Interferon-induced protein with tetratricopeptide repeats 3 may be a key factor in primary biliary cholangitis

Accumulating studies suggest that senescent biliary epithelial cells (BECs) produce senescence-associated secretory phenotypes (SASPs) and play various roles in the pathogenesis of primary biliary cholangitis (PBC) and other cholangiopathies. We examined comprehensive profiles of senescent BECs and its contribution to the pathogenesis of PBC taking advantage of microarray analysis. cDNA microarray analysis revealed that 1841 genes including CCL2, IFIT3, CPQ were commonly up-regulated in senescent BECs cultured in serum depleted media or media with glycochenodeoxycholic acid. Knockdown of IFIT3 significantly suppressed cellular senescence (p < 0.01) and significantly increased apoptosis (p < 0.01) in BECs treated with serum depletion or glycochenodeoxycholic acid. Significantly increased expression of IFIT3 was seen in senescent BECs in small bile ducts showing cholangitis and in ductular reactions in PBC, compared to control livers (p < 0.01). An inadequate response to UDCA was inversely correlated to the increased expression of IFIT3 in small bile duct in PBC (p < 0.05). In conclusion, the expression of various genes related to immunity and inflammation including SASPs were increased in senescent BECs. Upregulated IFIT3 in senescent BECs may be associated with the pathogenesis of PBC and may be a possible therapeutic target in PBC.

In this study, we examined the effects of knockdown of IFIT3 on cellular senescence, proliferation and apoptosis in cultured BECs. We also examined the expression of IFIT3 and its association with senescent markers p16 INK4a and p21 WAF1/Cip1 in human PBC and control livers.

Culture study. A number of genes was upregulated in senescent BECs induced by serum depletion and GCDC
treatment. Upregulated genes in senescent BECs, which show more than twofold change compared to control, were 2870 and 2789 genes in serum depletion and GCDC treatment for 7 days, respectively. 1841 genes were commonly upregulated in both 2 conditions. Major genes commonly upregulated in senescent BECs induced by serum depletion and GCDC treatment included CCL2, IFIT3, CPQ, NUPR1 and CIB2 (Table 1, Supplementary  table S1). Various inflammatory genes (chemokines and cytokines) including CCL2, CCL20, IL3, IL11, IL15 were upregulated in senescent BECs as SASPs (data not shown) in accordance with our previous study 8,16 . We put focus on IFIT3 and further examined its role in cell proliferation, apoptosis and cellular senescence. GSEA was also performed on top 500 up-regulated genes derived from senescent BECs induced by serum depletion and GCDC treatment. Various gene-sets were upregulated in each condition and a gene set: GO_RESPONSE_TO_ BIOTIC_STIMULUS was commonly upregulated in both condition and IFIT3 was included in several gene sets including this gene set (Supplementary Table S2).

IFIT3 expression was increased in senescent cells in mRNA and protein level.
We examined the mRNA expression of IFIT3 in cultured BECs treated with serum depletion and GCDC for 7 days. The expression of IFIT3 mRNA was significantly increased in senescent BECs treated with serum depletion and GCDC, compared with control (p < 0.01) (Fig. 1A). The expression of IFIT3 at protein level was also significantly increased in senescent BECs treated with serum depletion and GCDC, compared with control (p < 0.01) (Fig. 1B).
IFIT3 knockdown increased cell proliferation, apoptosis and decreased cellular senescence in BECs treated with serum depletion and GCDC. Effective knockdown of IFIT3. We examined the effect of IFIT3 knockdown on BECs. An effective knockdown of IFIT3 using siRNA was confirmed in mRNA and protein levels ( Fig. 2A,B).
IFIT3 knockdown increased cell proliferation activity. Cell proliferation activity detected by BrdU assay after the induction of cellular senescence (serum depletion and GCDC) for 4 days with or without knockdown of IFIT3 using siRNA. BrdU-labelling index was significantly low in BECs treated with serum depletion and GCDC for 4 days (p < 0.05) (Fig. 2C). The knockdown of IFIT3 significantly increased cell proliferation activity in BECs treated with serum depletion and GCDC (p < 0.05) (Fig. 2C).
