Comprehensive Cytogenetic and Genomic Profiling of the Murine AML12 (Alpha Mouse Liver 12) Hepatocyte Cell Line

TL;DR

This study provides a detailed genetic and functional profile of the AML12 mouse liver cell line, showing it has a complex genome and partial loss of mature liver cell features.

Contribution

The first integrated cytogenetic, STR, and transcriptomic reference map for the AML12 cell line is presented.

Findings

AML12 cells have a near-tetraploid, highly aneuploid karyotype with structural rearrangements and copy number changes.

AML12 maintains hepatocyte lineage identity but shows partial de-differentiation and reduced metabolic gene expression.

A 16-locus STR barcode was developed to authenticate AML12 and distinguish it from other murine cell lines.

Abstract

What are the main findings?
We present the first integrated cytogenetic, STR-based, and transcriptomic reference map of the widely used murine hepatocyte line AML12. Our findings reveal a near-tetraploid, highly aneuploid karyotype with recurrent structural rearrangements and segmental copy number changes.Bulk RNA-seq, along with conventional and quantitative RT-PCR, Western blotting, immunofluorescence, light microscopy, and ultrastructural analyses, demonstrate that AML12 cells maintain hepatocyte lineage identity but exhibit partial de-differentiation, including ductular/progenitor- and stress-associated transcriptional signatures.

We present the first integrated cytogenetic, STR-based, and transcriptomic reference map of the widely used murine hepatocyte line AML12. Our findings reveal a near-tetraploid, highly aneuploid karyotype with recurrent structural rearrangements and segmental copy number changes.

Bulk RNA-seq, along with conventional and quantitative RT-PCR, Western blotting, immunofluorescence, light microscopy, and ultrastructural analyses, demonstrate that AML12 cells maintain hepatocyte lineage identity but exhibit partial de-differentiation, including ductular/progenitor- and stress-associated transcriptional signatures.

What are the implications of the main findings?
We combined a cytogenetic and STR profile that establishes a practical authentication and quality control framework that supports the reliable use of AML12 across laboratories and aids in detecting clonal drift over time.Overall, AML12 remains a useful and experimentally tractable murine hepatocyte model, particularly for studying hepatocyte signaling, stress responses, and epithelial plasticity at a mechanistic level. However, if fully mature metabolic functions or precise handling of bile acids/xenobiotics are required, our data suggest that AML12 should be used as a heuristic model and complemented with more differentiated systems (such as primary hepatocytes, organoids, or in vivo models). This should ideally be done with condition-dependent optimization and appropriate functional validation.

We combined a cytogenetic and STR profile that establishes a practical authentication and quality control framework that supports the reliable use of AML12 across laboratories and aids in detecting clonal drift over time.

Overall, AML12 remains a useful and experimentally tractable murine hepatocyte model, particularly for studying hepatocyte signaling, stress responses, and epithelial plasticity at a mechanistic level. However, if fully mature metabolic functions or precise handling of bile acids/xenobiotics are required, our data suggest that AML12 should be used as a heuristic model and complemented with more differentiated systems (such as primary hepatocytes, organoids, or in vivo models). This should ideally be done with condition-dependent optimization and appropriate functional validation.

The murine Alpha Mouse Liver 12 (AML12) cell line, established over four decades ago, is one of the most commonly used non-transformed hepatocyte models in basic and pre-clinical liver research. Despite its widespread use, a comprehensive and current molecular characterization has been lacking. In this study, we combined cytogenetics with high-resolution genomic technologies to establish a detailed genetic reference profile of AML12. Inverted DAPI banding and multicolor fluorescence in situ hybridization (m-FISH) revealed a complex yet stable, near-tetraploid karyotype featuring double X-chromosome deletions [del(X)(A3)×2], a recurrent derivative chromosome der(3)t(2;3)(A2;H4), biallelic deletions of 17D1, two dicentric chromosomes dic(X;17), and multiple whole-chromosome gains (e.g., +1, +6, +15, +19×4) and losses (e.g., −4, −12, −16, −18). Multicolor banding (mcb) further pinpointed cryptic inversions on chromosomes 7 and 11. Copy number imbalances were visualized as in silico array comparative genomic hybridization (aCGH)-style profiles inferred from these metaphase-based assays, and no independent array- or sequencing-based copy number variation (CNV) experiment was performed in this study. Short tandem repeat (STR) profiling created a unique 16-locus authentication barcode that unambiguously distinguishes AML12 from other murine cell lines in public databases. Bulk RNA sequencing (RNA-seq) further demonstrated a transcriptional profile in AML12 cells that is indicative of hepatocyte origin while also revealing partial de-differentiation and reduced expression of selected urea cycle, gluconeogenic, and xenobiotic-metabolizing transcripts, consistent with limited mature hepatocyte functions. These functional inferences are likely based on gene expression patterns rather than on direct physiological assays. In summary, our study provides (i) the first integrated cytogenetic, STR, and next-generation sequencing dataset for AML12, (ii) a practical authentication panel for routine laboratory use, and (iii) reference information that will enhance the interpretation, reproducibility, and translational relevance of future studies using this versatile hepatocyte model.

