Correlation of CERCAM expression levels with clinicopathological characteristics of HNSCC patients
We used information on clinicopathological characteristics from the TCGA-HNSCC dataset to examine the correlation between CERCAM expression levels and clinicopathological characteristics of HNSCC patients (as shown in Table 1). Although no differences were observed in the correlation between CERCAM expression and factors such as race, age, gender, clinical grade, pathological grade, and history of alcohol consumption, but significant differences were observed in the correlation with tumor T-stage, smoking history, and survival outcome OS (Fig. 2a-c).
*P < 0.05, **P < 0.01, ***P < 0.001
We next assessed the diagnostic value of CERCAM using the ROC curve. As shown in Fig. 4a, the area under the curve (AUC) of CERCAM was 0.893, indicating the satisfactory diagnostic value of CERCAM in HNSCC. The results of time-dependent ROC analysis curves showed that the AUC values for predicting the survival rate of HNSCC patients at 3, 7, and 10 years based on the expression level of CERCAM were above 0.55, as shown in Fig. 4b. In addition, we constructed a column line graph model including tumor clinical N stage and CERCAM expression level as factors (Fig. 4c). Based on univariate and multivariate Cox regression analysis, clinical N stage and CERCAM expression level were valuable independent prognostic predictors for OS, DSS and PFI. The column line graph model showed that the above factors showed high clinical significance in predicting the survival probability of HNSCC patients at 3, 7, and 10 years.
Prognostic significance of CERCAM in a clinicopathological subgroup of HNSCC patients
To investigate the prognostic impact of CERCAM expression in the clinicopathological subgroup of HNSCC patients, we did a Cox regression analysis of OS, DSS, and PFI indices in each subgroup of patients in TCGA-HNSCC, and the results are shown in Table 3, Fig. 5. High CERCAM expression levels were associated with poor PFI in male patients. In race, its high expression could be found to be associated with poor OS, DSS, and PFI in both blacks and whites. Age was a significant factor affecting prognosis, and the results showed that high expression of CERCAM in patients over the age of 60 years was associated with poor OS, DSS, and PFI. In patients with a history of smoking, high expression of CERCAM was associated with poor OS, DSS, and PFI. In patients without a history of alcohol consumption high CERCAM expression was only associated with poor OS, PFI. In contrast, high expression levels of CERCAM in patients with advanced T3 and T4, patients with lymph node involvement in stages N2 and N3, and M0 patients without distant metastases were found to be associated with poor OS, DSS, and PFI in clinical T, N, and M stages. Finally, high expression levels of CERCAM in histologically graded G3 and G4 patients were correlated with poor OS, DSS, and PFI.
Table 3
Prognostic significance of CERCAM in the clinicopathological subgroup of patients with HNSCC
Characteristics
|
N (%)
|
HR for OS
(95% CI)
|
HR for DSS
(95% CI)
|
HR for PFI
(95% CI)
|
Gender
|
|
|
|
|
Male
|
368(73.3%)
|
1.25(0.90–1.72)
|
1.42(0.94–2.12)
|
1.52(1.09–2.12)*
|
Female
