lncRNA, miRNA, and mRNA dynamics in cholangiocyte-to-immature hepatocyte differentiation in liver regeneration Authors: Pingping Huang, Xue Yu, Congcong Shen, Zanxia Cao, Bailing Zhou, Ruiping Wang, Jiarong Wang, Jihua Wang Published: 2026 ABSTRACT AbstractNon-coding RNAs play essential roles during liver regeneration, however, their functions in cholangiocyte-to-hepatocyte differentiation is still unclear. Emerging liver ductal organoids offer a powerful model to replicate the molecular mechanisms of the in vivo cholangiocyte-to-immature hepatocyte transition, potentially accelerating advancements in liver regenerative medicine. Here, we established the liver ductal organoids from three C57BL/6 mice, and induced the differentiation of cholangiocyte into immature hepatocyte to carry out the transcriptome sequencing. Then, the expression profiles of lncRNA, miRNA and mRNA in expansion (EM) and differentiation medium (DM) samples were explored. A total of 529,049,578 and 61,266,130 clean reads were generated from 6 RNA-seq and 6 miRNA-seq libraries, respectively. Through stringent filtering, we identified 24,676 lncRNAs, 1,018 miRNAs and 38,689 mRNAs transcripts. This dataset will contribute to the elucidation of the underlying molecular mechanisms associated with cholangiocyte-to-immature hepatocyte differentiation during liver regeneration, and provides clues for liver regenerative medicine.Subject terms: Gene expression, Non-coding RNAs, Transcriptomics Background & Summary The liver is the largest internal organ with high regenerative capacity and complex functions, which can regenerate its parenchyma after acute and chronic injury1. The liver mainly contains two kinds of epithelial cells, including hepatocytes and cholangiocytes. The cellular source for hepatocyte regeneration after parenchyma being injured is a fundamental and hot topic in life science and medicine. The cholangiocytes can function as facultative stem cells and differentiate into hepatocyte to restore the hepatocyte population during the impaired hepatocyte proliferation2,3. For example, in lower vertebrates (such as zebrafish), liver regeneration after severe liver injury is mostly achieved through the differentiation of cholangiocytes into hepatocytes when the hepatocyte proliferation is limited4. Uncovering the molecular mechanisms of how cholangiocytes differentiate into hepatocytes during liver regeneration is important for the liver biology and medicine. However, due to limitations in research methods, traditional experimental approaches are unable to directly investigate the molecular mechanisms involved in this process. Over the past decades, many signaling pathways, protein coding genes and transcription factors correlated with various aspects of liver regeneration after liver injury have been revealed5. For example, peroxisome proliferator-activated receptor α can regulate the liver regeneration and liver size by activating the yes-associated protein (YAP)-transcriptional enhancer factor domain family member (TEAD) signaling pathway6. Silencing the expression of Hippo core kinases MST1 and MST2 helps rescue the liver regeneration in aged mice following two-thirds partial hepatectomy7. Furthermore, the Wnt, TGF-β and Notch signaling pathways are correlated with the cholangiocyte-to-hepatocyte transition in vivo8,9. In addition to the protein coding genes, non-coding RNAs (ncRNAs), such as lncRNA and miRNA, are also widely involved in the regulation of liver biological processes, including liver regeneration. LncRNAs are RNA transcripts with length longer than 200 nt and lack protein-coding capacity, but they may have short open reading frames10. Recent ten years, the functions of lncRNAs had been verified by low-throughput experiments and it had been found that lncRNAs could regulate the expression of genes by interacting with mRNA, DNA, protein, miRNA and so forth11. The functional roles of lncRNAs in liver regeneration are also gradually been revealed. A total of 1,329 differentially expressed lncRNAs (DE lncRNAs) were discovered in the different proliferative phase of liver regeneration in six mice with liver fibrosis12. LncRNA-LALR1 could enhance the hepatocyte proliferation and accelerated cell cycle progression during liver regeneration through activating Wnt/β-catenin signaling by experiments in vitro and in vivo13. LncRNA MALAT1 was also an important molecule for liver regeneration, which could promote the proliferation of hepatocyte by accelerating cell cycle progression14. LncHand2 stimulates liver repopulation and regeneration by