Figures

Full text

1. Introduction

Cell lines are indispensable tools in contemporary biomedical research because they offer reproducible, genetically tractable, and cost-effective platforms for interrogating fundamental biological processes, disease mechanisms, and drug responses. In hepatology, the murine Alpha Mouse Liver 12 (AML12) cell line has emerged as a workhorse model that bridges the gap between primary hepatocytes, which are limited by rapid de-differentiation and donor variability, and tumor-derived lines that often harbor extensive oncogenic mutations. AML12 was originally established from hepatocytes of a 5-month-old male homozygous transgenic mouse (CD1 strain, line MT42) overexpressing human transforming growth factor-α under the control of the zinc-inducible metallothionein 1 promoter, a strategy that yielded non-tumorigenic, contact-inhibited cells that nonetheless retain many differentiated liver functions [1]. Consequently, AML12 cells synthesize albumin, α1-antitrypsin, and transferrin, express phase I and phase II drug-metabolizing enzymes, and respond to endocrine and nutritional cues in a manner reminiscent of normal murine hepatocytes [1,2]. Importantly, AML12 cells similar to primary mouse hepatocytes encompass features that closely replicate physiological aspects of energy metabolism [3].

Because of these favorable characteristics, AML12 has been widely adopted across diverse areas of liver research. Investigators have employed the line to delineate insulin receptor signaling, gluconeogenic control, and lipid droplet dynamics, thereby uncovering regulators of hepatic glucose homeostasis and lipid storage [4,5,6,7]. In the context of non-alcoholic fatty liver disease (NAFLD), AML12 has facilitated in vitro models of free fatty acid-induced steatosis, lipotoxicity, and endoplasmic reticulum stress that revealed critical roles for endoplasmic reticulum stress, autophagy, and mitochondrial dysfunction in disease progression [2,8,9]. Oxidative stress studies have capitalized on the robust antioxidant machinery of AML12 to dissect Nrf2-mediated cytoprotection and ferroptosis pathways in both metabolic overload and drug-induced injury settings [10,11]. Virologists have used AML12 to explore hepatitis virus tropism and antiviral responses. Specifically, the cells support productive murine hepatitis virus replication and permit heterologous expression of hepatitis C viral proteins, thus serving as a convenient surrogate for antiviral drug screening [12,13]. Furthermore, toxicologists rely on preserved xenobiotic-metabolizing repertoire of AML12 cells to evaluate hepatotoxic liabilities of pharmaceuticals, environmental contaminants, herbal extracts, and engineered nanoparticles [2,14,15]. Moreover, AML12 cells are a suitable model to study aspects of tumorigenesis and tumor suppressor gene function [16].

Despite the broad utility of this cell line, surprisingly little is known about its genomic architecture. This knowledge gap is becoming increasingly problematic, as accumulating evidence shows that hidden chromosomal aberrations or driver mutations can profoundly influence cell physiology, confounding experimental outcomes and undermining reproducibility. In the era of precision medicine and large-scale data integration, detailed genomic reference profiles are therefore essential for the critical evaluation and cross-comparison of studies that rely on established cell models.

Here, we present the first integrated genetic and molecular characterization of AML12. By combining cytogenetics with high-resolution multicolor fluorescence in situ hybridization (m-FISH), multicolor banding (mcb), short tandem repeat (STR) profiling, and next-generation sequencing and next-generation RNA sequencing, we define the karyotypic landscape, establish a unique authentication barcode, and characterize genome-wide gene expression patterns in AML12. This comprehensive dataset provides an invaluable reference for the scientific community, safeguards experimental rigor through authenticated cell line use, and lays the foundation for a more nuanced interpretation of past and future research employing this pivotal murine hepatocyte model.