IFIT3 knockdown did not change the growth curve assessed by cell number. Cell growth was significantly arrested in BECs by the treatment with serum depletion or GCDC (p < 0.01). IFIT3 knockdown did not change www.nature.com/scientificreports/ the growth curve in each condition in BECs (Fig. 2D). The increased proliferation and apoptosis may result in no change in the growth curve by treatments with serum depletion and GCDC.
IFIT3 knockdown resolved G1/S arrest. Cell cycle was analyzed using Cell-Clock cell cycle assay on BECs by the treatment with serum depletion or GCDC for 4 days with or without knockdown of IFIT3 using siRNA. G1/S arrest was induced by the treatment with serum depletion or GCDC (p < 0.05) (Fig. 2E). G1/S arrest was significantly resolved by IFIT3 knockdown in BECs treated with serum depletion or GCDC (p < 0.01) (Fig. 2E).
IFIT3 knockdown increased apoptosis. Caspase-3/7 activity was detected for assessment of apoptosis at 4 days after the induction of cellular senescence with or without knockdown of IFIT3 using siRNA. Caspase-3/7 activity with green fluorescence was detected in apoptotic cells (Fig. 2F). Apoptotic cells were significantly increased in senescent BECs with knockdown of IFIT3 (p < 0.01) (Fig. 2F).
IFIT3 knockdown decreased cellular senescence. Cellular senescence was assessed by the activity of SA-β-gal after treatment with serum deprivation or GCDC (500 nM) for 4 days with or without knockdown of IFIT3 using siRNA (Fig. 2G). SA-β-gal labelling index, a marker of cellular senescence, was significantly increased by treatment with serum depletion or GCDC. Cellular senescence was significantly decreased by knockdown of IFIT3 using siRNA in BECs treated with serum depletion or GCDC for induction of senescence (p < 0.01) (Fig. 2G).

Human study. Increased expression of IFIT3 in senescent BECs in damaged small bile ducts in PBC. IFIT3
was expressed in the cytoplasm in BECs, when present ( Fig. 3A-D). The expression of IFIT3 was significantly increased in BECs in small bile ducts involved in cholangitis in PBC (Fig. 3B,C). Table 2 is a summary of the extent of IFIT3 expression in small bile ducts in PBC and control livers. The expression of IFIT3 was significantly more extent in small bile ducts in PBC, compared with control livers (p < 0.01). The expression of IFIT3 was significantly correlated with cholangitis activity in PBC (p < 0.01). Similar tendency was confirmed in another cohort of 15 patients with PBC (data not shown). The increased expression of IFIT3 in small bile duct was inversely correlated to inadequate response to UDCA in PBC (p < 0.05).

Increased expression of IFIT3 in senescent bile ductular cells in ductular reaction in PBC.
The expression of IFIT3 was significantly increased in bile ductular cells in ductular reaction in PBC (Fig. 3D). Table 3 is a summary of the extent of IFIT3 expression in bile ductular cells in PBC and control livers. The expression of IFIT3 was significantly more extent in bile ductular cells in PBC, compared with control livers (p < 0.01). Similar tendency was confirmed in another cohort of 15 patients with PBC (data not shown).
Increased expression of IFIT3 in senescent BECs in damaged small bile ducts and bile ductules in PBC. Double immunostaining revealed that the expression of IFIT3 was frequently increased in BECs in the senescent small bile ducts showing expression of p21 WAF1/Cip1 or p16 INK4a in PBC (Fig. 4A). The increased expression of IFIT3 was also seen in senescent BECs with expression of p16 INK4a and p21 WAF1/Cip1 in ductular reactions in PBC (Fig. 4B).
