3.1 Retrieval Results
A total of 433 non-paired samples were retrieved from the TCGA database, while 188 datasets were obtained from GEO. After reviewing dataset descriptions against inclusion criteria, 165 datasets were excluded, with an additional 2 datasets excluded based on exclusion criteria. The exclusion of dataset GSE25097 was due to the absence of GBA annotation in chip information, and dataset GSE33814 was excluded as an HCC sample gene expression matrix could not be located. Ultimately, 21 datasets were deemed suitable for analysis, encompassing 1003 liver cancer samples for correlation analysis.
3.2 Analysis of GBA Expression Across Various Cancers
Examination of liver cancer specimens obtained from the TCGA database indicated that GBA exhibits elevated expression levels in ten distinct cancer types, namely BLCA, BRCA, CHOL, ESCA, HNSC, KIRC, LIHC, LUAD, READ, STAD, and UCEC, while demonstrating reduced expression in three other cancer types: COAD, KICH, and THCA (Fig. 1).
3.3 Analysis of the Impact of GBA on Survival Rates in HCC
In order to evaluate the influence of GBA on the survival outcomes of patients with HCC, we conducted an analysis of the impact of GBA expression on time-to-survival rates across various demographic subgroups, including gender, age (≥ 60 years and < 60 years), and body mass index (BMI) categories (> 25 kg/m2 and < 25 kg/m2). Given the larger sample sizes in stages I and II, patients were aggregated into two groups: stages I and II combined, and stages III and IV combined.The study found that individuals with higher GBA expression levels exhibited a greater survival rate, as indicated by a hazard ratio (HR) of 1.53 (95% CI: 1.08, 2.16) (Fig. 2A). Furthermore, in patients aged 60 and above, those with low GBA expression levels demonstrated a significantly higher overall survival probability with an HR of 2.61 (95% CI: 1.60, 4.26), p < 0.001, while no statistically significant differences were observed in patients under the age of 60 (Figs. 2B and 2C).There were no statistically significant differences in overall survival rates across various BMI stratifications (Figs. 2D and 2E). However, among patients with a BMI > 25 kg/m2, individuals with high GBA expression had a hazard ratio of 1.72 (95% CI: 0.99, 3.00), p = 0.05 compared to those with low GBA expression. Similarly, no significant disparities were observed in overall survival rates based on different levels of GBA expression among different genders (Figs. 2F and 2G). Nevertheless, in male patients, those with high GBA expression had a hazard ratio of 1.53 (95% CI: 0.98, 2.40), p = 0.063 compared to those with low GBA expression.
3.4 Gene Sets Related to GBA in HCC Cases in the GEO Database
A total of 188 datasets were initially obtained from the Gene Expression Omnibus (GEO) database. Following a thorough review of dataset descriptions against predetermined inclusion criteria, 165 datasets were deemed ineligible for further analysis. Additionally, 2 datasets were excluded based on specific exclusion criteria; dataset GSE25097 was excluded due to the absence of GBA annotation in chip information, while dataset GSE33814 was excluded due to the unavailability of an HCC sample gene expression matrix. Ultimately, 21 datasets met the criteria for inclusion in the analysis, encompassing a total of 1003 liver cancer samples for correlation analysis, as detailed in Table 2.
Table 2
Basic Information of GSE Datasets Included in the Study
GSE number
|
Etiology Description
|
Paired Samples
|
Number of HCC Samples
|
Number of Control Samples
|
Control Tissue Status
|
Peritumoral Tissue Status
|
Viral Infection Status
|
Number of GBA-related Genes
|
GSE5975
|
HBV
|
N
|
238
|
0
|
NR
|
Cirrhosis
|
HBV
|
8846
|
GSE17548
|
NR
|
N
|
17
|
20
|
Cirrhosis
|
Cirrhosis
|
HBV
|
582
|
GSE45434
|
NR
|
N
|
16
|
0
|
NR
|
NR
|
NR
|
1846
|
GSE45435
|
NR
|
N
|
31
|
0
|
NR
|
NR
|
NR
|
3230
