LAMA3 (11%) and LAMB1 (11%). Laminin gene amplification and mRNA alteration occurred relatively more often in OC patients. The LAMB2 and LAMC1 mRNA expression levels were significantly and positively correlated with their relative linear copy number alteration (CNA) values. However, the LAMA1/A3/A5, LAMB3, and LAMC2 mRNA expression levels were only weakly correlated with their relative linear CNA values (Supplementary Figure S1). The foregoing results indicated that genetic changes in laminins were correlated with poor OS (Figure 1B, p = 0.0459) and DFS (Figure 1C, p = 0.0227) in OC patients. They also demonstrated that changes in laminin genes and copy numbers occurred in certain patients with OC and might be related to poor prognosis.

The mRNA expression levels of 12 laminin members were evaluated for OC patients registered in the ONCOMINE database. ONCOMINE includes data on extensive independent studies conducted using various patient cohorts. Figure 2A shows that the mRNA expression levels of 12 laminin family members were detected using the ONCOMINE database and were compared against normal tissues. At least one dataset indicated that the expression levels of LAMA5 and LAMC2 mRNAs were significantly upregulated. In contrast, the expression levels of LAMA2/A4 and LAMB2 mRNAs were significantly downregulated. However, certain studies demonstrated that LAMA3, LAMB1/B3 and LAMC1 were highly expressed in OC whereas other reports presented the opposite findings (Supplementary Table S1).

We used GEPIA2 to evaluate laminin mRNA expression in OC and normal ovary tissues. The mRNA expression levels of LAMA5, LAMB3, and LAMC2 were higher, while those of LAMA1/A2/A4 and LAMB1/B2 were lower in OC tissues than normal ovary tissues (Figure 2B). Our findings suggested that LAMA5, LAMB3, and LAMC2 were overexpressed, whereas the expression levels of LAMA2/A4 and LAMB1/B2 were downregulated in OC tissues.

We used the Human Protein Atlas (HPA) to explore laminin protein expression patterns in OC (Figure 3). Images depicting the results of immunohistochemical (IHC) staining performed for 11 or 12 samples were used for comparison. This revealed the expression of LAMA1 (six positive, six negative), LAMA2 (zero positive, twelve negative), LAMA3 (eleven positive, one negative), LAMA4 (five positive, seven negative), LAMA5 (one positive, eleven negative), LAMB1 (seven positive, five negative), LAMB2 (seven positive, five negative), LAMB3 (twelve positive, zero negative), LAMB4 (nine positive, two negative), LAMC1 (nine positive, two negative), LAMC2 (twelve positive, zero negative), and LAMC3 (zero positive, 12 negative) laminin proteins. The number of LAMA1-positive and LAMA1-negative cases was similar for the OC samples, but LAMA1 was not expressed in normal ovary tissues. LAMA3 and LAMB1/B3 proteins were expressed at medium or low levels in OC tissues but protein expression was not detected in normal ovary tissues. LAMB2 and LAMC1/C2 were expressed at high, moderate, or low levels in OC tissues but protein expression was not detected in normal ovary tissues. LAMB4 protein was expressed at medium or low levels in OC tissues, but their expression was not detected in normal ovary tissues. The LAMA4 protein expression levels were equally low in both OC and normal ovary tissues. LAMC2 proteins were expressed at high, moderate, or low levels in OC tissues but were not expressed in normal ovary tissues. However, no LAMA2 or LAMC3 protein expression was detected in OC or normal ovary tissues.

We also investigated the associations between laminin mRNA and protein expression. LAMA3 mRNA and protein expression were strongly correlated. There were moderate correlations between the mRNA and protein levels of LAMA1/A2/A4, LAMB1/B3, and LAMC2, LAMA5 and LAMC1 mRNA and protein expression levels were weakly correlated. There was no correlation between LAMB2 mRNA and LAMB2 protein expression, and no data were available for LAMB4 or LAMC3 mRNA and protein correlations (Supplementary Figure S2).

The preceding results suggested that the proteins expression levels of LAMA3, LAMB1/B2/B3, and LAMC1/C2 were higher in OC than normal tissues and there were varying degrees of correlation between the protein and mRNA expression levels of LAMA1/A2/A3/A4/A5, LAMB1/B3, and LAMC1/C2.