inducing the expression of Nkx1-2 and then triggers c-Met signaling15. Moreover, the lncRNA SNHG12 was upregulated at different time-points after 2/3 partial hepatectomy in mice, which could enhance hepatocyte proliferation in vitro and in vivo16. However, the roles of lncRNA in cholangiocyte-to-immature hepatocyte differentiation during liver regeneration are far from known. miRNAs are single-stranded RNAs with length varied from 21 to 25 nucleotides, which can inhibit the expression of mRNAs and regulate the liver regeneration17–19. The former study revealed the function of miR-199a in hepatocyte proliferation during liver regeneration and found that it could accelerate the hepatocyte proliferation by targeting TNF-α signaling pathway20. The miR-194 knockout mice can keep liver regeneration ability and protect the liver from acetaminophen-induced acute liver injury21. In addition, alcoholic hepatitis patients attempt to promote liver regeneration by down-regulating specific miRNAs, such as miR-30b-5p and miR-146a-5p22. The mRNA and lncRNA can communicate to each other by using miRNA response elements according to the competing endogenous RNA (ceRNA) hypothesis23. The ceRNA-based regulatory mechanisms during liver regeneration have also been discovered. For example, a ceRNA regulatory network during proliferative phase of rat liver regeneration was constructed, which contained 32 miRNAs, 107 lncRNAs, and 270 mRNAs24. Overall, both lncRNAs and miRNAs are major regulators of liver regeneration after liver injury. However, the functional roles of them in cholangiocyte-to-immature hepatocyte differentiation are poorly understood. The organoids have been chosen as ‘Method of the year 2017’ by Nature Methods, which provide convenient tools to study molecular and cellular basis of life25. In 3D culture systems, bile ducts isolated from liver can self-organize into long-term-expanding liver ductal organoids, in which the cholangiocytes can differentiate into functional hepatocytes by regulating Wnt, TGF-β and Notch signaling pathways8. Liver ductal organoids thus enable the modeling of the in vivo cholangiocyte-to-hepatocyte transition at the molecular level. This approach overcomes the limitations of traditional 2D culture systems or animal models that cannot directly replicate this complex cellular process. Thus, it provides a more accurate and physiologically relevant system to study cholangiocyte-to-hepatocyte differentiation. In this study, we established the liver ductal organoids from three C57BL/6 mice, and cultivated them sequentially in the expansion (EM) and differentiation medium (DM) (Fig. 1). Then, we collected organoids in EM and DM group and performed RNA and small RNA sequencing, respectively. Critical steps in the experimental pipeline, such as the evaluation of construction and differentiation of liver ductal organoids, assessment of raw and clean reads, correlation analysis of sequencing libraries, and validation of lncRNA/miRNA/mRNA expression levels, received high scores and demonstrated exceptional reliability. This comprehensive dataset is a valuable resource for relevant researchers to exploring the molecular mechanisms underlying cholangiocyte-to-immature hepatocyte differentiation during liver regeneration. The experimental design and workflow of this study. Methods Animals Three 6 to 8-weeks-old C57BL/6 mice were purchased from the Jinan Pengyue Experimental Animal Breeding company. The experimental designs and animal care were conducted in accordance with the Guide of the Care and Use of Laboratory Animals, and approved by the Animal Experimentation Ethics Committee of Dezhou University, Dezhou, China. Establishment of liver ductal organoids and differentiation The establishment of liver ductal organoids and its differentiation was mainly based on the Broutier et al.8 and Hu et al.26. First, liver ductal organoids were established using the bioGenous™ Mouse Liver Ductal Organoid Kit (Expansion) (K2006-MLD) (bioGenous, China) according to the manufacturer’s instructions. Briefly, the liver tissues were isolated from the mice and placed in pre-cooled Primary Tissue Protection Solution K601005 (bioGenous, China), and washed with basic medium (bioGenous, China) for three times. Matrigel (M315066) was thawed on ice, and basal medium supplemented with 1% FBS was prepared. Then, the isolated tissues were cut into small pieces of roughly 0.5 mm3 and were transferred to a 15-ml centrifuge tube and added about 10 ml of ice-cold basic medium. To remove red blood cells and fat, we pipetted the sample up and down