Undifferentiated HepaRG human hepatic cells (HPR101) were obtained from Biopredict (Rennes, France) and were cultured and differentiated into fully functional hepatocyte-like cells according to the manufacture’s protocol. HPR101 cells were grown in William’s E media (Gibco, Grand Island, NY) supplemented with ×1 GlutaMAX (Gibco, Grand Island, NY) and 10% HepaRG Growth Medium Supplement (Biopredict, Rennes, France) until confluent. Cells were then switched into HepaRG Differentiation media for 2 weeks. Differentiated HepaRG cells were used directly for transfection.

HepG2 cells (the American Type Culture Collection, ATCC) were grown in Eagle’s Minimum Essential Medium containing 10% fetal bovine serum (FBS) (HyClone). HEK 293T were grown in Dulbecco’s modified Eagle’s medium containing 10% FBS. STF cells (ATCC) were cultured in 80% DMEM F12 Medium, (ATCC 30-2006), add 20% Bovine Calf Serum, Iron Fortified (ATCC 30-2030) and 200ug/ml G-418. All cells were cultured at 37°C with 5% CO₂ and used within 30 passages. Cells were transfected with plasmids using lipofectamine 2000 (Thermo scientific). Cells were transfected with siRNA using Lipofectamine RNAiMAX (Invitrogen). Specifically, for 6 well plates, each well used 4 µl Lipofectamine RNAiMAX reagent and 4 µl siRNA (10 µM).

Total RNA was purified using the QIAGEN RNeasy kit (Germantown, MD). Primers for the amplification of TSPYL1, CTNNB1 and CYPs were Prime Time pre-designed qPCR primers (IDT Inc., Coralville, Iowa). All samples were measured in technical triplicates using the SYBR green reagent with an ABI Prism 7000HT sequence detection system (Applied Biosystems). The results were quantified using the Comparative Ct (ΔΔCt) method with the housekeeping gene GAPDH as an internal reference control.

Protein levels were determined using Western blot analysis. The following primary antibodies were used: Flag (1:1,000, Cell Signaling Technology, cat. #F1804), GAPDH (1:1,000, Cell Signaling Technology, cat. #5174S), β-catenin (1:1,000, Cell Signaling Technology, cat. #9582S), TSPYL1 (1:1,000, Bethyl Laboratories, cat. #A304-852A-M), and TCF7L2 (1:1,000, Cell Signaling Technology, cat. #2569S). The following secondary antibodies were used: Peroxidase- conjugated AffiniPure Goat Anti-Mouse IgG, light chain specific (1:2,000 dilution, Jackson immunoRsearch) and Peroxidase- conjugated IgG fraction monoclonal mouse anti-rabbit IgG, light chain specific (1:2,000 dilution, Jackson Immuno Research).

For Co-IP, 293T cells were grown in 10-cm dishes and were transfected with the appropriate plasmids for 48 h. Cell lysates were incubated with 3 μg of Flag antibody on a rotator overnight at 4°C. The protein–antibody–protein A/G-agarose complexes were prepared by adding 50 μl of protein A/G-agarose beads (Thermo scientific) for 2 h at 4°C. After three washings with NETN lysis buffer, the immunoprecipitated complexes were resuspended in reducing sample buffer and elute at 50°C for 10 min. The supernatants were transferred to a clean tube and boiled for 5 min. Supernatants were subjected to SDS–polyacrylamide gel electrophoresis (PAGE) and IB.

Mass Spectrometry was done using differentiated HepaRG cells with IRES-TSPYL1, 2 or 4 plasmids transiently transfected using lipofectamine 2000 for 48 h. TSPYL interacting proteins were immunoprecipitated using Flag antibody. Normal mouse IgG was used as a negative control. The immunoprecipitated protein was eluted using 2X Laemmili buffer and resolved using SDS-PAGE electrophoresis. Protein bands were visualized by Coomassie Blue Staining and were dissected into sections equally. The dissected gel slices were sent for Mass spectrometric analysis at the Harvard Taplin Mass Spec facility. Proteins uniquely precipitated using Flag antibody but not IgG, and shared among all three TSPYLs were included for further pathway analysis. Pathway analysis was performed using the Panther pathway database.