We used the Kaplan Meier plotter to investigate the impact of laminin mRNA level on survival in patients with OC. Upregulated expression levels of LAMA1/A4, LAMB1/B2, and LAMC1 and downregulated expression levels of LAMC2 were significantly and positively associated with poor OS in OC patients (Figures 4A–F). While upregulated expression levels of LAMA4/A5, LAMB1, and LAMC1 mRNAs were significantly and positively associated with poor PFS, elevated LAMA1/A2 expression levels were significantly and positively associated with better PFS (Figures 4G–L).

In general, the foregoing results showed that increases in LAMA4, LAMB1 and LAMC1 mRNA expression levels were positively related to poor OS and PFS, whereas upregulated LAMA5 mRNA level was positively related to worse PFS. Thus, increased LAMC2 mRNA expression may be considered to predict longer OS and upregulated LAMA1/A2 mRNA level is associated with better PFS in patients with OC.

We used the Kaplan Meier plotter to investigate the associations among laminin mRNA expression, OS, and PFS in patients at different clinical stages and pathological grades of OC. Table 1 shows that for patients with stage I or stage II OC, the LAMA1 and LAMB1 mRNA expression levels were strongly associated with worse OS (HR > 1), and the LAMA3 and LAMB4 mRNA expression levels were significantly related to better OS (HR < 1). For stage III OC patients, the LAMA1/A4, LAMB1, and LAMB3 mRNA expression levels were markedly associated with unfavorable OS, while the LAMB4 and LAMC2 mRNA expression levels were strongly related to favorable OS. For stage Ⅳ OC patients, transcriptional LAMB3 expression was significantly associated with longer OS, while transcriptional LAMA1/A3/A5 expression were significantly correlated with shorter OS. The data showed that for patients with grade 1 or 2 OC, the LAMA3, LAMB4 and LAMC2 mRNA expression levels were significantly correlated with longer OS; however, the LAMA1/A4 and LAMB1 mRNA expression levels were significantly related to shorter OS. For grade 3 OC patients, the LAMA1/A4/A5 and LAMB1 mRNA expression levels were significantly associated with poor OS.

Table 2 shows that the LAMA2/A3/A4, LAMB1 and LAMC2 mRNA expression levels were significantly associated with longer PFS in stage I or stage II OC patients, while the LAMA5 mRNA expression levels was significantly related to shorter PFS in the same patients. In stage III OC patients, the LAMA4/A5, LAMB1/B2, and LAMC1/C3 mRNA expression levels were significantly related to poor PFS. The LAMA3 and LAMA5 mRNA expression level were significantly correlated with unfavorable PFS in stage IV OC patients. The LAMA2/A3 and LAMC2 mRNA expression levels were markedly correlated with longer PFS in grades 1 and 2 OC patients, whereas the LAMA4 and LAMA5 mRNA expression levels were significantly associated with shorter PFS in grades 1, 2, and 3 OC patients. The LAMB2 and LAMC1/C2/C3 mRNA expression levels were markedly correlated with worse PFS in grade 3 OC patients.

Based on the OS and PFS results shown in Tables 1 and 2, upregulated expression levels of LAMA1/A4/A5 and LAMB1 may indicate poor prognosis in OC patients and LAMA5 might be considered the main negative prognostic factor for all stages and grades of OC. These results also suggest that LAMC2 expression may be used to predict a relatively longer prognosis for patients with OC.

We explored clinical data and information on laminin mRNA expression for OC patients in TCGA and extracted data from GTEx for the corresponding normal tissues. Data on 515 cases were available and the dataset was used to investigate the relationships of specific laminins via ROC (receiver operating characteristic) curve regression analysis. Our results were evaluated based on ROC curve AUC >0.70. We found that LAMA2 (AUC: 1.000; 95% CI: 0.999–1.000), LAMA4 (AUC: 0.881; 95% CI: 0.851–0.912), LAMA5 (AUC: 0.894; 95% CI: 0.859–0.928), LAMB1 (AUC: 0.889; 95% CI: 0.854–0.924), LAMB2 (AUC: 0.994; 95% CI: 0.990–0.999), LAMB3 (AUC: 0.977; 95% CI: 0.955–0.999), and LAMC2 (AUC: 0.992; 95% CI: 0.985–0.999) were qualified (Supplementary Figure S3).

