3.1 SLC35A3 expression in CRC patients.
TIMER database was used to determine the overall expression level of SLC35A3 in different types of malignant tumors. The results showed that SLC35A3 showed different expression profiles in various cancers. Compared with normal tissues, SLC35A3 mRNA expression in colon adenocarcinoma(COAD) and rectal adenocarcinomawas (READ) significantly decreased (Figure 1A). To further determine the different expression levels of SLC35A3 between colorectal tumors and normal tissues, we analyzed TCGA-COADREAD data sets to collect RNA sequencing data and clinical information on 647 colorectal adenocarcinomas and 51 normal colorectal tissues. The results showed that SLC35A3 was significantly down regulated in CRC tissues (p<0.001, Figure 1B). In addition, we analyzed the expression level of SLC35A3 in 50 CRC tissues and their matched adjacent normal tissues, indicating that SLC35A3 expression in CRC tissues was low (p<0.001, Figure 1C). Consistent, immunohistochemical data from the HPA database showed that SLC35A3 protein levels in CRC tissues were significantly decreased (Figure 1I).
3.2 Low expression of SLC35A3 is associated with adverse clinicopathological features of CRC.
Table 1 shows that 644 of 647 colorectal adenocarcinoma patients collected from TCGA data sets have complete clinical and gene expression data. CRC patients were divided into high expression group (n=322) and low expression group (n=322) according to SLC35A3 expression relative to the average expression value. The correlation between the expression of SLC35A3 and different clinicopathological features of CRC patients was evaluated. The results showed that the low expression of SLC35A3 mRNA was significantly correlated with N phase (p=0.004), pathological stage (p=0.041) and lymph node invasion (p<0.001). However, there was no significant correlation between SLC35A3 mRNA expression and T stage, M stage, sex, age, CEA level, residual tumor, and neural invasion (p>0.05).
The same result was found in Figure 1D-H, which showed that the low expression of SLC35A3 was significantly correlated with N (N0 vs.N1/N2, p<0.01, Figure 1E), pathological stage (Phase I/Phase II/Phase IV, p<0.05, Figure 1G), and lymph node invasion (Yes vs.no, p<0.001, Figure 1H). However, SLC35A3 mRNA expression was not significantly correlated with T stage (T1/T2 vs. T3/T4, p>0.05, Figure 1D) and M stage (M0 vs. M1, p>0.05, Figure 1F).
In addition, univariate logistic regression analysis was performed (Table 2), indicating that SLC35A3 mRNA expression was closely related to N stage (OR=0.623, 95% confidence interval (CI): 0.454-0.853, p=0.003), pathological stage (OR=0.630, 95% CI: 0.457-0.865, p=0.004), l lymphatic invasion (OR=2.090, 95% CI: 1.493-2.938, p<0.001).
Table 2. Logistic regression analysis of correlation between SLC35A3 expression and clinical pathological parameters in CRC.
Characteristics
|
Total(N)
|
Odds Ratio(OR)
|
P value
|
T stage (T3&T4 vs. T1&T2)
|
641
|
0.782 (0.531-1.148)
|
0.211
|
N stage (N1&N2 vs. N0)
|
640
|
0.623 (0.454-0.853)
|
0.003
|
M stage (M1 vs. M0)
|
564
|
0.777 (0.490-1.223)
|
0.277
|
Pathologic stage (Stage III&Stage IV vs. Stage I&Stage II)
|
623
|
0.630 (0.457-0.865)
|
0.004
|
Gender (Male vs. Female)
|
644
|
0.916 (0.672-1.249)
|
0.580
|
Age (<=65 vs. >65)
|
644
|
0.975 (0.713-1.332)
|
0.873
|
CEA level (<=5 vs. >5)
|
415
|
0.928 (0.622-1.382)
|
0.712
|
Residual tumor (R1&R2 vs. R0)
|
510
|
0.933 (0.492-1.757)
|
0.829
|
Perineural invasion (No vs. Yes)
|
235
|
1.140 (0.630-2.054)
|
0.663
|
Lymphatic invasion (No vs. Yes)
|
582
|
2.090 (1.493-2.938)
|
<0.001
|
Colon polyps present (No vs. Yes)
|
323
|
1.076 (0.668-1.730)
|
0.762
|
CEA: carcinoembryonic antigen. Statistically significant values are shown in bold (P<0.05);
3.3 Compared with normal colon epithelial cells and colon tissues, verify the expression of SLC35A3 in colon cancer cell lines and tumor tissues.