|
GSE45436
|
NR
|
N
|
93
|
41
|
NR
|
NR
|
NR
|
6474
|
GSE46444
|
NR
|
N
|
88
|
48
|
Cirrhosis
|
NR
|
NR
|
2603
|
GSE62232
|
various etiologies
|
N
|
81
|
10
|
NR
|
NR
|
NR
|
4712
|
GSE89377[47]
|
NR
|
N
|
40
|
67
|
normal
|
NR
|
NR
|
6186
|
GSE101685
|
NR
|
N
|
8
|
24
|
normal
|
NR
|
NR
|
12827
|
GSE116174
|
NR
|
N
|
64
|
0
|
NR
|
NR
|
NR
|
3060
|
GSE164760
|
NR
|
N
|
53
|
29
|
adjacent HCC
|
NASH
|
NR
|
2790
|
GSE14811
|
HBV
|
Y
|
56
|
56
|
NR
|
NR
|
NR
|
1103
|
GSE45050
|
NR
|
Y
|
6
|
6
|
a
|
NR
|
NR
|
572
|
GSE46408
|
NR
|
Y
|
6
|
6
|
NR
|
NR
|
NR
|
2610
|
GSE57957
|
NR
|
Y
|
39
|
39
|
NR
|
NR
|
NR
|
2796
|
GSE59259
|
alcoholic
|
Y
|
8
|
8
|
NR
|
NR
|
NR
|
2231
|
GSE64041
|
NR
|
Y
|
60
|
60
|
NR
|
NR
|
NR
|
9671
|
GSE74656
|
NR
|
Y
|
5
|
5
|
NR
|
NR
|
NR
|
2306
|
GSE76427
|
NR
|
Y
|
52
|
52
|
NR
|
NR
|
NR
|
2987
|
GSE84005
|
NR
|
Y
|
38
|
38
|
NR
|
NR
|
NR
|
2885
|
GSE99807
|
NR
|
Y
|
4
|
4
|
steatosis
|
NR
|
NR
|
1387
|
a: 2 Cirrhosis and 2 fatty change and 3 normal;NR: Not Report; N: Yes; N: No; HBV: hepatitis B virus. |
3.5 GSEA Analysis to Explore Pathways Involving GBA-Related Genes
In order to investigate the influence of GBA on HCC progression, we conducted a correlation analysis between GBA and other genes within the gene expression profiles of HCC samples sourced from multiple datasets. Genes exhibiting significant correlations with GBA were organized into gene sets (referred to as GBA sets or GBAs) for subsequent Gene Set Enrichment Analysis (GSEA). Our analysis demonstrated enrichment of GBAs across ten datasets, as detailed in Table 3. Specifically, the GSEA results indicated enrichment of ten gene sets associated with hsa04142 (Lysosome), nine gene sets linked to hsa01100 (Metabolic pathways), and three gene sets related to hsa00600 (Sphingolipid metabolism).
Table 3
the results of GSE analysis of GBA-related genes in the dataset
GSE number
|
Pathway contain GBA
|
Set Size
|
Enrichment Score
|
NES
|
P value
|
rank
|
GSE5975
|
hsa04142
|
82
|
0.39
|
1.96
|
0.00
|
2705
|
GSE17548
|
none
|
|
|
|
|
|
GSE45434
|
hsa01100
|
147
|
0.25
|
2.73
|
0.00
|
900
|
hsa04142
|
24
|
0.45
|
2.43
|
0.00
|
367
|
GSE45435
|
hsa01100
|
201
|
0.25
|
3.04
|
0.00
|
775
|
hsa04142
|
37
|
0.55
|
3.55
|
0.00
|
702
|
GSE45436
|
hsa01100
|
465
|
0.19
|
1.74
|
0.00
|
2043
|
hsa04142
|
67
|
0.43
|
2.79
|
0.00
|
1254
|
GSE46444
|
none
|
|
|
|
|
|
GSE62232
|
hsa04142
|
37
|
0.35
|
1.92
|
0.01
|
632
|
hsa00600
|
12
|
0.54
|
1.98
|
0.00
|
87
|
GSE89377
|
hsa01100
|
399
|
0.17
|
1.72
|
0.00
|
1572
|
GSE101685
|
hsa04142
|
79
|
0.46
|
4.19
|
0.00
|
3802
|
hsa00600
|
26
|
0.32
|
1.85
|
0.02
|
483
|
hsa00511
|
10
|
0.53
|
1.96
|
0.01
|
833
|
GSE116174
|
hsa01100
|
216
|
0.28
|
3.18
|
0.00
|
1291
|
hsa04142
|
35
|
0.49
|
3.05
|
0.00
|
779
|
GSE164760
|
hsa01100
|
281
|
0.29
|
2.47
|
0.00
|
887
|
GSE57957
|
hsa04142
|
170
|
0.23
|
2.23
|
0.00
|
515
|
hsa01100
|
22
|
0.47
|
2.25
|
0.00
|
459
|
GSE59259
|
none
|
|
|
|
|
|
GSE64041
|
hsa01100
|
498
|
0.30
|
4.12
|
0.00
|
3311
|
hsa04142
|
47
|
0.48
|
3.33
|
0.00
|
3100
|
hsa00600
|
18
|
0.39
|
1.84
|
0.02
|
2541
|
GSE74656
|
hsa01100
|
211
|
0.32
|
5.04
|
0.00
|
871
|
GSE76427
|
none
|
|
|
|
|
|
GSE84005
|
hsa04142
|
39
|
0.36
|
2.18
|
0.00
|
1451
|
GSE99807
|
none
|
|
|
|
|
|
GSE14811
|
none
|
|
|
|
|
|
GSE45050
|
none
|
|
|
|
|
|
GSE46408
|
none
|
|
|
|
|
|
hsa04142:Lysosome; hsa01100: Metabolic pathways; hsa00600: Sphingolipid metabolism. |
3.6 Impact of GBA on Metabolism
In order to investigate the impact of GBA on metabolism in HCC samples, a co-expression analysis was performed on genes within the hsa01100 (Metabolic pathways) from GBAs across nine gene sets. The occurrence of each gene across the nine metabolic pathways was recorded, resulting in a total of 715 genes from the union of gene sets within these pathways. Within these pathways, 46 genes were found to appear at least five times, as detailed in Tables 4 and 5.