|
134(26.7%)
|
1.64(0.99–2.70)
|
1.86(0.91–3.78)
|
1.43(0.81–2.53)
|
Race
|
|
|
|
|
Asian
|
10(2.1%)
|
0.52(0.07–3.78)
|
0.90(0.08–10.22)
|
2.1(0.22–20.49)
|
Black or African American&White
|
475(97.9%)
|
1.36(1.03–1.79)*
|
1.52(1.06–2.18)*
|
1.46(1.09–1.96)*
|
Age
|
|
|
|
|
≤ 60
|
245(48.9%)
|
0.99(0.66–1.50)
|
1.00(0.60–1.66)
|
1.13(0.74–1.71)
|
༞60
|
256(51.1%)
|
1.79(1.24–2.58)**
|
2.28(1.38–3.77)**
|
1.79(1.20–2.67)**
|
Smoker
|
|
|
|
|
No
|
111(22.6%)
|
0.79(0.43–1.45)
|
0.64(0.30–1.40)
|
0.95(0.54–1.70)
|
Yes
|
381(77.4%)
|
1.60(1.18–2.19)**
|
2.04(1.34–3.10)**
|
1.70(1.21–2.38)**
|
Alcohol.history
|
|
|
|
|
No
|
158(32.2%)
|
1.84(1.15–2.94)*
|
1.73(0.88–3.40)
|
1.79(1.03–3.12)*
|
Yes
|
333(67.8%)
|
1.25(0.89–1.76)
|
1.46(0.96–2.21)
|
1.34(0.96–1.88)
|
Clinical.stage
|
|
|
|
|
I&II
|
114(23.4%)
|
1.18(0.66–2.11)
|
1.45(0.67–3.13)
|
1.42(0.76–2.69)
|
III&IV
|
374(76.6%)
|
1.45(1.07–1.98)*
|
1.60(1.07–2.38)*
|
1.41(1.02–1.96)*
|
T.stage
|
|
|
|
|
T1&T2
|
177(36.4%)
|
1.06(0.65–1.71)
|
1.17(0.60–2.28)
|
1.48(0.87–2.52)
|
T3&T4
|
310(63.7%)
|
1.51(1.09–2.10)*
|
1.85(1.21–2.82)**
|
1.46(1.03–2.06)*
|
N.stage
|
|
|
|
|
N0&N1
|
319(66.5%)
|
1.07(0.77–1.50)
|
1.05(0.67–1.65)
|
1.05(0.73–1.51)
|
N2&N3
|
161(33.5%)
|
2.12(1.31–3.43)**
|
2.63(1.45–4.77)**
|
2.26(1.38–3.73)**
|
M.stage
|
|
|
|
|
M0
|
472(99%)
|
1.42(1.08–1.87)*
|
1.65(1.15–2.37)**
|
1.44(1.07–1.93)*
|
M1
|
5(1%)
|
(-)
|
(-)
|
|
Histologic.grade
|
|
|
|
|
G1&G2
|
362(74.9%)
|
1.12(0.82–1.54)
|
1.33(0.88-2.00)
|
1.29(0.92–1.79)
|
G3&G4
|
121(25.1%)
|
2.08(1.18–3.68)**
|
2.04(1.02–4.08)*
|
2.26(1.24–4.11)**
|
*P < 0.05, **P < 0.01, ***P < 0.001
Silencing CERCAM contributes to inhibit the malignant progression of HNSCC cells
These above results suggest that CERCAM has biological properties that promote the malignant progression of HNSCC. To investigate whether suppressing the expression of this gene contributes to inhibit the malignant progression of HNSCC, and provide a theoretical basis for distant gene therapy. As shown in Fig. 6a, by cell transfection assay, we used small interfering RNA to knock down the expression of CERCAM in HNSCC cells, and the results of qRT-PCR showed that si-CERCAM#2 transfection was the most efficient and the mRNA expression level of CERCAM was the lowest. Therefore, we selected si-CERCAM#2 for subsequent experiments. We then examined the effect of CERCAM expression in HNSCC cells on cell proliferation by CCK-8 assay, and the results are shown in Fig. 6b. Compared with the control group, the proliferation level of HNSCC cells decreased with the silencing of CERCAM, and the difference was significant (24h:ns,48h:P < 0.05, 72h:P < 0.01, 96h:P < 0.001). Therefore, silencing CERCAM helps to inhibit the malignant progression of HNSCC cells.
CERCAM mutations in HNSCC tumors
To explore CERCAM mutations in HNSCC, we used the cBioPortal platform to analyze its genetic change status based on TCGA data. The results showed the high CERCAM amplification in HNSCC (Fig. 7a). Gene mutation analysis identified missense as its gene mutation type, we found missense mutation R382H within the Glyco_transf_25 domain was mutated (Fig. 7b). Figure 7c also shows the 3D structure of the protein with the R382H mutation in CERCAM.