at least 10 times using a 10-ml pipette and discarded the supernatant. The pieces were performed with pre-warmed Digestion Medium (K601008) (bioGenous, China). Digestion was performed at 37 °C with gentle shaking (200 rpm) for no more than 30 min, until ductal structures were observed under a microscope. The digestion was stopped by transferring the digested sample to a 50 mL centrifuge tube containing 40 mL of epithelial organoid basal medium with 1% FBS. The tube was allowed to stand for 2 min, and the supernatant was transferred to a new tube. The supernatant was centrifuged at 300 g for 3 min at 4 °C, discarded, and then washed with basic medium for two times. After discarding the supernatant, the pellet was resuspended in Matrigel and seeded into 24-well plates at 25–30 µL per well. The plate was incubated at 37 °C in a 5% CO₂ incubator for 15–25 min to allow Matrigel polymerization. Complete medium (K2006-MLD) (bioGenous, China) was prepared according to the manufacturer’s instructions, and 500 µL of medium was added to each well after Matrigel solidification. The cultures were maintained in a humidified incubator at 37 °C with 5% CO₂. The medium was replaced daily with fresh pre-warmed complete medium. Mouse liver ductal organoids typically formed between days 5 and 8 after seeding. To obtain enough RNA for RNA sequencing, we passaged the liver ductal organoids for three times. For passaging, the medium was removed, and an equal volume of basal medium was added. The Matrigel and organoid mixture was gently scraped off using the tip of a pipette and transferred to a 1.5 mL microcentrifuge tube. The mixture was gently pipetted to dissociate the organoids from the Matrigel and centrifuged at 300 × g for 3–5 min. The supernatant (medium and Matrigel) was removed. Then, 200 µL of Organoid Passage Digestion Solution (E238001) (bioGenous, China) was added. The pipette tip was rinsed with Organoid Recovery Anti-Adhesion Rinse Solution (E238002) (bioGenous, China) before resuspending the organoid pellet. The tube was incubated at 37 °C for 2 min. The digestion was diluted by adding 1 mL of basal medium (5-fold volume dilution) and centrifuged at 300 × g for 3 min. The pellet was washed by resuspending in 1 mL of basal medium followed by centrifugation. The supernatant was discarded, and the pellet was resuspended in an appropriate volume of Matrigel. The resuspended mixture was seeded into 24-well plates at 25 µL per well, and incubated at 37 °C for 20 min to solidify the Matrigel. Complete medium for expansion was prepared by adding 200 µL of Mouse Liver Ductal Organoid Supplement B (50×) and 40 µL of Mouse Liver Ductal Organoid Supplement C (250×) to 9.76 mL of Mouse Liver Ductal Organoid Medium A. After Matrigel solidification, 500 µL of the prepared complete medium was added to each well. The medium was replaced every 3 days with fresh pre-warmed complete medium. Organoids were passaged again when they reached a diameter exceeding 100 µm. For liver organoid differentiation, liver ductal organoids from three mice were cultured in the EM using the bioGenous™ Mouse Liver Ductal Organoid Kit (Expansion) (K2006-MLD) for three days, some of them were used for EM group RNA isolation. The rest liver ductal organoids were successively cultured with differentiation medium (DM) I K2006-MLH (bioGenous, China) for nine days and cultured with DM II for three days, and then these organoids were used for DM group RNA isolation. DM I was prepared by adding 200 µL of Mouse Liver Ductal Organoid Differentiation Supplement B (50×) and 40 µL of Mouse Liver Ductal Organoid Differentiation Supplement C (250×) to 9.76 mL of Mouse Liver Ductal Organoid Differentiation Medium A, followed by thorough mixing. DM II was prepared by adding 200 µL of Mouse Liver Ductal Organoid Differentiation Supplement B (50×), 40 µL of Mouse Liver Ductal Organoid Differentiation Supplement C (250×), and 10 µL of Mouse Liver Ductal Organoid Differentiation Supplement D (1000×) to 9.76 mL of Mouse Liver Ductal Organoid Differentiation Medium A, followed by thorough mixing. The differentiation levels of organoids were examined by using marker gene expression. The typical gene markers used for cholangiocyte identification were Spp1, Krt7 and Sox9, while that for hepatocyte identification were Afp, Apoa1, Sult1a1, Alb, Ttr and Cyp3a11. RNA extraction Total RNA was isolated from three EM and DM organoids by RNAprep Pure Micro Kit (Tiangen, China) according to the manufacturer’s instructions, respectively. The quality of RNA was assessed with NanoDrop2000 and Agilent 