CD44 localized in cytoplasm and membrane of tumor cell and significantly associated with cancer
Gene and disease network interactions revealed that CD44 had 151 gene functional partners and the top 15 interacted genes were shown in Fig. 1A. The Open Target platform showed that total 980 diseases or phenotypes were associated with CD44 expression. Among these, solid cancer, such as breast cancer, hepatocellular carcinoma, glioblastoma, urinary bladder carcinoma non-small cell lung carcinoma and chronic myelogenous leukemia displayed a relationship with CD44 expression (Fig. 1B). To evaluate the distribution and expression of CD44 in tumor cell, we retrieved the immunofluorescence results of A-431, U-2OS, and U251-MG from the HPA database. The endoplasmic reticulum (ER) and microtubules were marked with yellow and red, respectively. It was observed that CD44 overlapped with ER and microtubules but displayed no staining in the nuclei, suggesting that CD44 colocalized with these markers in the cytoplasm of tumor cell (Fig. 1C). In addition, our immunofluorescence staining results also demonstrated the location of CD44 in the colon cancer cell lines (HT29 and HCT116) (Fig. 1D). Furthermore, compared with the adjacent tissue, the number of cancer cells expressing CD44 were increased in cancer tissue of colon cancer patients (Fig. 1E). These results suggests that CD44 expression showed a relationship with various cancers and colon cancer patients had increased number of tumor cell expressing CD44 in tumor tissue.
CD44 is significantly aberrantly expressed in various tumors tissues and associated with tumor stages
To further clarify the association of CD44 and cancer, the CD44 expression were evaluated in all types of cancer from TCGA databased. Compared with the normal tissues, the significantly overexpressed CD44 was observed in 25 tumors. The top 5 items were glioblastoma multiforme (GBM) (T:7.59 ± 1.15, N:2.70 ± 1.59, p = 1.8e- 83), glioma (GBMLGG)(T:6.26 ± 1.59,N:2.70 ± 1.59, p = 8.4e-204),brain lower grade glioma (LGG)(T:5.86 ± 1.48,N:2.70 ± 1.59,p = 2.4e-159),breast invasive carcinoma(BRCA)(T:7.52 ± 1.18, N:7.02 ± 0.82, p = 7.7e-13), and cervical squamous cell carcinoma and endocervical adenocarcinoma(CESC) (T:7.61 ± 1.17, N:7.04 ± 0.89, p = 0.04). In contrast, we observed significant downregulation in six tumors such as uterine corpus endometrial carcinoma (UCEC) (T: 5.71 ± 1.42, N: 6.59 ± 1.19, p = 2.4e-3), lung adenocarcinoma (LUAD) (T:6.87 ± 1.05, N:7.58 ± 0.64, p = 8.6e-29), and prostate adenocarcinoma (PRAD)(T:6.10 ± 1.08, N:6.59 ± 0.69,p = 1.6e-6) (Fig. 2A).In addition, we found that CD44 expression is significantly increased from stage I to IV of PAAD(Stage I = 21,II = 147,III = 3,IV = 4,p = 0.01), and stomach and esophageal carcinoma(STES)(Stage I = 76,II = 201,III = 230,IV = 57,p = 0.02), pan-kidney cohort (KIPAN)(Stage I = 464,II = 107,III = 189,IV = 103,p = 0.01).Of note, BRCA(Stage I = 182,II = 617,III = 248,IV = 20,p = 1.4e-3) displayed a relative decrease in the stage IV(Fig. 2B).Then, we analyzed the CD44 expression in G1-G3 or G4 of pan-cancer. We found that CD44 expression had an continuous upregulation from Grade 1 to Grade4 of GBMLGG(G2 = 247,G3 = 260,p = 1.8e-5),LGG(G2 = 247,G3 = 260,p = 1.8e-5),ESCA(esophageal carcinoma)(G1 = 18,G2 = 74,G3 = 49,p = 8.8e-3),STES(G1 = 30,G2 = 222,G3 = 294,p = 0.04),KIPAN(G1 = 14,G2 = 228,G3 = 206,G4 = 74,p = 5.2e-5),STAD(stomach adenocarcinoma)(G1 = 12,G2 = 148,G3 = 245,p = 0.01), KIRC(kidney renal clear cell carcinoma)(G1 = 14,G2 = 228,G3 = 206,G4 = 74,p = 5.2e-5),PAAD(G1 = 31,G2 = 95,G3 = 48,p = 1.4e- 5).While, HNSC (head and neck squamous cell carcinoma) (G1 = 61, G2 = 304, G3 = 124, G4 = 7,p = 2.1e-3) showed the reverse change with the grade increase(Fig. 2C). These above results suggested that CD44 significantly upregulated and increased in the advanced stage of most types of tumors. CD44 may evolve with the progress and development of cancer and may affect the survival outcome of patients with cancer.