TSPYL1 expression, cholesterol and BMI data from patients were downloaded from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) by GEO ID GSE48452 and GSE130991. Probe IDs were labeled in the figure legends. For the BMI analysis, samples were divided by TSPYL1 median expression level. The GSE130991 dataset was used for the age-specific LDL-cholesterol level test. Samples were grouped by both the median expression of TSPYL1 and the population median age. Statistical significance was tested by the Mann-Whitney test.

SuperTopFlash HEK293 cell line (STF) reporter cells were split into 24-well plates and were transfected 24 h later with 1ug of TSPYL overexpression plasmid or siRNA and 0.8 ng of the transfection control Renilla luciferase plasmid pTK-RL (Promega) using lipofectamine 2000 (Thermo scientific). 48 h after transfection, the cells were washed with PBS and luciferase activities were measured with a Dual-Luciferase Assay Kit (Promega, cat# E1960) according to the manufacturer’s protocols. Luciferase assays were performed at least in triplicate.

SMARTpool siRNA duplexes specific for CTNNB1 (Catalog ID: M-003482-00-0005), TSPYL1 (Catalog ID: M-028592-01-0005), TSPYL2 (Catalog ID: M-013880-01-0005), TSPYL4 (Catalog ID: M-017980-00-0005), and a Non-Targeting siRNA pool (siCTRL, Catalog ID:D-001206-13-05) were purchased from Dharmacon. Recombinant human Wnt-3a was purchased from R&D Systems. Human TSPYL cDNA constructs and empty vector, Pcmv6-XL4, were purchased from Origene Technologies (Rockville. MD).

HepaRG cells were used to perform ChIP assays to determine possible TSPYL1 binding to the promoter regions of CYP1B1 or CYP7A1. Primer sets (see Supplementary Table S2) which covered TCF7L2 binding sites of CYP1B1 or TCF7 binding sites of CYP7A1, were designed to “screen” potential TSPYL1-DNA binding sites. ChIP assay was performed using the SimpleChIP Enzymatic Chromatin IP kit (Cat. #91820, cell signaling technology) followed by quantitative PCR using the TB Green Premix Ex Taq™ PCR master mix reagent (Cat. #RR420A, TaKaRa).

The concentration of 20-HETE was determined by an ELISA kit (20H39-K01, Eagle Biosciences) according to the manufacturer’s instructions. For measurement of 20-HETE in cultured HepaRG cells, cells were trypsinized in 0.05% trypsin after washing with PBS. Half of the cells were used for PCR. The remainder of the cells were homogenized and sonicated with RIPA buffer containing triphenylphosphine. 50 µl was used for the 20-HETE assay on the day of extraction. The remaining homogenate was frozen at −80 C for protein assay.

A two-tailed Student-t test was used for statistical analysis for changes across conditions, if not specified. A p value < 0.05 was considered to be statistically significant. Statistical significance is indicated by asterisks. *p < 0.05; **p < 0.01; ***p < 0.001, ****p < 0.0001.

To study mechanisms and potential signaling pathways that interact with TSPYLs, we performed flag-affinity purification of 3XFLAG-TSPYL1, 2 and 4 from HepaRG cells after IRES plasmid transient transfection to isolate TSPYL1,2 and 4-associated proteins. A total of 2272 proteins were pulled down for these three TSPYL isoforms as shown by the Venn diagram in Figure 1A. The pie charts and Venn diagram of panther pathway analysis for each individual pull down by TSPYL1, 2 and 4 are shown in Supplementary Figure S1. Panther pathway analysis was performed for these 2272 proteins. Included among the top signaling pathways associated with TSPYLs on the basis of mass spectrometric analysis were Wnt signaling pathway proteins (Figures 1A,B). The other top 4 pathways were the Integrin signaling pathway, the Gonadotropin- releasing hormone receptor pathway, the Angiogenesis pathway, and Inflammation mediated by chemokines and cytokines. All of the protein pull-down information is available in a Supplementary Excel File.

β-Catenin is a key component of Wnt canonical signaling (Mosimann et al., 2009). It is known that β-catenin binds to TCF/LEF family members to activate the transcription of Wnt signaling target genes (Mosimann et al., 2009). Therefore, we confirmed the mass spectrometric results by performing co-immunoprecipitation using 293T cells and found that endogenous β-catenin (Figure 1C) could be co-precipitated with Flag-tagged-TSPYL1.

