Table 1. The relationship between ASF1B expression and some clinicopathological features
|
ASF1B Expression
|
|
Total(N=84)
|
High(N=45)
|
Low(N=39)
|
p
|
|
|
|
|
|
Age (year)
|
|
|
|
|
<65
|
44(52.4%)
|
22(48.9%)
|
22 (56.4%)
|
0.64
|
≥65
|
40(47.6%)
|
23(51.1%)
|
17(43.6%)
|
|
Gender
|
|
|
|
|
female
|
15(17.9%)
|
8(17.8%)
|
7(17.9%)
|
0.79
|
male
|
69(82.1%)
|
37(82.8%)
|
32(82.1%)
|
|
Stage
|
|
|
|
|
i
|
10(11.9%)
|
7(15.6%)
|
3(7.7%)
|
0.50
|
ii
|
15(17.9%)
|
6 (13.3%)
|
9 (23.1%)
|
|
iii
|
43(51.2%)
|
24(53.3%)
|
19(48.7%)
|
|
iv
|
16(19.0%)
|
8(17.8%)
|
8 (20.5%)
|
|
OS
|
|
|
|
|
Dead
|
72(85.7%)
|
41(91.1%)
|
31(79.5%)
|
0.13
|
Alive
|
12(14.3%)
|
4(8.9%)
|
8(20.5%)
|
|
Asbestos Exposure
|
|
|
|
|
No
|
14(16.7%)
|
7(15.6%)
|
7(17.9%)
|
0.77
|
Yes
|
70(83.3%)
|
38(84.4%)
|
32(82.1%)
|
|
Laterality
|
|
|
|
|
Bilateral
|
3 (3.6%)
|
2(4.5%)
|
1 (2.6%)
|
0.72
|
Left
|
31 (36.9%)
|
15(33.3%)
|
16(41.0%)
|
|
Right
|
50 (59.5%)
|
28(62.2%)
|
22 (56.4%)
|
|
3.1 ASF1B expression in normal and tumor tissue
The clinical characteristics, and the Chi-square test results for the comparison of MM with different ASF1B expression is presented in Table 1.The Chi-square test revealed that the malignant mesothelioma was more prevalent in male (82.1%), mainly attacking on the right side (59.5%) and people with asbestos exposure history (83.3%), The incidence ratio of male to female was found to be 4.6:1. Most of patients were in stage iii (51.2%), however, significant difference was observed in none of those subgroups. (p>0.05).
In three GEO databases (GSE112154, GSE51024, GSE42977), ASF1B expression was found to be significantly higher in the MM patients compared to the paired normal lung tissues, normal pleura or peritoneum samples. (p<0.05) (Yingjie Jia Figure1A, 1C, 1D).
The receiver operating characteristic (ROC) curve of the test cohort is presented in Yingjie Jia Figure1B. Area under the curve (AUC) was 0.9326 (p<0.0001) and the cut-off value was 7.635 (Specificity:87.27%, Sensitivity:95.12%). The specificity and AUC were calculated for the discrimination of mesothelioma from the paired normal controls in GSE51024.
3.2 The relationship between ASF1B expression and some clinicopathological features
In TCGA-MM, there were no compared tissue of mesothelioma so we could not determine the differential expression of ASF1B between normal and tumor tissues. However, the survival analysis was performed, and it was found that in low-expression group, patients tend to have longer OS than high-expression group. (p<0.05, Fig 2).
Through COX univariate analysis, it was found that the reduced expression of ASF1B was significantly related to longer overall survival. (P<0.001) This means that the low ASF1B expression level is a significant protective factor for the survival. (Table 2). Other clinical information like stage, gender, laterality and histology showed no significant difference.
Previous studies have shown that pathological type is an independent predictor of survival of malignant mesothelioma. The degree of differentiation was also an independent predictor of survival of malignant mesothelioma.23 Although there is no significant difference (p>0.05), our univariate Cox regression analysis indicated that sarcomatoid mesothelioma might be a hazard factor as compared to the epithelioid type (HR=2.00, 95%CI=[0.79, 5.11]).
A correlation analysis was performed on 84 patients with clinical case characteristics (age of the parents, tumor histology, status of nodal metastasis, clinical stage) using the UALCAN (Fig. 3).
The results showed that the increase in ASF1B expression was significantly correlated with histology (Yingjie Jia Figure3D, 3E, 3F). The results were verified by the data in GEO (Yingjie Jia Figure3E)(p<0.05), Oncomine (Yingjie Jia Figure3F)(p<0.05). The results showed that significant upregulation of ASF1B in biphasic mesothelioma and sarcomatoid mesothelioma, which are considered to be more malignant. And this could be one explanation of how ASF1B induces the progression of mesothelioma.
Table 2. COX univariate analyses of clinical characteristics and survival.
Clinicopathologic variable
|
Univariate analysis
|
HR95%CI
|
p value
|
Age(≥65 vs. <65)
|
1.25[0.77, 2.05]
|
0.37
|
Gender (male vs. female)
|
1.20[0.64, 2.26]
|
0.58
|
TNM stage (II vs. I)
|
0.55[0.23, 1.31]
|
0.18
|
TNM stage (III vs. I)
|
0.57[0.27, 1.19]
|
0.13
|
TNM stage (IV vs. I)
|
0.54[0.23, 1.26]
|
0.15
|
Asbestos exposure (yes vs. no)
|
1.40[0.74, 2.65]
|
0.30
|
Laterality (left vs. right)
|
0.65[0.39, 1.07]
|
0.09
|
Laterality (bilateral vs. right)
|
1.15[0.35, 3.74]
|
0.82
|
Histology (sarcomatoid vs. epithelioid)
|
2.00[0.79, 5.11]
|
0.15
|
Histology (biphase vs. epithelioid)
|
1.68[0.97, 2.94]
|
0.07
|
ASF1B
|
1.65[1.29, 2.11]
|
<0.001
|
H3K9me1 is the most prevalent methylation mark with ASF1B. The correlation between ASF1B methylation level and the clinicopathological features, was investigated and it was found that the degree of ASF1B methylation was lower in 81-100 years old patients (Yingjie Jia Figure4A, 4B)(p<0.05). With the cbioportal, it was found that the ASF1B methylation level is negatively related to the ASF1B mRNA (R=-0.27, p=0.01), and this result is presented in Yingjie Jia Figure4C.
