Two Novel Pathogenic Variants of TJP2 Gene and the Underlying Molecular Mechanisms in Progressive Familial Intrahepatic Cholestasis Type 4 Patients

Progressive familial intrahepatic cholestasis (PFIC) is an autosomal recessive inherited disease that accounts for 10%–15% childhood cholestasis and could lead to infant disability or death. There are three well-established types of PFIC (1–3), caused by mutations in the ATP8B1, ABCB11, and ABCB4 genes. Biallelic pathogenic variants in the tight junction protein 2 gene (TJP2) were newly reported as a cause for PFIC type 4; however, only a limited number of patients and undisputable variants have been reported for TJP2, and the underlying mechanism for PFIC 4 remains poorly understood. To explore the diagnostic yield of TJP2 analysis in suspected PFIC patients negative for the PFIC1–3 mutation, we designed a multiplex polymerase chain reaction-based next-generation sequencing method to analyze TJP2 gene variants in 267 PFIC patients and identified biallelic rare variants in three patients, including three known pathogenic variants and two novel variants in three patients. By using CRISPR-cas9 technology, we demonstrated that TJP2 c.1202A > G was pathogenic at least partially by increasing the expression and nuclear localization of TJP2 protein. With the minigene assay, we showed that TJP2 c.2668-11A > G was a new pathogenic variant by inducing abnormal splicing of TJP2 gene and translation of prematurely truncated TJP2 protein. Furthermore, knockdown of TJP2 protein by siRNA technology led to inhibition of cell proliferation, induction of apoptosis, dispersed F-actin, and disordered microfilaments in LO2 and HepG2celles. Global gene expression profiling of TJP2 knockdown LO2 cells and HepG2 cells identified the dysregulated genes involved in the regulation of actin cytoskeleton. Microtubule cytoskeleton genes were significantly downregulated in TJP2 knockdown cells. The results of this study demonstrate that TJP2 c.1202A > G and TJP2 c.2668-11A > G are two novel pathogenic variants and the cytoskeleton-related functions and pathways might be potential molecular pathogenesis for PFIC.