5400. The isolated total RNA were reverse-transcribed into cDNA with GoScript™ Reverse Transcription System (Promega, USA). RNA-seq library construction and sequencing Approximately 1 μg of total RNA per sample was used for RNA-seq library preparations. Then, the six strand-specific libraries were constructed by NEBNext® UltraTM RNA Library Prep Kit for Illumina® (NEB, USA) according to the manufacturer’s instructions. Briefly, ribosome RNA was depleted from total RNA. Fragmentation was carried out using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer(5X). First strand cDNA was synthesized with the random hexamer primer and M-MuLV Reverse Transcriptase (RNase H-), while the second strand cDNA was synthesized with the DNA Polymerase I and RNase H. Then, the cDNA fragments with length of 250–300 bp were selected by AMPure XP system (Beckman Coulter, Beverly, USA). Moreover, the USER Enzyme (NEB, USA) was used to digest the second strand with size-selected, adaptor-ligated cDNA and then the PCR was carried out using the Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. Furthermore, the library quality was assessed and then sequenced on the Illumina NovaSeq6000 platform based on the 150 bp paired-end sequencing strategy. High throughput sequencing was performed by Novogene Bioinformatics Technology Co., Ltd. The quality of all raw data was examined by performing FASTQC software (v. 0.11.9). The adaptors, ploy-N or with low quality reads were trimmed. The filtering criteria were as follows: (1) reads containing adapter sequences were removed; (2) reads with more than 10% ambiguous bases (N, representing undetermined bases) were discarded; (3) reads were filtered out if more than 50% of the bases had a Phred quality score ≤ 5 (Qphred ≤ 5). Indexes of the GRCm38 reference genome (https://ftp.ensembl.org/pub/release-102/fasta/mus_musculus/dna/) were built using HISAT2 (v2.0.5) with default parameters27, and the command was hisat2-build reference.fa ref_index. Then, the remaining high-quality clean data was mapped to the GRCm38 reference genome by HISAT2 (v2.0.5) (hisat2 -x ref_index -p 4 --dta -t --phred33 --rna-strandness RF -1 sample_1.clean.fq.gz -2 sample_2.clean.fq.gz --un-conc-gz sample.unmap.fq.gz 2 > sample_align.log | samtools sort -O BAM --threads 4 -o sample.bam -). The mapped reads from per sample was assembled (stringtie sample.bam -p 4 -G genome.gtf -o sample.gtf --rf) and quantified (stringtie -e -p 8 -G all_transcripts.gtf -A sample.genes.xls -o sample.transcript.gtf sample.bam --rf) by StringTie (v1.3.3b)28. The annotation file was downloaded from https://ftp.ensembl.org/pub/release-102/gtf/mus_musculus/. The unknown transcripts were employed to identify lncRNAs. The filtering processes mainly included two steps: (1) the short (≤200 nt) and single-exon transcripts were filtered; and (2) transcripts with protein-coding potential as predicted by one of the three softwares, CPC2 (v3.2.0)29, CNCI (v2.0)30 and Pfam (v1.6)31 were excluded, the remaining transcripts without protein-coding potential made up the candidate set of novel lncRNAs. miRNA library construction and sequencing For small RNA sequencing, approximately 3 μg of total RNA per sample was used for library preparations. Then, the six small RNA libraries were constructed by NEB Next® Multiplex Small RNA Library Prep Set for Illumina® (NEB, USA). Briefly, 3′ and 5′ adaptors were sequentially ligated to the 3′ and 5′ ends of small RNA molecules. First-strand cDNA was then synthesized following hybridization with a reverse transcription primer. Double-stranded cDNA libraries were generated by PCR enrichment. After purification and size selection, libraries containing inserts of 18–40 bp were selected. Library quality was assessed by using Qubit fluorometer and qRT-PCR. Quantified libraries were pooled according to effective concentration and data amount required, and then sequenced on the Illumina NovaSeq6000 platform based on the 50 bp sequencing strategy according to the manufacturer’s instructions. To obtain the clean data, the reads containing poly-N, with 5′ adapter contaminants, without 3′ adapter or the insert tag, containing ploy A or T or G or C and low quality reads were trimmed from the raw data. The small RNA tags were mapped to reference sequence by Bowtie software (v1.0.1) (parameter: -v 0 -k 1)32. Then, the mapped small RNA tags were used to detect known miRNA by miRBase20.0, and to predict novel miRNA by a combination of miREvo (v1.1)33 and mirdeep2 software34. Differential expression analysis Differentially expressed (DE) mRNAs, lncRNAs