2. Materials and Methods

2.1. Literature Search

To identify biomedical studies that utilized the AML12 hepatocyte cell line, we conducted a comprehensive search of PubMed/MEDLINE, Web of Science Core Collection, Scopus, Embase, and Google Scholar from the inception of the databases up to the date of our final search (last conducted on 6 January 2026). Our search strategies combined controlled vocabulary and free-text terms for the cell line and its various names along with hepatocyte/liver and murine descriptors. These strategies were tailored to the syntax of each database. An example of a search query used in PubMed was: “AML12” OR “AML-12” OR “AML 12” OR “Alpha Mouse Liver 12” NOT “leukemia”, excluding terms to eliminate false positives unrelated to the cell line. Additionally, we consulted cell line registries such as vendor catalogs and Cellosaurus to identify additional synonyms, which were then integrated into our iterative searches.

2.2. Cell Culture

AML12 cells were purchased from the American Type Culture Collection (CRL-2254, ATCC, Manassas, VA, USA). The cells were grown in a humidified incubator at a temperature of 37 °C with 5% CO_2_ in Dulbecco’s Modified Eagle Medium (DMEM, high-glucose, #D6171-500ML, Sigma-Aldrich, Merck, Taufkirchen, Germany). The medium was supplemented with 2 mM sodium pyruvate solution (#S8636-100ML), 10% fetal bovine serum (FBS, #F7524), and 2 mM L-glutamine (#G7513), all obtained from Sigma-Aldrich. Additionally, penicillin–streptomycin antibiotic solution (P0781-100ML) from Sigma Life Science (Lonza, Cologne, Germany) was added to the growth medium. These culture conditions differ from the current ATCC-recommended AML12 medium (DMEM/F-12 supplemented with insulin, transferrin, selenium, and dexamethasone). Consequently, the transcriptomic and functional characteristics described in this study reflect AML12 cells maintained in high-glucose DMEM + 10% FBS and should be interpreted in the context of these basal culture conditions. The medium was changed every 2–3 days. Cells were passaged at 70–80% confluence using Accutase solution (#A6964-100ML, Sigma-Aldrich) and reseeded at 1:5–1:10 split ratios on tissue culture plastic plates. Cells between passages 5–20 were used for experiments, and cultures were routinely confirmed to be mycoplasma-negative.

2.3. Short Tandem Repeat (STR) Profiling

The mouse cell line AML12 (RRID: CVCL_0140) underwent STR profiling using the CellCheckTM Mouse system. This system includes 19 species-specific consensus STR markers recommended by the Consortium for Mouse Cell Line Authentication for STR profiling in mice [17,18]. An STR similarity search was conducted using the STR profile of AML12 with the Cellosaurus STR Similarity Search Tool CLASTR 1.4.4 (release 41.0) and the Cellosaurus mouse STR database [19,20]. The search parameters were as follows: scoring algorithm: Tanabe, mode: Non-empty markers, score filter: 40%, and min. markers: 8.

2.4. Preparation of AML12 Metaphase Chromosomes, Karyotyping, and Molecular Cytogenetics

Preparation of AML12 metaphase chromosomes, karyotyping, and molecular cytogenetics were conducted. Metaphase spreads were prepared from logarithmically growing AML12 cultures by transient mitotic arrest using a microtubule inhibitor (KaryoMAX colcemid solution, #15212012, Gibco, Thermo Fisher Scientific, Schwerte, Germany), hypotonic treatment, and fixation in methanol:acetic acid, as previously described [21]. Multicolor FISH (m-FISH) was carried out with the 21XMouse probe set for mouse chromosomes (MetaSystems, Altlussheim, Germany), following the manufacturer’s protocol to assign distinct colors to each homologous chromosome pair. Detailed analysis of 20 metaphases was conducted using a Zeiss Axioplan fluorescence microscope (Carl Zeiss, Jena, Germany) coupled to a CCD camera. Images were captured and processed with the ISIS software, version 6.1.1 (MetaSystems) to derive the consensus AML12 karyotype. Conventional karyotyping was performed on the same metaphases as used for m-FISH, taking advantage of 4′,6-diamidino-2-phenylindole (DAPI) counterstaining. The latter leads to a DAPI banding pattern, which can be simply transformed to an inverted DAPI banding, similar to GTG banding using the ISIS software. FISH banding was also performed on 20 metaphases each, using 21 chromosome-specific probe sets of the murine multicolor banding (mcb) as previously described [22]. Clonal aberrations were identified based on established cytogenetic criteria, and karyotypes were annotated using the standard murine nomenclature.

