Increased expression of TNFRSF14 and LIGHT in biliary epithelial cells of patients with primary sclerosing cholangitis

Background and aims: There is a lack of biliary epithelial molecular markers for primary sclerosing cholangitis (PSC). We analyzed candidates from disease susceptibility genes identi�ed in recent genome-wide association studies. Methods: Expression was quanti�ed using immunohistochemistry in biliary epithelia in liver biopsy samples from patients with PSC (N = 45) and controls (N = 12). Samples from patients with primary biliary cholangitis (PBC) were used as disease controls (N = 20). Results: The levels of hepatic expression of ATXN2, HHEX, PRDX5, MST1, and TNFRSF14 were signi�cantly altered in the PSC group. We focused on the immune-related receptor, TNFRSF14. Immunohistochemistry revealed that TNFRSF14 positivity was signi�cantly higher in biliary epithelia in the PSC group (96 %) than in the control (42 %) and PBC (55 %) groups. High expression of TNFRSF14 was observed only in patients with PSC. Moreover, the expression of LIGHT, which encodes a TNFRSF14-activating ligand, was increased in PSC liver. Immunohistochemistry showed that high expression of LIGHT was more common in PSC biliary epithelia (53 %) than in the PBC (15 %) or control (0 %) groups; moreover, it was positively associated with �brotic progression. Conclusions: TNFRSF14 and LIGHT are attractive candidate markers for PSC.


Introduction
Primary sclerosing cholangitis (PSC) is an intractable idiopathic hepatobiliary disease characterized by multifocal bile duct strictures due to in ammation and brosis, resulting in liver failure (1).The pathogenesis of PSC remains unclear, and thus the identi cation of speci c biomarkers is urgently needed.
Molecular markers expressed in the biliary epithelia of patients with PSC remain unidenti ed.Recent studies have revealed that cellular senescence and the senescence-associated secretory phenotype (SASP) are characteristics of biliary epithelial cells in PSC (2).In explanted liver samples, the expression of p16, interleukin 6 (IL-6), and interleukin 8 (IL-8) is increased in the biliary epithelia of PSC rather than in those of primary biliary cholangitis (PBC), whose pathological condition, characterized by autoimmune bile duct injury, is similar to that of PSC (2).IL-8 was also expected to be a progressive or prognostic marker for PSC (3).However, IL-8 is only expressed at low levels in biliary epithelial cells in early PSC, whereas its expression is prominent in the advanced stage (4).The identi cation of uniquely occurring early pathological markers in the biliary epithelia of PSC requires the analysis of biopsy samples obtained at diagnosis instead of explanted samples, which re ect the terminal conditions of the disease.
Recent genome-wide association studies (GWAS) have revealed a strong association between PSC and the human leukocyte antigen (HLA) region on chromosome 6, indicating that PSC is an autoimmune disorder (5)(6)(7)(8)(9)(10)(11).Additionally, more than 20 non-HLA disease susceptibility genes have been identi ed (5,11).Despite their weaker association (odds ratios < 1.5) compared with that of HLA (odds ratio [3][4][5], each of them is a candidate potentially involved in the pathology of PSC through the regulation of the in ammatory immune response or bile-acid homeostasis (5,11).A recent study showed that downregulation of Takeda G protein-coupled receptor-5 (TGR5), one of the GWAS-derived risk genes regulating epithelial barrier functions, contributed to the development of cholangitis, speci cally in PSC biliary epithelia (12).However, the expression pro les of other genes in patient tissues, particularly in bile ducts, remain unknown.
In this study, we aimed to identify molecular markers with altered expression in PSC biliary epithelia, focusing on GWAS risk genes.First, we analyzed gene expression in liver biopsy samples from patients with PSC, with the resected liver tissues adjacent to the tumors as controls.Next, using immunohistochemistry, we validated these expression pro les in biliary epithelial cells and hepatocytes.
We further compared these results with those from control patients and patients with PBC as disease controls.

Human liver tissue
Needle liver biopsy samples were obtained from patients diagnosed with PSC and PBC at the University of Tokyo Hospital.Resected liver specimens were obtained from patients who underwent surgery at our hospital.This study was approved by the Ethical Committee for Clinical Research at our institutions (approval numbers 11902, Institute of Medical Science, Asahi Life Foundation; 2019140NI and 1302, The University of Tokyo).Informed consent was obtained from patients with PSC and liver tumors in the form of an opt-out on the website (2019140NI).Written informed consent was obtained from all patients with PBC (1302).

