Introduction
Colorectal cancer, the third most prevalent malignant disease in the world, has had a substantial effect on people's health and increased the burden on patients' families. Annually, there are more than 1.85 million cases of CRC and 855,000 deaths [
1,
2]. Numerous studies have elucidated the clinical features and molecular mechanisms of CRC, as well as the key signaling pathways associated with tumor growth and metabolism [
3], the 3- or 5-year survival rate in Stage IV is far from expectation [
4]. Metastasis has been confirmed as major cause of CRC fatality. Due to the liver's extensive portal and arterial blood supply, CRC metastasis occurs there the most frequently. [
5]. According to the previous studies [
6], approximately 25–30% CRC cases are clinically diagnosed as liver metastases and about 50% of CRC carriers will develop symptoms of liver metastases. Liver metastases may occur in some CRC patients following radical resection, resulting in more than 50% patient death [
7]. With the increasing incidence of colon cancer [
8], National Comprehensive Cancer Network guidelines proposal the use of surgery in combination with radiation and chemotherapy for CRC, and it is currently the standard of care agreed upon worldwide [
9]. Though the treatment of CRC has improved, its prognosis is still unsatisfactory. Numerous previous studies, there are many biological markers related to the prognosis of CRC, such as
CXCL8,
KLK8,
IMPAT [
10‐
12]. Nevertheless, these biological markers and therapeutic targets have not significantly enhanced CRC patient survival. With the exploration of targeted drugs in recent years, immune checkpoint inhibitors (ICI) have achieved satisfactory safety and efficacy in the treatment of CRC patients. A variety of immunotherapy drugs, including pembrolizumab, nibulumab and ipilimumab, have been approved for the treatment of advanced CRC [
13]. Early screening for CRC is very important to improve its prognosis [
14]. As a consequence, the search for new biomarkers and their therapeutic targets is crucial to improve the prognosis of CRC.
Grid2 interacting protein (GRID2IP), a postsynaptic scaffold protein at synapses in parallel fibrous Purkinje cells, may link GRID2 to the actin cytoskeleton and various signaling molecules. According to previous reports, GRID2IP may be related to Alzheimer's disease [
15], nervous system development [
16] and hemophilia A [
17]. Yao et al. indicated that GRID2 in whole blood samples of Parkinson's disease (PD) patients was significantly higher than that in healthy controls, and there was evidence that Grid2 was a biological indicator of PD [
18]. In addition, GRID2IP can also increase the risk of irritable bowel syndrome [
19]. Huang, et al. recently confirmed that GRID2 is closely associated to gut microbiota. Specifically, GRID2 deficiency in the mouse model resulted in changes in the species richness and composition of the gut microbiota, which affected the function of the gut microbiota and caused disorder of the neuroactive ligand–receptor interaction [
20]
. Therefore, we hypothesized that GRID2IP is associated with the occurrence and prognosis of intestinal tumors. Based on the TCGA database, we found that GRID2IP in many kinds of cancer patients were significantly different from that in normal controls. Specially, it was up-regulated in breast cancer and lung cancer but low expressed in head and neck squamous cell carcinoma (HNSC) and renal chromophobe cell carcinoma (KICH). In spite of this, few studies have reported the relationship between GRID2IP gene and CRC. The effect of GRID2IP on the prognosis of CRC and its effect on tumor microenvironment remains unclear.
In our study, we used transcriptome data retrieved from the TCGA database to identify the effect of GRID2IP on prognosis and tumor-associated immune cells in CRC. Functional enrichment analysis and protein interaction network were also utilized to explain the prognostic mechanism of GRID2IP. In addition, the K–M survival curves were used to visualize the relationship between GRID2IP and overall survival in CRC. Finally, a prognostic nomogram and its correction curve were established. Taken together, GRID2IP maybe a potential biomarker and drug target for predicting the prognosis of CRC, which is conducive to guiding clinical treatment strategy in CRCs.