Since intracytoplasmic localization of p21 WAF1/Cip1 (nucleus) or p16 INK4a (nucleus and cytoplasm) are different from IFIT3 (cytoplasm), the co-localization was not indicated by yellow color in the merged images (Fig. 4A,B). However, the expression was seen in same BECs. Figure 4C shows a view of double fluorescent stain in control normal liver. www.nature.com/scientificreports/

Discussion
The findings are summarized as follows; (1) Senescent BECs induced by both serum depletion or a treatment with GCDC for 7 days showed common upregulation of 1841 genes (fold change > 2), which included CCL2, IFIT3, CPQ and a number of chemokines and cytokines; (2) Cell proliferation and apoptosis were significantly increased (p < 0.01) and cellular senescence was significantly decreased (p < 0.01) by knockdown of IFIT3 in BECs treated with serum depletion or GCDC; (3) Small bile ducts showing cholangitis and in ductular cells in ductular reactions expressed IFIT3 in significantly higher level in PBC, compared to control livers; (4) Cholangitis activity was significantly correlated with the expression of IFIT3 in small bile ducts in PBC (p < 0.01); (5) Senescent BECs showing the expression of p16 INK4a and p21 WAF1/Cip1 in bile duct lesions in PBC expressed IFIT3 in high level; (6) Inadequate response to UDCA was inversely correlated to the increased expression of IFIT3 in small bile duct in PBC (p < 0.05).
In this study, we examined comprehensive profiles of senescent BECs relating to cholangiopathies. The results showed that the expression of various chemokines and cytokines are increased as SASPs in agreement with our previous studies 8, 15,16 . For example, CCL2, which we reported an increased expression and possible roles in the pathogenesis of PBC 8,16 , is included in the top 5 commonly-upregulated genes in senescent BECs in this study. It is now well known that senescent BECs express various chemokines and cytokines as SASPs in PBC and PSC, which may modulate inflammatory cell infiltration and fibrosis in cholangiopathies 8,15,16,25 . Altered expression of various inflammatory factors is also known in senescent cells 9,13,14,26 .
Among these inflammatory factors, we put focus on one of the top5 commonly-upregulated genes; IFIT3 [17][18][19] in senescent BECs in this study. In vitro study confirmed the increased expression of IFIT3 in mRNA and protein levels in this study, confirming the data of mRNA array study. In the present study, we found for the first time the increased expression of IFIT3 relating to cellular senescence in BECs in PBC. Increased expression of IFIT3 and its association with pathophysiology were reported in several diseases such as SLE 27 , RA 28 and degenerative annulus fibrosus 29 . IFIT3 is an IFN-induced protein and upregulates cGMP-AMP synthase (cGAS)-signal via stimulator of IFN genes protein (STING) pathway 27,29 . cGAS-STING pathway is known to play roles in the process of cellular senescence such as the expression of SASPs 30 . Taken together, IFIT3 may also contribute to the Ifit3 mRNA was significantly increased in BECs treated with serum depletion (Dep) compared to the control (p < 0.01) and the increase was significantly suppressed by a treatment with Ifit3 siRNA compared to the control siRNA (p < 0.01). Data are expressed as the means ± SD. *p < 0.01 vs. control + Cont siRNA, # p < 0.01 compared to Dep + Ifit3 siRNA. n = 3 for each group. (B) The protein level expression of IFIT3 assessed by immunoblotting in BECs treated with serum depletion and Ifit3 siRNA or control siRNA for 4 days. The protein level expression of IFIT3 was significantly increased in BECs treated with serum depletion compared to the control (p < 0.05) and the increase was significantly suppressed by a treatment with Ifit3 siRNA compared to the control siRNA (p < 0.05). *p < 0.05 vs. control + Cont siRNA; # p < 0.05 compared to Dep + Ifit3 siRNA, n = 3 for each group. (C) Cell proliferation was increased by a treatment with Ifit3 siRNA. Cell proliferation activity was detected by BrdU assay. Cell proliferation activities of BECs after the induction of cellular senescence (serum depletion and GCDC) for 4 days with a treatment with Ifit3 siRNA or control siRNA. BrdU-LI is significantly lower in BECs treated with serum depletion or GCDC compared to the control (p < 0.01) and significantly higher in BECs treated with Ifit3 siRNA. The data are expressed as the mean ± SD. *p < 0.05 vs. Cont siRNA; # p < 0.05. n = 5 for each group. (Con-Consi, 7.75 ± 2.95; Con-Ifit3si, 7.75 ± 2.50; Dep-Contsi, 0.72 ± 1.10; Dep-Ifit3si, 3.71 ± 1.84; GCDC-Contsi, 1.53 ± 1.94; Dep-Ifit3si, 7.60 ± 3.42). (D) Cell growth curve assessed by cell number was not changed by a treatment with Ifit3 siRNA. Cell number was assessed by WST assay after the induction of cellular senescence with serum deprivation (Dep) or GCDC (500 nM) with Ifit3 siRNA or control siRNA for 1, 2, 4 and 7 days. The data are expressed as the mean ± SD. *p < 0.01 vs. Con-Cont siRNA. n = 4 for each group. (E) G1/S arrest was resolved by a treatment with IFIT3 siRNA. Cell cycle was analyzed using Cell-Clock cell cycle assay on BECs by the treatment with serum depletion (Dep) or GCDC for 4 days with or without knockdown of IFIT3 using siRNA. Cells become yellow in G1, green in S/G2 and blue in M phase. The data are expressed as the mean ± SD. *p < 0.01 vs. Con-ConsiRNA; # p < 0.05 between IFIT3 siRNA and ConsiRNA in each condition.  