RNA extraction
Total RNA was extracted from formalin-xed para n-embedded liver biopsy samples using the Recover All-TM Total Nucleic Acid Isolation kit (Invitrogen, Carlsbad, CA, USA).Total RNA was also extracted from frozen resected liver specimens using an RNeasy Mini Kit (Qiagen, Hilden, Germany).

Real-time quantitative reverse transcriptase polymerase chain reaction (RT-qPCR)
For the synthesis of complementary DNA from RNA, ImProm-II Reverse Transcriptase (Promega, Tokyo, Japan) was used.qPCR was performed using the THUNDERBIRD SYBR qPCR Mix (Toyobo Co. Ltd., Osaka, Japan).All values were normalized to the levels of expression of β-actin mRNA.Primers are listed in Supplementary Table 1.

Immunohistochemistry
Para n-embedded tissue blocks were cut into 5-µm sections and mounted on glass slides.After dewaxing and rehydration, antigen retrieval was performed by heating slides in 10 mM citrate buffer (pH 6.0) at 120°C for 5 min.After blocking, sections were incubated overnight at 4°C with anti-TNFRSF14 (10138-1-AP, 1:200 dilution, Proteintech Japan, Tokyo, Japan) or anti-LIGHT (PA5-82467, 1:50 dilution, Invitrogen).Peroxidase-conjugated anti-rabbit immunoglobulin G (NICHIREI BIOSCIENCES, Tokyo, Japan) was used as secondary antibody.Peroxidase activity was visualized using Histo ne Simple Stain 3,3diaminobenzidine (DAB) solution (415171, Nichirei Bioscience).We grouped TNFRSF14-or LIGHTexpressing cells by staining intensity as high-or low-positive.A sample was considered negative when the frequency of stained cells was < 1 % in each cellular fraction.Stained slides were validated by 2 gastroenterologists (SK and HF) who were blinded to patient identities and clinical information.

Statistical analysis
The Mann-Whitney U test was used to compare levels of gene expression in liver tissues and laboratory data between groups.Fisher's exact or chi-square tests were used to compare the immunohistochemical staining data between groups.Differences were considered statistically signi cant at P < 0.05.

Patient characteristics
The clinical characteristics of patients in all groups are shown in Table 1.The PSC group included 47 patients, of whom 19 were men and 28 were women; the median age at diagnosis was 48 years (range, 14-82).We evaluated the disease stage at the time of diagnosis using Ludwig's classi cation for histological liver brosis.Approximately half of patients were diagnosed at an early stage (stage 1, N = 11; stage 2, N = 11; stage 3, N = 20; and stage 4, N = 5).We found that the prevalence of in ammatory bowel disease (IBD) was 36% (17/47), comparable to that in a recent study in Japan (13).The PBC group included 6 men and 14 women; the median age at diagnosis was 51 years (range, 35-67).According to Ludwig's classi cation, most patients were at early stages (stage 1, N = 5; stage 2, N = 12; stage 3, N = 3; stage 4, N = 0).The control group included 7 men and 5 women who had undergone surgery for intrahepatic cholangiocarcinoma (N = 9), intraductal papillary neoplasm of the bile duct (N = 1), or liver metastasis of colon cancer (N = 2).None of them had viral hepatitis.Their resected livers were histologically evaluated using the New Inuyama classi cation (14); they were all classi ed as F0, A1, or A2.We found that serum alkaline phosphatase and gamma-glutamyl transpeptidase levels in the PSC group were signi cantly higher than those in the control group and equivalent to those in the PBC group.
In all three groups, most patients (> 90%) were classi ed as Child-Pugh class A. Altered expression of risk genes identi ed in GWASs in the livers of patients with PSC To screen for potential candidate genes, we performed RT-qPCR on liver biopsy samples from the PSC group and resected liver samples from the control group.We chose 9 candidate genes from a previous GWAS study for analysis (11): ATXN2, CCL20, FOXP1, HHEX, IL2RA, MST1, NFKB1, PRDX5, and TNFRSF14.We accordingly collected 10 samples from the PSC group that met the inclusion criteria: (1) con rmed expression of the housekeeping gene β-actin, and (2) quantitation of the expression of each gene using a standard curve.We found that all 12 control samples satis ed both criteria.As shown in Fig. 1, the expression of ATXN2, HHEX, PRDX5, and TNFRSF14 was signi cantly upregulated, whereas that of MST1 was downregulated in the PSC group compared with that of controls.However, we did not detect any signi cant difference in the expression of CCL20, FOXP1, NFKB1, or IL2RA between the two groups (Fig. 1).
Biliary expression of TNFRSF14 was increased in PSC but not in PBC Among the ve candidate genes identi ed, we focused on TNFRSF14 because it has also been reported to be a disease susceptibility gene in ulcerative colitis, the most common comorbidity of PSC (5,15).TNFRSF14 encodes tumor necrosis factor (TNF) receptor superfamily member 14 (TNFRSF14), also known as herpesvirus entry mediator (HVEM).It is expressed in stromal, myeloid, lymphoid, and epithelial cells, and is involved in the in ammatory signal transduction through its interactions with multiple TNFrelated ligands and immunoglobulin-superfamily proteins (16).
To validate the TNFRSF14 expression pro les, we performed immunohistochemical staining of liver samples from control patients and patients with PSC.We classi ed expression levels using staining intensity as TNFRSF14-high, -low, and -negative (Fig. 2A).We assessed a total of well-preserved 45 liver biopsy samples from patients with PSC.We also evaluated the levels of expression in hepatocytes and compared the results with those of liver biopsy samples from patients with PBC.

