Available via license: CC BY 4.0
Content may be subject to copyright.
Page 1/24
Comprehensive characterization of pyroptosis
reveals novel molecular typing of biliary atresia as
well as contributes to precise treatment
Tengfei Li
Tianjin Medical University
Qipeng Zheng
Children’s Hospital Capital Institute of Pediatrics
Xueting Wang
Tianjin Medical University
Qianhui Yang
Tianjin Medical University
Mengdi Li
Tianjin Medical University
Xiaodan Xu
Tianjin Medical University
Yilin Zhao
Tianjin Medical University
Fangyuan Zhao
Tianjin Medical University
Ruifeng Zhang
Tianjin Medical University
Zhiru Wang
Tianjin Medical University
Rongjuan Sun
Tianjin Medical University
Shaowen Liu
Tianjin Medical University
Jiayinaxi Musha
Tianjin Medical University
Yanran Zhang
Tianjin Medical University
Jianghua Zhan ( zhanjianghuatj@163.com )
Tianjin Children’s Hospital
Page 2/24
Research Article
Keywords: biliary atresia, pyroptosis, molecular type, immune microenvironment, Cell-cell communication
Posted Date: March 13th, 2023
DOI: https://doi.org/10.21203/rs.3.rs-2665698/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
Page 3/24
Abstract
Background: Biliary Atresia (BA) is a devastating pediatric cholangiopathy affecting the bile ducts of the
liver. Current research has found a variety of causes for BA, with inammation and brosis is more
studied. However, these etiological mechanisms are not present in all patients. Pyroptosis has been
increasingly appreciated as a programmed cell death process but is less studied in BA. We have re-
classied BA by integrating gene microarray data and scRNA-seq data to support individualized clinical
treatment and mechanistic studies.
Methods: The BA microarray dataset GSE122340 was downloaded from the Gene Expression Omnibus
(GEO) database. GSE46960 and GSE15235 and sequencing data of identical twins as validation cohorts.
Through retrospective analysis, 17 differential pyroptosis genes (DEPRGs) were used for typing research.
An effective method for identifying BA typing through machine learning algorithms. Subsequently, we
performed drug prediction for the pyroptosis subtype to enable individualized treatment. Pyroptosis-score
was constructed and combined with scRNA-seq datasets to reveal immune cells and pathways activated
during pyroptosis.
Results: Two novel subtypes of pyroptosis were identied. The C1 subtype shows activation of
pyroptosis, enhanced inammatory response, and increased inltration of monocytes and neutrophils.
The C2 subtype exhibits cell cycle activation, low pyroptosis, and a milder inammatory response.
Macrophage pyroptosis may promote the inltration of more immune cells and the release of
inammatory factors, further aggravating the occurrence of hepatic pyroptosis, which in turn leads to a
poorer prognosis in inammatory BA.
Conclusion: In summary, we have dened two novel subtypes of pyroptosis and offered the possibility of
identifying them and individualizing treatment. The role of macrophages, neutrophils, and plasma cells in
the pyroptosis process of BA is worthy of further study.
Introduction
Biliary atresia (BA) is a neonatal disease characterized by inammation and brosis (1). Patients may not
have the same outcomes for the same surgical and pharmacological treatments due to different
pathogenesis (2, 3). Studies have shown that children with BA with brosis as the main factor have
poorer prognostic outcomes, later onset, and higher rates of postoperative liver transplantation compared
to inammatory BA (4). Therefore, new protocols for classifying children with BA as well as precise
treatment of different subtypes are needed to improve the overall prognosis of children with BA (5).
Pyroptosis is a form of programmed cell death mediated by two main pathways (6). Canonical
inammatory vesicles are intracellular multiprotein complexes that assemble and activate caspase-1,
which matures IL-1β, IL-18, and IL-37, upon receiving signals to induce the process of pyroptosis. Non-
canonical inammatory vesicles activate caspase-4/5/11 to induce pyroptosis (7). As a brous
Page 4/24
inammatory cholangiopathy, BA has been studied in several forms of cell death, but the process of
pyroptosis remains less studied (8–10).
KPE surgery is the treatment of the rst choice for BA, but the postoperative outcome of patients varies
widely, and this may be due to different pathogenesis (11). Therefore, precision pharmacotherapy may be
a novel strategy for the treatment of BA. Meanwhile, there is an urgent need to develop new markers to
predict the staging because of the diculty to distinguish the specic staging of BA.
At present, with standardised Kasai surgery widely available, there has not been a highly signicant
improvement in jaundice clearance and autologous liver survival after Kasai. Therefore, the differences in
the condition of livers have to be taken into account, leading to different prognostic outcomes. Previous
studies have conrmed that CMV infection, which affects the liver, in turn leads to poor prognostic
outcomes (12). It has also been shown that B-cell clearance can improve the immune status of the liver
and reduce the progression of liver brosis (13).
In this study, we constructed two novel BA subtypes from 17 DEPRGs and explored the characteristics of
each of these two subtypes, as well as the appropriate drug therapy. The pyroptosis score was
constructed and combined with scRNA-seq data sets to reveal that the pyroptosis process may be
associated with four interacting pathways of macrophages and plasma cells. Finally, ve hub genes may
be the most meaningful biomarkers for the occurrence of pyroptosis in BA.
Methods
Data Collection
The gene microarray data (GSE122340(14), GSE46960(15), GSE15235(4), GSE163650(16)) were
downloaded from the Gene Expression Omnibus (GEO) database. The sequencing data of identical twins
(17) was performed by our team, including one BA and one NC identical twin. GSE122340 used as a
training cohort included 171 BA samples and 7 NC samples. Meanwhile, GSE46960 consisting of 64 BA
samples and 7 NC samples, and GSE15235 consisting of 47 BA samples, were used for validation. In
scRNA-seq GSE163650, we included the sequencing data of CD45 + immune cells from two BA samples.
We provide information about each sample in the supplementary materials (Table S1).
