Systemic analyses of expression patterns and clinical features for GIMAPs family members in lung adenocarcinoma

Absent expression of GIMAPs family members in patients with LUAD

In order to reveal the different potential therapeutic value of GIMAPs in LUAD patients, the mRNA expression of GIMAPs in 20 types of cancers were initially measured and compared to normal tissues by Oncomine database (Figure 2, Table 1). The under- expression of GIMAP1/2/4/5/6/7/8 were found in lung cancer tissues consistently, while neither GIMAP3 nor GIMAP1-GIMAP5 presented available data. In Okayama Lung dataset, GIMAP1 under-expression was found in LUAD tissues compared with normal tissues with a fold change of -3.359 (p=1.37E-13) [9], while Hou observed a 2.848-fold decrease in GIMAP1 mRNA expression in LUAD samples (p=4.73E-18) [10]. Similarly, significant down-regulation of GIMAP4 was also found in Selamat Lung [11], Okayama Lung [9], Hou Lung [10] database. GIMAP5 was also found in Su Lung [12], Selamat Lung [11], Okayama Lung [9], Hou Lung [10] database. GIMAP6 was also found in Landi Lung [13], Su Lung [12], Okayama Lung [9], Hou Lung [10] database. GIMAP7 was also found in Selamat Lung [11], Okayama Lung [9], Hou Lung [10] database. GIMAP8 was also found in Selamat Lung [11] database. However, GIMAP2 was found in the large cell lung carcinoma not in the LUAD with lower mRNA expression. Next, the mRNA expression patterns of GIMAPs were further validated by UALCAN using the resource based on RNA-seq from TCGA database. The mRNA expressions of other GIMAPs members were all found to be significantly down-regulated in LUAD tissues compared to normal samples (Figure 3).

Furthermore, the protein expression patterns of GIMAPs were examined to describe gene expression from protein level which is closer to the most primitive form of the disease (Figure 4). Firstly, the protein expression patterns of GIMAPs were measured by UALCAN using the resources based on Clinical Proteomic Tumor Analysis Consortium (CPTAC) database. Unfortunately, there was only phosphoprotein not total protein data of GIMAP5, and the protein expression has no significant difference, while other GIMAPs have a high significant difference. Next, the Human Protein Atlas was used to explore the immunohistochemistry images of GIMAPs. GIMAP1/2 proteins were low expressed in normal lung tissues, whereas not expressions of them were observed in LUAD tissue (Figure 5A, 5B). Also, medium protein expressions of GIMAP4/6 were expressed in normal lung tissues, while not or low protein expressions of them was observed in LUAD tissue (Figure 5C, 5D). Moreover, GIMAP7/8 would have high protein expressions in normal lung tissues compared with that in LUAD sample (Figure 5E, 5F).

Generally, all the results above showed that transcriptional and protein expressions of GIMAPs were under-expressed in patients with LUAD.

Predicted functions and pathways of GIMAPs in LUAD patients

To explore the biological function of GIMAPs, we constructed a network of GIMAPs and their 20 related genes by GeneMANIA (Figure 6A). Most of the related genes were in the category named shared protein domains, which means two gene products are linked if they have the same protein domain.

Moreover, GO functions and pathways of GIMAPs and their 20 related genes were analyzed by WEB-based Gene SeT AnaLysis Toolkit (WebGestalt). The biological processes such as GO:0051052 (regulation of DNA metabolic process), GO:0038034 (signal transduction in absence of ligand), GO:0090559 (regulation of membrane permeability), and GO:0006260 (DNA replication) were remarkably regulated by the GIMAPs in LUAD (Figure 6B and Table 2). Cellular components, including GO:0019867 (outer membrane), GO:0005759 (mitochondrial matrix), GO:0005657 (replication fork), GO:0005811 (lipid droplet) and GO:0098798 (mitochondrial protein complex) were significantly associated with the GIMAPs alterations (Figure 6C). Besides, GIMAPs also prominently involved in the molecular functions, such as GO:0001882 (nucleoside binding), GO:0019001 (guanyl nucleotide binding), GO:0004386 (helicase activity), GO:0019205 (nucleobase-containing compound kinase activity) and GO:0140097 (catalytic activity, acting on DNA) (Figure 6D).

