Activation of canonical Wnt signaling is required for efficient direct reprogramming into human hepatic progenitor cells

Summary

Direct reprogramming is a technique for elucidating the mechanisms that control cell-fate decisions and holds promise as a therapeutic strategy. We previously showed that a specific combination of three transcription factors (FOXA3, HNF1A, and HNF6) can induce direct reprogramming of human umbilical vein endothelial cells (HUVECs) into human induced hepatic progenitor cells (hiHepPCs). However, low reprogramming efficiency limits their application in research and therapy. Here, we show that activation of the canonical Wnt signaling pathway increases the reprogramming efficiency of HUVECs to hiHepPCs by rapidly inducing chromatin remodeling and gene expression changes in the transduced HUVECs. Moreover, endogenous Wnt activation, mainly mediated by WNT2B, is required for the initiation of direct reprogramming from HUVECs to hiHepPCs. Wnt activation that allows rapid induction of hiHepPCs enables efficient production of a large amount of hiHepPCs, which is an advantage in research and clinical applications using hiHepPCs and their descendants.

Highlights

• •Wnt activation increases reprogramming efficiency into hepatic progenitor cells
• •Wnt activation rapidly induces chromatin remodeling and gene expression changes
• •Endogenous Wnt activation required for reprogramming is mediated primarily by WNT2B

In this article, Suzuki and colleagues demonstrate that activation of the canonical Wnt signaling improves the efficiency of direct reprogramming into human hepatic progenitor cells (hHepPCs) by rapidly inducing chromatin remodeling and gene expression changes. They also show that endogenous Wnt activation via WNT2B during the early stages of reprogramming is essential for the progressive induction of hiHepPCs.

Introduction

Recent advances have enabled the direct conversion of a type of somatic cell from another type of somatic cell without passing through a pluripotent state by the forced expression of defined transcription factors. This is generally called direct reprogramming and considered not only a novel technology for elucidating mechanisms controlling cell-fate decision but also a promising therapeutic strategy for the treatment of diseases (Horisawa and Suzuki, 2020, 2023; Wang et al., 2021).

Previously, we found that forced expression of three transcription factors (FOXA3, HNF1A, and HNF6) can induce conversion of human umbilical vein endothelial cells (HUVECs) and peripheral blood-derived endothelial cells into cells resembling human hepatic progenitor cells (hHepPCs) in vitro (Inada et al., 2020). These induced hHepPCs (hiHepPCs) propagate in long-term monolayer culture and give rise to functional hepatocytes and cholangiocytes by forming cell aggregates and cystic epithelial spheroids, respectively, in three-dimensional (3D) culture. When hiHepPC-derived hepatocytes are transplanted into the liver of a mouse model of acute liver failure with a high mortality rate (20% survival rate one week after injury), they rebuild and function as human liver parenchymal tissue in the recipient mouse liver without any tumor formation, showing a high lifesaving effect (80% survival rate one week after injury). Moreover, hiHepPC-derived cholangiocytes engrafted in the liver of mice treated with 3,5-diethoxycarbonyl-1,4-dihydrocollidine and formed ductal structures composed of biliary epithelial cells. Thus, hepatocytes and cholangiocytes differentiated from hiHepPCs have potential applications in cell transplantation, drug discovery research, and pathological analysis as alternatives to hepatocytes and cholangiocytes derived from the liver.

hiHepPCs have advantages in the development of biological and medical research on the liver, because they can produce a large number of hepatocytes and cholangiocytes repeatedly while maintaining high cell proliferation capacity over a long period of time. However, the low induction efficiency of hiHepPCs is still a major problem when considering their use. If induction efficiency of hiHepPCs can be increased, a large number of hiHepPCs can be produced in a short time, and hepatocytes and cholangiocytes obtained from them can be used in large quantities. Thus, in this study, we sought to find small-molecule compounds that can improve the reprogramming efficiency of HUVECs to hiHepPCs.

Our present data demonstrate that activation of the canonical Wnt signaling pathway increases the efficiency of direct reprogramming from HUVECs to hiHepPCs by rapidly inducing chromatin remodeling and gene expression changes in transduced HUVECs. Moreover, we found that the early stages of direct reprogramming from HUVECs to hiHepPCs requires Wnt activation, and that WNT2B expressed in the transduced HUVECs is the main cause of such endogenous Wnt activation.