Materials and methods
Data collection and preprocessing
Since most CRC are colorectal adenocarcinoma or colorectal adenocarcinoma, we downloaded their transcriptome data information from the TCGA database and combined them as colorectal adenocarcinoma (COADREAD). A total of 521 COADREAD cases with gene expression data (HTSeq-FPKM) and the detailed clinical information from COADREAD samples were downloaded from TCGA Repository, which includes 51 colorectal paracancerous tissues and 647 colorectal tissues. Incomplete or invalid data were excluded. Clinical features that were unavailable or unknown were considered missing values. In addition, transcripts per million reads (TMP) were derived from the level 3 HTSeq-FPKM data, which were used to the further study. Transcriptome data were divided into two subgroups based on the GRID2IP of each sample: GRID2IP-high and GRID2IP-low. Moreover, the normal colon tissues were extracted from GTEx database. Since all data for the study were obtained from freely accessible online databases, ethical approval and patient consent were not mandatory. The characteristics of all patients' clinical data are summarized in Additional file
2: Table S1. The complete research process diagram is summarized in Additional file
1: Fig. S2.
The expression of GRID2IP between CRC and normal tissue
The differential expression of GRID2IP was calculated using disease status (tumor or normal) as a variable and presented in box plots and scatter plots. Receiver operating characteristic (ROC) curve analysis was utilized to assess GRID2IP's diagnostic value. The "pROC" and "ggplot2" packages were used to visualize the ROC curve.
Quantitative real-time PCR
The total cell RNA of CCD18, HT29 and SW480 were extracted by Trizol reagent (Takara, Japan). Quantitative PCR results were performed by Hieff@ Qpcr SYBR®Green Master Mix (Yeasen, shanghai, China) and CFX96 Touch Real-Time PCR System (Bio-Rad, Hercules, CA, USA) on the basis of the manufacturer’s instructions. Gene-specific primer sequences: human GRID2IP: gene-specific primer sequences: human GRID2IP: (Forward primer 5’-CCATTTGCCAGTGACTCCGA-3’, Reverse primer 5’-TGTGCTGGAAGAAGCTCTCG-3’).
Immunoblot analysis
CCD18, HT29 and SW480 cells were lysed by NP40 lysis buffer (1 × protease inhibitor mixture), centrifuged with 15000g, 12 min. Supernatants were added 1xloading buffer and then boiled 15 min. Mixtures were subjected to 6–10% SDS PAGE gel electrophoresis, transferring to PVDF membranes. The membrane was incubated with 5% skim milk for 1 h to block irrelevant proteins, following incubated with indicated antibody overnight at 4 °C. Next, the membrane was treated with secondary antibody for at room temperature 1.5 h. Proteins were visualized by Chemiluminescent reagents kits (Thermo Fisher Scientific, Waltham, MA, USA) and detected by FluorChem E (Cell Biosciences, USA), the antibody of GRID2IP purchased from abcam (Britain), GAPDH purchased from proteintech (China).
Gene set enrichment analysis (GSEA)
GESA which is a computer algorithm based on gene expression matrix [
21] can use a predefined genes concentrated to assess and the sorting of phenotypic correlation gene distribution trend, so as to determine its contribution to the phenotype. In our study, GSEA generated an ordered gene sequence of all genes according to their correlation with GRID2IP expression and then analyzed the significant survival differences between the GRID2IP-high and GRID2IP-low subgroups by GSEA (Each group was analyzed 1000 times). The expression profiles of our samples were input into GSEA as phenotypic markers, and pathways with GRID2IP enrichment in each phenotype were ranked using nominal
P values and normalized enrichment score (NES). The “ClusterProfiler” package [
22] was used to analyze the GSEA enrichment and visualization.
DEGs analyze and enrichment of GRID2IP
According to the median GRID2IP expression levels (0–50% and 50–100%), a total of 647 COADREAD patients were divided into high GRID2IP and low GRID2IP subgroups. The R package "DESeq2" [
23] was used to conduct Wilcoxon rank-sum test to identify DEGs groups between the high and low expressing GRID2IP, with log-fold change > 1.5 and adjusted
P value 0.05 as the threshold. The R packages "Enhanced Volcano" and "PheatMap" were also applied to draw heatmap and volcano plot. Besides, R package ggplot2 was used to visualize the enrichment analysis of differential genes in paracancerous and tumor tissues analyzed by “ClusterProfiler” package. The “GOplot” package is applied to calculate the zscore [
24].