Analysis of CERCAM methylation levels in HNSCC tumors
We first used the UALCAN platform to analyze the DNA methylation levels of CERCAM genes in patient tumor tissues with their matched normal tissues in the TCGA-HNSCC dataset (Fig. 8a), we found that the CERCAM gene showed lower methylation levels in tumor tissues, with a significant difference (P < 0.05). We then applied the MethSurv platform to visually analyze the DNA methylation levels of CPG islands in the CERCAM gene, and we found that five CPG islands including cg13889221, cg06619282, cg27627570, cg08325021, and cg19324791 showed altered methylation levels (Fig. 8b). Combined with the above results that CERCAM expression correlated with poorer prognosis, suggesting that hypomethylation levels of CERCAM in tumors are accompanied with poorer prognosis, we hypothesize that CERCAM may function as an oncogene in tumors.
Correlation analysis of CERCAM and immune cell infiltration in HNSCC
Many studies have shown that immune cells in the tumor microenvironment are involved in regulating the process of tumor development. Based on this, we used the Timer database to explore the correlation between CERCAM expression and immune cell infiltration in HNSCC.As shown in Fig. 9, we found that the expression of CERCAM correlated most significantly with the infiltration of macrophages in tumors (Cor = 0.328, P = 1.5e-13), followed by CD4+ T cells (Cor = 0.243, P = 7.22e-8) and DC cells (Cor = 0.215, P = 1.81e-6). However, the expression level of CERCAM correlated weakly or not significantly with B cells, CD8+ T cells, and neutrophils in tumors.
To further investigate the correlation between CERCAM and different types of immune cell subpopulations, we analyzed the relationship between CERCAM expression in HNSCC and immune markers of different types of immune infiltrating cells (as shown in Table 4).The expression levels of CERCAM significantly correlated with marker genes of B cells, NK cells, monocytes, centrophages, DC cells, macrophages, M2 type macrophages, and T cells (including CD8 + T cells, Th1, Th2, Th9, Th17, Th22, and Tregs). And no significant correlation or weak correlation with Tfh and M1 macrophage marker genes.
Table 4
Correlation between CERCAM and different types of immune cell subpopulation marker genes
Cell Type
|
Gene marker
|
None
|
Purity
|
Cell type
|
Gene marker
|
None
|
Purity
|
|
|
Cor
|
P
|
Cor
|
P
|
|
|
Cor
|
P
|
Cor
|
P
|
B cell
|
CD19
|
-0.156
|
***
|
-0.165
|
***
|
|
IL23R
|
-0.053
|
|
-0.045
|
|
|
CD20(KRT20)
|
-0.033
|
|
0.003
|
|
|
IL17A
|
-0.158
|
***
|
-0.155
|
***
|
|
CD38
|
-0.028
|
|
-0.008
|
|
Th22
|
CCR10
|
0.063
|
|
0.075
|
|
NK cell
|
XCL1
|
0.114
|
**
|
0.128
|
**
|
|
AHR
|
0.3
|
***
|
0.296
|
***
|
|
CD7
|
-0.046
|
|
-0.054
|
|
Treg
|
FOXP3
|
0.144
|
***
|
0.149
|
***
|
|
KIR3DL1
|
-0.116
|
**
|
-0.114
|
*
|
|
CD25(IL2RA)
|
0.22
|
***
|
0.227
|
***
|
CD8 + T cell
|
CD8A
|
-0.116
|
**
|
-0.119
|
**
|
|
CCR8
|
0.173
|
***
|
0.181
|
***
|
|
CD8B
|
-0.108
|
*
|
-0.108
|
*
|
Macrophage
|
CD68
|
0.291
|
***
|
0.273
|
***
|
Tfh
|
BCL6
|
0.003
|
|
0.012
|
|
|
CD11b(ITGAM)
|
0.141
|
**
|
0.144
|
**
|
|
ICOS
|
0.055
|
|
0.061
|
|
M1
|
INOS(NOS2)
|
-0.086
|
*
|
-0.068
|
|
|
CXCR5
|
-0.08
|