Results

Activators of the canonical Wnt signaling pathway increase reprogramming efficiency of HUVECs to hiHepPCs

To screen for small-molecule compounds improving the reprogramming efficiency of HUVECs to hiHepPCs, we selected 10 candidate compounds that are known to affect cellular reprogramming (Chen et al., 2011; Duan et al., 2019; Guan et al., 2022; Hou et al., 2013; Ichida et al., 2009; Lluis et al., 2008; Xu et al., 2013; Zhu et al., 2010). The small-molecule compounds used in this study are the MEK inhibitors PD98059 and PD0325901, the glycogen synthase kinase-3 inhibitors (also known as activators of canonical Wnt signaling) CHIR99021 and BIO, the retinoic acid receptor agonist TTNPB, the bone morphogenetic protein signaling inhibitor LDN193189, the phosphatidylinositol 3-kinase inhibitor LY294002, the S-adenosyl-l-homocysteine hydrolase inhibitor DZNep, the histone deacetylase inhibitor VPA, and the transforming growth factor β (TGF-β) receptor I/activin receptor-like kinase 5 inhibitor RepSox.

HUVECs infected with retroviruses expressing hiHepPC-inducing factors, such as FOXA3, HNF1A, and HNF6, were cultured with or without each small-molecule compound, and the number of hiHepPC colonies formed by albumin⁺ (ALB⁺) cells was counted at 9 and 16 days after retrovirus infection (Figure 1A). ALB⁺ colonies were only observed in culture with CHIR99021 or BIO after 9 days of culture, and the number of colonies were increased further in the following week (Figures 1B and 1C). Flow cytometric analysis also revealed a significant increase in hiHepPCs expressing the epithelial cell marker E-cadherin in culture with CHIR99021 (Figure S1). On the other hand, other small-molecule compounds had little positive effect on colony formation compared to cultures without any compounds, and in fact had more negative effects (Figures 1B and 1C). CHIR99021 has the function of enhancing the progenitor state of hiHepPCs and facilitating their expansion, while exerting little effect on enhancing their differentiation potential into hepatocytes and cholangiocytes (Figure S2). These data demonstrated that activation of the canonical Wnt signaling pathway specifically increases the efficiency of direct reprogramming from HUVECs to hiHepPCs.

Figure 1.

CHIR99021 and BIO increase the reprogramming efficiency of HUVECs to hiHepPCs

(A) Schematic diagram of the experimental procedure. P, passage number of HUVECs after infection with retroviruses expressing hiHepPC-inducing factors. Cells (1 × 10⁵) were passaged in wells of 6-well plates.

(B and C) Immunofluorescence staining of ALB was conducted for transduced HUVECs in culture with or without the indicated small-molecule compounds 9 and 16 days after retrovirus infection (B), and the number of ALB⁺ colonies formed in each well of 6-well plates were counted (C). DNA was stained with DAPI. Scale bars, 100 μm. Data represent the mean ± SD (n = 3 independent experiments). Statistical difference was determined by one-way ANOVA followed by Dunnett’s test. ^∗p < 0.05 and ^∗∗p < 0.01 (vs. control).

Gene expression changes and chromatin remodeling in the reprogramming of HUVECs to hiHepPCs proceeds more rapidly in culture with CHIR99021

To investigate the effects of Wnt activation on gene expression changes and chromatin remodeling during the reprogramming of HUVECs to hiHepPCs, we conducted 3ʹ untranslated region sequencing (3ʹUTR-seq), a method of RNA sequencing, and assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) for HUVECs and those transduced with hiHepPC-inducing factors in culture with or without CHIR99021 at 9, 16, and 44 days after retrovirus infection (Figure 2A). 3ʹUTR-seq revealed that genes related to hepatic functions associated with various metabolic processes and HUVEC-enriched genes were rapidly up- and downregulated, respectively, in culture with CHIR99021 during the direct reprogramming from HUVECs to hiHepPCs (Figure 2B). ATAC-seq analysis demonstrated that the chromatin at the regulatory regions of the ALB gene underwent earlier and more substantial opening in the presence of CHIR99021 (Figure S3).

Figure 2.