and miRNAs between DM and EM groups were detected by DESeq2 package (v1.24.0) in R. The resulting P value was adjusted based on the Benjamini and Hochberg’s approach to control the false discovery rate. Genes or transcripts with Padj < 0.05 and |log2 (fold change)| > 1 calculated by Deseq2 were assigned as differentially expressed. Quantitative real-time PCR (qRT-PCR) Expression levels of randomly selected mRNA, miRNAs, lncRNAs and organoids marker genes were validated by qRT-PCR. All primer pairs are listed in Table 1. qRT-PCR reactions were performed with Taq SYBR® Green qPCR Premix (Lablead) on the QuantStudio 5 Real-Time PCR System (Applied Biosystems). Primer sequences for cholangiocyte (Spp1, Krt7 and Sox9) and immature hepatocyte/hepatocyte marker genes (Afp, Apoa1, Sult1a1, Alb, Ttr and Cyp3a11) were synthesized according to Wang et al.35 and Sun et al.36. Each reaction for marker gene contained 2 μl template cDNA, 0.5 μl forward and reverse primer, 10 μl 2X Sybr Green qPCR master Mix and 7 μl nuclease-free water. The reactions were amplified based on the following conditions: 95 °C for 10 min, 40 cycles of 95 °C for 30 s, 58 °C for 30 s, 72 °C for 30 s, and melting curve analysis was performed on 65 °C for 0.5 s and 95 °C for 5 s. Table 1Primers used for qRT-PCR.PrimersSequence (5′-3′)Krt7Forward AGGAGATCAACCGACGCACReverse GTCTCGTGAAGGGTCTTGAGGSox9Forward GCCAGATGGACCCACCAGTATReverse TCCAAACAGGCAGGGAGATTCAlbForward ATGTTACCAAGTGCTGTAGTReverse AATCTGCTTCTCCTTCTCTGTtrForward CTCACCACAGATGAGAAGReverse GGCTGAGTCTCTCAATTCCyp3a11Forward TACTGTGATGGAGATGGAATACReverse GGTGAAGAGCATAAGATGGASpp1Forward AGCAAGAAACTCTTCCAAGCAAReverse GTGAGATTCGTCAGATTCATCCGAfpForward CTTCCCTCATCCTCCTGCTACReverse ACAAACTGGGTAAAGGTGATGGApoa1Forward GGCACGTATGGCAGCAAGATReverse CCAAGGAGGAGGATTCAAACTGSult1a1Forward AACATGGAGCCCTTGCGTAAAReverse ATGAGCACATCATCAGGCCAGGm15889-202Forward GTGCTGGGCGATTGAACGAReverse GGGAGGAACAAGGGCTTAGACAGm42047-202Forward CCAATAACCCACAGGTGACTCTTCReverse GAGCCACCGTTGTCCATCTTTmmu-miR-664-5pReverse Transcription (RT) Primer CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCCAGTCTAForward ACACTCCAGCTGGGCTGGCTGGGGAAAATReverse TGGTGTCGTGGAGTCGmmu-miR-667-5pReverse Transcription (RT) Primer CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCGTGCTCAForward ACACTCCAGCTGGGCGGTGCTGGTGGAGCAGTGReverse TGGTGTCGTGGAGTCG Primers used for qRT-PCR. Forward AGGAGATCAACCGACGCAC Reverse GTCTCGTGAAGGGTCTTGAGG Forward GCCAGATGGACCCACCAGTAT Reverse TCCAAACAGGCAGGGAGATTC Forward ATGTTACCAAGTGCTGTAGT Reverse AATCTGCTTCTCCTTCTCTG Forward CTCACCACAGATGAGAAG Reverse GGCTGAGTCTCTCAATTC Forward TACTGTGATGGAGATGGAATAC Reverse GGTGAAGAGCATAAGATGGA Forward AGCAAGAAACTCTTCCAAGCAA Reverse GTGAGATTCGTCAGATTCATCCG Forward CTTCCCTCATCCTCCTGCTAC Reverse ACAAACTGGGTAAAGGTGATGG Forward GGCACGTATGGCAGCAAGAT Reverse CCAAGGAGGAGGATTCAAACTG Forward AACATGGAGCCCTTGCGTAAA Reverse ATGAGCACATCATCAGGCCAG Forward GTGCTGGGCGATTGAACGA Reverse GGGAGGAACAAGGGCTTAGACA Forward CCAATAACCCACAGGTGACTCTTC Reverse GAGCCACCGTTGTCCATCTTT Reverse Transcription (RT) Primer CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCCAGTCTA Forward ACACTCCAGCTGGGCTGGCTGGGGAAAAT Reverse TGGTGTCGTGGAGTCG Reverse Transcription (RT) Primer CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCGTGCTCA Forward ACACTCCAGCTGGGCGGTGCTGGTGGAGCAGTG Reverse TGGTGTCGTGGAGTCG miRNA expression was quantified using stem-loop reverse transcription followed by SYBR Green qPCR. For each miRNA, a gene-specific stem-loop RT primer was used for reverse transcription with a commercial kit (Servicebio®RT First Strand cDNA Synthesis Kit) in a 20 μL reaction, containing 5 × Reaction Buffer, RT primer mix (1.0 μM each), Servicebio®RT Enzyme Mix, and total RNA. The reaction was incubated at 25 °C for 5 min, 42 °C for 30 min, and 85 °C for 5 sec. Then, qRT-PCR was performed using a specific forward primer (complementary to the miRNA sequence) and a universal reverse primer. U6 was used as an internal reference for quantifying miRNAs. Immunofluorescence Organoids were embedded in paraffin, and sectioned by freezing microtome. Sections were dewaxed, subjected to antigen retrieval for 30 min, and blocked with 5% BSA. Sections were incubated with rabbit anti-Albumin primary antibody (Servicebio, GB152080, 1:200) overnight at 4 °C, followed by incubation with CY3-conjugated goat anti-rabbit secondary antibody (Servicebio, GB21303, 1:300) for 50 min at room temperature in the dark. Nuclei were counterstained with DAPI for 10 min, followed by autofluorescence quenching for 5 min. Images were acquired using a