Further studies are mandatory to confirm an involvement of IFIT3 on signal transduction pathways. Accumulating evidences suggest that IFN pathways may participate in the pathogenesis of PBC 20-24 , however, there has been no report regarding IFIT3 in PBC to our knowledge. Recent integrated analysis using GWAS, and mRNA microarray data sets predicted that IFNG and CD40L are the central upstream regulators in both  www.nature.com/scientificreports/ disease susceptibility and activity of PBC 24 . The interplay of types I and II IFN is also implicated as a cause of human PBC 23 and a murine model of autoimmune cholangitis 22 . A previous paper implicated type I IFN signaling as a necessary component of the sex bias in the murine model of autoimmune cholangitis with chronic IFN-γ stimulating 22 . The previous paper also suggested that drugs that target the type I IFN signaling pathway would have potential benefit in the earlier stages of PBC 22 . IFN-γ but not IFN-β, TGF-β or TNFα was found to up-regulate STING expression in keratinocytes 31,32 . Taken together, the present study suggests that IFIT3 may be a key molecule, which participates in the crossroads of IFN signaling, cGAS-STING pathway and cellular senescence in PBC.
In the present study, cellular senescence was decreased, and cell proliferation was increased by a knockdown of IFIT3 in BECs. Furthermore, G1/S arrest was resolved by a knockdown of IFIT3 in BECs treated with serum depletion or GCDC. A previous study reported significant accumulation of cells at G1/S transition in human monocytic cells with ectopic expression of IFIT3 18 . IFIT3 has been shown to have anti-proliferative activity by enhancing the expression of cell cycle negative regulators such as p27 and p21 by downregulating c-Myc 18 . The findings in the present study agree with previous study, suggesting that IFIT3 may contribute to induce cellular senescence.
A knockdown of IFIT3 in BECs treated with serum depletion or GCDC significantly increased apoptosis in the present study. Previous study suggested that upregulation of IFIT3 plays a protective role in lung epithelial cells in dengue virus infection by an inhibition of apoptosis 33 . IFIT3 knockdown induced apoptosis and suggested that the apoptotic effects of IFIT2 could be negatively regulated by IFIT3 34 . The findings in the present study agree these previous studies and confirmed anti-apoptotic roles of IFIT3 overexpression. It is well known that senescent cells are resistant to apoptosis 26,35 . "Senescent cells and anti-apoptotic pathways" (SCAPs) including Bcl-xL are thought to have responsibility for this resistant mechanism 26,35 . Since IFIT3 is upregulated in senescent cells and inhibits apoptosis 33,34 , increased expression of IFIT3 in bile duct lesions in senescent BECs in PBC may contribute to the anti-apoptosis as one of SCAPs. Taken together, IFIT3 may serve as a novel therapeutic target for clearance of senescent cells and for blocking the production of type I IFN and other proinflammatory cytokines by the cGAS/STING signaling pathway in PBC.
In conclusion, senescent BECs showed increased expression of various genes related to immunity and inflammation including SASPs. The increased expression of IFIT3 in BECs may be involved in the pathogenesis of PBC and could be a therapeutic target in PBC.

Methods
Culture study. Cell culture and treatments. Mouse intrahepatic BECs were isolated from 8-week-old female BALB/c mice and were purified and cultured as described previously 36,37 . All methods were carried out in accordance with relevant guidelines and regulations. The cell density of the cells was less than 80% during experiments. Cellular senescence was induced in BECs cultured in vitro incubation in serum depleted media or treatment with media containing 200 μM GCDC for 4-7 days, as described previously [37][38][39] . A treatment with GCDC causes cellular senescence via the induction of endoplasmic reticulum stress and dysregulated autophagy, as demonstrated in previous studies 15,[37][38][39] .
