Analysis of the role of Frizzled 2 in different cancer types

Frizzled 2 (FZD2) is an important receptor in the Wnt pathway, which is highly expressed in malignant tumors and helps regulate multiple tumor behaviors. Its expression level is related to prognosis. Here, bioinformatic analysis was performed to understand the expression of FZD2 in different tumors. We examined FZD2 expression using pan‐cancer data of 33 cancer types from The Cancer Genome Atlas (TCGA). Differential expression analysis (Wilcoxon's test) was used to compare tumor and normal tissues. Univariate Cox proportional hazard regression was performed to compare gene expression and overall patient survival. COSMIC, cBioPortal, and CCLE were used to examine FZD2 mutations in human cancers. Dryness index was calculated using one‐class logistic regression (OCLR). Spearman's correlation was performed based on gene expression and dryness score and used to analyze the correlation between gene expression and stemness score, matrix score, immune score, estimated score, tumor mutation burden (TMB), microsatellite instability (MSI), and drug sensitivity. STRING website was used to construct an FZD2 protein interaction network and identify genes that interact with FZD2. We report that FZD2 is highly expressed in most tumors, differing between cancer types. Expression was related to patient overall survival (OS), disease‐specific survival, disease‐free interval (DFI), mutations, drug sensitivity, tumor microenvironment, immune cell infiltration, immune checkpoint gene expression, immunotherapy indicators (TMB, MSI), and tumor cell stemness. FZD2 influenced drug sensitivities, including cobimetinib (r = −0.553, P < 0.001), selumetinib (r = −0.539, P < 0.001), bafetinib (r = −0.538, P < 0.001), tamoxifen (r = −0.523, P < 0.001), alvespimycin (r = −0.520, P < 0.001), and nilotinib (r = −0.502, P < 0.001). FZD2 has the most significant correlation with ROR2 (r = 0.4, P < 0.001), Wnt2 (r = 0.37, P < 0.001), and Wnt4A (r = 0.34, P < 0.001). The results confirm the importance of FZD2 expression in cancer prognosis and treatment, and provide new clues for treatment strategies.

Frizzled receptors (FZDs) are seven-span membrane proteins belonging to a subclass of the G protein-coupled receptor family [1]. There are 10 FZDs in human cells (FZD1-FZD10) [1]. There are 19 members of the Wnt family that can bind to these 10 members of the FZD family to activate the Wnt/b-catenin pathway [2]. Abnormal activation of the Wnt pathway plays an important role in cell carcinogenesis, tumorigenesis, and invasion. Abnormally activation of Wnt through its signaling pathway may cause tumors [3]. FZD is usually affected in these cases, and abnormal expression can be seen in a variety of malignant tumors; therefore, inhibiting this pathway may engender new breakthroughs in the treatment of tumors [4].
Frizzled 2 (FZD2) is a newly discovered tumor marker. It is one of the important receptors of the Wnt signaling pathway and is mainly involved in nonclassical pathway signal transduction [5]. It is highly expressed in a variety of malignant tumors and participates in the regulation of various tumor behaviors [6][7][8][9]. Its expression level is closely related to patient prognosis, and therefore, it is expected to become a new prognostic indicator and therapeutic target for a variety of cancers [10,11]. Here, bioinformatic analysis was performed to understand the expression of FZD2 in different tumors and its possible connection with cancer. This study used TCGA data to conduct a comprehensive analysis of FZD2 expression characteristics, prognostic value, correlation of tumor-infiltrating immune cells, and drug sensitivity, to provide more information to better understand the importance of FZD2 in pan-cancer.

TCGA pan-cancer data
On March 23, 2020, data on different types of cancer were downloaded from the Xena Browser (https://xenabrowser. net/datapages/), including gene expression RNA-Seq (HTSeq-FPKM), clinical data, and survival data. The pancancer data of 33 primary tumors are described in Table 1.

Differential expression analysis of FZD2 between normal and tumor samples
For all TCGA tumor types, the 'ggpubr' R software package was used to perform differential expression analysis (Wilcoxon's test) between tumor and normal tissues. Only tumor types with more than five normal samples were included. In the heat map, the difference in FZD2 gene expression in pan-carcinoma is presented in the form of log 2 fold change (log 2 FC).

Clinical correlation analysis
The correlation between high and low levels of FZD2 expression and overall survival (OS), disease-specific survival (DSS), and disease-free interval (DFI) was analyzed using an R software package (Kaplan-Meier diagram) using phenotype and survival data of 33 TCGA cancers from the GDC TCGA collection in the UCSC Xena database (http://xena.ucsc.edu/). According to the median expression level of FZD2, these were divided into high expression and low expression groups. In addition, Cox proportional hazard regression analysis was used to obtain the hazard ratio of FZD2 in each TCGA tumor type. Furthermore, the differential analysis was used to detect differences in FZD2 expression characteristic levels at different stages of the 33 cancers. P < 0.05 was considered statistically significant.