CHIR99021 induces rapid gene expression changes in the reprogramming of HUVECs to hiHepPCs

(A) Schematic diagram of the experimental procedure. P, passage number of HUVECs after infection with retroviruses expressing hiHepPC-inducing factors. Cells (5 × 10⁴) were passaged in wells of 12-well plates.

(B) The genes up- or downregulated after the induction of hiHepPCs from HUVECs were extracted from 1,000 differentially expressed genes (DEGs) that were identified between HUVECs and those transduced with hiHepPC-inducing factors in culture with CHIR99021 at 44 days after retrovirus infection. Heatmap images from 3ʹUTR-seq data showing change in the expression of up- or down-regulated genes during conversion of HUVECs to hiHepPCs in culture with or without CHIR99021 9, 16, and 44 days after retrovirus infection. Top ten most significantly enriched biological processes associated with up- or downregulated genes are shown.

Next, we identified the differentially accessible regions (DARs) between HUVECs and those transduced with hiHepPC-inducing factors in culture with or without CHIR99021 by comparing the signal values of ATAC-seq data. The number of DARs acquired or lost in transduced HUVECs, designated DAR (gain) or DAR (lose), respectively, was larger when cultured with CHIR99021 compared to parental HUVECs, especially on day 9 of culture (Figure S4A). Enrichment analyses revealed that gene ontology (GO) terms associated with liver development, material transport, and metabolic processes were enriched among DAR (gain), while HUVEC-related GO terms were enriched among DAR (lose) (Figure S4B). On the other hand, the number of DARs between transduced HUVECs cultured with or without CHIR99021, designated DAR (+C) or DAR (–C), respectively, was higher in those with CHIR99021 at 9 days of culture, but higher in those without CHIR99021 at 16 and 44 days of culture (Figure 3A). Among DAR (+C), liver-related GO terms associated with material transport, bile acid secretion, and metabolic processes were enriched (Figure 3B). However, even in transduced HUVECs, HUVEC-related GO terms were enriched among DAR (–C) (Figure 3B). In the comparison of HUVECs and those transduced with hiHepPC-inducing factors, HUVEC-related GO terms were enriched among DAR (lose) in the same way whether CHIR99021 was in the medium or not (Figure S4B). Thus, it is suggested that CHIR99021 works to more effectively close chromatin structures associated with the expression of HUVEC-enriched genes.

Figure 3.

CHIR99021 induces rapid chromatin remodeling in the reprogramming of HUVECs to hiHepPCs

(A) MA plot images from ATAC-seq data showing the change in DAR (–C) and DAR (+C) during conversion of HUVECs to hiHepPCs in culture with or without CHIR99021 9, 16, and 44 days after retrovirus infection.

(B) Top ten most significantly enriched biological processes related to the DAR-associated genes.

Taken together, our data demonstrated that activation of the canonical Wnt signaling pathway in the reprogramming of HUVECs to hiHepPCs enables prompt up- and downregulation of liver- and HUVEC-related genes, respectively, by rapidly inducing the opening and closing of chromatin structures in specific regions.

Wnt activation has a significant impact on the early stages of direct reprogramming from HUVECs to hiHepPCs

To identify when Wnt activation has an effect on hiHepPC induction, HUVECs transduced with hiHepPC-inducing factors were cultured with CHIR99021 for 2, 4, 6, and 9 days after retrovirus infection, and the number of hiHepPC colonies formed by ALB⁺ cells was counted on day 9 of culture (Figure 4A). The results showed that the number of ALB⁺ colonies increased significantly when CHIR99021 was supplemented with the culture medium for at least 4 days, but did not increase any more than that when CHIR99021 was added for more than 4 days (Figures 4B and 4C).

Figure 4.

Wnt activation is effective in the early stages of direct reprogramming from HUVECs to hiHepPCs

(A) Schematic diagram of the experimental procedure. P, passage number of HUVECs after infection with retroviruses expressing hiHepPC-inducing factors. Cells (1 × 10⁵) were passaged in wells of 6-well plates. EC-1 to 5 indicate experimental cases 1–5.

(B and C) Immunofluorescence staining of ALB was conducted for transduced HUVECs in culture with or without CHIR99021 9 days after retrovirus infection (B), and the number of ALB⁺ colonies formed in each well of 6-well plates were counted (C). DNA was stained with DAPI. Scale bars, 100 μm. Data represent the mean ± SD (n = 3 independent experiments). Statistical difference was determined by one-way ANOVA followed by Turkey’s HSD test. ^∗∗p < 0.01.