fluorescence microscope (Nikon Eclipse C1). Data Records The RNA and miRNA sequencing raw data is available in the Genome Sequence Archive (https://ngdc.cncb.ac.cn/gsa/) with the accession number CRA02452037. We also deposited the raw RNA and miRNA sequencing data in the NCBI Sequence Read Archive (SRA) under the accession number SRP68928738. The expression profiles of mRNA, lncRNA and miRNA, and DE transcript lists are provided in Supplementary Table S1-7. Technical Validation Liver ductal organoids establishment and differentiation We successfully constructed the liver ductal organoids from three C57BL/6 mice and differentiated them into immature hepatocytes as shown in Fig. 2A. The qRT-PCR further validated that EM group exhibited significantly higher expression of cholangiocyte marker genes (Sox9, Spp1), while the DM group showed significantly higher expression of hepatocyte marker genes (Apoa1, Sult1a1, Alb and Ttr) and immature hepatocyte marker gene (Afp) (Fig. 2B,C). Immunofluorescence staining revealed that Alb was undetectable in EM group, with only DAPI-stained nuclei visible (Fig. 2D). In contrast, DM group exhibited clear Alb positivity (red) co-localizing with DAPI, confirming successful hepatocyte lineage commitment at the protein level (Fig. 2D). Establishment of 3D liver ductal organoids and its differentiation. (A) Liver ductal organoids cultured with expansion medium (EM), differentiation medium (DM) I and DM II. (B) Expression levels of cholangiocyte (Krt7, Spp1 and Sox9) marker genes, and immature hepatocyte marker gene (Afp) in EM and DM group by qRT-PCR. (C) Expression levels of hepatocyte (Alb, Ttr, Apoa1, Sult1a1 and Cyp3a11) marker genes in EM and DM group by qRT-PCR. (D) Immunofluorescence validation of Albumin expression. Representative images of EM and DM organoids stained with DAPI (blue, nuclei) and anti-Albumin antibody (red). The statistical data represent mean ± S.D., * p < 0.05, ** p < 0.01, ***p < 0.001. RNA quality and integrity assessment RNA purity was assessed using a NanoDrop2000 (Thermo Fisher Scientific, USA), and RNA integrity was evaluated using an Agilent 5400 Fragment Analyzer (Agilent Technologies, USA). All samples had A260/280 ratios between 1.9 and 2.1 and RQN value 10.0. RNA-Seq data quality To identify key genes/transcripts associated with cholangiocyte-to-immature hepatocyte differentiation, we constructed 6 RNA-seq libraries and 6 miRNA-seq libraries from liver ductal organoids in EM and DM groups. A total of 537,612,208 raw reads were generated from 6 RNA-seq libraries. After removing adapter sequences and low-quality sequences, 529,049,578 clean reads were retained for further lncRNA/mRNA analysis. In each sample, the percentage of clean reads varied from 98.12 to 98.65%, and majority (93.46–95.42%) of reads were mapped to the reference genome. Q20, Q30 and GC content of the clean data were 96.69 ± 0.13%, 91.39 ± 0.27% and 48.21 ± 0.63% (Table 2). In the end, there were 38,689 mRNA transcripts (16,655 genes) and 24,676 lncRNAs were generated from six samples. The expression profiles of lncRNA and mRNA were shown in Table S1 and Table S2, respectively. The transcripts/gene with read count higher than 1 at least one sample were defined as expression. Table 2RNA-seq read statistics.SampleRaw readsClean readsQ20%Q30%GC%Total mapUnique mapMulti mapEM1815213307998609096.7191.4249.3574819341 (93.54%)69919323 (87.41%)4900018 (6.13%)DM1887789748712821496.9291.8748.6382789679 (95.02%)74727868 (85.77%)8061811 (9.25%)EM2828795908164376096.7691.5548.1376612629 (93.84%)72802399 (89.17%)3810230 (4.67%)DM2959568129465910096.6691.2347.5990319898 (95.42%)83279424 (87.98%)7040474 (7.44%)EM3951376989356866896.5691.1648.0187450418 (93.46%)83338201 (89.07%)4112217 (4.39%)DM3933378049206374696.5491.0947.5487570960 (95.12%)81584139 (88.62%)5986821 (6.5%)GC(%), the total number of bases G and C as a percentage of the total number of clean bases; Q20/30(%), the percentage of bases with Phred values greater than 20/30 in the total base. RNA-seq read statistics. GC(%), the total number of bases G and C as a percentage of the total number of clean bases; Q20/30(%), the percentage of bases with Phred values greater than 20/30 in the total base. For 6 miRNA-seq libraries, a total of 62,469,214 raw reads were generated from six samples. After removing low-quality reads, 61,266,130 clean reads were retained for further miRNA analysis. The percentage of clean reads varied from 97.58 to 98.58% at each sample, and most (71.74–95.45%) of reads were mapped to the reference genome. Q20, Q30 and GC content