Mutation analysis
The catalog of somatic mutations in cancer (COSMIC) database (https://cancer.sanger.ac.uk/cosmic/) collects millions of coding mutations, noncoding mutations, genome rearrangements, fusion genes, copy number abnormalities, and gene expression variations in the human genome [12]. In this study, COSMIC was used to examine FZD2 mutations in human cancers. cBioPortal (http://cbioportal.org) is an open resource that can be used to interactively explore multiple sets of cancer genomic data [13]. In this study, cBioPortal was used to analyze the mutation rate and distribution of FZD2 in different exons in TCGA pancancer data. The Cancer Cell Line Encyclopedia (CCLE) project dataset is a compilation of gene expression data from human cancer cell lines and was used to analyze FZD2 mutations in various cancer cell lines [14].

Correlation analysis between tumor mutation burden and microsatellite instability
Tumor mutation burden (TMB) is defined as the total number of somatic gene coding errors, base substitutions, insertions, or deletions detected per million bases. The correlation between tumor mutation load and FZD2 gene expression was calculated using Spearman's test; this was also used to calculate the correlation between microsatellite instability and FZD2 expression. The result was represented by the R 'fmsb' package radar chart (***P < 0. 001; **P < 0. 01; *P < 0.05).
TIMER TIMER (https://cistrome.shinyapps.io/timer) provides cancer researchers with a comprehensive analysis network tool for analyzing immune cell infiltration in a variety of cancers [15]. The database uses statistical methods validated by pathological examinations to evaluate the immune infiltration of B cells, CD4+ T cells, CD8+ T cells, neutrophils, macrophages, and dendritic cells (DCs) on tumors. This database was used to analyze the correlation between FZD2 expression and a large number of immune infiltrations.

Stemness indices and tumor microenvironment in pan-cancer
The tumor microenvironment mainly includes tumor cells, mesenchymal cells, and the extracellular matrix. These play an important role in tumor growth, angiogenesis, tumor invasion, and metastasis [16]. The ESTIMATE method was used to analyze the correlation between FZD2 expression in TCGA tumor samples and the ratio of stromal cells and immune cells [17]. The ESTIMATE score is calculated based on gene expression characteristics, which can reflect the purity of the tumor; it also has good prediction accuracy. By using the estimation package and the limma package, a Spearman correlation analysis was performed between the expression level of FZD2 and the matrix score.
To further analyze the relationship between FZD2 and pan-cancer stemness, a one-class logistic regression (OCLR) machine learning algorithm was used to calculate the stemness index of TCGA tumor samples, and Spearman's correlation was performed based on gene expression and stemness score analysis [18]. The dryness indices based on DNA methylation (DNAss) and on mRNA expression (RNAss) were obtained.

Analysis of drug sensitivity in pan-cancer
The cancer cell line platform established by the National Cancer Institute (NCI) has been widely used in drug screening based on related gene expression. NCI-60 is a collection of 60 human cancer cell lines from nine different cancer types (leukemia, colon cancer, lung cancer, cancers of the central nervous system, kidney cancer, melanoma, ovarian cancer, breast cancer, and prostate cancer). NCI-60 expression data were obtained from CellMiner. The Pearson correlation coefficient was calculated to analyze the relationship between mRNA expression and the 50% growth inhibitory concentration of the drug.

Establishment of protein-protein interaction (PPI) network
The Search Tool for the Retrieval of Interacting Genes/ Proteins (STRING) website was used to construct the FZD2 protein interaction network and obtain the genes that are mainly related to FZD2. TCGA was used for correlation analysis of genes related to FZD2.

FZD2 gene expression in human cancers
Tumor samples from the TCGA database were integrated to analyze FZD2 mRNA expression characteristics. When only tumors in the TCGA and adjacent tissues were included, FZD2 was found to be upregu-

Correlation analysis between FZD2 expression level and prognosis
Using data from the TCGA database, univariate Cox regression analysis was used to evaluate the correlation between FZD2 mRNA expression levels and OS and DSS in different types of cancer. When the median expression value of each cancer type was classified, it was found that upregulation of FZD2 expression was related to shorter OS and DSS in KIRC, LGG, MESO, SARC, and UVM. In contrast, upregulation of FZD2 expression was related to the longer OS and DSS in UCS ( Fig. 2A-M). The hazard ratios for FZD2 were significant for KICH, KIRC, LGG, MESO, PAAD, SARC, and STAD, among which FZD2 had the highest risk effect in KICH ( Fig. 2O-P). The correlation between FZD2 expression and DFI was analyzed using Cox regression, and a significant hazard ratio was found for STAD (Fig. 2N). According to the median expression of FZD2 across the different cancer types, patients were divided into either a high or low expression group; when analyzed, it was found that the survival difference between the high and low expression groups was significant and that patients with high FZD2 expression had earlier recurrence after tumor resection (Fig. 2Q).