We next examined whether Wnt activation is required for the reprogramming of HUVECs to hiHepPCs. To this end, HUVECs transduced with hiHepPC-inducing factors were cultured with the Wnt production inhibitor IWP2 for 2, 4, 6, and 16 days after retrovirus infection to block endogenous Wnt activation, and the number of ALB⁺ colonies was counted on day 16 of culture (Figure 5A). The data showed that the number of ALB⁺ colonies decreased significantly when IWP2 was supplemented with the culture medium for at least 2 days, and colony formation was almost completely prevented when IWP2 was added for more than 4 days (Figures 5B and 5C).

Figure 5.

Wnt activation is required for direct reprogramming from HUVECs to hiHepPCs

(A) Schematic diagram of the experimental procedure. P, passage number of HUVECs after infection with retroviruses expressing hiHepPC-inducing factors. Cells (1 × 10⁵) were passaged in wells of 6-well plates. EC-6 to 10 indicate experimental cases 6–10.

(B and C) Immunofluorescence staining of ALB was conducted for transduced HUVECs in culture with or without IWP2 16 days after retrovirus infection (B), and the number of ALB⁺ colonies formed in each well of 6-well plates were counted (C). DNA was stained with DAPI. Scale bars, 100 μm. Data represent the mean ± SD (n = 3 independent experiments). Statistical difference was determined by one-way ANOVA followed by Turkey’s HSD test. ^∗∗p < 0.01.

Taken together, our data demonstrated that the initial stage of direct reprogramming from HUVECs to hiHepPCs requires Wnt activation, and the addition of CHIR99021 at that time promotes the reprogramming efficiency.

WNT2B plays a central role in endogenous Wnt activation during conversion of HUVECs to hiHepPCs

Our present data showed that endogenous Wnt activation is a critical event in the early stages of direct reprogramming from HUVECs to hiHepPCs. To identify specific Wnt ligands secreted from HUVECs after transduction with hiHepPC-inducing factors, we conducted 3ʹUTR-seq for HUVECs and those transduced with hiHepPC-inducing factors at 4 and 8 days after retrovirus infection (Figure 6A). 3ʹUTR-seq revealed that, in addition to upregulation of genes related to protein modification and cell migration and downregulation of HUVEC-related and cell division-related genes, genes associated with the regulation of Wnt signaling pathway were enriched in transduced HUVECs compared to parental HUVECs (Figures 6B and S5). In these Wnt-related genes, more than 80% of genes were identified as genes associated with the canonical Wnt signaling pathway (Figure S6). These data supported the importance of canonical Wnt signaling activation in the early stages of direct reprogramming from HUVECs to hiHepPCs that we found in this study.

Figure 6.

Activation of the endogenous Wnt signaling pathway in the early stages of hiHepPC induction

(A) Schematic diagram of the experimental procedure. P, passage number of HUVECs after infection with retroviruses expressing hiHepPC-inducing factors. Cells (1 × 10⁵) were passaged in wells of 6-well plates.

(B) Heatmap image from 3ʹUTR-seq data showing the genes significantly differentially expressed among HUVECs and those transduced with hiHepPC-inducing factors at 4 and 8 days after retrovirus infection. GO term enrichment analysis (biological process) was performed for genes with expression levels higher in transduced HUVECs than in parental HUVECs.

The signal values of 3ʹUTR-seq data indicated that WNT2B, encoding a canonical Wnt ligand, was expressed in HUVECs and that only its expression level increased after induction of the reprogramming of HUVECs to hiHepPCs (Figure 7A). In fact, WNT2B is included in the top cluster of 196 transcripts shown in the heatmap in Figure 6B, whose expression levels were increased by direct reprogramming into hiHepPCs. Thus, we next examined whether WNT2B is required for the direct reprogramming from HUVECs to hiHepPCs. To this end, we used CRISPR interference (CRISPRi) to repress the expression of WNT2B in HUVECs after transduction with hiHepPC-inducing factors (Figure 7B). To suppress WNT2B expression immediately after the transduction of HUVECs, we reduced the number of retrovirus infections to one and performed CRISPRi-based WNT2B repression from 2 days after retrovirus infection. It took longer for ALB⁺ cells to emerge because one-time retrovirus infection reduced the reprogramming efficiency. Quantitative polymerase chain reaction (qPCR) analyses revealed that the expression levels of WNT2B and ALB, but not those of genes encoding other canonical Wnt ligands expressed in HUVECs and those transduced with hiHepPC-inducing factors, were significantly decreased in transduced HUVECs under the CRISPRi-based WNT2B repression at 16, 23, and 30 days after retrovirus infection (Figures 7C and S7). Moreover, a significant decrease in the number of ALB⁺ cells was observed when WNT2B expression was suppressed (Figure 7D). Thus, our data demonstrated that endogenous Wnt activation is mainly caused by WNT2B expressed in the transduced HUVECs, which is required for the induction of direct reprogramming from HUVECs to hiHepPCs.