In KEGG analysis, these pathways including hsa01521 (EGFR tyrosine kinase inhibitor resistance), hsa05210 (colorectal cancer), hsa04210 (apoptosis), hsa05014 (amyotrophic lateral sclerosis) and hsa05213 (endometrial cancer) were associated with the functions of GIMAPs in LUAD (Figure 6E and Table 3). While in the Reactome analysis, the most significantly enriched terms were involved in R-HSA-109606 (intrinsic pathway for apoptosis), R-HSA-111453 (BH3-only proteins associate with and inactivate anti-apoptotic BCL-2 members), R-HSA-114452 (activation of BH3-only proteins), R-HSA-109581 (apoptosis), and R-HSA-5357801 (programmed cell death) (Figure 6F and Table 4).

After GO and pathway analyzing, immune infiltration of GIMAPs was estimated by Tumor IMmune Estimation Resource (TIMER). The scatterplots of GIMAPs showed a purity-corrected value and statistical significance. All GIMAPs had negative associations with tumor purity, which means these genes were highly involved in the progression of immune infiltration in LUAD (Figure 7 and Table 5).

Taken together, the results showed that GIMAPs played a potential promotive role in the regulation of the DNA metabolic process and involved in apoptosis, as well as immune infiltration.

Association of transcriptional and protein expression of GIMAPs family members with clinicopathological features of LUAD patients

Further, we analyzed a relationship between GIMAPs’ expression with the clinicopathological characteristics of LUAD patients by UALCAN, including individual cancer stage, patient’s gender, and tumor grade. The mRNA expression of GIMAPs based on RNA-seq was from TCGA database and protein expression from CPTAC. The mRNA expressions of GIMAPs were remarkably correlated with patients’ individual cancer stages, and patients who were in more advanced cancer stages tended to express lower mRNA expression of GIMAPs (Figure 8). Similarly, mRNA expressions of GIMAPs were significantly related to patient’s gender, and male patients tended to express lower mRNA expression (Figure 9). Besides that, protein expression of GIMAPs were significantly related to tumor grade, as tumor grade increased, the protein expression of GIMAPs tended to be higher (Figure 10). In short, these results suggested that 8 GIMAPs family members were significantly associated with clinicopathological features in LUAD patients.

Prognostic value of mRNA expression of GIMAPs in LUAD patients

To analyze the prognostic value of GIMAPs in LUAD patients, we carried out a prognostic evaluation for the mRNA expression of GIMAPs in LUAD patients by GEPIA. As the results showed, mRNA expressions of all GIMAPs family members were significantly associated with LUAD patients’ prognosis, higher mRNA expression of GIMAP1 (HR=0.68, and p=0.0098), GIMAP2 (HR=0.69, and p=0.016), GIMAP4 (HR=0.7, and p=0.017), GIMAP5 (HR=0.63, and p=0.0026), GIMAP6 (HR=0.58, and p=0.00035), GIMAP7 (HR=0.57, and p=0.00018), GIMAP8 (HR=0.57, and p=0.00023), GIMAP1-GIMAP5 (HR=0.63, and p=0.0029) were significantly associated with longer overall survival of LUAD patients (Figure 11). These results indicated that higher mRNA expressions of GIMAPs could increase the survival time of LUAD patients and they may be exploited as potentially useful biomarkers to predict the LUAD patients’ survival.

The molecular-disease association of GIMAPs with multiple diseases

After analyzing the prognostic value of GIMAPs in LUAD patients, we further explore the relationship of GIMAPs with various diseases by the Open Targets Platform. GIMAP1/4/6/7/8 were predicted to have a higher association with LUAD filtered by RNA expression in this study, while GIMAP2 was predicted to have an association with lung carcinoma and GIMAP2 was predicted to have an association with sepsis (Figure 12). All in a word, GIMAPs had a close relationship with LUAD and they may be predicted as potential therapeutic targets for LUAD patients.

Identification of differentially expressed GIMAPs at transcriptional level

Oncomine (https://www.oncomine.org/resource/login.html) provides robust, peer-reviewed analysis methods and a powerful set of analysis functions to compute gene expression signatures [26]. In this study, Oncomine database is used to identify the transcriptional expressions of GIMAPs family members in LUAD tissues with their corresponding adjacent normal control samples. The data in our study was compared by t-test and the threshold statistical parameters were that, p value < 0.0001, fold change = 2, gene rank = 10%, and data type was mRNA.