Progressive familial intrahepatic cholestasis (PFIC) is an autosomal recessive inherited disease that accounts for 10%-15% childhood cholestasis and could lead to infant disability or death. There are three well-established types of PFIC (1-3), caused by mutations in the ATP8B1, ABCB11, and ABCB4 genes. Biallelic pathogenic variants in the tight junction protein 2 gene (TJP2) were newly reported as a cause for PFIC type 4; however, only a limited number of patients and undisputable variants have been reported for TJP2, and the underlying mechanism for PFIC 4 remains poorly understood. To explore the diagnostic yield of TJP2 analysis in suspected PFIC patients negative for the PFIC1-3 mutation, we designed a multiplex polymerase chain reactionbased next-generation sequencing method to analyze TJP2 gene variants in 267 PFIC patients and identified biallelic rare variants in three patients, including three known pathogenic variants and two novel variants in three patients. By using CRISPR-cas9 technology, we demonstrated that TJP2 c.1202A > G was pathogenic at least partially by increasing the expression and nuclear localization of TJP2 protein. With the minigene assay, we showed that TJP2 c.2668-11A > G was a new pathogenic variant by inducing abnormal splicing of TJP2 gene and translation of prematurely truncated TJP2 protein. Furthermore, knockdown of TJP2 protein by siRNA technology led to inhibition of cell proliferation, induction of apoptosis, dispersed F-actin, and disordered microfilaments in LO2 and HepG2celles. Global gene expression profiling of TJP2 knockdown LO2 cells and HepG2 cells identified the dysregulated genes involved in the regulation of actin cytoskeleton. Microtubule cytoskeleton genes were significantly downregulated in TJP2 knockdown cells. The results of this study demonstrate that TJP2 c.1202A > G and TJP2 c.2668-11A > G are two novel pathogenic variants and the cytoskeleton-related functions and pathways might be potential molecular pathogenesis for PFIC.
The TJP2 gene was first discovered and reported by Duclos et al. (1994). It is located at chromosome 9q21.11 and has a total of 140,901 base pairs comprising 23 exons (NM_004817.3). The product of TJP2 gene is the tight junction protein 2, also called zona occludens 2 (ZO-2), which belongs to the membraneassociated guanylate cyclase family and is involved in the connection between epithelial cells and endothelial cells (Jesaitis and Goodenough, 1994). The TJP2 protein not only binds to the C-terminus of various transmembrane binding proteins but also interacts with multiple nuclear proteins, and participates in the regulation of gene expression and cell proliferation (Gonzalez-Mariscal et al., 2012;Traweger et al., 2013). It is a ubiquitous tight junction scaffold protein present in liver (Jesaitis and Goodenough, 1994). Tight junctions are intercellular barriers that control paracellular solute diffusion across cell layers and separate bile from plasma in liver (McCarthy et al., 2000;Tsukita et al., 2001).
Progressive familial intrahepatic cholestasis type 4 (PFIC4, OMIM #615878) was first described by Sambrotta et al. (2014). The deficiency of TJP2 protein may lead to the disruption of intercellular connections, which facilitates the bile to enter the liver parenchyma through the paracellular space and leads to cholestasis (Sambrotta and Thompson, 2015). Although an increasing number of pathogenic variants have been revealed in the TJP2 gene (Vitale et al., 2018;Wei et al., 2020;Zhang et al., 2020a), the molecular mechanism by which those variants cause PFIC needed to be further characterized (Sambrotta and Thompson, 2015;Amirneni et al., 2020).
In our studies, we designed a multiplex polymerase chain reaction (PCR)-based next generation sequencing method to analyze TJP2 gene variants in 267 PFIC patients negative for PFIC 1-3 mutations and identified three known pathogenic variants and two novel variants. Using CRISPR-cas9 and minigene technologies, we confirmed that the two novel variants are likely pathogenic. To better understand the pathogenicity of PFIC 4, we used small interfering RNA (siRNA) technology to downregulate the expression of TJP2 protein and investigated its impact on the cell proliferation, apoptosis, cytoskeleton, and overall structure in LO2 and HepG2 cell lines. Meanwhile, we performed global gene expression profiling of TJP2 knockdown LO2 cells and control LO2 cells to understand the signaling pathways associated with the pathogenesis of TJP2 deficiency. These results are helpful to understand the pathogenesis and may guide the treatment of PFIC.

Patient Enrollment and Ethical Conduct of Research
Two hundred and sixty-seven PFIC patients with normal γ-glutamyl transpeptidase and free of pathogenic variants in ABCB11, ATP8B1, and ABCB4 were recruited globally by Baylor College of Medicine. We excluded those patients with disease reasonably attributed to drug exposure, autoimmune hepatitis, infection, and biliary atresia. A total of 267 Caucasian patients are sporadic cases and unrelated, including 138 females and 129 males, aged up to 30 years. All patients have signed the written informed consent before participating in this project. This study has been approved by the Institutional Review Board for Human Subject Research at the Jiangmen Maternity and Child Health Care Hospital (Jiangmen, IRB number: 2019053).

Sample Collection and DNA Extraction
Two milliliters of peripheral blood was collected from each PFIC patient. The genomic DNA was extracted from peripheral blood leukocytes with a commercially available DNA isolation kit (Gentra Systems Inc., Minneapolis, MN), according to the manufacturer's instructions.