Functional enrichment analysis of Immune microenvironment
Single sample GSEA from R package “GSVA” [
25] was powered to evaluate the immune infiltration level. The relative tumor invasion level of immune cells was quantitatively assessed by integrating the type signature gene list of published gene expression levels [
26]. The 24 types of immune cells were performed to evaluate the immune cells enrichment in tumor tissues. To detect immune cell infiltration and expression levels of different GRID2IP mRNA, Wilcoxon rank sum test and Pearson correlation analysis were performed. TISCH online database(
http://tisch.comp-genomics.org/) is applied to analyze the relationship between GRID2IP and TME. In the aspect of designing database parameter, we focus on the Maj-Lineage (Cell type). CRC (Cancer type); No parameter set (Cell type); All (Lineage for calculating correlation); No treatment(Treatment); Primary(Primary/Metastatic). In addition, the stromal score, ESTIMATE score and immune score was calculated using "estimateScore" algorithm. Furthermore, lollipop plot was generated using "ggplot2" package to delineate the correlation between GRID2IP and immune checkpoint inhibitor-related genes.
Clinical features analysis and prognosis assessment
All statistical analyses of this study were performed in the R package (v.4.2.1). Wilcoxon rank sum test and logistic regression were applied to clarify the relationship between clinical features and GRID2IP. Cox regression and Kaplan–Meier methods were used to investigate the clinical features associated with OS, DSS and PFI in TCGA database patients. A multivariate Cox analysis was employed to examine the influence of GRID2IP expression level on survival and other clinical features (stage, subtype, status of distant metastases, and histological grade). The median determined the cutoff value for the GRID2IP expression. The Kaplan–Meier analysis and the two-sided log-rank test were used to calculate the differences in 10-year OS, PFI, and DSS between the high and low GRID2IP subgroups. The nomogram based on the Cox regression model was created using the independent prognostic indicators discovered by multivariate analysis to estimate the odds of survival for 1, 3, 5 years, respectively. The "rms" package was employed to create nomograms with calibration plots and clinical features. Calibration plots were performed to analyze the nomogram prediction probabilities against observed events. Of note, the 45° line represented the actual value. C index represents the accuracy of nomogram prediction and the accuracy of different prognostic factors. All statistical tests in this investigation were two-tailed.
Statistical analysis
The data of quantitative PCR were presented with mean ± SD with at least three independent experiments. GraphPad Prism (v.8.0) was used to conduct a student's t test between groups, and p < 0.05 (two-sided) was considered statistically significant. The Wilcoxon test was applied to examine the differences between the two groups' expression. Chi-square test was used to investigate the correlation between GRID2IP expression and clinical features between the two subgroups.
Discussion
As a Purkinje fiber postsynaptic scaffold protein, GRID2IP is closely related to glutamate receptor delta 2 (GRID2). Base on previous studies, GRID2 may be a biological marker for the prognosis of gastric cancer, and upregulated GRID2 is an independent risk factor for gastric cancer [
32]. Therefore, we focused on the relationship of GRID2IP in CRC prognosis. High GRID2IP levels were found to be strongly associated with CRC prognosis in our study. In addition, we concentrated on the expression profile, clinicopathological significance, and clinical prognostic value of GRID2IP in CRC patients by analyzing TCGA cohort. As the result in our study, GRID2IP was substantially overexpressed in a variety of malignancies, including CRC, BRCA, CHOL, and LUAD. The ROC curve confirmed that using the expression level of GRID2IP to distinguish between normal tissue and tumor tissue has a certain specificity (AUC = 0.728). According to the GRID2IP expression level, we divided all samples into high and low GRID2IP subtypes and analyzed the DEGs between the two groups. Then, we enriched signaling pathways associated with MF, BP, CC using GO and KEGG. GSEA analysis showed that DEGs with high expression of GRID2IP were significantly enriched in GABA receptor, calcium ion and striated muscle contraction signaling pathway.
In previous studies, the occurrence and metastasis of cancer are inseparable from the immune system. Since the discovery of a major breakthrough in the use of immunotherapy to prevent and treat cancer in 1891, immunotherapy has gradually become a new cancer treatment method, such as surgery, radiotherapy, chemotherapy, and targeted therapy [
33]. Each of these immune cells has its own unique function and is interconnected. It was found that DC cells, as antigen-presenting cells, can further initiate tumor-related immune responses by activating CD8 + T cells [
34]. Macrophages, as indispensable immune cells in innate immunity, play a crucial role in the development and metastasis of cancer. Different types of macrophages function differently. Tumor cells are phagocytosed by pro-inflammatory M1 macrophages, whereas anti-inflammatory M2 macrophages promote tumor growth and invasion [
35]. Neutrophils are not only innate immune phagocytes that act an indispensable role in immune defense but also the link between inflammation and cancer, which plays an indispensable role in the development and spread of cancer [
36,
37]. In addition, Th1 cells and Th2 cells can exert potent anticancer effects by encouraging tumor interstitial remodeling and tumor tissue repair [
38].
