3.1 Clinical information
Albumin (ALB), globulin (GLO), alanine transaminase (ALT), aspartate aminotransferase (AST), cholinesterase (CHE), alkaline phosphatase (ALP), total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IBIL), glutamyl transferase (GGT), creatinine (Cr), alpha fetoprotein (AFP), prothrombin time (PT), international normalized ratio (INR), white blood cell (WBC), hemoglobin (Hb), and platelet count (PLT), were measured and analyzed by ANOVA. ALB, ALT, CHE, ALP, TBIL, DBIL, IBIL, PT, and INR differed significantly (P<0.05) among the HC, CHB and ACLF groups (Table 1).
Table 1. Clinical parameters in the HC, CHB and ACLF groups and ANOVA statistics
Parameter
|
Unit
|
CHB
|
ACLF
|
HC
|
ANOVA
|
Age
|
years
|
39.33±4.73
|
40.33±7.02
|
33.67±3.21
|
0.313
|
ALB
|
g/L
|
40.90±0.44
|
35.20±2.16
|
51.11±1.00
|
<0.001
|
GLO
|
g/L
|
24.60±3.40
|
25.93±6.49
|
27.53±1.12
|
0.716
|
ALT
|
μmol/L
|
112.33±149.47
|
853.00±138.62
|
14.00±1.00
|
<0.001
|
AST
|
μmol/L
|
36.67±19.76
|
304.67±271.88
|
10.00±1.73
|
0.112
|
CHE
|
U/L
|
7782.33±1050.24
|
5092.00±361.96
|
8136.00±485.45
|
0.003
|
ALP
|
U/L
|
65.00±11.79
|
163.00±52.60
|
52.33±3.21
|
0.009
|
TBIL
|
μmol/L
|
14.00±12.49
|
425.67±172.69
|
14.00±2.65
|
0.003
|
DBIL
|
μmol/L
|
6.33±5.86
|
251.67±93.18
|
2.00±1.00
|
0.002
|
IBIL
|
μmol/L
|
7.67±6.66
|
174.00±94.87
|
12.00±2.65
|
0.016
|
GGT
|
μmol/L
|
48.67±54.50
|
92.67±38.89
|
15.67±1.53
|
0.125
|
Cr
|
μmol/L
|
75.67±17.21
|
59.67±15.04
|
71.67±4.04
|
0.376
|
AFP
|
μg/L
|
6.33±4.29
|
83.47±64.43
|
3.27±0.49
|
0.065
|
PT
|
s
|
11.77±0.87
|
38.50±5.30
|
11.14±0.89
|
<0.001
|
INR
|
-
|
1.03±0.08
|
3.68±0.49
|
1.01±0.05
|
<0.001
|
WBC
|
109/L
|
5.73±1.37
|
6.20±1.87
|
5.80±0.46
|
0.904
|
Hb
|
g/L
|
151.33±24.01
|
148.33±5.13
|
160.00±3.61
|
0.610
|
PLT
|
109/L
|
164.67±21.01
|
103.00±39.04
|
207.67±16.29
|
0.009
|
3.2 Overview of circRNA profiles
High-throughput sequencing revealed a total of 21,101 circRNAs in the HC, CHB and ACLF subjects. Among the 21,101 circRNAs, 13,024 were detected in HCs, 13,172 in CHB patients, 12,166 in ACLF patients, and 6,445 in all three groups (Figure. 1B). There are no outliers of circRNA expression in the HC, CHB, or ACLF group due to instrument or operational anomalies. The distribution of circRNAs on the genome is shown in Supplemental Figure 2.
Approximately 86.61% of circRNAs had a predicted spliced length of less than 2,000 nt; 48.44% had a length less than 500 nt and 26.53% had a length between 500 nt and 1,000 nt (Figure. 1A).