Multiplex PCR-Based Next Generation Sequencing Analysis of TJP2 Gene
Twenty-three pairs of TJP2 gene primers were designed to amplify TJP2 exonic regions (Supplementary Table 1). Multiplex PCR preamplification of the TJP2 gene was performed using TaqMan R PreAmp Master Mix (Cat no. 4391128, Applied Biosystems) following the manufacturer's instructions. The PCR products were fragmented and indexed separately for the subsequent sequencing, as described previously (Li et al., 2016), Equal molar ratios of 48 indexed samples were pooled and sequenced on a MiSeq using reagent kit V2 (Illumina, San Diego, CA) with 150-cycle single-end reads. The raw data in basic calling files (.bcl format) were converted to qseq files before demultiplexing with CASAVA software version 1.7 (Illumina). Demultiplexed sequence reads were aligned to the TJP2 gene reference sequence NM_004817.3, and variants were detected by the NextGENe software version 2.3 (SoftGenetics, State College, PA). Sanger sequencing was used to validate the sequence variants identified by NGS as previously described (Li et al., 2015). Confirmed variants were annotated through a public population database and literature databases including 1000 Genomes (version phase3), dbSNP (build 148), GnomAD 1 , Clinvar (version 201706), HGMD (version 20164), and OMIM 2 . Common variants (frequency > 1% in database) were discarded. Rare or novel variants were further investigated on its amino acid conservation, protein structure, and function. Variants that change located in conserved, structural, or functional regions of protein were regarded as potential pathogenic variants. Variant interpretation was performed according to the American College of Medical Genetics (ACMG) guidelines (Richards et al., 2015).

Pathogenicity Prediction of Novel Variants and Experimental Validations
For the novel variants, c.1202A > G (p.Glu401Gly) and c.2668-11A > G, their functional impacts were predicted by multiple in silico tools software firstly. Then, for missense variant c.1202A > G (p.Glu401Gly), CRISPR-cas9 technology was utilized to construct mutant cell strain (HepG2) to confirm pathogenicity by reverse transcription polymerase chain reaction (RT-PCR), Western blot analysis, cell counting kit-8 assay analysis, and immunofluorescence protein analysis; Minigene assay was utilized to explore the impact of c.2668-11A > G variants on the splicing pattern of TJP2 gene. All experimental procedures and parameters were provided in the supplementary methodology section (Supplementary Methods).

Functional Validation and Molecular Mechanism Exploration of the Deficiency of TJP2 Protein
Using siRNA technology for downregulated the expression of TJP2 protein in LO2 and HepG2 cell lines separately. For these two TJP2 knockdown cell lines, we analyzed the expression changes of P53 protein and Actin protein by Western blot method, cell proliferation ability by cell counting kit-8 assay, and cell apoptosis rate by flow cytometry. Microtubules and microfilament in LO2 cells were visualized by confocal microscopy using anti-F-actin (1:500; Cytoskeleton, Inc. Cat. # PHDG1-A) and anti-β-tubulin (1:300; Abcam, Cat. #ab195883). Detailed information regarding experimental operations and parameters were provided in the supplementary methodology section (Supplementary Methods).

RNA-seq Library Preparation, Sequencing, and Analysis
RNA was extracted following the Trizol reagent manual. RNA was precipitated by 1:1 isopropanol (v/v) and 1 µl of glycogen at -20 • C overnight. Two parallel mRNA libraries were constructed using VAHTS mRNA-seq V3 Library Prep Kit following the manufacturer's instructions for each cell line separately. Libraries were sequenced on an Illumina NovaSeq 6000 sequencer for 318 cycles with a strategy of paired end 150. Reads that passed the Illumina quality filters were kept for the subsequent analyses. Adapters were trimmed from the reads, and reads shorter than 17 nt were discarded. The reads were mapped to the human mRNA reference database using FANSe3 algorithm on Chi-Cloud NGS Analysis Platform (Chi-Biotech Co. Ltd., Shenzhen, China). Differentially expressed genes (DEGs) were determined by R package edgeR (Version: 3.28.1) at the following cutoffs: p-value < 0.01 and absolute log2 fold change > 1 ( Robinson et al., 2009;Rajkumar et al., 2015). Kyoto Encyclopedia of Genes and Genomes (KEGG) signaling pathway enrichment analyses were performed for DEGs by clusterProfiler (Version: 3.14.3) (Alexa et al., 2006;Yu et al., 2012). Kyoto Encyclopedia of Genes and Genomes pathways with the Benjamini-Hochberg-adjusted p-value < 0.05 were considered statistically significant.