2.5. In Silico Comparative Genomic Hybridization-Style Copy-Number Mapping

Copy number imbalances were determined using mFISH and mcb data in conjunction with inverted DAPI banding in AML12 metaphases. The affected murine chromosomal segments were then assigned approximate genomic coordinates using the UCSC Genome Browser and the mouse genome annotation for Build 37 (GRCm37/mm9). These coordinates were then projected onto the human reference genome (GRCh38.p13/hg38) through UCSC cross-genome mapping to identify orthologous regions with corresponding gains or losses. No microarray-based hybridization experiment was conducted. Instead, the data was visualized as in silico aCGH-style profiles based on mFISH-/mcb-inferred copy number changes. The resulting copy number alteration (CNA) set was subsequently compared with relevant reference datasets as specified elsewhere in the study. Therefore, all copy number profiles presented in this work should be considered as in silico reconstructions derived from metaphase-based cytogenetic data rather than as results of independent array- or sequencing-based CNV assays.

2.6. Next-Generation Sequencing and Data Analysis

Next-generation sequencing (NGS) of AML12 cells was conducted following established workflows [21]. The cells were cultured under basal conditions until they reached approximately 80% confluence. Total RNA was isolated from 9 individual plates (3 × 3) and lysed in guanidine thiocyanate. Total RNA was isolated using a cesium chloride cushion and then precipitated with ethanol and resuspended in nuclease-free water. RNA yield and integrity were evaluated using UV spectrophotometry and an Agilent 4200 TapeStation (Agilent, Santa Clara, CA, USA). The samples were depleted of rRNAs, reverse-transcribed and indexed with NEBNext Multiplex Oligos for Illumina (Index Primers Set 1, New England Biolabs, Ipswich, MA, USA).

Based on our strategy, this study includes a single biological sample (AML12), and no biological replicates were generated. All library preparation and sequencing were performed in a single batch, so no batch effects were present, and no batch correction was applied. Paired-end RNA sequencing (2 × 150 bp) was carried out on an Illumina MiSeq platform using a MiSeq Reagent Kit v2 (300 cycles). Sequencing generated approximately 7.1 million read pairs (7.8 million total sequences). After processing with the nf-core/rnaseq pipeline (v3.12.0), 92.0% of reads aligned to the GENCODE GRCm39 v35 reference genome, with 81.5% properly paired reads. The overall mapping rate was 100%, and the estimated sequencing error rate was 0.38%. rRNA content was low (2.12%), duplication levels were modest (4.7%), and GC content (~49–50%) was consistent with the expected mouse transcriptome composition. The 5′–3′ bias (1.41) indicated acceptable transcript coverage uniformity. Raw sequencing data (FASTQ files: AML12_S2_R1_001.fastq.gz and AML12_S2_R2_001.fastq.gz) and processed transcript abundance tables have been submitted to the Gene Expression Omnibus (GEO). GEO and associated Sequence Read Archive (SRA) accession numbers are currently in process and will be provided upon release.

The expression levels of individual genes were compared to curated cell-type-enriched marker sets for hepatocytes, hepatic stellate cells, cholangiocytes, and Kupffer cells taken from the PanglaoDB database [23], supplemented with additional lineage markers commonly used in liver research.

2.7. Electron-Microscopic Cell Analysis

For ultrastructural analysis, AML12 cells were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer and then post-fixed in 1% osmium tetroxide. They were dehydrated through an ethanol series and embedded in epoxy resin. Ultrathin sections (approximately 70–90 nm) were cut on an ultramicrotome, mounted on copper grids, and contrasted with uranyl acetate and lead citrate. Samples were examined on a Zeiss Leo 906 transmission electron microscope operating at 60 kV. Images acquired at magnifications ranging from 2156× to 27,800× were used to assess subcellular architecture, including mitochondrial morphology.