The expression level of SLC35A3 in human normal colon epithelial cell line NCM460 and human colon cancer cell lines HCT116, HT-29, and SW620 was determined using qRT-PCR. Compared to normal colon epithelial cells, colon cancer cell lines showed significant downregulation of SLC35A3 mRNA expression (Figure 2A). In five pairs of randomly selected CRC tissues and their adjacent normal tissues, qRT-PCR was used to detect SLC35A3 expression. As shown in Figure 2B, compared with adjacent normal tissues, the expression of SLC35A3 in CRC tissues was significantly down regulated in mRNA. The above findings suggest that SLC35A3 may be involved in the initiation and progression of cancer in CRC patients.
3.4 Low expression of SLC35A3 predicts poor prognosis in CRC patients.
Based on the TCGA-COADREAD data set, Kaplan Meier survival analysis was performed to verify the correlation between SLC35A3 expression and prognosis. The low expression of SLC35A3 is positively correlated with poor overall survival (OS) (HR=0.62, 95% CI=0.44-0.88, p=0.008, Figure 3A). Similarly, we observed that the decreased expression of SLC35A3 was significantly associated with disease specific survival (DSS) (HR=0.59, 95% CI=0.37-0.93, p=0.022, Figure 3B). However, the low expression of SLC35A3 was not significantly associated with poor progression free interval (PFI) (HR=0.79, 95% CI=0.58-1.07, p=0.132, Figure 3C). In univariate analysis, T stage, N stage, M stage, pathological stage, age, SLC35A3 and CEA expression level affected the prognosis of CRC patients (all p<0.05). Moreover, multivariate Cox regression identified that M stage, pathological stage, age, and SLC35A3 expression level as independent risk factors for poor overall survival (OS) of CRC patients (Table 3).
3.5 SLC35A3 expression is a potential diagnostic biomarker in CRC patients.
ROC analysis was used to evaluate the potential value of SLC35A3 expression in differentiating CRC tissue from normal tissue in the TCGA-COADREAD data sets. The results showed that SLC35A3 could be used as a good diagnostic biomarker for patients with CRC, with an AUC of 0.857 (95%CI: 0.812-0.902, Figure 3D).
Table 3. Univariate and multivariate analysis of clinical pathological factors related to OS in CRC patients.
Characteristics
|
Total (N)
|
Univariate analysis
|
|
Multivariate analysis
|
Hazard ratio (95% CI)
|
P value
|
Hazard ratio (95% CI)
|
P value
|
T stage(T1/T2 vs. T3/T4)
|
640
|
2.468 (1.327-4.589)
|
0.004
|
|
1.900 (0.655-5.516)
|
0.238
|
N stage (N0 vs. N1/N2)
|
639
|
2.627 (1.831-3.769)
|
<0.001
|
|
0.391 (0.114-1.346)
|
0.137
|
M stage (M0 vs. M1)
|
563
|
3.989 (2.684-5.929)
|
<0.001
|
|
2.250 (1.066-4.751)
|
0.033
|
Pathologic stage (Stage I/Stage II ) vs. (Stage III/Stage IV)
|
622
|
2.988 (2.042-4.372)
|
<0.001
|
|
5.909 (1.371-25.477)
|
0.017
|
Gender (Female vs. Male)
|
643
|
1.054 (0.744-1.491)
|
0.769
|
|
|
|
CEA level (≤5 vs.>5)
|
414
|
2.620 (1.611-4.261)
|
<0.001
|
|
1.506 (0.800-2.832)
|
0.204
|
SLC35A3 ( High vs. Low)
|
643
|
0.620 (0.437-0.882)
|
0.008
|
|
0.486 (0.276-0.855)
|
0.012
|
Age(≤65 vs.>65)
|
643
|
1.939 (1.320-2.849)
|
<0.001
|
|
4.067 (2.101-7.872)
|
<0.001
|
CRC, colorectal cancer; OS, overall survival; CI, confidence in interval; T: topography distribution; N: lymph node metastasis; M: distant metastasis. Statistically significant values are shown in bold (P<0.05);