Quantitative Real-Time PCR
Total RNA was isolated with TRIzol reagent according to the manufacturer's instructions. Three parallel RNA samples were reverse transcribed using random hexamer primers in the presence of RNase inhibitor (Takara Bio, Shiga, Japan). qRT-PCR was performed with SYBR Premex EX Taq (Takara Bio) using the 7300 Sequence Detection System (Applied Biosystems, Foster City, CA). A relative quantification analysis was performed using the Ct method (Livak and Schmittgen, 2001), with actin as endogenous references. Relative gene expression is presented as the ratio of the target gene to reference.

Statistical Analysis
Statistical analyses were performed with SPSS v. 19.0. The data were performed in triplicate and analyzed by using mean ± SD, Student's unpaired t-test. The value of n is mentioned in the figure legends and always stands for separate biological replicates. All comparisons between groups were made by unpaired twotailed Student's t-test. Differences with * p < 0.05, * * p < 0.01, and * * * p < 0.001 were considered statistically significant. n.s., not significant.

Variants Detected in Patients by Multiplex PCR-Based Sequencing of TJP2 Gene and Clinical Features
Five TJP2 gene variants were detected in the 267 PFIC patients, including two novel variants and three known pathogenic variants ( Table 1). The two novel variants are c.1202A > G (p.Glu401Gly) and c.2668-11A > G, which were confirmed by Sanger sequencing (Supplementary Figure 1). Two compound heterozygous variants were detected in patient 46384 and patient 54691. The homozygous variants were identified in patient 62966. Table 1 summarizes clinical features and ACMG classification information.
Patient 62,966 at 5 years old has a weight of 17.50 kg, a height of 105.50 cm, a BMI of 15.72, a weight percentile of 15, a height percentile of 3, and a BMI percentile of 63, has cholestatic jaundice (idiopathic) and hepatitis, fat-soluble vitamin deficiency of ADEK, splenomegaly, hypervascular lesions in liver, elevated AFP (most recently normal), and scleral icterus. Seven-year-old patient 62,966 has idiopathic cholestatic neonatal hepatitis with striking hepatomegaly and splenomegaly. She has a prolonged INR, which may be related to vitamin K deficiency or worsening synthetic function, despite vitamin K replacement. Her albumin is also low, raise concern for worsening synthetic function. Her absorption of all fat-soluble vitamins is very poor despite high intake, as seen in disorders where there is poor bile acid delivery to the gut. Her declining platelet count and increasing spleen size indicated hypersplenism due to portal hypertension. CT examination found three masses. In the anterolateral dome of the liver, corresponding to the junction between Couinaud segment 8 and segment 4A, there is a hypervascular liver mass measuring 2.5 × 2.4 × 2.1 cm. During portal venous phase imaging, the mass becomes iso-attenuating to liver parenchyma. During delayed phase imaging, there is subtle central lower attenuation with a rim of continued slightly higher attenuation relative to surrounding liver parenchyma. We also conducted hematoxylineosin staining assays on patient 62966 ( Table 1).
The TJP2 c.1202A > G Is a Novel Pathogenic Variant in PIFC The missense variant TJP2 c.1202A > G (p.Glu401Gly) was predicted as probably damaging (HDIV score: 0.95, HVAR score: 0.714) by PolyPhen-2,Disease_causing (1.000) by MutationTaster, deleterious (score: -4.05) by PROVEAN, and damaging (score: 0.013) by SIFT. Based on the GERP++, PhyloP, and PhastCons methods, this missense variant position probably belongs to a conserved element as the conservation scores are 5.75, 4.794, and 1 separately. We used CRISPR-cas9 technology to construct a mutant HepG2 cell line and performed various experiments to further characterize the functional impact of the variant. The expression levels of TJP2 mRNA and protein were upregulated (Figures 1A,B). Cell proliferation was enhanced in c.1202A > G mutated HepG2 cells as compared to wild-type HepG2 cells at 48 h after transfection (p-value < 0.05, Student's t-test, Figure 1C). The immunofluorescence assay demonstrated that TJP2 protein expressed at both cytoplasm and nucleus in WT HepG2 cells. Almost the TJP2 protein is expressed at nucleus in c.1202A > G mutated HepG2 cells (Figure 1D). These results suggest that TJP2 c.1202A > G is pathogenic at least partially by increasing the expression and nuclear localization of TJP2 protein. According to the ACMG guidelines, our prediction results and functional experimental discoveries above could supply PS3, PM2 (only one heterozygote in gnomAD, 4.0E-06), PP3, and PP4 evidence for the TJP2 c.1202A > G variant. Combining with all these evidence categories, we suggest that TJP2 c.1202A > G is a likely pathogenic variant.
The TJP2 c.2668-11A > G Variant Leads to Translation of Prematurely Truncated TJP2 Protein The TJP2 c.2668-11A > G variant is located in the intron region of TJP2 gene and may have an impact on splicing. We utilized Alamut Visual mutation analysis software (version 2.14, Interactive Biosoftware, Rouen, France), including five different algorithms, SpliceSiteFinder-like, MaxEntScan, NNSPLICE, GeneSplicer, and Human Splicing Finder, to evaluate the splicing effects of the variant. Alamut Visual mutation analysis software predicted that the TJP2 c.2668-11A > G variant created a new acceptor site and a 10-bp intron region was retained in the transcript, causing premature translational termination of TJP2 protein (Supplementary  Figure 2). The conservation scores predicted by the PhyloP and PhastCons methods are 0.669 and 0.903, indicating that the variant position might be a conserved element. Then, we performed a minigene experiment to further validate the prediction. The results of the minigene assay showed the plasmid containing wild-type TJP2 could transcribe exon 19 of the TJP2 gene. In addition to exon 19, the plasmid containing mutant TJP2 expressed an additional 10-bp intronic region (ATTCCTCTAG) (Figure 2, Supplementary Figure 3). These results suggest that the TJP2 c.2668-11A > G variant causes abnormal splicing of TJP2 mRNA; translation of 897 amino acids prematurely truncated TJP2 protein (normal is 1190 amino acids, Supplementary Table 4). According to the ACMG guidelines, our prediction results and functional experimental discoveries above could supply PS3, PM2 (only two heterozygotes in gnomAD, 9.0E-06), PP3, and PP4 evidence for the TJP2 c.2668-11A > G variant. Combining with all these evidence categories, we suggest that TJP2 c.2668-11A > G is a likely pathogenic variant.