FZD2 mutations in pan-cancer
COSMIC provides information about FZD2 mutations in different cancers, including missense mutations, nonsense mutations, and synonymous mutations (Figs 3A and Fig. S1). Synergistic mutations were obvious in breast cancer, endometrial cancer, large intestine cancer, liver cancer, lung cancer, skin cancer, and stomach cancer, while nonsense mutations were rare (Fig. 3A). The sample size of other tumor mutations was small, and different types of mutations also appeared (Fig. S1). C>T and G>A mutations were found to be the most common in the FZD2 coding chain, while A>T and T>A mutations were rare. Fig. 3B,C shows the mutation result of cBioPortal, illustrating the mutation level of FZD2 in the TCGA cancer database. A total of 106 mutation sites were found in FZD2 through the cBioPortal database, located between amino acids 0 and 565 (Fig. 3B). Among these, the mutation rate was higher in esophagogastric adenocarcinoma and endometrial carcinoma (Fig. 3C). Missense mutations and silent  mutations were also found in cancer cell lines (Fig. 3D).

The relationship between FZD2 mRNA expression and tumor immune microenvironment
After determining the prognostic value of FZD2, the relationship between FZD2 and tumor-infiltrating immune cells in cancer was explored. The ESTIMATE method was used to analyze the correlation between FZD2 expression in TCGA tumor samples and the ratio of both stromal cells and immune cells (Fig. 4A). In COAD, DLBC, LGG, LIHC, PCPG, PRAD, READ, and UVM, it was found that FZD2 significantly positively correlated with stromal score, immune score, and estimated score. FZD2 had the highest correlation with stromal score in TCGT (r = 0.71, P < 0.001), while the highest correlation with immune score (r = 0.65, P < 0.001) and estimate score (r = 0.68, P < 0.001) was found in UVM. FZD2 expression and immune cell infiltration were also analyzed using the TIMER database correlation between levels, where the expression of FZD2 had the highest correlation with macrophages, DCs, and T-cell CD4+ cells (Fig. S2).

Correlation between FZD2 expression and certain immune checkpoint gene expression in certain cancers
Immune checkpoints are a class of inhibitory molecules that play a protective role in the human immune system, preventing excessive activation of T cells from causing damage to themselves. Tumor cells can exploit this protective mechanism by overexpressing the checkpoint molecules, inhibiting the antitumor response of the immune system to achieve immune escape. Immune checkpoint inhibitors act to block the interaction of immune checkpoints and their ligands, break immune tolerance, enhance immune cell activity, and promote immune clearance of tumor cells, thereby inhibiting the occurrence and development of tumors. The mRNA sequence database allows us to assess whether there is a link between FZD2 expression and the expression of such checkpoint genes. The correlation analysis between FZD2 and checkpoint gene expression revealed a high correlation in VISR, CD200, TNFRSF4, TNFRSF 14, NRP1, and CD44 in various types of cancer (P < 0.05). In addition, significant co-expression of FZD2 and other immune checkpoint genes was detected in Adrenocortical carcinoma (ACC), BRCA, DLBC, KICH, LGG, PCPG, PRAD, TGCT, and THYM. However, in TGCT and THYM, the expression of FZD2 was negatively correlated with most immune checkpoint molecules (Fig. 4B).

Relationship between FZD2 mRNA expression, and TMB and MSI in some cancers
The relationship between TMB and MSI and FZD2 expression was examined in various cancer types. The results showed that the expression of FZD2 correlated significantly with TMB in ACC, BLCA, CESC, COAD, HNSC, KICH, LGG, LIHC, PCPG, PRAD, STAD, SKCM, THYM, and UCEC (P < 0.05), and that KICH, COAD, and THYM had the highest coefficients, while LIHC had the lowest (Fig. 4C). The coefficient value indicates that FZD2 expression was positively correlated with high mutation status in KICH, COAD, and THYM, and positively correlated with low mutation status in LIHC. The correlation between FZD2 expression and MSI was analyzed in 33 cancers, and expression of FZD2 was significantly correlated with MSI in BLCA, BRCA, COAD, KICH, LUSC, PAAD, PCPG, and STAD (P < 0.05; Fig. 4D). The coefficient of KICH was the highest, indicating a positive correlation between FZD2 expression and MSI in this type. In contrast, the expression of FZD2 had the lowest coefficients in PAAD, PCPG, and STAD, indicating that in these types there is a significant negative correlation between FZD2 expression and MSI.