RNA extraction and cDNA microarray. Total RNA was extracted from the cells with a QIAGEN RNeasy Mini kit (QIAGEN, Hilden, Germany), as described previously 37 according to the manufacturer's protocol. Genomewide expression profiling was performed using a 3D-Gene scanner with 3D-Gene Oligo chip 24 k (Toray Indus-  www.nature.com/scientificreports/ tries, Inc., 23,522 distinct genes) and the supplier's protocol. Hybridization signals were scanned 3D-Gene Scanner (Toray Industries) and processed by 3D-Gene Extraction software (Toray Industries). The raw data of each spot was normalized by subtraction with a mean intensity of the background signal determined by all blank spots' signal intensities of 95% confidence intervals. The raw data intensities greater than 2 standard deviations (SD) of the background signal intensity were considered to be valid. cDNA microarray was performed using pooled RNA samples from 2 independent experiments for each condition. Second cDNA microarray was performed using different sets of samples for each condition and major up-regulated genes including IFIT3 were confirmed. Data analysis and functional analysis was performed by Gene Ontologies and KEGG pathway analysis. GSEA was performed according to the instructions.
Data deposition. Microarray data are deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (accession number GSE168052).
Real-time quantitative reverse transcriptase-polymerase chain reaction. After cDNA was synthesized, quantitative real-time PCR was performed to measure Ifit3 and β-actin mRNA, as described previously 37 according to a standard protocol using the SYBR Green PCR Master Mix (Toyobo, Tokyo, Japan). Forward and reverse primers 5′-GAG TGC TGC TTA TGG GGA GA and 5′-AGA GCA GTT TGT CAG CAA TCC, respectively, were used for Ifit3 and 5′-CCA CCG ATC CAC ACA GAG TA and 5′-GGC TCC TAG CAC CAT GAA GA for β-actin as an internal control. Each experiment was performed twice in triplicate, and the mean was calculated for each of the experiments.
Immunoblotting. The cell lysate samples (10 μg) were resolved by SDS-PAGE and transferred to a nitrocellulose membrane as described previously 37 . After transfer, the membranes were processed for immunoblotting as described previously 37 . The primary antibodies used were shown in supplementary table S2. Densitometry of the resulting bands was performed using Image-J software and normalized to the loading control.
Assay for cell proliferation. Cell proliferation activity was assessed on day 4 after treatment by using a 5-bromo-2′-deoxy-uridine (BrdU) Labeling and Detection Kit (Roche, Nonenwald, Germany), according to manufactures' protocol. The nuclei were simultaneously stained with DAPI. At least 1 × 10 3 total cells were checked and counted to assess the BrdU-labeling index with a conventional fluorescence microscope (Olympus).
Assay for cell number. BECs were seeded into 96-well microplates (1 × 10 4 cells/well) and incubated in a final volume of 100 μl medium. The cell number was assessed on days 1, 2, 4 and 7 after treatment using a Cell Proliferation Reagent WST-1 (Roche, Basel, Switzerland) according to manufacturer's recommendation.
Assay for cell cycle. The cell cycle alteration was detected on day 4 after treatment by a Cell-Clock Assay Kit (Biocolor, Northern Ireland, UK), according to manufacturer's protocol. Images of the cells were acquired using a conventional microscope (Olympus) and analyzed using Image J software.
Assay for apoptosis. The apoptotic cells in each condition were assessed after the induction of cellular senescence and a treatment with senolytic reagents by using CellEvent Caspase-3/7 Green Detection Reagent (Life Technologies, Carlsbad, CA) as described previously 39 according to manufacturer's protocol. The nuclei were simultaneously stained with DAPI. At least 1 × 10 3 total cells were checked and counted to assess the percentage of apoptotic cells showing Caspase-3/7 activity with a conventional fluorescence microscope.  www.nature.com/scientificreports/ Assay for cellular senescence. The activity of senescence-associated β-galactosidase (SA-β-gal) was detected after the induction of cellular senescence and a treatment with senolytic reagents by using the senescence detection kit (Bio Vision, Mountain View, CA) according to manufacturer's protocol 40 . The proportion of senescent cells was assessed by counting SA-β-gal-positive cells in at least 1 × 10 3 total cells.