Knockdown of TJP2 Protein Inhibits Cell Proliferation and Induces Apoptosis
To investigate the molecular mechanism underlying the pathogenesis of TJP2 deficiency, we performed knockdown   of TJP2 gene expression by two different siRNAs (1573 and 1673). As compared to the negative control, the siRNAs caused marked reduction in TJP2 protein expression in both LO2 and HepG2 cell lines. As a result, the knockdown of TJP2 protein led to increased expression of TP53 protein in the two cell lines (Figures 3A,B). The CCK-8 assay and flow cytometry assay demonstrated that the downregulation of TJP2 protein expression inhibited cell proliferation at 48 h and 72 h after transfection of siRNAs (Figures 3C,D) and promoted cell apoptosis in both LO2 and HepG2 cells (p < 0.05 for all cases, Student's t-test, Figures 3E-H).

Knockdown of TJP2 Causes Cell Structural Disorder
To analyze the impact of deficiency of TJP2 protein on the cytoskeleton and overall structure, the cytoskeleton was stained by immunofluorescence with β-tubulin and F-actin antibody, respectively. The cytoskeletons of LO2 and HepG2 cells were observed by confocal laser microscopy. The results showed that silencing TJP2 protein expression did not significantly change the tubulin of LO2 and HepG2 cells. In contrast, the downregulation of TJP2 protein expression resulted in dispersed F-actin and disordered microfilaments (Figure 4), indicating the role of TJP2 in F-actin regulation.