Screening and Expression of Differential Genes
The "Limma" package (version 3.52.2) (18) was used to screen differential genes (DEGs) with adj.P < 0.05
and llogFCl > 1 as the analysis threshold, and the "ggplot2" package (version 3.3.6) (19) was used to plot
heatmap and volcano map for visualization. A total of 161 pyroptosis genes (PRGs) were obtained from
previous studies (20). The DEPRGs were the intersection of 161 PRGs and 1565 DEGs.
Consistency Clustering
Page 5/24
Based on the expression levels of differential pyroptosis genes, unsupervised consistent clustering was
performed in the BA group using the "ConsensusClusterPlus" package (version 1.61.0) (21), and the k
value with higher cluster stability was selected according to the clustering effect. Select clusterAlg as "hc",
distance as "pearson", and seed as 1262118388.71279. LASSO regression analysis was performed using
the "glmnet" package (version 4.1.4) (22) on the genes differing between BA subtypes, and signature
genes were selected by taking 0.01322484 as the optimal λ. The signature genes were put into the
Random Forest for further screening. The 900 "trees", Mean Decrease Accuracy > 10, and Mean Decrease
Gini > 1 were selected as standard. The obtained results were included in the binary logistics regression
analysis model to quantify the BA subtypes identication. Finally, we constructed the genetic model
formula that is available for the identication of BA subtypes. The ROC curves were used to assess their
diagnostic ecacy in the training and validation cohorts.
Function and Pathway Enrichment Analyses
Based on the "cluster Proler" package (version 4.5.2) (23) and "ReactomePA" package (version 1.41.0)
(24), functional enrichment analysis was performed to obtain the enrichment results of GO, KEGG, and
Reactome pathways, which were annotated and visualized. The "GSVA" package (version 1.45.5) (25)
was used to obtain enrichment scores for DEPRGs as pyroptosis scores, which was used to represent the
pyroptosis signature.
Drug sensitivity analysis
The top 48 highly expressed genes of the C1 subtype and the top 150 highly expressed genes of the C2
subtype were included in the CMap database (26) to identify relevant small molecule compounds or
drugs. The results were ranked by "raw_cs" to lter the top six positively correlated small molecule
compounds or drugs of each BA subtype.
Immunocyte Inltration analysis
By CIBERSORT (https://cibersortx.stanford.edu/), the BA subtypes were assessed for immune cell
inltration. The "estimate" package (version 1.0.13) (27) was used to calculate the immune score, stromal
score, and estimate score. The enrichment abundance of the 17 immune pathways of the samples was
assessed according to the GSVA package and the ggplot2 package was used to show the differences
between subtypes. The "ggcor" package (version 0.9.4.3) was used to calculate the correlation between
apoptosis scores and 22 immune cells as well as 17 DEPRGs. The "rstatix" package (version 0.7.1) was
used to explore the correlation between hub pyrolysis genes and 22 immune cells as well as six common
immune checkpoint (ICP) genes (CD274, CTLA4, HAVCR2, LAG3, PDCD1, and TIGIT). The "ggplot2"
package was used to plot the correlation heatmap and "circlize" package (version 0.4.15) (28) was used
to draw the circosplot.
PPI Network Construction and Hub Gene Analysis
Page 6/24
DEPRGs were imported into the STRING website (https://string-db.org/) to construct PPI networks, and
the results were imported into Cytoscape software for visualization. CytoHubba plug-in was used to
select the top 5 hub genes based on the "Degree" algorithm to generate hub gene networks. Differential
expression of hub pyroptosis genes between the BA and NC groups was assessed by rank sum tests in
both the GSE122340 and GSE46960 cohorts, and boxplots were plotted using the "ggplot2" package.
Identical twin data were also validated for hub pyroptosis genes and the heatmap was created. Based on
the "pROC" package (version 1.18.0) (29), the diagnostic ecacy of hub pyroptosis genes for BA was
assessed and visualized by "ggplot2" package.
The scRNA-seq analysis
The scRNA-seq data from BASM and iBA of GSE163650 were integrated by removing batch effects via
the "tidyverse" package (version 1.3.2) (30). Afterward, the pyroptosis-score will be calculated for each
cell using 17 DEPRGs via the "GSVA" package (version 1.45.5). Cells will be divided into high and low
pyroptosis-score according to the median of the pyroptosis-score. Two groups of cells were normalized
by the "Seurat" package (version 4.1.1) with the parameters nFeature_RNA > 200, nFeature_RNA < 2500,
and per cent.mt < 5 (31). Ultimately, 4189 cells were included in the low pyroptosis-score and 4280 cells
were included in the high pyroptosis-score. Cells were re-annotated and cell proportions were calculated
via the "dplyr" package (version 1.0.10), the CellMarker website (http://xteam.xbio.top/CellMarker/), and
previous studies (16). The "CellChat" package (version 1.5.0) (32) is used to calculate intercellular
communication relationships and the visualization is performed with the "patchwork" package (version
1.1.2). Finally, we observed the expression of ve hub genes in scRNA-seq through the "Seurat" package
(version 4.1.1).
Statistical Analysis
The data were analyzed through the use of R software (version 4.1.3). Data conforming to the normal
distribution were evaluated using the unpaired Student’s t-test and data conforming to the non-normal
distribution were evaluated using the Wilcoxon test, and the statistical signicance threshold was set at p
< 0.05. Pearson correlation analysis for two sets of normal quantitative data. Spearman correlation
analysis for two sets of skewed quantitative data. The statistical signicance threshold was set at p <
0.05.
Results
Differential Gene Analysis
The GSE122340 dataset was analyzed by "Limma" package, and 1565 DEGs (1253 high and 312 low)
were identied by using adj. p < 0.05 and |logFC| > 1 as thresholds (Table S2). The top 20 highly and top
20 lowly expressed genes were selected to demonstrate the expression differences between BA and NC
groups (Fig.1A). The volcano map was performed to assess the expressed changes of DEGs between BA
and NC groups (Fig.1B). 17 DEPRGs were obtained by taking the intersection of DEGs and PRGs (Fig.1C)
Page 7/24
(Table S3). Among the 17 DEPRGs, except STAT3, the expression levels of the other 16 genes were
signicantly up-regulated in BA (Fig.1D). In conclusion, these 17 DEPRGs may be involved in the
progression of BA and promote the occurrence of the pyroptosis process.