Biliary expression of LIGHT is increased in PSC
Recent studies have revealed that the TNF family member LIGHT (lymphotoxin-like, which exhibits inducible expression and competes with herpes simplex virus glycoprotein D for herpes virus entry mediator, a receptor expressed by T lymphocytes), an activating ligand of TNFRSF14, is involved in brosis in many tissues, including the lungs, skin, and liver (17)(18)(19).The expression of both TNFRSF14 and LIGHT was increased in the synovial tissues of patients with other autoimmune disorders, such as rheumatoid arthritis (20).As the expression of LIGHT was signi cantly elevated in PSC (Fig. 3A), we next measured the LIGHT expression.
As in the case of TNRSF14, we performed immunohistochemistry in liver tissues from control patients, patients with PSC, and patients with PBC using antibodies against LIGHT.We classi ed the levels of expression of biliary epithelial cells and hepatocytes based on their staining intensity as LIGHT-high, -low, and -negative (Fig. 3B).We found that the fraction of LIGHT-positive (LIGHT-high and -low) biliary epithelial cells in patients in the PSC group was signi cantly higher than that in the control group (100% [45/45] vs. 83% [10/12], P = 0.041) (Fig. 3C).However, we did not detect any signi cant differences in staining between the PSC and PBC groups (95% [19/20], P = 0.308) (Fig. 3C).Interestingly, we observed that LIGHT-high cells were more common in PSC (53% [24/45]) than in PBC (15% [3/20], P = 0.006), whereas they were not observed in the control group (Fig. 3C).Likewise, we found that the proportion of LIGHT-positive hepatocytes was also signi cantly higher in PSC than in controls (PSC 100% [45/45], control 67% [8/12], P = 0.001) (Fig. 3C).We did not observe any signi cant differences in positivity between PSC and PBC (100% [20/20], P = 1) (Fig. 3C).We noticed that LIGHT-high hepatocytes were equally common in PSC and PBC (PSC: 64% [29/45], PBC: 55% [11/20], P = 0.655), whereas they were not found in control tissues (Fig. 3C).These observations suggested that the expression of LIGHT was increased in the biliary epithelial cells and hepatocytes of patients with PSC.We more frequently detected high expression of LIGHT in PSC than in PBC biliary tissues, although the levels of expression in hepatocytes were comparable between these groups.

High expression of LIGHT in the bile duct is correlated with brotic progression of PSC
We found that patients with PSC showed various expression patterns of TNFRSF14 and LIGHT (Supplementary Fig. 1).Therefore, we determined whether there was a relationship between increased expression of TNFRSF14/LIGHT and clinical characteristics or laboratory data in patients with PSC (Tables 2 and 3).As shown in Table 3, we detected more LIGHT-high biliary epithelial cells (71%, P = 0.027) and hepatocytes (69%, P = 0.012) in later stages of the disease, whereas other clinical factors, including age, sex, and comorbid IBD, did not correlate with high expression of LIGHT in either cell type.We found that TNFRSF14 expression was not associated with disease stage (Table 2).Considering that Ludwig's classi cation is based on the histological grading of liver brosis, these results suggest that high expression of LIGHT in biliary epithelial cells and hepatocytes is correlated with brotic disease progression in PSC.
Regarding the relationship with laboratory data, the following results were obtained with statistical signi cance.We found that patients with TNFRSF14-high hepatocytes tended to have lower levels of PT-INR (P = 0.023) (Table 2).In addition, we more often detected LIGHT-high biliary epithelial cells or hepatocytes in patients with lower levels of albumin (P = 0.042) or higher levels of total bilirubin (P = 0.008), respectively (Table 3).