Table 4
Genes Appearing at Least Five Times in Various Metabolic Pathways
symbol
|
frequency
|
average correlation
|
|
symbol
|
frequency
|
average correlation
|
GBA
|
10
|
1.00
|
|
FDPS
|
6
|
0.41
|
PMVK
|
10
|
0.61
|
|
GALE
|
6
|
0.43
|
DPM3
|
9
|
0.53
|
|
GGPS1
|
6
|
0.44
|
FLAD1
|
9
|
0.57
|
|
GNS
|
6
|
0.50
|
HEXB
|
8
|
0.47
|
|
GPX1
|
6
|
0.49
|
NDUFS2
|
8
|
0.61
|
|
LPIN1
|
6
|
0.47
|
NEU1
|
8
|
0.52
|
|
PIGM
|
6
|
0.59
|
SDHC
|
8
|
0.58
|
|
PYCR2
|
6
|
0.40
|
ACOT8
|
7
|
0.52
|
|
QPRT
|
6
|
0.50
|
ACSM1
|
7
|
0.44
|
|
SEPHS2
|
6
|
0.50
|
ATP6V1E1
|
7
|
0.43
|
|
AKR1C3
|
5
|
0.46
|
B4GALT3
|
7
|
0.51
|
|
ALG1
|
5
|
0.44
|
COX8A
|
7
|
0.54
|
|
ALG9
|
5
|
0.50
|
GMPPA
|
7
|
0.36
|
|
ATP6V0A1
|
5
|
0.42
|
PI4KB
|
7
|
0.50
|
|
BPGM
|
5
|
0.59
|
PIGC
|
7
|
0.56
|
|
CANT1
|
5
|
0.38
|
PIP4K2C
|
7
|
0.48
|
|
CERS2
|
5
|
0.56
|
PPOX
|
7
|
0.55
|
|
CYP2R1
|
5
|
0.44
|
THTPA
|
7
|
0.47
|
|
GLB1
|
5
|
0.40
|
AMDHD2
|
6
|
0.44
|
|
GPAA1
|
5
|
0.47
|
ATP6AP1
|
6
|
0.45
|
|
IMPA2
|
5
|
0.38
|
ATP6V0B
|
6
|
0.48
|
|
NDUFA1
|
5
|
0.48
|
ATP6V0C
|
6
|
0.44
|
|
NME7
|
5
|
0.42
|
ATP6V0E2
|
6
|
0.49
|
|
NUDT5
|
5
|
0.52
|
ATP6V1D
|
6
|
0.49
|
|
PDE6D
|
5
|
0.46
|
BPNT1
|
6
|
0.42
|
|
PIGU
|
5
|
0.42
|
DOLK
|
6
|
0.43
|
|
TK1
|
5
|
0.49
|
Table 5
Genes Appearing at Least Five Times in Lysosome Pathways
symbol
|
frequency
|
average correlation
|
|
symbol
|
frequency
|
average correlation
|
GBA
|
11
|
1.00
|
|
MCOLN1
|
6
|
0.48
|
HEXB
|
10
|
0.51
|
|
AP1B1
|
5
|
0.42
|
CLN3
|
9
|
0.57
|
|
ATP6V0A1
|
5
|
0.45
|
GNS
|
9
|
0.53
|
|
ATP6V0C
|
5
|
0.46
|
NEU1
|
9
|
0.52
|
|
ATP6V0D1
|
5
|
0.52
|
ATP6AP1
|
8
|
0.44
|
|
CTNS
|
5
|
0.40
|
CTSA
|
8
|
0.57
|
|
CTSF
|
5
|
0.53
|
TPP1
|
8
|
0.53
|
|
DNASE2
|
5
|
0.50
|
ATP6V0B
|
7
|
0.47
|
|
GLB1
|
5
|
0.53
|
CTSD
|
7
|
0.51
|
|
GNPTG
|
5
|
0.41
|
SLC17A5
|
7
|
0.48
|
|
IGF2R
|
5
|
0.43
|
3.7 High-Frequency Gene BP Interaction Analysis
The biological processes mediated by GBA-related genes are predominantly involved in metabolic pathways within HCC samples, as illustrated in Fig. 3. These processes encompass cellular metabolic activities, nitrogen compound metabolism, lipid metabolism, carbohydrate metabolism, small molecule metabolism, phosphorus metabolism, peptide metabolism, and sulfur metabolism. The interactions occurring within these processes have a significant impact on fundamental metabolic pathways, including monosaccharide and N-acetylglucosamine metabolic processes, ceramide metabolic processes, glycosphingolipid metabolic processes, carboxylic acid metabolic processes, and glutathione metabolic processes. These interactions ultimately influence the GPI anchor biosynthetic process and ATP biosynthetic and metabolic processes, potentially leading to alterations in metabolic homeostasis within organisms.