Because of the nuclear localization and the association among TSPYLs, β-catenin and TCF/LEF, we next tested the possibility of a functional link between TSPYLs and Wnt signaling. We used SuperTopFlash (STF) HEK293 reporter cells to determine whether the knockdown of TSPYLs could affect Wnt signaling. TSPYL1 knockdown increased luciferase reporter transcription (Figure 2). TSPYL2 and TSPYL4 knockdown also increased Wnt activity 1.25 and 1.55 fold, respectively, but less than that observed for TSPYL1 (1.73 fold) (Figure 2). These observations raised the possibility that these TSPYLs might be negative regulators of Wnt/β-catenin signaling.

CYPs play an important role in the maintenance of cholesterol homeostasis (Pikuleva, 2006). For example, CYP7A1 is the first and the rate-limiting enzyme in the classic bile acid synthesis pathway in which cholesterol is metabolized to form bile acids, while CYP11A1 catalyzes a step in a pathway by which cholesterol serves as a substrate for the synthesis of steroid hormones (Pikuleva, 2006). CYP1B1 also has a significant impact on cholesterol metabolism. Consistent with our previous observations (Qin et al., 2018), we found that CYP1B1 was upregulated by TSPYL1 knockdown (Figures 3A,B) and that it was downregulated by TSPYL1 overexpression in HepaRG cells (Figures 3D,E). CYP7A1 was significantly downregulated by TSPYL1 knockdown (Figure 3C) and was upregulated by TSPYL1 overexpression in those cells (Figure 3F). PCR results had previously shown that CYP11A1 was not significantly regulated by TSPYL1 knockdown (Qin et al., 2018), so we did not perform follow up studies on CYP11A1. We next analyzed the correlation of TSPYL1 and CYP1B1 expression, using expression quantitative trait loci (eQTL) data from human hepatic tissue (Innocenti et al., 2011). The expression of TSPYL1 and CYP1B1 were negatively correlated, with a correlation coefficient of -0.26 (Figure 4A, p = 1.35e-08). When we analyzed the correlation of TSPYL1 with CYP7A1 expression, we observed that TSPYL1 and CYP7A1 expression were positively correlated, with a correlation coefficient of 0.25 (Figure 4A, p = 5.54e-08).

According to the TCF7L2 ChIP-seq data in the ENCODE database, TCF7L2 binds to the CYP1B1 promoter in Panc1 cells (Figure 5A). We hypothesized that TSPYL1, by interacting with β-catenin and TCF7L2, might bind to the same CYP1B1 promoter region as does TCF7L2 in HepaRG cells. Therefore, we performed CHIP assays with antibodies to both TSPYL1 and TCF7L2 using HepaRG cells because of the high protein expression level of TSPYL1 in HepaRG cells. Specifically, a series of PCR primers was designed to amplify TCF7L2 binding sites in the CYP1B1 promoter. Our CHIP results showed that TSPYL1 bound to the same CYP1B1 promoter region as did TCF7L2 (Figure 5B). We next performed co-immunoprecipitation to study possible interactions among TCF7L2-FLAG, TSPYL1 and endogenous β-catenin. The results showed that TSPYL1 overexpression dramatically reduced β-catenin binding to TCF7L2 (Figure 6). These results suggested that the expression of TSPYL1 might influence the interaction between TCF7L2 and β-catenin, specifically that TSPYL1 might interfere with β-catenin binding to TCF7L2.

A putative bile acid responsive element, BAREII (DR1), is present in the human CYP7A1 gene promoter. TCF7 binds to BAREII to regulate CYP7A1 gene transcription in HepG2 cells based on ENCODE data (see Supplementary Figure S2A). Therefore, we performed a CHIP assay to determine whether TSPYL1 might bind to the BAREII element in the CYP7A1 promoter. Four pairs of primers for PCR of the CYP7A1 promoter region were used. We failed to detect significant TSPYL1 recruitment to the CYP7A1 promoter region in HepaRG cells (see Supplementary Figure S2B).