|
-0.089
|
*
|
|
IRF5
|
-0.009
|
|
-0.013
|
|
TH1
|
T-bet(TBX21)
|
-0.077
|
|
-0.078
|
|
|
COX2(PTGS2)
|
-0.074
|
|
-0.058
|
|
|
STAT4
|
0.228
|
***
|
0.239
|
***
|
M2
|
CD163
|
0.309
|
***
|
0.293
|
***
|
|
IL12RB2
|
0.037
|
|
0.024
|
|
|
VSIG4
|
0.351
|
***
|
0.336
|
***
|
|
WSX1(IL27RA)
|
0.162
|
***
|
0.184
|
***
|
|
CD206(MRC1)
|
0.388
|
***
|
0.322
|
***
|
|
STAT1
|
0.044
|
|
0.029
|
|
TAM
|
CCL2
|
0.255
|
***
|
0.26
|
***
|
|
IFN-γ(IFNG)
|
-0.17
|
***
|
-0.173
|
***
|
|
CD80
|
0.238
|
***
|
0.248
|
***
|
|
TNF-α(TNF)
|
0.072
|
|
0.085
|
|
|
CD86
|
0.255
|
***
|
0.264
|
***
|
Th2
|
GATA3
|
0.138
|
**
|
0.14
|
**
|
|
CCR5
|
0.01
|
|
0.014
|
|
|
CCR3
|
0.233
|
***
|
0.235
|
***
|
Monocyte
|
CD14
|
0.265
|
***
|
0.257
|
***
|
|
STAT6
|
-0.005
|
|
-0.004
|
|
|
CD16(Fcgr3B)
|
0.11
|
*
|
0.092
|
*
|
|
STAT5A
|
0.001
|
|
0.011
|
|
|
CD115(CSF1R)
|
0.297
|
***
|
0.303
|
***
|
Th9
|
TGFBR2
|
0.473
|
***
|
0.464
|
***
|
Neutrophil
|
CD66b
|
-0.112
|
*
|
-0.097
|
*
|
|
IRF4
|
-0.042
|
|
-0.045
|
|
|
CD15(FUT4)
|
0.341
|
***
|
0.356
|
***
|
|
PU.1(SPI1)
|
0.231
|
***
|
0.235
|
***
|
DC cell
|
CD1C
|
0.086
|
*
|
0.092
|
*
|
Th17
|
STAT3
|
-0.052
|
|
-0.045
|
|
|
CD141(THBD)
|
0.054
|
|
0.04
|
|
|
IL21R
|
0.076
|
|
0.078
|
|
|
CD11C(ITGAX)
|
0.211
|
***
|
0.207
|
***
|
*P < 0.05, **P < 0.01, ***P < 0.001.
CERCAM induces macrophage M2 polarization immune infiltration in HNSCC
The above results revealed that the expression level of CERCAM correlated most significantly with macrophages in immune infiltrating cells, and the correlation analysis of cellular marker gene expression in its subpopulation revealed that CERCAM correlated only with M2 macrophages, but not with M1 macrophages (Fig. 10a), this is an interesting finding. So, as shown in Fig. 10b, we further explored the correlation of CERCAM expression with the infiltration levels of M1macrophages and M2 macrophages in HNSCC, HPV(+)-HNSCC, and HPV(-)-HNSCC patients with using two algorithms based on CIBERSOFT and CIBERSOFT-ABS in the TIMER2.0 database. We found that in all three types of HNSCC patients, the results of both algorithms showed a significant positive correlation between the expression level of CERCAM and the infiltration level of M2 macrophages, while there was no significant correlation with M1 macrophages. Therefore, we conjecture that CERCAM may be involved in inducing macrophage M2 polarization immune infiltration in HNSCC.
To verify our conjecture, we induced THP-1 into M0 macrophages in vitro (Fig. 10c), M0 macrophages in a separate group as a negative control, and then established a co-culture system of HNSCC cells and M0 macrophages, knockdown of CERCAM gene expression levels in HNSCC by transfection experiments and compared to the si-NC group, then tested the expression levels of M2 macrophage marker genes (CD163, CD206, VSIG4), which represent the levels of induced polarization. As shown in Fig. 10d, we found that the si-CERCAM group had significantly lower mRNA expression levels of CD163, CD206, and VSIG4 genes compared to the NC group, and the difference was significant(P༜0.05). Therefore, the above experimental results suggest that CERCAM in HNSCC may induce macrophages to polarize toward M2.