Function and Pathway Enrichment Analyses
To explore which functions are associated with the 17 DEPRGs, we performed functional enrichment
(Table S4). KEGG analysis revealed the signicant enrichment of DEPRGs in infection and inammatory
processes, including salmonella infection and the NOD-like receptor signaling pathway (Fig.2A). The
NOD-like signaling pathway was found to activate the NF-κB signaling pathway, which in turn activates
the classical pyroptosis process (33). GO enrichment analysis revealed that DEPRGs were associated
with bacterial infection and immune cell responses, including response to molecule of bacterial origin,
leukocyte proliferation, and azurophil granule lumen. Further, Reactome analysis conrmed the role of
DEPRGs in immune activation, and pyroptosis enrichment (Fig.2B). IL10 regulates immune processes in
both directions and can be produced by macrophages in response to inammatory factors (34). IL4 and
IL13 play a pro-inammatory role in allergic diseases (35). IL1 is mainly produced by activated
mononuclear macrophages and its precursor IL1β requires caspase-1 cleavage to induce inammation
(36). Therefore, it is reasonable to assume that 17 DEPRGs may mediate inammatory and pyroptosis
processes via monocyte macrophages and leukocytes.
Consensus Clustering
To further investigate the characteristics of DEPRGs, we performed the consistent cluster analysis of 171
BA patients from GSE122340 (Fig.3A-B). Two pyroptosis subtypes of BA were redened, with C1
containing 100 BA samples and C2 containing 71 BA samples (Table S5). A total of 525 DEGs were
screened between subtype C1 and subtype C2 (Fig.3C) (Table S5). The 48 genes highly expressed in C1
were considered to be its signature genes and similarly, the 477 genes highly expressed in C2 were
identied as its signature genes. To explore the differences in signaling pathways between the two
subtypes, functional enrichment analysis was performed (Table S5). The enrichment results showed that
subtype C1 was mainly enriched in immune and inammation-related aspects such as defense response
to fungus, antimicrobial humoral response, RAGE receptor binding, and Staphylococcus aureus infection
(Fig.3D). However, subtype C2 was mainly enriched in transcriptional translation and oxidative
metabolism-related aspects such as RNA splicing, ribonucleoprotein complex assembly, antioxidant
activity, and RNA transport (Fig.3E). GSEA enrichment analysis also showed that C1 subtype was
signicantly higher than C2 subtype in angiogenic, metabolic and inammatory pathways (Fig.3F).
Based on the same clustering and enrichment methods, we obtained the same results in the independent
external validation cohort GSE46960 (Figure S1A). Two pyroptosis subtypes of BA were dened, with C1
containing 29 BA samples and C2 containing 35 BA samples (Table S6). There are 43 DEGs in the
validation set, including 12 genes with high expression of the C1 subtype and 31 genes with high
expression of the C2 subtype (Table S6). The C1 subtype was signicantly associated with viral infection,
whereas the C2 subtype was signicantly enriched in the cell cycle (Figure S1D-E). GSEA enrichment
Page 8/24
analysis also showed that C1 subtype was signicantly higher than C2 subtype in angiogenic, metabolic
and inammatory pathways (Figure S1F). These two pyroptosis subtypes have different regulatory
patterns, which may help to understand the heterogeneity of BA mechanisms. The C1 exhibits an
immune-related subtype, whereas the C2 exhibits the cell cycle and immunosuppression subtype.
Pyroptosis Typing Immune Microenvironment
Afterward, we analyzed the immune microenvironment of two pyroptosis subtypes. it was found that
dendritic cells resting (P < 0.01), eosinophils (P < 0.05), monocytes (P < 0.001), neutrophils (P < 0.001), and
T cells CD4 + memory resting (P < 0.001) had higher inltration levels in subtype C1 (Fig.4A) (Table S7).
While, the mast cells resting (P < 0.001), NK cells activated (P < 0.001), T cells CD4 naïve (P < 0.01), T cells
follicular helper (P < 0.001), and T cells regulatory (P < 0.001) had higher inltration levels in subtype C2
(Table S7). The external validation cohort also veried that the two pyroptosis subtypes had similar
immune microenvironments (Figure S1B) (Table S8). Immune scores (P < 0.001), stromal scores (P <
0.001), and estimate scores (P < 0.001) were higher for subtype C1 than for subtype C2 (Fig.4B-D) (Table
S7). The validation cohort demonstrates higher immune score for the C1 subtype (P < 0.05) (Figure S1C)
(Table S8). 17 immune-related pathways scored higher in the C1 subtype than in the C2 subtype (P <
0.01) (Fig.4E) (Table S7). In conclusion, subtype C1 exhibits the immune activation state with higher
monocyte and neutrophil abundance. Subtype C2 shows an immunosuppressed state with increased T
cells regulatory inltration.
Machine Learning
Since it is currently dicult to identify the subtypes, we have developed a completely novel formula to
solve this problem through machine learning. First, the LASSO regression analysis was used to screen for
genes associated with pyroptosis subtyping from the 525 differential genes in the C1 and C2 subtypes in
the GSE123340 cohort, and 35 genes were retained in the model (Fig.5A-B). These genes were screened
by constructing the Random Forest and 5 genes were included as candidates (Fig.5C-D). The candidate
genes were included in the Binary Logistics Regression Analysis, and the output of the pyroptosis
formula: -4.4783*GSDMB + 3.5155*CTSG-6.3545*HEXIM2. The ROC curves were used to analyze the
diagnostic ecacy of the three important genes and the pyroptosis formula. The AUCs were 0.849, 0.93,
0.929, and 0.995, respectively (Fig.5E). This formula was used to validate the pyroptosis subtyping of
GSE46960 and the AUC was 0.623 (Fig.5F). The 3-gene pyroptosis formula may help in the identication
of typing and is useful for clinical studies.