Related genes with FZD2 and their interacting protein network
STRING was used to analyze the PPI with FZD2 (Fig. 7A). The main interactions with FZD2 in the PPI network were LRP5, ROR2, Wnt2B, Wnt11, Wnt5A, Wnt1, Wnt4, Wnt2, CTNNB1, and Wnt3A. Using the TCGA to analyze the correlation between FZD2 and these genes (Fig. 7B-K), the results showed that FZD2 had the most significant correlation with ROR2 (r = 0.4, P < 0.001), Wnt2 (r = 0.37, P < 0.001), and Wnt4A (r = 0.34, P < 0.001). The correlation between FZD2 and stromal scores, immune scores, and ESTIMATE scores in pancancer. Spearman's correlation tests were used for testing, and P < 0.05 was considered significant. (B) Correlation between FZD2 mRNA expression levels and acknowledged immune checkpoints' mRNA expression in multiple tumors from TCGA database. The lower triangle in each tile indicates coefficients calculated by Pearson's correlation test, and the upper triangle indicates log 10 -transformed P-value. *P < 0.05, **P < 0.01, ***P < 0.001. (C) Correlation between TMB and FZD2 expression. Spearman's correlation test was used for testing, P < 0.05 was considered significant. (D) Correlation between MSI and FZD2 expression. Spearman's correlation test was used for testing, P < 0.05 was considered significant.