CD44 is a potential prognostic marker in pan-cancer
Based on TCGA RNA-seq data and clinical information, a Cox proportional hazards regression model was used to determine the association of CD44 expression levels with the prognosis of patients with various cancer. We found that total 9 types of cancer with higher CD44 expression had poorer overall survival and cancer specific survival. These TCGA cancer types were GBMLGG, LGG, KIRP, KIPAN, HNSC, PAAD, and TGCT (testicular germ cell tumor) (Fig. 3A). On the contrary, increased CD44 mRNA level uniquely related with considerable outcome in UVM. The significant Kaplan-Meier survival curves of OS and CSS were showed in Fig. 3B and Fig. 3C.
Gene Set Enrichment Analysis
Considering the correlation between CD44 expression and prognosis, Gene set enrichment analysis (GSEA) was used for the Gene Ontology (GO) and ImmuneSigDB gene sets from Molecular Signatures Database (MSigDB) analyses of high-CD44 and low-CD44 group in pan-cancer (Fig. 4A-B). We found that many CSCs related gene sets and immune associated sets were enriched in the high-risk group. GO enrichment analysis suggested that these genes were mainly concentrated in cell adhesion and migration, epithelial-mesenchymal transition (EMT), Notching signaling pathway, phosphatidylinositol phosphate biosynthesis process (Fig. 4A). GO and ImmuneSigDB enrichment analysis commonly displayed that upregulated CD44 involved in MHC II biosynthesis, cytokine, chemokine, and interleukin production, T cell and B cell proliferation, and regulations on T cell activation.
CD44 expression is related with immune cell infiltration in human cancers
The subtypes and amounts of infiltrating lymphocytes in tumor are important predictors of the survival of patients with cancer(12).Hence, we obtained the content of 11 specific immune cells in each sample of total 44 cancer types. Then, based on TIMER database, we used the Spearman rank correlation coefficient to explore the relationship between CD44 expression and these immune cell scores. Our results indicated that CD44 expression significantly correlated with tumor purity in 43 cancer types. The heat map described the results at p < 0.005. The upregulation of CD44 expression is correlated notably positively with Treg, macrophages M1 and M2 in most types of analyzed cancers, especially in TGCT (R = 0.68, R = 0.71, respectively) (Fig. 5A). While CD44 expression showed a negative relationship with amounts of infiltrated monocytes. In addition, CD44 expression was correlated with the infiltration levels of CD4+T cells in 17 cancer types, B cells in 17 cancer types, CD8+T cells in 22 cancer types, neutrophils in 25 cancer types, and dendritic cells in 20 cancer types. KIPAN, GBMLGG, LGG, and TCGT displayed a strong correlation between CD44 expression and quantities of immune infiltrated cells.