Human study. Classification of intrahepatic biliary tree. The intrahepatic biliary tree is classified into intrahepatic large and small bile ducts (septal and interlobular bile ducts) by their size and distributions in the portal tracts 41 . Bile ductules, which are characterized by tubular or glandular structures with a poorly defined lumen and located at the periphery of the portal tracts 41,42 , are not included in the small bile ducts and evaluated separately.
Liver tissue preparation. A total of 171 liver tissue specimens (all were biopsied or surgically resected) were collected from the liver disease file of our laboratory and affiliated hospitals. All methods were carried out in accordance with the Declaration of Helsinki, relevant guidelines and regulations. The Ethics Committee of Kanazawa University approved this study. The liver specimens in this study were 70 PBC, 61 chronic viral hepatitis (CVH), 12 PSC, 10 extrahepatic biliary obstruction (EBO) and 18 "histologically normal" livers. All PBC specimens were from patients fulfilling the clinical, serological and histological characteristics consistent with the diagnosis of PBC 2 and histologically classified according to Nakanuma classification 43 . Table 4 is a summary of the clinicopathological features of the PBC patients included in the present study. Seventeen PBC livers were after UDCA therapy and 17 were UDCA non-responders. Thirty-eight and 23 CVH patients were regarded as F0-2 and as F3, 4, respectively 44 . Ten and 61 of CVH cases were serologically positive for hepatitis B surface antigen and anti-hepatitis C viral antibody, respectively. Causes of EBO were obstruction of the bile duct at the extrahepatic bile ducts or the hepatic hilum due to stone or carcinoma, and the duration of jaundice was less than 1 month. "Histologically normal" livers were obtained from surgically resected livers for metastatic liver tumor or traumatic hepatic rupture. Normal liver tissues were obtained from an area apart from the tumor and carcinoma tissues were not evaluated.
Liver tissue samples were fixed in 10% neutral-buffered formalin and embedded in paraffin. More than twenty serial sections, 4 μm-thick, were cut from each block. Several sections were processed routinely for histologic study, and the remainder was processed for the subsequent immunohistochemistry.
Immunohistochemistry. The expression of IFIT3 and senescence-related markers p16 INK4a , p21 WAF1/Cip1 were examined, as described previously 12 . The primary antibodies used are shown in supplementary table S3. A similar dilution of the control mouse or rabbit Immunoglobulin G (Dako) was applied instead of the primary antibody as a negative control. Positive and negative controls were routinely included. Histological analysis was performed in a blinded manner. BECs in small bile ducts and bile ductules were separately evaluated.
Extent of IFIT3 expression in small bile ducts and bile ductules. The extent of expression was evaluated as follows: 1+, focal, positive cells are detected in one third or fewer portal tracts; 2+, moderate, positive cells are detected in small bile ducts of more than one third of portal tracts; 3+, extensive, positive cells are detected in small bile ducts of more than two thirds of portal tracts. Double immunofluorescence. Double immunofluorescence for IFIT3 with senescent markers (p16 INK4a and p21 WAF1/Cip1 ) was also performed. In brief, either of p16 INK4a or p21 WAF1/Cip1 was detected using Vector Red Alkaline Phosphatase Substrate Kit (Vector Lab, Burlingame, CA), followed by second staining for Ifit3 using Alexa-488-labeled anti-rabbit IgG. The sections were counterstained with DAPI and evaluated under a conventional fluorescence microscope. www.nature.com/scientificreports/ Statistical analysis. Statistical analysis of differences was performed using the Kruskal-Wallis test with Dunn's posttest. When the number of groups is 2, statistical analysis of difference was performed using the Mann-Whitney test. The Chi-square test or Fisher's exact test was used to analyze categorical data. The correlation coefficient of 2 factors was evaluated using Spearman's rank correlation test. When the P value was less than 0.05, the difference was regarded as significant. All analyses were performed using the GraphPad Prism software (GraphPad Software, San Diego, CA, USA).