Drug Sensitivity Analysis
By importing the signature genes of the pyroptosis subtypes into the CMap database, the top six
positively correlated small molecule compounds or drugs of each subtype were selected. The candidate
drugs of subtype C1 were BRD-A63236097, SNX-2112, tanespimycin, PD-168393, thioridazine, and
triciribine. While, the candidate drugs of subtype C2 were VU-0418947-2, carbamazepine, myricitrin,
epothilone-b, BRD-K78844995, and BRD-K76274772 (Fig.5G). The analysis of the applicable drugs
Page 9/24
revealed that the C1 subtype was signicantly associated with the activation of the TNF pathway, HSP
pathway, and EGFR pathway, while the C2 subtype was associated with HIF function, protein kinase
activation, and microtubulin. In addition to this, the genetic targets of drug action may become the basis
for future drug development.
Construction of apoptosis scores
To further investigate the role played by the pyroptosis process in BA, we constructed pyroptosis scores
for subsequent studies. The C1 subtype had a higher pyroptosis score than the C2 subtype in both the
training cohort and the validation cohort (P < 0.05) (Fig.6A-B). In another independent cohort, GSE15235,
inammatory BA had a higher apoptosis score than brotic BA (P < 0.01) (Fig.6C). Pyroptosis-score was
strongly correlated with the 17 DEPRGs, indicating that it was able to represent the characteristics of
these 17 DEPRGs (Table S9). Further, the pyroptosis-score showed a signicant positive correlation with
neutrophils (r = 0.42, P < 0.001), eosinophils (r = 0.24, P < 0.01), monocytes (r = 0.41, P < 0.001) and M0
macrophages (r = 0.22, P < 0.01), while it showed a negative correlation with M2 macrophages (r=-0.24, P
< 0.01) (Fig.6D) (Table S9). These results suggest that the apoptotic process may be accompanied by an
imbalance between pro- and anti-inammatory processes. This is evidenced by an increased M0
inltration and a decreased potential for differentiation to M2, which fails to eliminate the inammatory
process, while neutrophils and eosinophils exacerbate the inammatory process.
The scRNA-seq analysis
A total of six cell types were annotated, including Monocyte Macrophage, T cell and NKT cell, B cell,
Diving, Plasma cell, and other myebid, similar to previous studies (Fig.7A-B). Cells with the high
pyroptosis-score had more mononuclear macrophage inltration, but less T-cell and NKT cell inltration
(Fig.7C). The results of intercellular communication showed that the number and strength of the effects
of intercellular communication were higher in the high pyroptosis-score (Fig.7D). Among these, the role
of macrophages and plasma cells is particularly signicant, probably as a result of macrophage
pyroptosis developing. In contrast, plasma cells show an increase in intensity without a signicant
change in number, which may be due to the simultaneous presence of inammatory processes caused by
autoimmunity (Fig.7E-F). Afterward, we showed that four pathways, including the MIF signaling network,
TGFB signaling network, SPP1 signaling network, and GALECTIN signaling network, were highly
expressed in the high pyroptosis-score, which could be novel targets for blocking pyroptosis in BA
(Fig.7G-J). In summary, the process of pyroptosis is accompanied by inammation, macrophage, and
plasma cell activation. However, the activation of pyroptosis was accompanied by the absence of NK
cells, which is consistent with the results of immune inltration. The four highly expressed pathways
could be novel targets for BA research and drug development.
Identication of hub genes
The PPI network was constructed by 17 DEPRGs using Cytoscape, and hub genes were extracted through
the plugin CytoHubba with the DEGREE scored system (Figure S2A). Five genes were selected as hub
Page 10/24
genes, including IL18, IL1B, STAT3, PYCARD, and IFI16 (Figure S2B). ROC curves were plotted to assess
the ecacy of these ve hub pyroptosis genes for diagnosing BA, and the AUC values of hub genes were
0.886, 0.820, 0.878, 0. 996, and 0.932, respectively (Figure S2C). In the external validation cohort
GSE46960, the AUC values of the ve hub genes were 0.903, 0.829, 0.758, 0.781, and 0.891, respectively
(Figure S2D). In addition to this, ve hub genes also showed signicant differences in expression
between the BA and NC in the validation cohort (Figure S2E). High expression of IL1B, IL18 and IFI16 was
also observed in the identical twin cohort. However, STAT3 and PYCARD expression were not consistent
with our ndings, which need to be further explored (Figure S2F). In addition to this, we show the
chromosomal locations where key genes are located (Figure S2G). IFI16 and PYCARD are involved in the
formation of pyroptosis inammasome, while STAT3 is an important regulatory gene. IL1B and IL18 are
important end-products of pyroptosis and promote inammation upon activation.
Relationship of hub genes with Immune Microenvironment
We further analyzed the interaction between ve hub genes and immune cells by integrating gene
microarray data and scRNA-seq data set. First, we analyzed the correlation between ve hub genes and
22 immune cell types (Figure S3A). M0 macrophages showed a positive correlation with STAT3 (r = 0.301,
p < 0.001), IL1B (r = 0.279, p < 0.001), IL18 (r = 0.313, p < 0.001), and IFI16 (r = 0.270, p < 0.001). M1
macrophages showed a positive correlation with PYCARD (r = 0.287, p < 0.001), IL18 (r = 0.244, p < 0.01),
and IFI16 (r = 0.189, p < 0.05). Neutrophils showed a positive correlation with STAT3 (r = 0.556, p < 0.001),
IL1B (r = 0.578, p < 0.001), and IFI16 (r = 0.264, p < 0.001). However, M2 macrophages showed a negative
correlation with STAT3 (r=-0.174, p < 0.05) and IL1B (r=-0.245, p < 0.01). In addition, these ve hub genes
are signicantly associated with six common ICP genes (Figure S3B). Finally, we validated the expression
of ve hub genes in scRNA-seq data set from the high apoptosis-score group (Figure S4A). IL1B, IL18,
and PYCARD are mainly expressed on monocyte macrophages (Figure S4B-C, F). IFI16 and STAT3 are
expressed in almost all immune cells (Figure S4D-E). Next, the relationship between the ve hub genes
and the immune cells is further described. IL1B, IL18, and PYCARD expression was elevated mainly in
monocyte macrophages, whereas IFI16 and STAT3 were mainly expressed in B-cells and T-cell NKT cells.