of the clean data were 99.29 ± 0.13%, 97.65 ± 0.29% and 49.08 ± 0.88% (Table 3). In the end, 1,018 miRNAs were detected in all the samples, including 960 known miRNAs and 58 novel miRNAs. The expression profiles of miRNA were shown in Table S3. Table 3miRNA-seq read statistics.SampleTotal readsClean readsQ20%Q30%GC %Total sRNAMapped sRNA“+” Mapped sRNA“−” Mapped sRNAEM11032742210181235 (98.58%)99.3997.8947.8396162549178942 (95.45%)1745109 (18.15%)7433833 (77.30%)DM1100440179880347 (98.37%)99.397.6149.8296886238621934 (88.99%)3682410 (38.01%)4939524 (50.98%)EM21048907210289629 (98.10%)99.3797.949.2198289718692418 (88.44%)2708276 (27.55%)5984142 (60.88%)DM21120431410985288 (98.05%)99.0197.0448.461073683710067422 (93.77%)3517714 (32.76%)6549708 (61.00%)EM31029958210050156 (97.58%)99.3197.7150.48100199327187921 (71.74%)1986109 (19.82%)5201812 (51.91%)DM3101048079879475 (97.77%)99.3497.7548.6795239078875710 (93.19%)3216590 (33.77%)5659120 (59.42%)GC(%), the total number of bases G and C as a percentage of the total number of clean bases; Q20/30(%), the percentage of bases with Phred values greater than 20/30 in the total base. miRNA-seq read statistics. GC(%), the total number of bases G and C as a percentage of the total number of clean bases; Q20/30(%), the percentage of bases with Phred values greater than 20/30 in the total base. The density maps of mRNA, lncRNA and miRNA expression are shown in Fig. 3A, reflecting a high consistency in mRNA, lncRNA and miRNA expression patterns across different samples. The Pearson correlation test indicated that the similarity of expression patterns of samples within DM or EM group was high, the highest correlation coefficient was close to 0.972, and the lowest was 0.805 (Fig. 3B). Expression profiles of mRNAs, lncRNAs and miRNAs. (A) FPKM/TPM density distribution curve. (B) Pearson correlation coefficient between samples. (C) Volcano map of differential mRNAs and ncRNAs. (D) Heat map of differential mRNAs and ncRNAs. To evaluate the consistency and variability among the three biological replicates, principal component analysis (PCA) was also performed based on the expression levels of mRNAs, lncRNAs, and miRNAs, respectively (Fig. 4). For mRNAs, PCA revealed that PC1 and PC2 explained 35.1% and 19.6% of the total variance, respectively. For lncRNAs, PC1 and PC2 explained 30.8% and 23.7% of the variance, respectively. For miRNAs, PC1 and PC2 explained 36.5% and 21.9% of the variance, respectively. As expected, PCA separated the EM and DM samples along the first principal component for all three molecular data types (mRNAs, lncRNAs, and miRNAs), confirming that the differentiation condition is the major source of transcriptional variation. Collectively, the PCA results demonstrate that differentiation condition (EM vs. DM) is the major source of transcriptional variation, as reflected by the separation along PC1. The PCA analysis of mRNA, lncRNA and miRNA. The correlation coefficients among replicates were as low as approximately 0.805 for some transcript types, suggesting moderate variability. This variability is further observed in PCA plots. Inherent heterogeneity of organoid cultures during expansion and differentiation, and biological variation among the three individual mouse donors, contributed to these results. Despite this variability, all replicates clustered within their respective experimental groups with no outliers, and the clear separation between EM and DM groups confirms that the differentiation treatment drives consistent transcriptional reprogramming across replicates. Therefore, we consider the three biological replicates to be acceptable for downstream analyses. Verification of differentially expressed transcripts From the expression profiles, DE mRNAs, lncRNAs and miRNAs transcripts were generated by comparing DM vs. EM group as showed in volcano plot (Fig. 3C) and heatmap (Fig. 3D). In total, there were 4,073 DE mRNAs (2,164 upregulated and 1,909 downregulated), 1,653 DE lncRNAs (751 upregulated and 902 downregulated) and 180 DE miRNAs (85 upregulated and 95 downregulated) in the liver ductal organoids of DM group compared to those of EM group based on the RNA/miRNA-seq data (Table S4-6). Furthermore, we also analyzed the DE mRNA genes between DM vs. EM group, and found that 3,396 mRNA genes were differentially expressed in two groups. Among these DE mRNA genes, 1,716 were upregulated in DM group and 1,680 were downregulated in DM group (Table S7). To validate the expression patterns of DE mRNAs, lncRNAs, and miRNAs, we randomly selected six upregulated and