3.3. GO enrichment, KEGG pathways and comparable mutant genes
The GO enrichment includes three parts-molecular function, biological process, cellular component, and the ASF1B mainly function in histone binding in nucleus and chromatin (Table 3).
MSigDB enrichment analysis was performed for this study, to screen for signaling pathways with major variations caused by the expression of ASF1B (p < 0.05). Subsequently, the KEGG pathway analysis by the LinkedOmics platform (Yingjie Jia Figure. 5) was compared with the GSEA(Yingjie Jia Figure 6), and those pathways mainly functioned in the proliferation of cells.
Table 3. GO enrichment on ASF1B
Category
|
Accession
|
Term
|
|
Cellular Component
|
GO:0005634
|
nucleus
|
|
|
|
GO:0000785
|
chromatin
|
|
|
|
GO:0005654
|
nucleoplasm
|
|
|
GO:0032991
|
protein-containing complex
|
Biological Process
|
GO:0006335
|
DNA replication-dependent nucleosome assembly
|
|
|
GO:0006333
|
chromatin assembly or disassembly
|
|
|
GO:0030154
|
cell differentiation
|
|
|
GO:0006325
|
chromatin organization
|
|
|
GO:0001835
|
blastocyst hatching
|
|
|
GO:0007275
|
multicellular organism development
|
|
|
GO:0007283
|
spermatogenesis
|
|
|
GO:0006334
|
nucleosome assembly
|
Molecular Function
|
GO:0042393
|
histone binding
|
|
|
GO:0005515
|
protein binding
|
The main pathways modulated by ASF1B were mainly related to cell cycle and include chromosome segregation, organelle fission, spindle organization, DNA replication and mitotic cell cycle phase transition. At the same time, ASF1B also could downregulate the process of cargo loading into vesicle, protein activation cascade, response to interleukin-6 and platelet-derived growth factor receptor signaling pathway.
In GSEA between the high and low AFS1B expression datasets to screen out the differentially activated signaling pathways, ASF1B modulated the gene replication, nucleotide excision repair, and base excision repair to name a few.
As revealed by the previous studies, ASF1B in proliferating cells handles the pool of replicative histone H3.1 thereby acting as the prominent histone acceptor/donor during DNA replication24.
3.4 Related genes and their function by STRING
In our analysis, BDH2, KLHL9, C13orf33, CMAH, TM4SF1, TUSC1, BBS12, ANXA8, SLC4A4, SERPING1, LMBRD1, IFITM2, YPEL3, SYNE1, DYNC2LI1, RICH2, CALCOCO1, PIP5K1B, ITLN1, KBTBD3, NPR1 were negatively related to the expression of ASF1B in MM patients,
Meanwhile, CDCA5, RAD54L, STMN1, BIRC5, TROAP, CCNB2, CDCA3, FOXM1, KIFC1, UBE2C, SPC25, PKMYT1, MCM7, KIF18B, ORC1L, SPAG5, MCM2, KIF2C, GINS4, POLD1, AURKB, RAD51 were positively related to ASF1B in MM patients. (Yingjie Jia Figure 7A-7B)
STRING shows that these closely related genes were all function in the process of TATA box binding protein associated factor (TAF), Centromere kinetochore component CENP-T histone fold and core histone H2A/H2B/H3/H4(Yingjie Jia Figure 7C).
We investigated the relative expression of miRNAs and found that hsa-miR-503, hsa-miR-130b, hsa-miR-301b, hsa-miR-196b were positively correlated to the expression of ASF1B and they all could significantly shorten the survival of MM patients. In contrary, hsa-miR-29c, hsa-miR-195, hsa-miR-100, hsa-miR-30d were negatively correlated to the expression of ASF1B and could prolong the OS time. (Yingjie Jia Figure 8)
3.5 Association between ASF1B expression and composition of TIICs
In the database pf TCGA, macrophage M1, macrophage M2, CD8 T cell, T cell follicular helper and regulatory T cell showed a significant increase in ASF1B high-expression group. In contrary, the NK cells resting, monocytes, eosinophils, neutrophils, dendritic cells activated, CD4 T memory resting showed significant decrease under high expression of ASF1B.
In the database of GEO, the regulatory T cells (Tregs), macrophage M1, macrophage M0, dendritic cells resting showed a significant increase in ASF1B high-expression group. In contrary, the B cells naïve, plasma cells, monocytes, dendritic cells activated showed a decrease under the effect of high ASF1B expression (Yingjie Jia Figure 9 and Yingjie Jia Figure 10). The detail proportion of immune cells data can be obtained in the supplement table 2(GEO) and supplement table 3(TCGA).
The results of the present study revealed that under the regulatory influence of ASF1B, the microenvironment changed a lot. These results were verified by GEO and TCGA. Two databases showed that under the effect of ASF1B, the changes of DC cells, Tregs and the subgroup of macrophages indicated the immune suppression in the tumour microenvironment. More than that, many other immune cells need further investigated.