3.3 Identification of differentially expressed circRNAs and mRNA
According to the circRNA criteria (fold change > 2 and difference of P<0.05 between two groups), 550 differentially expressed circRNAs in the CHB and ACLF groups were identified compared with the HC group. Among them, 176 differentially expressed circRNAs were selected in the CHB group (89 up-regulated and 87 down-regulated) and 374 differentially expressed circRNAs were selected in the ACLF group (181 up-regulated and 193 down-regulated). The differentially expressed circRNAs between the ACLF and HC groups are shown in a volcano plot (Figure. 2A). The unsupervised hierarchical cluster analysis results are shown in Figure 2B.
The differentially expressed candidate circRNAs among the three groups showing progressive up- or down-regulation were further analyzed. There were 4 up-regulated and 14 down-regulated circRNAs (Table 2, Figure 3). mRNAs were analyzed by the same method. The results showed there were 7 up-regulated and 9 down-regulated mRNA. The expression information and statistic differences of circRNA and mRNA were shown in the Supplemental Table 1,2.
Table 2. Differentially expressed circRNAs (ACLF vs. HC)
CircRNA
|
CHB vs. HC
|
ACLF vs. HC
|
Regulation
|
Chromosome
|
Length
|
Strand
|
Type
|
|
FC2
|
P_val
|
FC2
|
P_val
|
|
|
|
|
|
circRNA_00389
|
1.6887
|
0.1679
|
3.5835
|
0.0006
|
down
|
chr1
|
363
|
+
|
sense-overlapping
|
circRNA_02808
|
1.6185
|
0.4137
|
3.6685
|
0.0080
|
down
|
chr10
|
361
|
-
|
sense-overlapping
|
circRNA_03646
|
1.3686
|
0.4386
|
5.8769
|
0.0003
|
down
|
chr11
|
6085
|
-
|
intergenic
|
circRNA_04205
|
1.5300
|
0.2323
|
3.2813
|
0.0112
|
down
|
chr12
|
1368
|
-
|
sense-overlapping
|
circRNA_05781
|
1.7071
|
0.1450
|
3.2003
|
0.0131
|
down
|
chr14
|
3199
|
-
|
intergenic
|
circRNA_06400
|
1.5643
|
0.1009
|
3.3205
|
0.0000
|
down
|
chr14
|
369
|
-
|
exonic
|
circRNA_07485
|
1.9656
|
0.1531
|
3.2457
|
0.0107
|
down
|
chr16
|
845
|
+
|
exonic
|
circRNA_07734
|
1.7126
|
0.1422
|
2.8129
|
0.0077
|
down
|
chr16
|
313
|
+
|
sense-overlapping
|
circRNA_08533
|
2.7321
|
0.0232
|
13.2602
|
0.0000
|
down
|
chr17
|
939
|
+
|
sense-overlapping
|
circRNA_09331
|
1.4930
|
0.2115
|
7.4732
|
0.0004
|
down
|
chr18
|
658
|
+
|
exonic
|
circRNA_17523
|
1.5966
|
0.1871
|
3.7990
|
0.0074
|
down
|
chr6
|
640
|
-
|
sense-overlapping
|
circRNA_17526
|
2.4262
|
0.0631
|
6.1841
|
0.0091
|
down
|
chr6
|
439
|
-
|
sense-overlapping
|
circRNA_19214
|
1.8125
|
0.1170
|
7.6139
|
0.0001
|
down
|
chr8
|
258
|
-
|
sense-overlapping
|
circRNA_04487
|
0.5663
|
0.2698
|
4.4060
|
0.0010
|
up
|
chr12
|
455
|
+
|
sense-overlapping
|
circRNA_15699
|
0.6673
|
0.3637
|
3.6601
|
0.0351
|
up
|
chr5
|
1923
|
+
|
exonic
|
circRNA_16083
|
0.7123
|
0.5781
|
5.4313
|
0.0000
|
up
|
chr5
|
5262
|
+
|
exonic
|
circRNA_16987
|
0.8639
|
0.8767
|
7.9830
|
0.0000
|
up
|
chr6
|
527
|
-
|
sense-overlapping
|
Note: FC, fold‑change, absolute ratio (non-log scale) of normalized intensities between two groups (threshold, 2.0). P‑value calculated using a paired t‑test (threshold, 0.05). circRNA, circular RNA; hsa, Homo sapiens; +, positive strand; ‑, negative strand.