Discussion
Abnormal activation of the Wnt signaling pathway causes abnormal accumulation of b-catenin in tumor cells, leading to abnormal cell proliferation and tumor occurrence [3]. As the receptors of the Wnt signaling pathway, FZDs activate downstream signaling by binding to Wnt ligands, further regulating cell proliferation, differentiation, migration, tissue polarity, and tumor development. FZDs have been found to be specifically expressed on the cell plasma membrane, and FZD2 is one of the most important receptors in the noncanonical Wnt pathway; it is highly expressed in many cancers and is a marker of poor prognosis [9,19,20]. Studies have found that the FZD2 receptor protein can combine with Wnt3A activated by ROR2 molecules to initiate the Wnt signaling classical pathway and act as a cancer-promoting factor in lung cancer [21]. Similarly, our study found that FZD2 has a significant correlation with ROR2. Gene expression profile analysis revealed that FZD2 plays a key role in the occurrence of gastric cancer (GC) [22]. In addition, the latest research has found that FZD2 is more highly expressed in hepatocellular carcinoma tissues than in adjacent tissues, and the recurrence-free survival rate of patients with high FZD2 expression is significantly lower than that of patients with low expression. Furthermore, FZD2 expression is significantly correlated with the mesenchymal phenotype in HCC cell lines, and knocking out FZD2 can inhibit the migration and invasiveness of liver cancer cells [23]. Studies have shown that FZD2 can promote OSCC cell migration and invasion by regulating the STAT3 pathway [24]. From the results of this study, according to the TCGA database, FZD2 was highly expressed in a variety of cancers and was closely related to patient survival and clinical stage. Therefore, it was hypothesized that FZD2 may act as an oncogene in most tumors.
Cancer stem cells (CSCs) are a small group of cells in tumors that have self-renewal ability, strong tumorforming ability, and resistance to chemotherapy drugs and radiotherapy [25,26]. They are the root of tumorigenesis, drug resistance, recurrence, and metastasis. The Wnt/b-catenin signaling pathway regulates the self-renewal of liver stem cells and liver CSCs [27][28][29][30]. As the receptor of Wnt, it has been confirmed that some family members of FZD are related to tumor stem cells and drug resistance [31]; for example, FZD7 can regulate the function of stem cells in the stomach and intestinal epithelium, and FZD7 expression increases in GC cells and tissues [32]. Knockout of FZD7 or use of Wnt/b-catenin inhibitors has been shown to reduce the stemness and chemoresistance of GC cells [33]. FZD8 is highly expressed in human lung cancer tissue samples and cell lines, and knockout of FZD8 can increase the sensitivity of lung cancer cells to paclitaxel [34]. In addition, the analysis done in this study found that FZD2 was significantly correlated with Wnt2 and Wnt4A. It appears that FZD2 may affect the drug resistance of tumor cells through the Wnt signaling pathway.
Studies have shown that FZD2 promotes migration and invasion of OSCC cells by regulating the STAT3 pathway [24]. The IL-6/ STAT3 signaling pathway is related to the stemness of breast cancer cells [35,36]; both cancer cells and stromal cells in the tumor microenvironment can produce IL-6, which promotes breast cancer cell invasion, stemness, and drug resistance by activating STAT3 [37]. This study also found that FZD2 is associated with DNAss, RNAss, and stemness in some tumors, indicating that FZD2 may play a role in stemness maintenance. Further analysis also found that FZD2 is related to a variety of drug sensitivities, such as cobimetinib, selumetinib, bafetinib, tamoxifen, alvespimycin, and nilotinib. As the expression of FZD2 increases, the sensitivity of cells to these drugs also decreases. This could mean that FZD2 is related to chemotherapeutic drug resistance, and it regulates tumor cell stemness through the Wnt signaling pathway to cause drug resistance. These issues warrant further study for confirmation.
The Wnt pathway plays a vital role not only in cell development, survival, and proliferation, but also in immunity [38,39]. DCs are antigen-presenting cells that play an important role in the initiation and regulation of acquired immunity, and regulate the immune tolerance process. In the tumor microenvironment, Wnt binds to the co-receptors LRP5/LRP6 of the Wnt classic signaling pathway (expressed by DC cells), activates the classic Wnt signaling pathway, mediates immune tolerance, inhibits the immune response of effector T cells, and alters antitumor effects [40]. This study found that FZD2 was significantly related to DCs and the tumor microenvironment in a variety of tumors. In addition, other studies have shown that Wnt1 molecules bind to the transmembrane receptor Frizzled, and co-receptors LRP5/LRP6 and CD36 on the cell surface and upregulate the expression of CD36 on macrophages by activating the classic Wnt signaling pathway and PPAR-c to promote macrophages. The function of these cells is to take up low-density lipoproteins, thereby affecting the physiological activity of macrophages [41]. This study found that FZD2 had a significant correlation with macrophage and T-cell CD4+ in a variety of tumors. In addition, FZD2 had a significant correlation with multiple immune checkpoints in various types of cancer. Further study is needed to determine whether FZD2 affects the proliferation or drug resistance of tumor cells by affecting the tumor microenvironment or cellular immunity, and this conclusion needs to be further studied for confirmation.
Tumor mutation burden is an independent biomarker that has been discovered in a variety of tumor immunotherapies in recent years, and can be used to predict the efficacy of immunotherapy [42,43]. Those with high TMB expression have been shown to benefit more from immune checkpoint inhibitor therapy [44]. TMB reflects the total number of replacement and insertion/deletion mutations per megabase in the exon coding region of the evaluated gene in the tumor cell genome. Driving gene mutations can lead to tumors, and a large number of somatic mutations can produce new antigens, which can activate T cells and cause immune responses [45]. Therefore, when the number of gene variants accumulates, more new antigens are produced, and there is a greater possibility of recognition by the immune system. Previous research in our group found that FZD2 is related to tumor immunity. In this study, further analysis of the correlation between FZD2, TMB, and MSI was performed, and the results show that there is a link between FZD2 expression and TMB and MSI in certain cancer types. Studies have shown that frameshift mutations of AXIN2 and TCF7L2 are common in GC with high MSI, and these mutations may promote the development of GC through the control of Wnt signaling [46]. MSI is now considered an indicator to distinguish the types of tumors in patients with COAD. It was also found that FZD2 was mutated in breast, endometrial, large intestine, liver, lung, skin, and stomach cancer. In addition, the expression of FZD2 was significantly correlated with MSI in BLCA, BRCA, COAD, KICH, LUSC, PAAD, PCPG, and STAD.
Although this study confirmed the involvement of FZD2 in tumorigenesis, drug sensitivity, and tumor cell immunity, it does have some limitations. The data come entirely from open databases and have not been verified experimentally. Also, FZD2 is highly expressed in a variety of tumors and is associated with poor prognosis, but despite this, the specific mechanism behind this action has not been verified. The expression of FZD2 also has a certain correlation with drug sensitivity, tumor microenvironment, tumor immunity, TMB, and MSI, but there is lack of data confirming their correlation.

Conclusions
FZD2 was found to be highly expressed in various tumors, and this high expression is related to poor survival and disease progression. The expression of FZD2 was also related to tumor drug sensitivity, tumor microenvironment, immune cell infiltration, immune checkpoint gene expression, and immunotherapy indicators (TMB, MSI). In summary, these results confirm the importance of FZD2 expression in cancer prognosis and treatment and provide new clues for cancer treatment strategies.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article.