downregulated transcripts for qRT-PCR analysis (Fig. 5). Among these, Sult1a1, Gm15889-202, and miR-664-5p were upregulated, while Spp1, Gm42047-202, and miR-667-5p were downregulated in the DM group compared with the EM group, as confirmed by both RNA-seq/miRNA-seq and qRT-PCR methods. Verification of differentially expressed mRNA (A), lncRNA (B) and miRNA (C). * p < 0.05, ** p < 0.01, ***p < 0.001. Supplementary information Supplementary Table S1 Expression profiles of lncRNA (read count) Supplementary Table S2 Expression profiles of mRNA (read count) Supplementary Table S3 Expression profiles of miRNA (read count) Supplementary Table S4 All differentially expressed mRNAs transcripts (DMvsEM) Supplementary Table S5 All differentially expressed lncRNAs (DMvsEM) Supplementary Table S6 All differentially expressed miRNAs (DMvsEM) Supplementary Table S7 All differentially expressed mRNAs genes (DMvsEM) Acknowledgements This research was supported from the National Natural Science Foundation of China (62371091, 62071085), Natural Science Foundation of Shandong of China (ZR2023MC186, ZR2021QF143, ZR2021QC135), Talent Introduction Project of Dezhou University of China (2020xjrc216), Enterprise project (HXKT2023211), Dezhou Municipal Research and Development Program Project (2024dzkj26), The Youth Innovation Team Lead-education Project of Shandong Educational Committee. Author contributions P.P.H. designed the experiments, analyzed the data and wrote the manuscript, X.Y. and C.C.S. conducted experiments, Z.X.C., J.R.W. and R.P.W. participated in experiments and data analysis, B.L.Z. participated in data analysis and conceived the project, J.H.W conceived the project. Data availability The sequencing data supporting this study are available in the Genome Sequence Archive (GSA) under accession number CRA024520 (https://ngdc.cncb.ac.cn/gsa/browse/CRA024520) and in the NCBI Sequence Read Archive (SRA) under accession number SRP689287 (https://identifiers.org/ncbi/insdc.sra:SRP689287). Code availability The following software’s and tools were used in this work. No custom code was utilized during the analysis of the study. FASTQC software (v. 0.11.9) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). HISAT2 (v2.0.5) (https://www.ccb.jhu.edu/software/hisat/index.shtml). StringTie (v1.3.3b) (https://ccb.jhu.edu/software/stringtie/). CPC2 (v3.2.0) (http://cpc2.cbi.pku.edu.cn/; https://github.com/gao-lab/CPC2_standalone). CNCI (v2.0) (https://github.com/www-bioinfo-org/CNCI). Pfam (v1.6) (http://pfam.xfam.org/). miRbase database (https://www.mirbase.org). miRanda (v3.3a) (http://www.microrna.org/microrna/home.do). DESeq2 (1.24.0) (https://bioconductor.org/pack-ages/release/bioc/html/DESeq2.html). Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Pingping Huang, Xue Yu. Contributor Information Bailing Zhou, Email: blzhou@126.com. Jihua Wang, Email: jhw25336@126.com. Supplementary information The online version contains supplementary material available at 10.1038/s41597-026-07427-5. References Associated Data Data Citations Supplementary Materials Supplementary Table S1 Expression profiles of lncRNA (read count) Supplementary Table S2 Expression profiles of mRNA (read count) Supplementary Table S3 Expression profiles of miRNA (read count) Supplementary Table S4 All differentially expressed mRNAs transcripts (DMvsEM) Supplementary Table S5 All differentially expressed lncRNAs (DMvsEM) Supplementary Table S6 All differentially expressed miRNAs (DMvsEM) Supplementary Table S7 All differentially expressed mRNAs genes (DMvsEM) Data Availability Statement The sequencing data supporting this study are available in the Genome Sequence Archive (GSA) under accession number CRA024520 (https://ngdc.cncb.ac.cn/gsa/browse/CRA024520) and in the NCBI Sequence Read Archive (SRA) under accession number SRP689287 (https://identifiers.org/ncbi/insdc.sra:SRP689287). The following software’s and tools were used in this work. No custom code was utilized during the analysis of the study. FASTQC software (v. 0.11.9) (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). HISAT2 (v2.0.5) (https://www.ccb.jhu.edu/software/hisat/index.shtml). StringTie (v1.3.3b) (https://ccb.jhu.edu/software/stringtie/). CPC2 (v3.2.0) (http://cpc2.cbi.pku.edu.cn/; https://github.com/gao-lab/CPC2_standalone). CNCI (v2.0) (https://github.com/www-bioinfo-org/CNCI). Pfam (v1.6) (http://pfam.xfam.org/). miRbase database (https://www.mirbase.org). miRanda (v3.3a) (http://www.microrna.org/microrna/home.do). DESeq2 (1.24.0) (https://bioconductor.org/pack-ages/release/bioc/html/DESeq2.html).