Single-cell level expression pattern of CERCAM in HNSCC
Single-cell transcriptome sequencing is an emerging technology that has been developed in recent years and can clearly demonstrate gene expression patterns at the single-cell level, enabling us to better understand the differences in gene expression patterns between different single cells in the tumor microenvironment. Therefore, we used the TISCH2 platform to analyze the expression pattern of CERCAM at the single cell level based on one of the largest HNSCC single cell transcriptome sequencing datasets (HNSC_GSE103322) in the GEO database. As shown in Fig. 11.a-d, it was found that CERCAM was most expressed in fibroblasts, plasma cells, and tumor cells, followed by expression in myofibroblasts, monocytes/macrophages, endothelial cells, and mast cells, while no or no significant expression in CD4convT cells, CD8T cells, CD8Tex, and Myocyte.
CERCAM gene co-expression analysis in HNSCC and PPI network construction with Hub gene screening
We next analyzed the genes co-expressed with CERCAM in HNSCC as well as used functional enrichment analysis to assess the role of CERCAM in the development of HNSCC and its potential molecular mechanisms. We used the LinkedOmics database based on TCGA-HNSCC data to test genes associated with CERCAM expression by Pearson, and we obtained a total of 11,087 genes associated with CERCAM expression (false discovery rate [FDR] < 0.05). Positive correlations (red dots) and negative correlations (green dots) are shown in Fig. 12a. The top50 genes significantly positively and negatively correlated with CERCAM are shown in Fig. 12b and c.
We then used Cytoscape software to construct PPI networks from these above strongest co-expressed genes, and the PPI networks consisted of 58 nodes and 265 edges (Fig. 13a). We then applied Cytoscape's MCODE plugin to extract the most important modules among them, including 19 nodes and 153 edges, as shown in Fig. 13b (Score = 12.42). We then applied the cytoHubba plugin to screen 10 hub genes, including COL1A1, COL1A2, COL3A1, COL5A1, COL5A2, COL6A1, COL6A2, FN1, MMP, and POSTN (Fig. 13c). As shown in Fig. 13.d, we further examined the expression of these hub genes in the TCGA-HNSCC dataset, and we found that these genes were all highly expressed in tumor tissues compared to normal tissues.
CERCAM gene enrichment analysis in HNSCC
We selected the genes with Pearson coefficient greater than 0.3 among the above 11087 genes associated with CERCAM expression and obtained a total of 1931 co-expressed genes. We subjected these genes to KEGG and GO enrichment analysis to identify the functions and pathways of CERCAM co-expressed genes involved in HNSCC. As shown in Fig. 14a, the results of functional enrichment analysis revealed significant correlations with functions related to cell adhesion function, including "extracelluar structure organization", "extracelluar matrix organization", "collagen-containing extracellular matrix", "cell-substrate junction ", "cell-substrate adherens junction", "focal adhesion", "cell adhesion molecule binding", "extracelluar matrix structural constituent". As shown in Fig. 14b, the results of the pathway enrichment analysis showed that co-expressed genes were mainly enriched in cell adhesion-related pathways including "Focal adhesion" and "Rap1", and we also found that they were most enriched in "PI3K-Akt" and "MAPK". Nowadays, numerous studies have shown that PI3K-Akt and MAPK pathways are two of the most frequently mutated oncogenic pathways in cancer and are involved in cancer development(40–42).
Subsequently, we screened the differential genes in HNSCC by dividing the median value of CERCAM expression into two sample groups, high and low. We then performed GSEA enrichment analysis to further identify the pathways associated with CERCAM expression. After screening terms for conditions meeting NOM P < 0.05, FDR < 0.25, and |NES| > 1, Fig. 14c-l shows the top 10 associated terms in the CERCAM high expression group. We found that the CERCAM high expression group was mainly enriched in gene sets related to adhesion and remodeling of cells to the extracellular matrix, including "NABA_CORE_MATRISOME", "REACTOME_EXTRACELLULAR_ MATRIX_ORGANIZATION", "NABA_ECM_GLYCOPROTEINS", "REACTOME_DEGRADATION_OF_THE_EXTRACELLULAR_MATRIX", "REACTOME_COLLAGEN_FORMATION", "KEGG_ECM_RECEPTOR_INTERACTION". Taken together, the results of gene enrichment analysis suggest that high expression of CERCAM may be involved in functional signaling pathways related to cell adhesion in HNSCC to promote cancer development.