3.6 Construction and verification of nomographs based on SLC35A3 expression level.
A nomograph was developed to help clinicians determine the prognosis of CRC patients. The nomogram is based on clinical features independently related to patient survival in multivariate analysis (M stage, pathological stage, age, SLC35A3) (Figure 4A). The C index of the nomogram is 0.738. The calibration chart used to verify the reliability of the prediction model is shown in Figure 4B. The deviation correction line displayed in the calibration chart is close to the ideal curve (also known as the 45 degree line), which indicates the consistency between the observed value and the predicted value. These results indicate that nomography is a model superior to a single prognostic factor and can establish the long-term survival rate (1, 3 and 5 years) of CRC patients.
3.7 SLC35A3 mutation in CRC.
The accumulation of gene mutations will promote the development of cancer. To explore the mutation of SLC35A3 gene in CRC, based on TCGA data, we analyzed its mutation state through cBioPortalTM platform, the results showed that the mutation rate of SLAC5A3 in CRC was 1.3% (Figure 5A). Missense mutation and truncation mutation are the main mutation types in SLC35A3 (Figure 5B). Missense mutation Y117F in the Nuc_sug_transp domain was detected in CRC cases, the three-dimensional structure of the SLC35A3 protein shows the Y117F mutation in Figure 5C. V44A mutation, M82T mutation, Y113N mutation, R207S mutation and E229 mutation in SLC35A3 protein structure were also detected in other CRC patients. These findings reveal that SLC35A3 gene changes may play a key role in the pathogenesis of CRC.
3.8 Promoter methylation level of SLC35A3 in CRC.
Promoter DNA methylation has been proved to affect transcriptional inhibition and participate in tumorigenesis [28]. We compared the methylation level of SLC35A3 promoter in CRC and adjacent normal tissues. Our analysis results show that the methylation level of SLC35A3 promoter is significantly reduced in colon cancer (Figure 6A) and rectal cancer (Figure 6B), with a statistically significant difference. These results suggest that the decreased expression of SLC35A3 in CRC may be due to the change of promoter methylation.
3.9 Potential Biological Functions and Pathways of SLC35A3 in CRC.
Next, we analyzed the potential biological function of SLC35A3 in CRC, and selected the differential genes (| logFC |>1, P.adj<0.05) between the low SLC35A3 group and the high SLC35A3 group for enrichment analysis. GO analysis of biological process (BP) shows that metal ion transmembrane transport, drug transport, movement behavior, positive regulation of synaptic transmission, adenylate cyclase activated G protein coupled receptor signal pathway were significantly enriched (Figure 7A). GO analysis of cell components (CC) showed that endoplasmic reticulum lumen, presynaptic membrane, components of synaptic membrane, dopaminergic synapses, etc. were significantly enriched (Figure 7B). Molecular function (MF) analysis showed that receptor ligand activity, channel activity, metal ion transmembrane transporter activity and passive transmembrane transporter activity were significantly enriched (Figure 7C). KEGG analysis showed that the interaction of neuroactive ligand receptors was the most important significant enrichment pathway (Figure 7D). In conclusion, the results suggest that SLC35A3 may be involved in the changes of cell membrane potential, transmembrane transporter activity, cell communication and the "neuroactive ligand receptor interaction" pathway, thereby regulating the proliferation and invasion of CRC.
In addition, GSEA was performed based on normalized enrichment fraction (NES) and FDR (false detection rate) q values to clarify the possible biological pathway of SLC35A3 regulation between high and low expression groups. As shown in Figure 8 and Table 4, several biological pathways are significantly enriched in the SLC35A3 overexpression group, including starch and sucrose metabolism pathway (Figure 8A), cell cycle G1 phase, S phase (Figure 8B), DNA double strand break repair (Figure 8C), base excision repair (Figure 8D), epigenetic regulation of gene expression (Figure 8E) and histone arginine methylation (Figure 8F). In addition, WNT signal pathway (Figure 8G) and cancer pathway (Fig. 5H) in the low expression group of SLC35A3 are significantly enriched in multiple cancer invasive features (Figure 8I). These results suggest that SLC35A3 may affect the progression of CRC by regulating energy metabolism, cell cycle, DNA repair, epigenetic regulation and carcinogenic pathway.
Table 4 Results of Gene Set Enrichment Analysis (GSEA).