The effect of the WNT/β-catenin pathway on the regulation of the expression of major human P450 enzymes in HepaRG cells has been studied extensively (Thomas et al., 2015), but CYP1B1 had not been studied in that context. To obtain additional information on the effect of the WNT/β-catenin pathway in hepatic cells, we studied an additional cell line, HepG2. To identify CYP genes that were transcriptionally regulated by Wnt, we determined the effect of Wnt pathway activation or inhibition on the expression of a series of CYPs in HepG2 cells. As a first step, we compared the effect of β-catenin siRNA and control siRNA knockdown on CYP expression. siRNA-mediated knockdown of β-catenin resulted in an approximate 75% decrease in the expression of β-catenin (CTNNB1) mRNA and its downstream target genes (see Supplementary Figure S3C). We also tested the effect of stimulation with the canonical Wnt ligand, Wnt-3a. Wnt-3a treatment significantly induced the expression of Wnt signaling downstream genes compared with BSA treatment as a control (see Supplementary Figure S3D). Changes in CYP mRNA expression in HepG2 cells were also determined. Specifically, PCR analysis revealed that CYP1B1, CYP2A6 and CYP2B6 were transcriptionally downregulated more than 2-fold by β-catenin knockdown in HepG2 cells (see Supplementary Figure S3A), whereas expression of CYP2C9, CYP2C18, CYP3A7, CYP7A1, CYP7B1, and CYP27A1 were upregulated by β-catenin knock down in HepG2 cells (see Supplementary Figure S3A). By contrast, CYP1B1, CYP2A6 and CYP2B6 expression was significantly upregulated after Wnt-3a stimulation, while the expression of CYP2C9, CYP2C19 and CYP3A7 was downregulated more than 2-fold by Wnt activation (see Supplementary Figure S3B). These results support the conclusion that the expression of CYP1B1, together with that of other CYPs, is regulated by Wnt/β-catenin signaling in HepG2 cells.

We had demonstrated that TSPYL1 negatively regulated CYP1B1 expression by blocking β-catenin binding to TCF7L2 on the CYP1B1 promoter. After β-catenin knockdown, CYP1B1 expression was down regulated in HepaRG cells (Figures 7A,B). In summary, TSPYL1 and β-catenin knockdown regulated CYP1B1 expression in opposite directions.

Therefore, we anticipated that by inhibiting the activity of the Wnt/β-catenin pathway, we could “rescue” TSPYL1 knockdown-dependent CYP1B1 upregulation. To test that hypothesis, we conducted a double knockdown experiment. Our PCR results supported the hypothesis that CYP1B1 expression could be rescued by β-catenin and TSPYL1 double knockdown (Figure 7C). 20-HETE, a downstream metabolite of CYP1B1, was also restored by TSPYL1 and CTNNB1 double knockdown (Figure 7D). These results supported the conclusion that TSPYL1 regulates CYP1B1 expression through Wnt/β-catenin signaling.

Transgenic mice overexpressing CYP7A1 in the liver have been reported to be resistant to high-fat diet induced obesity, fatty liver disease and diabetes (Li et al., 2010), while decreased CYP1B1 expression has been correlated with altered lipid metabolism, especially lysophosphatidylcholines, contributing to protection against the development of obesity (Li et al., 2014). Since TSPYL1 knockdown in hepatic cells resulted in increased CYP1B1 and decreased CYP7A1 expression, we asked whether TSPYL1 expression might be associated with obesity and/or plasma cholesterol concentrations in humans. As an initial step toward systematically testing the hypothesis that hepatic TSPYL1 expression might be associated with obesity, we examined publically available data from two clinical studies addressing obesity in which hepatic mRNA expression had been determined. The first trial entitled “A Biological Atlas of Severe Obesity (Biological Tissue Collection (ABOS)” (ClinicalTrials.gov Identifier: NCT01129297) involved 897 obese patients. Figure 8A shows the results of transcriptomic profiling of hepatic tissue from those 897 obese patients (Mean BMI 46.7). Among those patients, 173 received statin treatment, and differential expression analysis was performed between samples from patients who were treated with or without statins to test possible statin effects on gene expression. TSPYL1 expression levels for those patients were negatively associated with plasma LDL-cholesterol levels. A separate cohort from the “Human Liver Biopsy of Different Phases of Control to NASH” (GSE48452) study involved 73 severely obese patients with various degrees of non-alcoholic fatty liver disease (NAFLD). Those patients also had transcriptomic profiling of hepatic tissue, and hepatic TSPYL1 expression in this group was negatively associated with BMI (Figure 8B). Both of these datasets were analyzed using non-parametric two-group testing (Mann-Whitney test). These two examples are, of course, merely suggestive, and studies designed to specifically test hypotheses arising from the present series of studies will have to be performed in the future.