Figure 7.

WNT2B secreted from transduced HUVECs is critical for the induction of hiHepPCs

(A) The signal values of 3ʹUTR-seq data showing the expression levels of genes encoding canonical and non-canonical Wnt ligands in HUVECs and those transduced with hiHepPC-inducing factors at 4 and 8 days after retrovirus infection.

(B) Schematic diagram of the experimental procedure. Cells (5 × 10⁴) were passaged in wells of 12-well plates.

(C) qPCR analyses of the expression of WNT2B and ALB in HUVECs transduced with hiHepPC-inducing factors at 16, 23, and 30 days after retrovirus infection. Lentivirus-mediated CRISPRi was performed 2 days after retrovirus infection. All data were normalized to the values for CRISPRi (mock)-conducted transduced HUVECs 16 days after retrovirus infection and are depicted as fold-changes.

(D) Immunofluorescence staining of ALB was performed for CRISPRi-conducted transduced HUVECs 30 days after retrovirus infection. Graph at right depicts percentages of ALB⁺ cells. DNA was stained with DAPI. Scale bars, 100 μm. Data represent the mean ± SD (n = 3 independent experiments). Statistical difference was determined by two-sided Student’s t test. ^∗p < 0.05 and ^∗∗p < 0.01.

Discussion

Here, we show that activation of the canonical Wnt signaling pathway improves reprogramming efficiency of HUVECs to hiHepPCs. In particular, CHIR99021 exhibits a higher improvement in the reprogramming efficiency than BIO. In cultures with CHIR99021, HUVECs transduced with hiHepPC-inducing factors more quickly up- or downregulate transcription of liver- or HUVEC-related genes, respectively, compared to cultures without CHIR99021, by inducing more rapid opening and closing of chromatin structures in specific regions. Moreover, we found that endogenous Wnt activation during the early stages of direct reprogramming of HUVECs to hiHepPCs is a fundamental incident for the progressive induction of ALB⁺ hiHepPCs from HUVECs. In this case, among the WNT gene family, only WNT2B increases gene expression levels and acts as an essential factor for endogenous Wnt activation. Thus, Wnt activation is required for the direct reprogramming from HUVECs to hiHepPCs, and additional Wnt activation by small-molecule activators for canonical Wnt signaling improves the reprogramming efficiency of HUVECs to hiHepPCs.

In combination with other small molecules, CHIR99021 positively affects cellular reprogramming for induction of human pluripotent cells (Guan et al., 2022), neuronal cells (Hu et al., 2015), and cardiomyocytes (Cao et al., 2016). In studies on the liver, CHIR99021 facilitates hepatic reprogramming of mouse embryonic fibroblasts in cooperation with the TGF-β type I receptor inhibitor A-83-01 and BMP4 (Lim et al., 2016). Also, a chemical cocktail of CHIR99021 and A-83-01 with or without the Rho-associated kinase inhibitor Y-27632 allowed dedifferentiation of adult mouse or human infant hepatocytes, respectively, into hepatic progenitor-like cells in culture (Katsuda et al., 2017, 2019). These findings suggest that the canonical Wnt signaling pathway activated by CHIR99021 is essentially necessary for inducing the cell-fate conversion, which is also supported by our present data. However, the roles of activation of the canonical Wnt signaling pathway during cellular reprogramming are still largely unknown. In this study, we showed that CHIR99021 induces rapid chromatin remodeling and gene expression changes in HUVECs transduced with hiHepPC-inducing factors and increases the reprogramming efficiency of HUVECs to hiHepPCs. Thus, in other cases of cellular reprogramming, activation of the canonical Wnt signaling pathway is suggested to play a role similar to that of hiHepPC induction.