In conclusion, we conrm that macrophages may play an important role in BA pyroptosis and may further
contribute to the amplication of the inammatory response.
Discussion
Biliary atresia has multiple potential causative factors, but the main causative factors are different in
each child (37). Clinical practice has found that differences in the main causative factors may lead to
different clinical outcomes in different children (38). Currently, the effectiveness of drug treatment
options for BA is limited, and alterations in the immune microenvironment and genes may become a
novel basis for drug treatment of BA. Therefore, precise treatment by analyzing the characteristics of
different subtypes of BA is an urgent problem (39).
Page 11/24
In our study, we constructed two BA subtypes. The C1 subtype exhibited a high cell pyroptosis score and
high immune inltration. By analysis of scRNA-seq and GSE15235 data, C1 was hypothesized to have a
high-intensity immune cell interaction as well as an inammatory phenotype. However, C2 exhibits cell
cycle and low immune inltration. Similarly, it tended to have a brotic phenotype and lower cellular
interactions. In addition, to enable the identication of typing in the clinic by sequencing technology, we
also constructed the pyroptosis typing formula by machine learning. Its excellent results were obtained in
both the training set and the external validation set. CMap database analysis showed that the C1
phenotype can be treated with drugs such as BRD-A63236097 (TNF inhibitor), SNX-2112 (HSP inhibitor),
and PD-168392 (EGFR inhibitor). It was shown that in RRV-induced BA mice, hepatic macrophages
expressing TNF-α were signicantly increased and exhibited signicant inammation (40). Several
studies have shown that several genes in the HSP family can promote the expression of inammatory
mediators such as NFKB, TNF-α, IL-6, and IL-8 (41–43). In experimental mouse models, the EGFR
pathway has been shown to promote liver brosis and inammatory progression (44). C2 is suitable for
treatment with VU-0418947-2 (HIF modulator), carbamazepine (Carboxamide antiepileptic), myricitrin
(protein kinase inhibitor), and epothilone-b (tubulin inhibitor). It has been suggested that HIF upregulation,
which inhibits iron death, further promotes liver brosis progression (45). Carbamazepine promotes liver
repair and regeneration after acute liver injury, which may be benecial in prolonging the survival time of
the patient's autologous liver (46). However, the liver damage it causes cannot be ignored. Myricitrin has
antioxidant, anti-inammatory and anti-brotic effects in a mouse model of liver injury (47, 48).
Epothilone-b promotes microtubule stabilization and inhibits brosis in spinal cord injury (49). In
conclusion, we believe this may provide support for the precision treatment of BA.
By combining transcriptomic and scRNA-seq data, the process of pyroptosis was found to be closely
associated with macrophages, especially M1 pro-inammatory macrophages. Bo Shu et al. found that
pyroptosis and inammation of macrophages occurred, which could further aggravate liver brosis (50).
Compared to macrophages, activation of pyroptosis may be associated only with enhanced interaction of
plasma cells without increasing their number of inltrations. Further, we found that Macrophage
interactions were signicantly activated on these four pathways, including the MIF signaling network,
TGFβ signaling network, SPP1 signaling network, and GALECTIN signaling network. The role of the TGFβ
signaling pathway in BA has been revealed, not only in the development of bile ducts but also in inducing
apoptosis of bile ducts and liver brosis progression (51, 52). MIF, SPP1, and GALECTIN are closely
associated with inammatory and immune responses in various diseases, but studies in BA are still
scarce (53–55). Finally, ve hub genes for pyroptosis were identied as novel biomarkers of biliary
atresia. IL18 is a pro-inammatory cytokine in the IL-1 family that is expressed by macrophages (56).
Studies have shown that IL18 promotes M1 macrophage polarization, which in turn promotes
inammatory responses (57). In addition to this, IL18 has been reported to be a susceptibility gene for BA
in children in southern China (58). STAT3 is a key regulator of the immune system and variations in its
polymorphisms are associated with a variety of autoimmune diseases (59). Ming Fu et al. showed that
inhibition of STAT3 expression induced apoptosis in BEC cells and enhanced IL-8 expression in the BA
mouse model (60). PYCARD exerts its apoptotic effects by encoding the cysteinase recruitment structural
Page 12/24
domain (ASC) (61). The inammatory vesicle component PYCARD was found to be essential for the
activation of cystein-1 and cystein-5, but remains poorly studied in BA (62). Our study identied ve hub
genes that were signicantly associated with macrophage M1 and neutrophils, which may be closely
related to the pathogenic mechanism of the C1 subtype.
In conclusion, we constructed two novel subtypes of BA and provided usable scoring formulas for typing
identication. Meanwhile, drug selection for high and low immunophenotypes may be the focus of future
clinical attention. Macrophages and neutrophils may be closely associated with the process of pyroptosis
that occurs in BA, which in turn leads to the progression of the inammatory phenotype. Finally, we
identied ve hub genes as possible marker genes for BA, who may play a key role in altering the immune
microenvironment. However, there are still many problems with our study. Public data was used for
analysis without experimental and clinical sample validation. Sequencing is costly and time-consuming
and still requires technological advances to achieve clinical application.