3.4 GO and KEGG analyses
The functions of the target genes of differentially expressed circRNAs in the ACLF group were further analyzed by GO enrichment and KEGG pathway analysis. Significant terms in the GO and KEGG pathway analyses (P<0.05) were selected and ranked by p-value.
The top-10 down-regulated GO terms, classified by BP, CC, MF, are shown in Supplemental Figure 3A. The most enriched BP, CC and MF terms were regulation of lymphocyte activation, integrin alphaL-beta2 complex and RNA polymerase I core binding, respectively. Among the up-regulated GO terms, the most enriched BP, CC and MF terms were positive regulation of macrophage tolerance induction, Golgi lumen, and extracellular matrix structural constituent, respectively (Supplemental Figure. 3B).
KEGG pathway analysis (Table 3, Supplemental Figure 4 ) revealed that the most enriched term in down-regulated circRNAs was primary immunodeficiency, while in up-regulated circRNAs, NOD-like receptor signaling pathway was the most enriched term. As shown in Table 3, KEGG pathway hsa05340 was the most important. The KEGG map is shown in Supplemental Figure 5.
Table 3. KEGG analysis of ACLF risk status in up- and down-regulated circRNAs ranked by p-value and enrichment score
CircRNA
|
Path-id
|
P(10-3)
|
Core
|
Term
|
Down-regulated circRNAs
|
|
|
circRNA_00389
|
hsa05340
|
0.0424
|
91.525
|
Primary immunodeficiency
|
circRNA_00389
|
hsa05166
|
0.0177
|
11.5368
|
HTLV-I infection
|
circRNA_07734
|
hsa05150
|
0.1249
|
53.8382
|
Staphylococcus aureus infection
|
circRNA_00389
|
hsa04650
|
0.1724
|
14.0808
|
Natural killer cell-mediated cytotoxicity
|
circRNA_07734
|
hsa05144
|
0.2829
|
35.8922
|
Malaria
|
circRNA_07734
|
hsa05323
|
0.3921
|
30.5083
|
Rheumatoid arthritis
|
circRNA_08533
|
hsa05219
|
0.3921
|
30.5083
|
Bladder cancer
|
circRNA_07734
|
hsa05416
|
0.9218
|
19.8967
|
Viral myocarditis
|
circRNA_08533
|
hsa05230
|
1.5902
|
15.1281
|
Central carbon metabolism in cancer
|
circRNA_07734
|
hsa04514
|
1.7781
|
14.3008
|
Cell adhesion molecules (CAMs)
|
Up-regulated circRNAs
|
|
|
circRNA_15699
|
hsa04621
|
0.4187
|
29.5242
|
NOD-like receptor signaling pathway
|
circRNA_16083
|
hsa04514
|
1.7781
|
14.3008
|
Cell adhesion molecules (CAMs)
|
circRNA_16987
|
hsa04915
|
3.1911
|
10.6424
|
Estrogen signaling pathway
|
circRNA_04487
|
hsa04722
|
8.5407
|
6.4454
|
Neurotrophin signaling pathway
|
3.5 circRNA-miRNA co-expression network
A network map of the most significant relationships between circRNAs and miRNAs is shown in Figure 4. Among the differentially expressed circRNAs selected from the ACLF group, circRNA_03646, circRNA_06400, circRNA_09331, circRNA_17523, circRNA_05781, circRNA_07734, circRNA_08533, circRNA_16562, circRNA_17526, circRNA_15699, circRNA_16083 underpinned the circRNA-miRNA co-expression network, as determined by Arraystar. Among them, thee expression of circRNA_03646, circRNA_06400, circRNA_07734, circRNA_08533, circRNA_16083 had statistical significance between ACLF and HC group. Moreover, circRNA_07734, circRNA_08533, and circRNA_16083 were also had statistical significance in KEGG analysis. circRNA_07734 only contacted two miRNA in the network, which less than circRNA_08533 and circRNA_16083.