We also showed that endogenous Wnt activation, mainly mediated by WNT2B, is required for the initiation of direct reprogramming from HUVECs to hiHepPCs. Upregulation of WNT2B, induced by transduction of HUVECs with hiHepPC-inducing factors, is observed before conversion of HUVECs to ALB⁺ hiHepPCs. Thus, transduced ALB⁻ HUVECs in the early stages of direct reprogramming may overexpress WNT2B themselves to assist conversion to ALB⁺ hiHepPCs. hiHepPC-inducing factors contribute both internally and externally to cell-fate conversion by not only regulating the expression of genes that characterize the types of cells, but also inducing the expression of secreted factors that are necessary for reprogramming.

The canonical Wnt signaling activated by addition of CHIR99021 to culture medium enables rapid induction of hiHepPCs from HUVECs. Thus, CHIR99021 can be used for producing large numbers of hiHepPCs more quickly, which is advantageous for basic research and clinical applications using hiHepPCs and hiHepPC-derived hepatocytes and cholangiocytes.

Methods

Cell culture

Cell culture was performed as described previously (Inada et al., 2020). Briefly, HUVECs (PromoCell) were cultured in HUVEC medium (1:1 mixture of Medium 200 [Thermo Fisher Scientific], supplemented with Low Serum Growth Supplement [Thermo Fisher Scientific], and FibroLife S2 Comp kit [Kurabo]). hiHepPCs were grown in hepato-medium (plus) composed of a 1:1 mixture of Dulbecco’s modified Eagle’s medium (DMEM) and F-12 (Nacalai Tesque), supplemented with 20% FibroLife S2 Comp kit, 4% fetal bovine serum, 1 μg mL⁻¹ insulin (Wako), 10⁻⁷ M dexamethasone (Sigma-Aldrich), 10 mM nicotinamide (Sigma-Aldrich), 2 mM L-glutamine (Nacalai Tesque), 50 μM β-mercaptoethanol (Nacalai Tesque), 20 ng mL⁻¹ recombinant human hepatocyte growth factor (PeproTech), 1 μM A-83-01 (Tocris), 2 μM SB431542 (Tocris), 5 μM Y-27632 (Wako), and penicillin/streptomycin (Nacalai Tesque). Small-molecule compounds, such as 10 μM PD98059 (Cayman), 1 μM PD0325901 (Cayman), 3 μM CHIR99021 (Tocris), 1 μM TTNPB (Sigma-Aldrich), 100 nM LDN193189 (Selleck Chemical), 5 μM LY294002 (Chemscene), 50 nM DZNep (Abcam), 0.5 mM VPA (Sigma-Aldrich), 1 μM BIO (Abcam), 10 μM RepSox (R&D Systems), and 5 μM IWP2 (Sigma-Aldrich), were supplemented with the hepato-medium (plus). Differentiation of hiHepPCs into hepatocytes and cholangiocytes was induced under 3D culture conditions as described previously (Inada et al., 2020).

Retrovirus production and transduction of cells

Production of retroviruses expressing hiHepPC-inducing factors and transduction of HUVECs were conducted as described previously (Inada et al., 2020).

Immunostaining

Immunofluorescence staining of ALB was conducted as described previously (Sekiya and Suzuki, 2011) for HUVECs after transduction with hiHepPC-inducing factors using a Goat anti-Human ALB primary antibody (Bethyl) and an Alexa Fluor 488-conjugated donkey anti-goat IgG secondary antibody (Invitrogen).

Flow cytometry

HUVECs transduced with hiHepPC-inducing factors were trypsinized, fixed using a Transcription Factor Buffer Set (BD Biosciences), and stained with an Alexa Fluor 647-conjugated mouse anti-E-cadherin antibody (BD Biosciences). The immunostained cells were analyzed using the Cell Sorter SH800S (Sony Biotechnology).