List Of Abbreviations
BA Biliary atresia
GEO Gene Expression Omnibus
DEPRGs Differential pyroptosis genes
KPE Kasai portoenterostomy
DEGs Differentially expressed genes
PRGs Pyroptosis genes
ROC Receiver operating characteristic curve
GO Gene Ontology
KEGG Kyoto Encyclopedia of Genes and Genomes
GSVA Gene set variation analysis
ICP Immune checkpoint
PPI Protein-protein interactions
AUC Area under the cure
Declarations
Data Availability Statement
The data used and analyzed during the current study are available from Gene Expression Omnibus (GEO)
(https://www.ncbi.nlm.nih.gov/geo/).
Page 13/24
Author Contributions
TFL, QPZ and XTW designed the study and analyzed the data. MDL, XDX, SWL and RFZ collected the
data. TFL, YLZ, FYZ and ZRW drafted and revised the manuscript. TFL, RJS, JYNX, YRZ and QHY revised
the images. TFL, QPZ and JHZ revised the manuscript. All authors contributed to the article and approved
the submitted version.
Funding details
This study was supported by grants from Tianjin Municipal Science and Technology Bureau Major
Projects (Grant No.21ZXGWSY00070) and Xinjiang Uygur Autonomous Region Science Foundation
Projects (Grant No.2022D01A27 and 2021D01A38)
Ethics approval and consent to participate
Not applicable.
Conicts of Interest
No author has a conict of interest or nancial ties to disclose.
References
1. Harpavat S, Garcia-Prats JA, Anaya C, Brandt ML, Lupo PJ, Finegold MJ, et al. Diagnostic Yield of
Newborn Screening for Biliary Atresia Using Direct or Conjugated Bilirubin Measurements. Jama.
2020;323(12):1141–50.
2. Muise AM, Turner D, Wine E, Kim P, Marcon M, Ling SC. Biliary atresia with choledochal cyst:
implications for classication. Clin Gastroenterol Hepatol. 2006;4(11):1411–4.
3. Bates MD, Bucuvalas JC, Alonso MH, Ryckman FC. Biliary atresia: pathogenesis and treatment.
Semin Liver Dis. 1998;18(3):281–93.
4. Moyer K, Kaimal V, Pacheco C, Mourya R, Xu H, Shivakumar P, et al. Staging of biliary atresia at
diagnosis by molecular proling of the liver. Genome Med. 2010;2(5):33.
5. Lendahl U, Lui VCH, Chung PHY, Tam PKH. Biliary Atresia - emerging diagnostic and therapy
opportunities. EBioMedicine. 2021;74:103689.
. Wei X, Xie F, Zhou X, Wu Y, Yan H, Liu T, et al. Role of pyroptosis in inammation and cancer. Cell Mol
Immunol. 2022;19(9):971–92.
7. de Carvalho Ribeiro M, Szabo G. Role of the Inammasome in Liver Disease. Annu Rev Pathol.
2022;17:345–65.
. Erickson N, Mohanty SK, Shivakumar P, Sabla G, Chakraborty R, Bezerra JA. Temporal-spatial
activation of apoptosis and epithelial injury in murine experimental biliary atresia. Hepatology.
2008;47(5):1567–77.
Page 14/24
9. Funaki N, Sasano H, Shizawa S, Nio M, Iwami D, Ohi R, et al. Apoptosis and cell proliferation in biliary
atresia. J Pathol. 1998;186(4):429–33.
10. Sasaki M, Nakanuma Y. Stress-induced cellular responses and cell death mechanisms during
inammatory cholangiopathies. Clin Res Hepatol Gastroenterol. 2017;41(2):129–38.
11. Hartley JL, Davenport M, Kelly DA. Biliary atresia. Lancet. 2009;374(9702):1704–13.
12. Fischler B, Czubkowski P, Dezso A, Liliemark U, Socha P, Sokol RJ, et al. Incidence, Impact and
Treatment of Ongoing CMV Infection in Patients with Biliary Atresia in Four European Centres. J Clin
Med. 2022;11(4).
13. Wang J, Xu Y, Chen Z, Liang J, Lin Z, Liang H, et al. Liver Immune Proling Reveals Pathogenesis and
Therapeutics for Biliary Atresia. Cell. 2020;183(7):1867-83.e26.
14. Luo Z, Shivakumar P, Mourya R, Gutta S, Bezerra JA. Gene Expression Signatures Associated With
Survival Times of Pediatric Patients With Biliary Atresia Identify Potential Therapeutic Agents.
Gastroenterology. 2019;157(4):1138-52.e14.
15. Bessho K, Mourya R, Shivakumar P, Walters S, Magee JC, Rao M, et al. Gene expression signature for
biliary atresia and a role for interleukin-8 in pathogenesis of experimental disease. Hepatology.
2014;60(1):211–23.
1. Taylor SA, Chen SY, Gadhvi G, Feng L, Gromer KD, Abdala-Valencia H, et al. Transcriptional proling
of pediatric cholestatic livers identies three distinct macrophage populations. PLoS One.
2021;16(1):e0244743.
17. Chen Y, Zhan J. Physical development and cognitive performance in a monozygotic twins for biliary
atresia: Report of a case and literature reviewing. Journal of Pediatric Surgery Case Reports.
2016;11:9–13.
1. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression
analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47.
19. Ito K, Murphy D. Application of ggplot2 to Pharmacometric Graphics. CPT Pharmacometrics Syst
Pharmacol. 2013;2(10):e79.
20. Peng S, Zhang T, Zhang S, Tang Q, Yan Y, Feng H. Integrated Bioinformatics and Validation Reveal
IL1B and Its Related Molecules as Potential Biomarkers in Chronic Spontaneous Urticaria. Front
Immunol. 2022;13:850993.
21. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with condence
assessments and item tracking. Bioinformatics. 2010;26(12):1572–3.
22. Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via
Coordinate Descent. J Stat Softw. 2010;33(1):1–22.
23. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, et al. clusterProler 4.0: A universal enrichment tool for
interpreting omics data. Innovation (Camb). 2021;2(3):100141.