2.8. Reverse Transcription Polymerase Chain Reaction (RT-PCR)

Conventional RT-PCR was performed on cDNA using the following thermal profile: an initial denaturation at 95 °C for 5 min; cycling steps of 1 min at 95 °C for denaturation, 1 min at 60 °C for annealing for Hnf4a, Foxa2, Krt18 and Ppia, 64 °C for, Alb, and Gapdh and a 3 min extension at 72 °C. This was followed by a final extension at 72 °C for 10 min. PCR products were resolved on 2% agarose gels supplemented with GelRed (Sct123, Merck Millipore, Burlington, VT, USA) in 1× TAE buffer (40 mM Tris base, 20 mM acetic acid, 1 mM ethylenediaminetetraacetic acid disodium salt dehydrate (EDTA)) and imaged using an iBright 750 system (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). Quantitative mRNA expression analysis was performed as previously described [24] with the following cycling conditions: 10 min at 95 °C for initial denaturation and 40 amplification cycles with 15 s at 90 °C and 1 min at 60 °C. Relative mRNA expression was normalized to the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (Gapdh) and calculated using the 2^−ΔΔCT^ method [25].

The primers used are listed in Table 1.

2.9. Western Blot Analysis

Proteins were extracted from AML12 cells, primary mouse hepatocytes and mouse liver using standard protocols [26]. Equal amounts of total protein (30 µg per lane) were denatured at 80 °C for 10 min under reducing conditions, resolved on 4–12% Bis-Tris NuPAGE gels, and transferred to 0.45 µm nitrocellulose in NuPAGE transfer buffer. Transfer quality was verified by Ponceau S staining. Membranes were blocked in TBS with 5% milk, probed with primary antibodies followed by HRP-conjugated secondaries, and visualized by chemiluminescence (#34075, SuperSignal West Dura, Thermo Fisher Scientific^TM^). Liver lysate from a male mouse (80 µg per lane) obtained from a previous study served as a positive control after homogenization with an MM400 mixer mill [25]. Additionally, protein extracts prepared previously from primary mouse hepatocytes (30 µg) were used as a further control [27,28]. All antibodies are listed in Table 2.

2.10. Phalloidin Stain and HNF4α Immunocytochemistry

We plated 4.0 × 10^4^ AML12 cells per well in a four-chamber, tissue-culture-treated glass slide (Falcon, #354104; BD Biosciences, Erembodegem, Belgium). A more detailed, stepwise protocol is available in [29]. After 24 h, the medium was aspirated, cultures were rinsed three times with 1× PBS, and cells were fixed for 20 min in 4% paraformaldehyde (phosphate-buffered, pH 7.4) protected from light. Following three additional PBS washes, cells were permeabilized on ice for 4 min with 0.2% Triton X-100 in PBS. Non-specific binding was blocked with 3% donkey serum in PBS for 1 h at room temperature. After further PBS washes, F-actin was labeled with Alexa Fluor 488^TM^ phalloidin (1×, #A12379, Invitrogen, Thermo Fisher Scientific) for 20 min in the dark, and nuclei were counterstained with DAPI (#D1306, Thermo Fisher Scientific) for 30 min in the dark. For immunocytochemistry of HNF4α, a primary antibody directed against HNF4α (1:100, #sc-6556, Santa Cruz Biotech., Santa Cruz, CA, USA) was diluted in blocking buffer and incubated overnight at 4 °C under shaking. The next day, donkey anti-goat Alexa Flour^TM^ 488 conjugate (1:500, #A32814, Invitrogen, Thermo Fisher) was diluted in blocking buffer and incubated for 1 h at room temperature. The samples were then washed three times with PBS and once with deionized water, mounted in Vectashield Antifade medium (#H-1000, Vector Laboratories, Burlingame, CA, USA), and imaged on a Nikon Eclipse E80i fluorescence microscope using the NIS-Elements software (version 3.22.01; Nikon Europe, Düsseldorf, Germany).

3. Results

3.1. Usage of AML12 Cells in Biomedical Research

AML12 cells are an immortalized, non-tumorigenic murine hepatocyte line that retain key hepatocyte features and hormone/xenobiotic responsiveness, making them a widely used in vitro model for liver biology. They are routinely employed to study hepatic lipid and glucose metabolism, lipotoxicity, oxidative/ER stress, nuclear receptor signaling, bile acid pathways, and toxicology. Additionally, they are amenable to genetic perturbation (e.g., CRISPR/Cas9 or RNAi), facilitating mechanistic studies that align with mouse in vivo models. To contextualize their use and relevance for this study, we systematically searched the literature in PubMed/MEDLINE, Web of Science Core Collection, Scopus, Embase, and Google Scholar to identify peer-reviewed publications employing AML12 cells across these research domains. We identified over 1500 articles that have used AML12 in biomedical research. For instance, a search conducted on the PubMed database resulted in 1463 articles, with a clear tendency that the number increases each year (Figure 1).