Functional Characterization of DEGs in TJP2 Knockdown LO2 and HepG2 Cells
To understand the signaling pathways associated with pathogenic variants in PIFC, we next performed global gene expression profiling on TJP2 knockdown LO2 cells and HepG2 cells as compared to control cells. In LO2 cells, a total of 5981 DEGs were detected (p < 0.01), of which 3,169 were downregulated and 2,812 were upregulated in TJP2 knockdown cells ( Figure 5A). In HepG2 cells, a total of 1,013 DEGs were detected (p < 0.01), of which 559 were downregulated and 454 were upregulated in TJP2 knockdown cells ( Figure 5C). We performed KEGG pathway analysis to find out the overlap pathway of these two cell lines. The overlap KEGG pathways that are significantly enriched for DEGs were pathways in cancer, regulation of actin cytoskeleton, and TGFβ signaling pathway. Related to the functional observation of TJP2 above, the regulation of actin cytoskeleton pathway stood out in the analysis (Figures 5B,D). We compared the LO2 transcriptome data with the data of HepG2 cells. The abnormally expressed genes involved in the regulation of the actin cytoskeleton pathway showed a significant overlap ( Figure 5E); these genes are PDGFRB, PIK3CB, ITGA2, TMSB4X, GSN, PIK3R3, and ITGAV. By performing qPCR analysis, we further confirmed the dysregulation of these genes in LO2 and HepG2 cells upon TJP2 knockdown ( Figure 5F).

DISCUSSION
In the present study, we identified three patients among 267 PFIC patients which were previously tested negative for ABCB11, ATP8B1, and ABCB4 gene variants carried homozygous or compound heterozygous disease-causing variant in TJP2 gene. The remaining negative 264 PFIC patients probably have variants outside these areas like the untranslated regions or the other genes (e.g. NR1H4, MYO5B, USP53), especially in patients who experienced cholestatic symptoms within the first year of life (Dröge et al., 2017). Two variants among all the five variants we have identified in TJP2 gene could be pathogenic based on our functional studies. TJP2 c.1202A > G was predicted as a damaging mutation by multiple software and could significantly upregulate the expression of TJP2 protein especially in the nucleus, which might lead to abnormal biological processes including enhancement of cell proliferation. Functional prediction and experimental results both indicated that TJP2 c.1202A > G is most likely a pathogenic variant. Splicing variant TJP2 c.2668-11A > G causes abnormal splicing of TJP2 mRNA, which could produce prematurely truncated TJP2 protein and cause the deficiency of TJP2 protein. Deficiency of TJP2 protein has been reported as the pathogenic mechanism of PFIC type 4 (Zhang and Yu, 2016). Generally, PFIC type 4 predominantly occurs in childhood (Vitale et al., 2018), but patient 46384 who has compound heterozygous pathogenic variants (c.1202A > G and c.817delG, which were inherited from her mother and father, respectively.) was 21 years old at the time of the sample collection. The compound heterozygous variants are a novel missense variant c.1202A > G (p.Glu401Gly) and a pathogenic variant c.817delG (p.Ala273Profs * 38) (Sambrotta et al., 2014;Zhou et al., 2015). TJP2 c.1202A > G could upregulate the expression of TJP2 protein and TJP2 c.817delG caused the deficiency of TJP2 protein. We speculate that the two functionally opposite heterozygous pathogenic variants might lead to a functional balance and partially explain why patient 46384 could survive so long.
Hepatocytes exhibit a unique cellular polarity and the maintenance of cellular polarity is essential for many functions of hepatocytes (Gissen and Arias, 2015). In a recent study, Itoh et al. inactivated the TJP2/ZO-2 gene in alive mouse livers and observed that the cellular polarity was compromised (Itoh et al., 2021). The specific transporters localized at basolateral and apical membrane domains are responsible for the uptake and secretion processes (Stapelbroek et al., 2010). A study from Sambrotta et al. shows that deficiency of TJP2 protein caused a failure of localization of claudin-1 at the canalicular membrane, with the disruption of the tight-junction structure leading to a leakage of the biliary components through the paracellular space into the liver parenchyma, causing damage to the surrounding hepatocytes and cholangiocytes (Sambrotta and Thompson, 2015). However, the effect of TJP2 protein deficiency on the cytoskeleton and overall cell structure has not been studied. We have used siRNA to study and found that the deficiency of the TJP2 protein could cause filament arrangement disorder and cell fusion in both HepG2 and LO2 cell lines. The mRNA-seq analysis also showed that the dysregulation genes are related to cytoskeleton organization or cytoskeletal binding function.
TJP2 protein often interacts with multiple nuclear proteins to regulate gene expression and cell proliferation (Gonzalez-Mariscal et al., 2012). According to Western blot analysis, we found that the deficiency of TJP2 protein could increase the expression of P53 protein. However, the increased expression of TP53 has turned out not to be statistically significant in the DEG analysis. Knockdown of TJP2 inhibits cell proliferation, induces apoptosis, and causes cell structural disorder. TJP2 is a multidomain molecule that binds to a variety of cell signaling proteins, to the actin cytoskeleton, and to gap, tight, and adherens junction proteins, and inhibits the Wnt signaling pathway, reduces cell proliferation, and promotes apoptosis (Gonzalez-Mariscal et al., 2012). Actin sedimentation studies showed TJP2 protein to interact directly with F-actin (Wittchen et al., 1999). We report that the downregulation of TJP2 protein expression resulted in dispersed F-actin and disordered microfilaments. Based on the results of mRNA-seq analysis in LO2 cells, we found that the deficiency of TJP2 protein could influence the expression of genes involved in intracellular membrane-related organelles and cytoskeleton, thereby affecting the binding function, metabolic processes, and cytoskeleton organization. These biological processes are closely related to the physiological processes related to cholestasis (Javitt, 1994;Chiang, 2009;Boyer, 2013). The KEGG pathway enrichment analysis showed that many DEGs were involved in the apoptosis pathway, which, to a large extent, explains why TJP2 deficiency increases cellular apoptosis.
Reported patients with TJP2 deficiency display severe progressive cholestatic liver disease in early childhood, which increases the risk of developing hepatocellular carcinoma (Zhou et al., 2015). Many studies have reported that infant liver cancer patients have TJP2 gene variants (Zhou et al., 2015;Parsons et al., 2016;Zhang and Yu, 2016). In our study, the signal pathways enriched for DEGs are mostly related to the small-molecule metabolic pathways and cancer-related signal (F) Quantitative PCR confirmation of dysregulated genes. The relative mRNA expression data are presented as mean ± standard deviation (n = 3, *p < 0.05, **p < 0.01, ***p < 0.001). siTJP2, siTJP2 (1673). pathways such as the mTOR signaling pathway, hepatocellular carcinoma, TNF signaling pathway, and mTOR signaling pathway. The dysregulated cancer-associated pathways caused by the deficiency of TJP2 protein might be attributable to the development of hepatocellular carcinoma. Therefore, the influence on cytoskeleton organization pathway and the filament arrangement disorder might provide new insight into the pathogenic mechanism of variants in TJP2 gene in PFIC type 4.