3.8 High-Frequency Gene GO Analysis
In order to delve deeper into the biological mechanisms associated with GBA-related genes, a ClueGO analysis was performed on these frequently occurring genes. The findings indicate that these genes are primarily involved in nine level-three GO terms, including primary active transmembrane transporter activity, lysosome organization, glycerophospholipid biosynthetic process, organophosphate biosynthetic process, membrane lipid metabolic process, organic hydroxy compound biosynthetic process, mannosylation, ribonucleoside bisphosphate metabolic process, and alpha-amino acid biosynthetic process (Figs. 4A). Significantly, GBA interacts with CTSD, PIP4K2C, CLN3, TPP1, HEXB, and PI4KB in vacuole organization and collaborates with CLN3 and TPP1 in the lysosomal protein catabolic process (Figs. 4B). Furthermore, GBA is involved in autophagosome organization in conjunction with PIP4K2C and CTSD, and participates in the respiratory electron transport chain with COX8A, NDUFA1, NDUFS2, and SDHC (Figs. 4C). In the context of membrane lipid metabolism, GBA is predominantly involved in various catabolic and biosynthetic processes related to glycosphingolipids, carbohydrates, ceramides, sphingolipids, lipoproteins, and proteins. Key genes implicated in these processes include NEU1, HEXB, and GLB1. Additionally, GBA plays a role in alcohol and polyol biosynthesis (Figs. 4D).
CC analysis reveals that GBA is predominantly localized within the Golgi apparatus, extracellular exosome, organelle subcompartment, endoplasmic reticulum, and membrane-bounded organelle (Figs. 4E). At the molecular function level, GBA primarily demonstrates glycosyltransferase activity and catalytic activity (Figs. 4F). Interaction analysis was further conducted on the 67 high-frequency genes, as depicted in Fig. 5, indicating that GBA exhibits co-expression relationships with 24 high-frequency genes, including shared protein domains with HEXB and GLB1.
3.9 Regulatory Effects of Induced High and Low Expression of GBA on Downstream Genes in MHCC-97H Cells
In the MHCC-97H hepatocellular carcinoma cell line, the overexpression of GBA expression levels demonstrated a positive correlation with the upregulation of lysosomal metabolic genes NEU1, CTSD, CTSA, GALNS, and CLN3, as well as non-lysosomal genes ACOT8, FDPS, PMVK, PIGC, and GALNS. Conversely, the downregulation of GBA expression was associated with a corresponding decrease in the expression levels of NEU1 and CLN3 (see Fig. 6).
3.10 Relationship between GBA-Regulated Gene Expression Levels and Survival Time in Liver Cancer Patients
Analysis of the expression levels of NEU1, CTSD, CTSA, GALNS, CLN3, ACOT8, FDPS, PMVK, and PIGC in liver cancer patients using the TCGA database indicated that upregulation of these genes is strongly correlated with a poor prognosis in this patient population (see Figs. 7A-K).
3.11 Impact of GBA on Downstream Genes Potentially Related to the Concentration of Its Substrate - Glucosylceramidase
In the studies described, it was found that GBA expression levels regulate the expression of ACOT8. Considering that the ACOT family genes are a group of isoenzymes involved in the hydrolysis of acetyl groups from medium and long-chain fatty acids, the impact of GBA on ACOT was explored by inducing high expression of UGCG (thereby increasing the concentration of glucosylceramide) and simultaneous high expression of GBA. The results showed that ACOT4, ACOT9, and ACOT11 were upregulated along with the upregulation of UGCG and GBA, (Fig. 8) .
In the aforementioned studies, it was determined that GBA expression levels play a role in regulating the expression of ACOT8. Given that the ACOT gene family consists of isoenzymes involved in the hydrolysis of acetyl groups from medium and long-chain fatty acids, the influence of GBA on ACOT was investigated by inducing high levels of UGCG (resulting in increased glucosylceramide concentration) and simultaneous high levels of GBA. The findings indicated that the upregulation of UGCG and GBA led to the upregulation of ACOT4, ACOT9, and ACOT11, as depicted in Fig. 8.