Overall, the ROC analysis of the laminins in the OC sets indicated that LAMA2/A4/A5, LAMB1/B2/B3, and LAMC2 could be considered to accurately differentiate between tumor and non-tumor tissues.

Epithelial ovarian cancer (EOC) is susceptible to immune recognition[26]. CD8+TIL expression is associated with better prognosis in advanced EOC patients[27]. Based on the existing literature, we used the online TIMER database to explore associations between laminins and immune infiltration levels in OC patients. There were negative correlations between OC tumor purity and LAMA4 and LAMB1. The correlation ratio was the most considerable for LAMA4 (r = −0.526; p = 8.23E-36). LAMA4 was positively weakly correlated with macrophage (r = 0.19; p = 2.89E-05), neutrophil (r = 0.142; p = 1.87E-03) and dendritic cell (r = 0.171; p = 1.64E-04), while LAMB1 was positively weakly correlated with only macrophage (r = 0.17; p = 1.76E-04). Details of the relationships between immune cell types and laminin members are illustrated in Figure 5 and Supplementary Figure S4.

LAMA1 was negatively correlated with CD8+ T cell (r = −0.692; p = 6.37E-11), B cell (r = −0.303; p = 1.21E-11), neutrophil (r = −0.317; p = 1.10E-12) and dendritic cell (r = −0.32; p = 6.95E-13). LAMA5 was significantly and negatively correlated with macrophage (r = −0.207; p = 4.84E-06), neutrophil (r = −0.24; p = 1.01E-7) and dendritic cell (r = −0.249; p = 3.08E-08). LAMC1 was significantly and negatively correlated with CD8+ T cell (r = −0.233; p = 2.53E-07). However, LAMA2/A3, LAMB2/B3/B4, and LAMC2/C3 were not associated with any specific immune cell types (Supplementary Figure S4).

The development of platinum resistance in tumors leads to inferior outcomes and remains an obstacle in efficacious anticancer therapy. To investigate whether laminin genes could be implicated in the development of endogenous chemotherapy resistance, we retrieved data from GEO regarding the responses of OC patients to chemotherapy. GSE131978 defined patients with tumor progression times >12 months as being relatively platinum-sensitive and those with tumor progression times < six months as being platinum-resistant. LAMA1 and LAMB3 expression levels were upregulated in platinum-resistant OC patients (Supplementary Table S2). We also explored the GSE58470 dataset[28]. It includes the parental cisplatin-sensitive IGROV-1 and the platinum-resistant variant IGROV-1/Pt1 OC cell lines. LAMC2 expression was significantly upregulated in the platinum-resistant mutants (Supplementary Table S2). Hence, LAMA1, LAMB3, and LAMC2 may play important roles in the development of platinum resistance.

In the gene interaction network generated based on the use of GeneMANIA[29], we observed that laminin family genes might be associated with ITG and netrin family genes (Figure 6A). We used STRING[25] to analyze the protein-protein interaction network (local clustering coefficient = 0.875). Laminin proteins were co-expressed with ITG, nidogen, dystrophin, dystroglycan, collagen types XVII and VII, and alpha 1 chain proteins (Figure 6B).

The GO and KEGG databases revealed laminin functions and the genes that were significantly associated with their alterations. Biological processes such as GO:0030198 (extracellular matrix organization), GO:0001568 (blood vessel development), and others were remarkably influenced by laminin mutations in OC (Supplementary Figure S5A). Molecular functions such as GO:0004060 (arylamine N-acetyltransferase activity), GO:0001228 (DNA-binding transcription activator activity, RNA polymerase II-specific), GO:0005201 (extracellular matrix structural constituent), GO:0005509 (calcium ion binding), GO:0005518 (collagen binding), GO:0050839 (cell adhesion molecule binding), and others were significantly associated with laminin alterations (Supplementary Figure S5B). Laminin mutations also significantly affected cellular components including GO:0031012 (extracellular matrix), GO:0070161 (anchoring junction), and GO:0005938 (cell cortex) (Supplementary Figure S5C). In the KEGG analysis, ten pathways including ko04512 (ECM-receptor interaction), ko04010 (MAPK signaling pathway), ko05412 (arrhythmogenic right ventricular cardiomyopathy (ARVC)), ko04210 (apoptosis), and others were affected by laminin mutations in OC (Supplementary Figure S5D).