Description
|
Set Size
|
EnrichmentScore
|
NES
|
P value
|
p.adjust
|
q values
|
Rank
|
KEGG_STARCH_AND_SUCROSE_METABOLISM
|
52
|
0.59773391
|
1.86860419
|
0.00040855
|
0.0085906
|
0.0061857
|
6357
|
REACTOME_BASE_EXCISION_REPAIR_AP_SITE_FORMATION
|
63
|
0.63736182
|
2.058913
|
0.00013193
|
0.0085906
|
0.0061857
|
5890
|
REACTOME_RMTS_METHYLATE_HISTONE_ARGININES
|
76
|
0.602562409
|
2.00222025
|
0.00012820
|
0.00859059
|
0.00618570
|
3941
|
REACTOME_EPIGENETIC_REGULATION_OF_GENE_EXPRESSION
|
146
|
0.50738425
|
1.83162574
|
0.00011663
|
0.00859059
|
0.00618570
|
8157
|
REACTOME_DNA_DOUBLE_STRAND_BREAK_REPAIR
|
166
|
0.44863791
|
1.64049479
|
0.00068564
|
0.00863341
|
0.00621653
|
9366
|
FISCHER_G1_S_CELL_CYCLE
|
198
|
0.4203582
|
1.56352259
|
0.00122754
|
0.01016656
|
0.00732048
|
10519
|
PID_WNT_SIGNALING_PATHWAY
|
27
|
-0.673145376
|
-2.1977280
|
0.00029832
|
0.00859059
|
0.00618570
|
6273
|
ANASTASSIOU_MULTICANCER_INVASIVENESS_SIGNATURE
|
64
|
-0.674435264
|
-2.6959558
|
0.00041528
|
0.00859059
|
0.00618570
|
8553
|
KEGG_PATHWAYS_IN_CANCER
|
325
|
-0.32795732
|
-1.6580184
|
0.00205761
|
0.01397440
|
0.01006233
|
6564
|
3.10 Correlation between SLC35A3 expression and immune infiltration level in CRC.
Tumor infiltrating lymphocytes are closely related to the improvement of cancer prognosis [26, 29]. Therefore, we further explored the relationship between SLC35A3 expression and CRC immune infiltration. The correlation between SLC35A3 and 24 immune cell subsets in CRC was analyzed using Spearman r's ssGSEA (Figure 9A). We found that SLC35A3 was positively correlated with T helper cells (R =0.427, p<0.001, Figure 9B), Th2 cells (R =0.390, p<0.001, Figure 9C), and Tcm cells (R =0.327, p<0.001, Figure 9D). In addition, SLC35A3 expression was associated with NK cells (R=-0.342, p<0.001, Figure 9E), TReg cells (R=-0.222, p<0.001, Figure 9F), pDC cells (R=-0.251, p<0.001, Figure 9G), DC cells (R=-0.201, p<0.001, data not shown), iDC cells (R=-0.156, p<0.001, data not shown), aDC cells (R=-0.114, p=0.004, data not shown), Neutrophils (R=-0.157, p<0.001, data not shown), NK CD56bright cells (R=-0.242, p<0.001, data not shown), NK CD56dim cells (R=-0.179, p<0.001, data not shown), Cytotoxic cells (R=-0.127, p=0.001, data not shown), and Tem (R=-0.129, p=0.001, data not shown) were negatively correlated.
3.11 The expression of SLC35A3 in CRC is related to immune checkpoint (ICP) gene.
Immune checkpoint (ICP) gene has an important influence on immune cell infiltration and immunotherapy [30]. We further studied the relationship between SLC35A3 expression and ICP gene in CRC to explore the potential of SLC35A3 in immunotherapy. The results showed that the expression of SLC35A3 was closely related to most of the 47 ICP genes (Figure 10A). Among them, the expression of SLC35A3 was positively correlated with CD274 (R=0.128, p<0.001), ICOS (R=0.166, p<0.001), TIGIT (R=0.143, p<0.001), CD40LG (R=0.094, p=0.017) . It was negatively correlated with CD40 (R=-0.144, p<0.001), PDCD1 (R=-0.097, p=0.013), LAG3 (R=-0.106, p=0.007), CD70 (R=-0.211, p<0.001) (Figure 10B).