Correlations between CD44 expression and immune marker sets, TMB, and MSI in cancers
Then, we explored whether the cancer immune response is related with the CD44 mRNA expression. We found that chemokines (41genes), immune receptors (18 genes), MHC (21genes), immunoinhibitors (24 genes), immunostimulators (46 genes) were obviously activated in the group with overexpressed CD44 (Fig. 5B). Among them, CD44 expression is positively correlated with immune chemokines, such as (chemokine C-C motif) CCL2, CCL5, CCL15, CCL20, CCL21, CXCL13, and their receptors, such as CCR1, CCR2, CCR5, CCR7, and CXCR3. These chemokines and receptors can improve the infiltration of CD8 + T cells, TH17 cells, and antigen-presenting cells. MHC associated genes that reveal the capacity of antigen presentation and processing also had a positive relationship with CD44 expression. It was also observed that ICIs (immune checkpoint inhibitors), such as PD-L1, CTLA-4, LAG-3, IDO1, and TIGIT had a positively strong correlation with CD44 expression in many types of cancer, especially in ESCA and UVM (Fig. 5C). Moreover, TMB, MSI, and neoantigen both play an essential association with the therapeutic efficacy of ICIs. Our results showed that there was a strongly positive association between CD44 expression and TMB, MSI, and neoantigen in COAD, READ, SARC, UCEC, and READ. In the contrast, LUAD (lung adenocarcinoma), LUSC (lung squamous cell carcinoma), BLCA (bladder urothelial carcinoma), CHOL (cholangiocarcinoma) showed a negative relationship with CD44 expression (Fig. 5D-F). These findings further revealed that CD44 is a potential predictor for the sensitivity of immunotherapy based on the immune checkpoint.
CD44 methylation related with T-cell dysfunctions and poor prognoses of cancer cohorts
We calculated the promoter methylation level of CD44 in pan-cancer. Different beta-values were hypermethylation (0.7–0.5) or hypomethylation (0.3–0.25). Many TCGA cancer types showed the hypermethylation of CD44. In LUSC, LUAD, ESCA, BRCA, the promoter methylation levels of CD44 were significantly higher than those in normal groups. While the cancer cohort had decreased CD44 methylation levels in PAAD, COAD, LIHC, TGCT (Fig. 6A). Hypomethylation of CD44 was positively associated with dysfunctional T cell phenotypes and survival outcomes in 20 cancer types (Fig. 6B). Then, Kaplan-Meier survival analyses were performed to explore the relationship between CD44 promoter methylation and patient prognosis. Brain cancer, PAAD, and HNSC cohort with higher methylation levels of CD44 had better survival prognosis. In contrast, the hypomethylation level of CD44 was positively associated with T-cell dysfunctions but was a protective factor in patients in LUAD (Fig. 6C). Together, these results indicated that epigenetic methylation of CD44 is associated with dysfunctional T-cell phenotypes via different mechanisms that ultimately result in poor prognoses of PAAD, COAD, LIHC, TGCT cohorts while prolonging the survival of LUAD cohort.
CD44 promoted the tumor growth and remodeled the immune environment of mouse colon cancer
To further verify the function of CD44 in cancer, we constructed stable MC38 cell line expressing shRNA -CD44 using lentiviral delivery approach. Then, the flow cytometry detected that the expression of CD44 was significantly decreased in MC38 shRNA-CD44 cell line (Fig. 7A). We found that CD44 knockdown reduced the proliferation, colony number and migration cells in MC38(Fig. 7B-D). Furthermore, CD44 knockdown also inhibited subcutaneous xenograft tumor growth in mice (Fig. 7E). The mouse xenografted with shRNA-CD44 MC38 had a prolonged survival outcome (Fig. 7F). Importantly, the immune environment of mouse colon cancer was detected by multicolor flow cytometry. Total CD45 positive cells were higher in the group xenografted with shRNA-CD44 MC38.Of these, the group xenografted with shRNA-CD44 MC38 had obviously increased CD4+, CD8+ T cells and MDSC (myeloid derived suppressor cell) in CD45 cells. In contrast, Treg cell in CD45 cells of
the mouse xenografted with shRNA-CD44 MC38 was significantly decreased. Dendritic cells, macrophages M2 and M1 in in CD45 cells showed no significant difference among two groups.