24. Yu G, He QY. ReactomePA: an R/Bioconductor package for reactome pathway analysis and
visualization. Mol Biosyst. 2016;12(2):477–9.
Page 15/24
25. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq
data. BMC Bioinformatics. 2013;14:7.
2. Lamb J, Crawford ED, Peck D, Modell JW, Blat IC, Wrobel MJ, et al. The Connectivity Map: using gene-
expression signatures to connect small molecules, genes, and disease. Science.
2006;313(5795):1929–35.
27. Becht E, Giraldo NA, Lacroix L, Buttard B, Elarouci N, Petitprez F, et al. Estimating the population
abundance of tissue-inltrating immune and stromal cell populations using gene expression.
Genome Biol. 2016;17(1):218.
2. Gu Z, Gu L, Eils R, Schlesner M, Brors B. circlize Implements and enhances circular visualization in R.
Bioinformatics. 2014;30(19):2811–2.
29. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package
for R and S + to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77.
30. Carpenter CM, Frank DN, Williamson K, Arbet J, Wagner BD, Kechris K, et al. tidyMicro: a pipeline for
microbiome data analysis and visualization using the tidyverse in R. BMC Bioinformatics.
2021;22(1):41.
31. Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, 3rd, et al. Comprehensive
Integration of Single-Cell Data. Cell. 2019;177(7):1888 – 902.e21.
32. Jin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, et al. Inference and analysis of cell-
cell communication using CellChat. Nat Commun. 2021;12(1):1088.
33. Li S, Sun Y, Song M, Song Y, Fang Y, Zhang Q, et al. NLRP3/caspase-1/GSDMD-mediated pyroptosis
exerts a crucial role in astrocyte pathological injury in mouse model of depression. JCI Insight.
2021;6(23).
34. Saraiva M, O'Garra A. The regulation of IL-10 production by immune cells. Nat Rev Immunol.
2010;10(3):170–81.
35. Bieber T. Interleukin-13: Targeting an underestimated cytokine in atopic dermatitis. Allergy.
2020;75(1):54–62.
3. Dinarello CA. Overview of the interleukin-1 family of ligands and receptors. Semin Immunol.
2013;25(6):389–93.
37. Bezerra JA, Wells RG, Mack CL, Karpen SJ, Hoofnagle JH, Doo E, et al. Biliary Atresia: Clinical and
Research Challenges for the Twenty-First Century. Hepatology. 2018;68(3):1163–73.
3. Asai A, Miethke A, Bezerra JA. Pathogenesis of biliary atresia: dening biology to understand clinical
phenotypes. Nat Rev Gastroenterol Hepatol. 2015;12(6):342–52.
39. Lakshminarayanan B, Davenport M. Biliary atresia: A comprehensive review. J Autoimmun.
2016;73:1–9.
40. Mack CL, Tucker RM, Sokol RJ, Kotzin BL. Armed CD4 + Th1 effector cells and activated
macrophages participate in bile duct injury in murine biliary atresia. Clin Immunol. 2005;115(2):200–
9.
Page 16/24
41. Qi J, Liu Y, Yang P, Chen T, Liu XZ, Yin Y, et al. Heat shock protein 90 inhibition by 17-
Dimethylaminoethylamino-17-demethoxygeldanamycin protects blood-brain barrier integrity in
cerebral ischemic stroke. Am J Transl Res. 2015;7(10):1826–37.
42. Märker T, Sell H, Zillessen P, Glöde A, Kriebel J, Ouwens DM, et al. Heat shock protein 60 as a
mediator of adipose tissue inammation and insulin resistance. Diabetes. 2012;61(3):615–25.
43. Qu J, Wang W, Zhang Q, Li S. Inhibition of Lipopolysaccharide-Induced Inammation of Chicken
Liver Tissue by Selenomethionine via TLR4-NF-κB-NLRP3 Signaling Pathway. Biol Trace Elem Res.
2020;195(1):205–14.
44. Perugorria MJ, Latasa MU, Nicou A, Cartagena-Lirola H, Castillo J, Goñi S, et al. The epidermal
growth factor receptor ligand amphiregulin participates in the development of mouse liver brosis.
Hepatology. 2008;48(4):1251–61.
45. Yuan S, Wei C, Liu G, Zhang L, Li J, Li L, et al. Sorafenib attenuates liver brosis by triggering hepatic
stellate cell ferroptosis via HIF-1α/SLC7A11 pathway. Cell Prolif. 2022;55(1):e13158.
4. Kawaguchi T, Kodama T, Hikita H, Tanaka S, Shigekawa M, Nawa T, et al. Carbamazepine promotes
liver regeneration and survival in mice. J Hepatol. 2013;59(6):1239–45.
47. Domitrović R, Rashed K, Cvijanović O, Vladimir-Knežević S, Škoda M, Višnić A. Myricitrin exhibits
antioxidant, anti-inammatory and antibrotic activity in carbon tetrachloride-intoxicated mice. Chem
Biol Interact. 2015;230:21–9.
4. Shen Y, Shen X, Cheng Y, Liu Y. Myricitrin pretreatment ameliorates mouse liver ischemia reperfusion
injury. Int Immunopharmacol. 2020;89(Pt A):107005.
49. Ruschel J, Hellal F, Flynn KC, Dupraz S, Elliott DA, Tedeschi A, et al. Axonal regeneration. Systemic
administration of epothilone B promotes axon regeneration after spinal cord injury. Science.
2015;348(6232):347–52.
50. Shu B, Zhou YX, Li H, Zhang RZ, He C, Yang X. The METTL3/MALAT1/PTBP1/USP8/TAK1 axis
promotes pyroptosis and M1 polarization of macrophages and contributes to liver brosis. Cell
Death Discov. 2021;7(1):368.
51. Chung-Davidson YW, Ren J, Yeh CY, Bussy U, Huerta B, Davidson PJ, et al. TGF-β Signaling Plays a
Pivotal Role During Developmental Biliary Atresia in Sea Lamprey (Petromyzon marinus). Hepatol
Commun. 2020;4(2):219–34.