UALCAN (http://ualcan.path.uab.edu) is a comprehensive and interactive web resource based on RNA-seq of 26 cancer types from TCGA database [27]. To determine the reliability of the differential expressed data, UALCAN database was chosen to further verify. In this study, the mRNA expressions of different GIMAPs between LUAD tissues with normal tissues were performed in the TCGA-LUAD dataset with a statically significant p < 0.001.

Identification of differentially expressed GIMAPs at protein level

Besides the transcriptional analysis from TCGA, UALCAN database could provide protein expression analysis using data from CPTAC Confirmatory/Discovery dataset for LUAD [28]. The proteome-based subtypes could be observed in external cancer proteomic datasets. The protein expressions of GIMAPs family members between LUAD tissues with normal tissues were performed with a statically significant p < 0.001.

The Human Protein Atlas (https://www.proteinatlas.org) is a website that contains protein expression data for near 20 highly common kinds of cancers based on representative immunohistochemistry data [29]. In order to avoid the result bias caused by data from a single database source, the Human Protein Atlas was applied to identify the tumor-type specific proteins expression patterns of GIMAPs that are differentially expressed in LUAD patients. The immunohistochemistry images of GIMAPs expressed in LUAD and normal tissues could be directly visualized.

Construction of related genes network and functional analysis

GeneMANIA (http://www.genemania.org) is a flexible web interface for generating hypotheses about gene function using available genomics and proteomics data [30]. In our study, the GIMAPs family members were submitted to the GeneMANIA to illustrate the functional association network among GIMAPs and their related genes. The advanced statistical options used were as follows: max resultant genes = 20, max resultant attributes = 10, using the automatically selected weighing method.

WebGestalt (http://www.webgestalt.org) is a functional enrichment analysis web tool with continuously updated and effectively reduce data redundancy [31]. GO functions and pathways of GIMAPs and their 20 related genes were enriched by WebGestalt. The Method of Interest is selected in Over-Representation Analysis (ORA). The GO functional enrichment was performed in the biological process no Redundant (BP), cellular component no Redundant (CC), molecular function no Redundant (MF). And the pathway analysis was performed in KEGG and Reactome pathway.

TIMER (https://cistrome.shinyapps.io/timer/) is a comprehensive web server using the microarray expression values for calculating a systematic analysis of immune infiltrates across diverse cancer types [32]. The immune infiltration estimation of GIMAPs was performed in LUAD by TIMER. The scatterplots of GIMAPs was generated to show the purity-corrected partial Spearman’s rho value and statistical significance. The positive purity value expected genes highly expressed in the tumor cells, and the opposite is expected for genes highly expressed in the microenvironment.

Clinicopathological analysis of GIMAPs in LUAD

Furthermore, UALCAN was used to analyze the association between the mRNA or protein expressions of GIMAPs in LUAD tissues with their clinicopathologic parameters such as individual cancer stage, patient’s gender, and tumor grade. The results could be got directly by selecting the clinicopathological grouping options integrated into the UALCAN database. In particular, only the tumor group could be divided into different clinicopathological groups. The statically significant p is less than 0.001.

Survival analysis of GIMAPs

The prognostic value of mRNA expression of distinct GIMAPs in LUAD was analyzed by using GEPIA (http://gepia.cancer-pku.cn/index.html) [33]. which could analyze the RNA sequencing expression data of 9,736 tumors and 8,587 normal samples from the TCGA and the GTEx projects. In this study, patients with LUAD were divided into high and low expression groups based on median values of mRNA expression. The statically significant difference was considered when a p-value < 0.05.

GIMAPs expression patterns under different biological conditions

The Open Targets Platform (https://www.targetvalidation.org) is a comprehensive and powerful data integration for accessing and visualizing potential drug targets related to diseases [34]. To further explore the relationship between GIMAPs with LUAD among multiple diseases, the Open Targets Platform was used to show the available evidence for GIMAPs-disease associations. Data type was filtered by mRNA expression in this study. All methods and Strategies above could be seen in the flow chart below (Figure 1).