Gene expression analysis

Total RNA was isolated from HUVECs and those transduced with hiHepPC-inducing factors using the ISOGEN II kit (Nippon gene) and the RNeasy Mini kit (QIAGEN) according to the manufacturer’s instructions and analyzed by 3ʹUTR-seq (Goya et al., 2022; Hashimshony et al., 2016; Inada et al., 2020)), which is a modified version of the single-cell RNA-seq technique CEL-seq2 for bulk transcriptome analysis, with cell barcodes added for each sample for demultiplexing data. The raw data were demultiplexed and converted into unique molecular identifier (UMI) counts with the celseq2 pipeline (Hashimshony et al., 2016). Based on the UMI count, transcript per million (TPM) for each gene was calculated using the following formula:

$$TP{M}_i=\frac{-4096\times \left(\mathit{ln}\left(4096-UM{I}_i\right)-\mathit{ln}4096\right)}{{\sum }_j\left[-4096\times \left(\mathit{ln}\left(4096-UM{I}_j\right)-\mathit{ln}4096\right)\right]}\times {10}^6$$

Heatmaps of differentially expressed genes between HUVECs and hiHepPCs were generated using heatmap.2 program in gplots package (v.3.1.3.1) based on normalized UMI count values obtained from iDEP program (v.0.96) (Ge et al., 2018) with default parameters, and also the genes were clustered with k-means algorithm on the iDEP program. For the enrichment analysis, DAVID program (v.6.8) (Huang et al., 2009) and the GO biological process annotation database were employed. All datasets were deposited in the Gene Expression Omnibus (GEO) database under accession number GEO: GSE279106. For the analysis shown in Figure 6B, we used our previously reported datasets (accession number GEO: GSE120732) to analyze the gene expression data of HUVECs. TPM values of Wnt family genes were calculated from the raw UMI values of the 3ʹUTR-seq data. cDNAs were synthesized from the total RNA as described previously (Suzuki et al., 2000), and qPCR was performed using the THUNDERBIRD SYBR qPCR Mix (Toyobo) according to the manufacturer’s instructions. qPCR primers for WNT2B, ALB, AFP, DLK1, CYP3A4, CYP2C9, WNT3, WNT9A, CK19, SOX9, and GAPDH are listed in Table S1. The values for GAPDH were used as normalization controls.

ATAC-seq

The cultured cells were collected at specified time points by trypsinization. Omni ATAC-seq was conducted with the standard protocol using 50,000 cells per reaction (Corces et al., 2017). The collected genomic DNA fragments were purified using QIAquick PCR purification kit (QIAGEN), and indexing and amplification of them were performed with the NEBNext High-Fidelity 2× PCR Master Mix (New England Biolabs). The amplified library was then purified and size-selected using AMPure XP (Beckman Coulter). The raw sequences were initially trimmed using Trim Galore (v.0.6.5) (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Subsequently, the trimmed sequences were aligned to the reference genome (GRCh38) with bowtie2 (v.2.3.5.1) (Langmead and Salzberg, 2012). Mitochondrial genome-derived reads and PCR duplicates were removed using samtools (v.1.14) (Li et al., 2009) and Picard (v.2.21.4) (https://broadinstitute.github.io/picard/), respectively. Minus-average (MA) plots of DARs were generated using DiffBind program (v.3.12.0) (Ross-Innes et al., 2012). For the enrichment analysis, GREAT webtool (v.4.0.4) (Tanigawa et al., 2022) and the GO biological process annotation database were employed. All datasets were deposited in the GEO database under accession number GEO: GSE279128.

CRISPRi-mediated knock down of WNT2B expression

A single-guide RNA (sgRNA) for WNT2B was designed using the CRISPR design tool CRISPOR (Concordet and Haeussler, 2018). The sgRNA oligonucleotides (forward: 5′-CAC CGG AGG CGG CTG CTA CAC CTA-3′ and reverse: 5′-AAA CTA GGT GTA GCA GCC GCC TCC-3′) were annealed, and the annealed sgRNA oligonucleotides were inserted into a lentiviral vector (lentiCRISPR v.2, Addgene). To knock down the WNT2B expression, HUVECs were transduced with the lentiviral vector containing the sgRNA for WNT2B.