3.6 Correlation analysis
Spearman's correlation analysis indicated that the RPM values ofcircRNA_07734, circRNA_08533, and circRNA_16083 were related to ALT, AST, TBIL, and IBIL (Table 4). circRNA_07734 andcircRNA_08533 showed strong negative correlation with ALT, AST, TBIL, and IBIL, which indicated if ALT, AST, TBIL, and IBIL increased higher, the expression of circRNA_07734 and circRNA_08533 would be lower. circRNA_16083 is different with circRNA_07734 and circRNA_08533, which had positive correlation with ALT, AST, TBIL, and IBIL. In clinical practice, ALT, AST, TBIL, and IBIL are used to evaluate the liver function. Commonly, the higher the ALT, AST, TBIL, and IBIL level were, the more severe the liver injury will be. Therefore, circRNA_07734, circRNA_08533, and circRNA_16083 may be useful as the sensitive indicators of liver injury.
Table 4. Spearman's correlations analysis of circRNA_07734, circRNA_08533, and circRNA_16083 with differentially expressed mRNA and alanine transaminase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), and indirect bilirubin (IBIL) (n = 9)
mRNA and
liver index
|
circRNA_07734
|
circRNA_08533
|
circRNA_16083
|
Coefficient
|
P
|
Coefficient
|
P
|
Coefficient
|
P
|
NM_000361.2
|
-0.417
|
0.265
|
-0.700*
|
0.036
|
0.65
|
0.058
|
NM_001005862.2
|
.733*
|
0.025
|
0.800**
|
0.01
|
-0.733*
|
0.025
|
NM_001199139.1
|
-.667*
|
0.05
|
-0.867**
|
0.002
|
0.65
|
0.058
|
NM_004633.3
|
-.683*
|
0.042
|
-0.883**
|
0.002
|
0.667*
|
0.05
|
NM_005204.3
|
-.750*
|
0.02
|
-0.883**
|
0.002
|
0.750*
|
0.02
|
NM_006720.3
|
.783*
|
0.013
|
0.867**
|
0.002
|
-0.900**
|
0.001
|
NM_014790.4
|
.700*
|
0.036
|
0.950**
|
<0.001
|
-0.733*
|
0.025
|
NM_080913.3
|
-.833**
|
0.005
|
-0.983**
|
<0.001
|
0.833**
|
0.005
|
XM_005257807.2
|
.733*
|
0.025
|
0.917**
|
0.001
|
-0.733*
|
0.025
|
XM_011511061.1
|
-.817**
|
0.007
|
-0.967**
|
<0.001
|
0.850**
|
0.004
|
XM_011520156.1
|
-0.267
|
0.488
|
-0.317
|
0.406
|
0.5
|
0.17
|
XM_011520271.1
|
0.567
|
0.112
|
0.833**
|
0.005
|
-0.617
|
0.077
|
XM_011524254.1
|
0.633
|
0.067
|
0.583
|
0.099
|
-0.717*
|
0.03
|
XM_011524699.1
|
.717*
|
0.03
|
0.800**
|
0.01
|
-0.750*
|
0.02
|
XM_011531698.1
|
-.800**
|
0.01
|
-0.867**
|
0.002
|
0.883**
|
0.002
|
XM_011537524.1
|
0.517
|
0.154
|
0.6
|
0.088
|
-0.533
|
0.139
|
ALT
|
-.733*
|
0.025
|
-.733*
|
0.025
|
.733*
|
0.025
|
AST
|
-0.469
|
0.203
|
-0.603
|
0.086
|
.669*
|
0.049
|
TBIL
|
-0.583
|
0.099
|
-.883**
|
0.002
|
.767*
|
0.016
|
IBIL
|
-0.65
|
0.058
|
-.883**
|
0.002
|
.850**
|
0.004
|