52. Kanzler S, Galle PR. Apoptosis and the liver. Semin Cancer Biol. 2000;10(3):173–84.
53. Gupta A, Sharma D, Gupta H, Singh A, Chowdhury D, Ganju L, et al. Severe heat stress modulated
nuclear factor erythroid 2-related factor 2 and macrophage migration inhibitory factor pathway in rat
liver. J Cell Commun Signal. 2022.
54. Fujisawa Y, Matsuda K, Uehara T. Osteopontin enhances the migration of lung broblasts via
upregulation of interleukin-6 through the extracellular signal-regulated kinase (ERK) pathway. Biol
Chem. 2020;401(9):1071–80.
55. Arsenijevic A, Stojanovic B, Milovanovic J, Arsenijevic D, Arsenijevic N, Milovanovic M. Galectin-3 in
Inammasome Activation and Primary Biliary Cholangitis Development. Int J Mol Sci. 2020;21(14).
Page 17/24
5. Kaplanski G. Interleukin-18: Biological properties and role in disease pathogenesis. Immunol Rev.
2018;281(1):138–53.
57. Zhang L, Li M, Wang Z, Sun P, Wei S, Zhang C, et al. Cardiovascular Risk After SARS-CoV-2 Infection
Is Mediated by IL18/IL18R1/HIF-1 Signaling Pathway Axis. Front Immunol. 2021;12:780804.
5. Liang J, Wen Z, Zhao J, Liang Q, Liu T, Xia H, et al. Association of IL18 genetic polymorphisms with
increased risk of Biliary atresia susceptibility in Southern Chinese children. Gene. 2018;677:228–31.
59. Hillmer EJ, Zhang H, Li HS, Watowich SS. STAT3 signaling in immunity. Cytokine Growth Factor Rev.
2016;31:1–15.
0. Fu M, Tan L, Lin Z, Lui VCH, Tam PKH, Lamb JR, et al. Down-regulation of STAT3 enhanced
chemokine expression and neutrophil recruitment in biliary atresia. Clin Sci (Lond).
2021;135(7):865–84.
1. de Souza JG, Starobinas N, Ibañez OCM. Unknown/enigmatic functions of extracellular ASC.
Immunology. 2021;163(4):377–88.
2. Martinon F, Burns K, Tschopp J. The inammasome: a molecular platform triggering activation of
inammatory caspases and processing of proIL-beta. Mol Cell. 2002;10(2):417–26.
Figures
Page 18/24
Figure 1
Differential Gene Analysis. (A) Heatmap of the expression levels of the top 20 highly and top 20 lowly
expressed DEGs between BA and NC groups. (B) Volcano map of differential genes. (C) The intersection
between DEGs and PRGs. (D) Expression of 17 DEPRGs between BA and NC groups.
Figure 2
Function and Pathway Enrichment Analyses. (A) The bar chart shows the results of GO and KEGG
enrichment analysis for 17 DEPRGs. (B) The circle graph shows the results of the Reactome enrichment
analysis.
Page 19/24
Figure 3
Training Cohort Pyroptosis Subtypes. (A,B) Consensus clustering analysis of 17 DEPRGs. (C) DEGs
volcano maps of two pyroptosis subtypes. (D) KEGG and GO enrichment analysis of highly expressed
genes of the C1 subtype. (E) KEGG and GO enrichment analysis of highly expressed genes of the C2
subtype. (F) Results of GSEA enrichment analysis.
Page 20/24
Figure 4
Pyroptosis Subtypes Immune Microenvironment. (A) Bar chart showing the difference between C1 and C2
immune inltration. (B) Differences in immune, stromal, and estimate scores in C1 and C2. (C)
Differential expression of 17 common immune pathways in C1 and C2.
Page 21/24
Figure 5
Machine Learning and Drug Sensitivity analysis. (A,B) LASSO regression analysis of 525 DEGs. (C,D) The
35 genes for Random Forest analysis. (E) The ROC curves indicate a strong distinguishing effect of the 3
gene formula in the training cohort. (F) The ROC curves indicate a strong distinguishing effect of the 3
gene formula in the external validation cohort. (G) Analysis of drugs for which C1 and C2 are applicable,
and the corresponding targets of action of the drugs.
Page 22/24
Figure 6
Construction of Pyroptosis Scores. (A) Differences in pyroptosis-score between C1 and C2 subtypes in the
training cohort. (B) Differences in pyroptosis-score between C1 and C2 subtypes in the external validation
cohort. (C) Differences in pyroptosis-score between inammatory and brotic BA in the GSE15235. (D)
The correlation between apoptosis-score and 22 immune cells as well as 17 DEPRGs.
Page 23/24
Figure 7
The scRNA-seq analysis. (A,B) High and low pyroptosis-score immune cell landscape. (C) High and low
pyroptosis-score immune cell ratios. (E,F) Differences in the strength and number of immune cell
interactions between high and low pyroptosis-score. (G-J) Four high expression interaction pathways in
pyroptosis-score.
Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.
FigureS1ValidationCohortPyroptosisTyping.tif
Page 24/24
FigureS2IdenticationofHubGenes.tif
FigureS3Immunoinltrationanalysisofhubgenes.tif
FigureS4ExpressionofhubgenesinscRNAseq.tif
TableS1.Samplesinformationaboutthecohortsused.docx
TableS2.DEGsbetweenBAandNCgroups.xlsx
TableS3.PRGsandDEPRGs.xlsx
TableS4.DEPRGsEnrichmentanalysis.xlsx
TableS5.TrainingCohortConsensusClustering.xlsx
TableS6.ValidationCohortConsensusClustering.xlsx
TableS7.TheImmuneMicroenvironmentofTrainingCohort.xlsx
TableS8.TheImmuneMicroenvironmentofValidationCohort.xlsx
TableS9.CorrelationanalysisofpyroptosisscorewithimmunecellsandDEPRGs.xlsx