Lentivirus production and transduction of cells

To produce recombinant lentiviruses, plasmid DNA was transfected into 293T cells (a gift from H. Miyoshi) using linear polyethylenimine (PEI) (Polysciences). At 3 days before transfection, 293T cells (1.6 × 10⁶) were plated on poly-L-lysine-coated 10-cm dishes and cultured in a medium for mouse embryonic fibroblast (MEF) (Sekiya and Suzuki, 2011). Next, 48 μL of 1 mg/mL PEI, 20 μg of lentiviral plasmid DNA, 15 μg of psPAX2 (Addgene), and 6 μg of pMD2.G (Addgene) were diluted in 1 mL of DMEM and incubated for 15 min at room temperature. The mixtures were added to the plated 293T cells in a drop-by-drop manner. After 16 h of incubation at 37°C under 5% CO₂, the medium was replaced with fresh medium for MEF, and the culture was continued. Supernatants from the transfected cells were collected at 24 h and 48 h after medium replacement, filtered through 0.2 μm cellulose acetate filters (Sartorius), and concentrated by centrifugation (10,000 × g for 16 h at 4°C). The virus pellets were resuspended in Hanks’ balanced salt solution (Nissui) at 1/520 of the initial supernatant volume, and stored at −80°C until use. HUVECs were incubated for 3 h in HUVEC medium containing the concentrated viral supernatants with 4 μg/mL polybrene (Nacalai Tesque).

Colony formation assay

hiHepPCs were plated on type I collagen-coated 6-well plates at 100 cells/well. At 4 days after plating, hiHepPCs were washed with phosphate-buffered saline (PBS) and fixed with 10% Formalin (Nacalai Tesque). Fixed cells were washed with PBS and stained with 0.01% crystal violet (Nacalai Tesque) at room temperature for 1 h. Following staining, the cells were washed with double distilled water and air-dried.

Statistics and reproducibility

Statistical significance was analyzed using two-sided Student’s t test, one-way analysis of variance (ANOVA) followed by Dunnett’s test, one-way ANOVA followed by Tukey’s honestly significant difference (HSD) test, quasi-likelihood F-test, Fisher’s exact test, binomial and hypergeometric enrichment tests, and permutation test. A difference at p < 0.05 was considered statistically significant. Cell and tissue images presented in the figures and supplementary figures were obtained from at least three independent experiments, and representative images are shown.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Atsushi Suzuki (suzukicks@bioreg.kyushu-u.ac.jp).

Materials availability

All the materials generated and used in this study will be available upon reasonable request.

Data and code availability

All datasets were deposited in the GEO database under accession numbers GEO: GSE279106 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE279106) and GSE279128 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE279128). The publicly available datasets (accession number GEO: GSE120732 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE120732]) were also used in this study.

Acknowledgments

We thank Drs. Toshio Kitamura, Masafumi Onodera, Hiroyuki Miyoshi, Yasuyuki Ohkawa, and Hitoshi Kurumizaka for sharing reagents and Mariko Tasai, Yoshimi Iwasaki, Mitsuhiro Kurata, Yuuki Honda, Kanako Ichikawa, Ryo Ugawa, and Emiko Koba for excellent technical assistance. This work was supported in part by the JSPS KAKENHI (grant numbers: JP21K18039 and JP23K17199 to S.M.; JP23K11851 to K.H.; and JP18H05102, JP19H01177, JP19H05267, JP20H05040, JP22H05634, JP22H04698, JP22H00592, JP23K18579, JP25K22908, and JP25H00445 to A.S.), the Program for Basic and Clinical Research on Hepatitis of the Japan Agency for Medical Research and Development (AMED) (JP24fk0210116 and JP25fk0210170 to A.S.), the Research Center Network for Realization of Regenerative Medicine of AMED (JP24bm1123005 to A.S.), the MEXT Promotion of Development of a Joint Usage/Research System Project: Coalition of Universities for Research Excellence Program (JPMXP1323015486 to A.S.) and Cooperative Research Project Program (to A.S.), the Medical Research Center Initiative for High Depth Omics (to A.S.), the Takeda Science Foundation (to S.M. and A.S.), the Uehara Memorial Foundation (to S.M., K.H., and A.S.), the Naito Foundation (to S.M. and A.S.), and the Shinnihon Foundation of Advanced Medical Treatment Research (to S.M.).

Author contributions

S.M., K.H., H.I., Y.S., and M.Y. performed experiments, collected data, and conducted data analyses. A.S. contributed to the conception, design, and overall project management and wrote the paper.

Declaration of interests

A.S. has filed patent applications related to this study as an inventor.

Published: October 30, 2025