Long-term survival in CRC patients has been linked to T-cell infiltration into the tumor bed, indicating a potential function for immune regulation in regulating tumor growth [
39‐
41]. Single-cell analysis suggested that GRID2IP was mainly expressed in immune cells, myofibroblasts and malignant cells. It is worth mentioning that the infiltration level of tumor-associated immunity, including T cells, neutrophils, macrophages, DC cells, Th1 cells, Th2 cells, T helper cells, and CD8 + T cells are significantly reduced in the GRID2IP-high group which points out that the high GRID2IP is mainly involved in the regulation of T-cell-dominated cellular immunity and macrophage-dominated innate immunity. In addition, the ESTIMATE and immune scores in the GRID2IP-high subgroup were lower than those in the GRID2IP-low subgroup, indicating a lower degree of immune cell infiltration in the GRID2IP-high subgroup. In addition, ESTIMATE score and immune score were lower in the GRID2IP-high subgroup than in the GRID2IP-low subgroup, indicating a lower degree of immune cell infiltration in the GRID2IP-high subgroup. Moreover, high GRID2IP expression was negatively correlated with most immune checkpoint inhibitor-related genes and HLA-related genes.
Another worth noting is that GRID2IP is differentially expressed in different clinical subgroups. We analyzed the differences in GRID2IP expression between different subgroups by Cox regression analysis, and found that GRID2IP was significantly correlated with some clinical features, including N-stage, M-stage, pathology stage, residual tumor, lymphatic invasion, CEA level, BMI. Moreover, the AUC diagnosed by the clinical ROC curve was 0.728, indicating that GRID2IP is a convincing biomarker for judging tumor tissue. The violin plot results indicated that patients with distant metastasis, lymph node metastasis or nodal invasion, wasting, CEA level > 5, and patients with residual tumor had poor prognosis, irrespective of gender, age and treatment effect. We also used Cox regression analysis to establish a GRID2IP-related model for predicting OS survival in CRC patients by calculating the cumulative total score for each independent prognostic factor score by nomogram. The results of the correction plots show that the model is reliable in predicting the survival rate of CRC patients.
Microsatellite instability a phenomenon in which defects in mismatch repair cause hypermutations, can be used to screen for Lynch syndrome and predict response to immune checkpoint inhibitors [
42]. It is a unique biological feature of CRC, which includes high mutation burden, tumor lymphocyte infiltration and increased production of mutation-associated neoantigens [
43]. Based on the previous studies, immune checkpoint inhibitors are highly effective in the treatment of CRC patients with MSI subgroups [
44]. The anti-programmed death protein 1 (PD1) inhibitor pembrolizumab has been approved for first-line treatment in mismatch repair defects (dMMR) and advanced CRC in MSI-H [
45].
To understand the relationship between GRID2IP and MSI status in CRC patients, we evaluated the correlation between MSI and TMB scores and GRID2IP in pan-cancer, and further visualized the correlation between MSI and TMB and GRID2IP in CRC patients with scatter plots. Specifically, we found that GRID2IP was negatively correlated with MSI and TMB in CRC patients. Subsequently, we also compared differences in somatic mutations between different GRID2IP expression subgroups by generating waterfall plot. Furthermore, we also analyzed the survival rate between GRID2IP-high and GRID2IP-low subgroups by Kaplan–Meier plot. The result show that the OS, PFI, and DSS of GRID2IP-high group are worse than the GRID2IP-low group. Grouped based on clinic pathology and independent prognostic factors, we observed T-stage (T3&T4), N-stage (N1&N2), pathology stage (T3&T4), residual tumor(R1&R2), lymphatic invasion, BMI (< 25) subgroups with high expression of GRID2IP had worse prognosis. It is further suggested that GRID2IP is a biological molecule strongly associated with the prognosis of CRC patients.
Although the above results suggest that GRID2IP is a biomarker for prognostic and therapeutic targets in CRC, our study also has certain limitations. In addition, a considerable number of cases lack complete clinical information. The therapeutic effect of GRID2IP as a therapeutic target lacks relevant reports or evidence support, and the signaling pathways involved in the occurrence and progression of CRC, as well as its upstream and downstream molecules, are not well-understood. Therefore, experimental investigation is required to determine the mechanism of action of GRID2IP on the progress of CRC.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.