CONCLUSION
This study showed the diagnostic yield of TJP2 analysis in suspected PFIC parents negative for identifiable PFIC1-3 mutations (up to 1.12%). We reported two novel pathogenic variants, TJP2 c.1202A > G and TJP2 c.2668-11A > G, and expanded the clinical and molecular spectrum of PFIC4. Our study enhanced understanding of the consequences of TJP2 deficiency in vitro and revealed that the cytoskeletonrelated functions and pathways might be potential molecular pathogenesis for PFIC.

DATA AVAILABILITY STATEMENT
The sequencing data has been deposited in Sequence Read Archive (SRA) with BioProject ID PRJNA702616. All data are released and can be accessed from this BioProject in the submission portal https://www.ncbi.nlm.nih.gov/Traces/study/ ?acc=PRJNA702616.

ETHICS STATEMENT
This study has been approved by the Institutional Review Board for Human Subject Research at the Jiangmen Maternity and Child Health Care Hospital (Jiangmen, IRB number: 2019053). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
JT, YD, and HT designed the study and revised the manuscript. JT and MT drafted the manuscript and interpreted the data. LL, JLL, FW, XM, ML, and JL reviewed and finalized the manuscript. WM, HD, SZ, HS, and XL conducted the study and collected the data. All the authors have read and approved the final version of the submitted manuscript.