Discussion
In this study, using liver biopsy samples obtained from patients at diagnosis, we found increased expression of the immune-related receptor TNFRSF14 in biliary epithelial cells of patients with PSC but not those with PBC.In addition, we observed increased expression of its activating ligand, LIGHT, in PBC biliary epithelial cells and demonstrated a positive correlation between the high LIGHT expression and brotic disease progression.The expression of TNFRSF14 and LIGHT was also upregulated in PSC hepatocytes.In addition to TNFRSF14, we con rmed the altered expression of several diseasesusceptibility genes previously reported from GWAS: ATXN2, HHEX, MST-1, and PRDX5 in the livers of patients with PSC.
Owing to their dependency on diagnostic imaging, liver biopsies are not routinely performed for PSC diagnosis (21).Therefore, previous studies have focused on liver resection specimens obtained during transplantation.Recent reports have shown increased expression of MAdCAM-1 and VAP-1, which are involved in the recruitment of T-lymphocytes, in PSC hepatic endothelial cells (22,23).This alteration of expression has been reported to be prominent at the end stage; however, it has not been con rmed in liver biopsy samples taken at diagnosis, which should not re ect the indirect effects of secondary liver damage (22)(23)(24).Even when using liver biopsy specimens, the altered expression of biliary epithelia could be masked by that of hepatocytes in the analysis, such as in RNA sequencing of bulk samples (25).To detect any initial changes in the biliary epithelia during the progression of this refractory disease, validation with immunohistochemical assays using liver biopsy samples is essential (Fig. 2B, 3C, Table 2, and 3).
Few reports have analyzed the expression of TNFRSF14 in hepatobiliary diseases.A previous immunohistochemical study demonstrated that the expression of TNFRSF14 was not elevated in any intrahepatic cellular fraction of patients with chronic viral hepatitis (26).Increased expression of TNFRSF14 was not observed in the biliary epithelia of patients with PBC, and its expression level was not correlated with that of serum ALP in patients with PSC (Fig. 2 and Table 2).These ndings suggest that upregulation of TNFRSF14 in biliary epithelial cells is a primary feature of the pathogenesis of PSC and not a secondary change downstream of cholestasis or in ammation.
We observed not only the upregulation of TNFRSF14, but also that of LIGHT (Fig. 3B and 3C).LIGHT is a lymphokine produced by hematopoietic cells, and its serum level is elevated and correlated with disease progression in patients with in ammatory diseases such as dermatitis (27).Considering the correlation between LIGHT expression and the histological brotic stage of PSC (Table 3), LIGHT should be further studied as a new biomarker re ecting PSC progression, although its functional signi cance has not been elucidated.In future studies, we will measure serum and bile LIGHT in patients with PSC and test for correlations with prognoses and/or comorbidities such as acute cholangitis and cholangiocarcinoma.This study had some limitations.(1) It was a single-center retrospective study with a limited number of cases; the sample size in the PSC group was inadequate for multivariate analysis.(2) There was a small number of cases at Ludwig's stage 3/4 (only 3) in the PBC group, which might have contributed to the low rate of LIGHT-high biliary epithelial cells (Table 1 and Fig. 3C).
In conclusion, we identi ed 2 novel genes that were upregulated in the biliary epithelia of PSC, that is, TNFRSF14 and LIGHT.The former was not upregulated in the biliary epithelia of PBC, while the latter was correlated with the histological degree of brotic progression of PSC.The interactions between the two in ammatory regulators might play important roles in the pathogenesis of PSC, and their combined detection is likely to be useful as a diagnostic or prognostic biomarker for this refractory disease.

Disclosures
The declare that they have no con icts of interest. Figures

Table 2
Relations between TNFRSF14 expressions and clinical characteristics and laboratory data