3.2. Short Tandem Repeat Analysis

Short tandem repeat profiling of our AML12 culture shows agreement with the Cellosaurus AML12 reference at multiple loci including MCA-4-2 (20.3), MCA-5-5 (14, 15), MCA-6-7 (12), MCA-12-1 (19), MCA-15-3 (21.3), MCA-18-3 (21), and MCA-X-1 (26). The only discrepancy is at MCA-6-4 (15.3 in our sample versus 16 in Cellosaurus) (Table 3).

Several additional markers in Cellosaurus lack reference values and could not be cross-checked (e.g., MCA-1-1, MCA-1-2, MCA-2-1, MCA-3-2, MCA-7-1, MCA-8-1, MCA-11-2, MCA-13-1, MCA-17-2, MCA-19-2, and MCA-9-2). All typed alleles fall within the published “known allele ranges” where available, except for MCA-X-1, whose allele 26 falls outside the range but matches the AML12 entry in Cellosaurus and is outside the range published by others [18]. Additionally, MCA-9-2 has no published range. However, these findings indicate a strong match between our AML12 stock and the reference profile, with only minor divergence at a single locus that may reflect technical or subline variability. Taken together, these STR data confirm that our AML12 culture is a properly authenticated derivative of the canonical AML12 line and can serve as a reliable reference stock for downstream cytogenetic and transcriptomic analyses.

3.3. Phenotypic Appearance of AML12 Cells

Under routine bright-field or phase-contrast microscopy, AML12 cultures exhibit an adherent epithelial phenotype with polygonal, hepatocyte-like cells forming a cobblestone monolayer when grown to confluency (Figure 2). Individual cells have centrally located round nuclei with one to two prominent nucleoli and a moderate cytoplasm-to-nucleus ratio. Bi-nucleated cells are frequently seen, consistent with hepatocyte characteristics. The cytoplasm appears finely granular, with occasional refractile vacuoles or lipid-like droplets depending on the culture conditions. Cells at the edges of colonies may appear slightly elongated and migratory, while those in the center are tightly packed with distinct cell–cell borders indicating junctional contacts. In densely packed areas, narrow intercellular clefts resembling bile canaliculi-like lumina may be observed. Overall, the morphology is consistent and non-fibroblastic, with strong adherence to the substrate under standard AML12 growth conditions. Phase-contrast imaging revealed that approximately 5 out of 100 cells (5%) of AML12 cells were binucleated.

3.4. Electron-Microscopic Analysis of AML12 Cells

Using transmission electron microscopy, AML12 cells exhibit a hepatocyte-like ultrastructure. The nuclei are round-to-oval in shape, with mostly euchromatic interiors, peripheral heterochromatin, and prominent nucleoli. Multiple-nucleoli cells were most common (Figure 3). The cytoplasm is filled with abundant rough endoplasmic reticulum organized in parallel cisternae, a well-developed Golgi apparatus, and numerous mitochondria with lamellar cristae. Additionally, lysosomes and occasional autophagic profiles can be observed. The cortical cytoskeleton is visible beneath the plasma membrane, indicating an epithelial, non-fibroblastic phenotype that retains partial polarity in 2D culture.

In addition, we visualized the organization of the actin cytoskeleton in AML12 cells by staining with phalloidin. Under these conditions, AML12 hepatocytes displayed a continuous F-actin network throughout the cell body, with prominent staining along the cell periphery. This is consistent with their polygonal epithelial morphology and tight cell–cell contacts observed in phase-contrast microscopy. At higher magnifications, individual stress-fiber-like structures and cortical actin belts became apparent, outlining the cobblestone monolayer and highlighting the close apposition of neighboring cells. Nuclei were clearly counterstained with DAPI, allowing for reliable correlation of nuclear position with cytoskeletal architecture in both sub-confluent and confluent areas (Figure 4). Overall, the phalloidin-based F-actin labeling confirms the non-fibroblastic, epithelial phenotype of AML12 cells and supports the presence of a well-organized cytoskeletal scaffold compatible with hepatocyte-like polarity in 2D culture. In quantitative terms, nearly all AML12 cells stained with phalloidin displayed prominent cortical F-actin belts outlining cell–cell contacts. Stress-fiber-like bundles traversing the cytoplasm were less pronounced, consistent with their epithelial, non-fibroblastic morphology.

3.5. Expression of Typical Hepatocyte Markers in AML12 Cells

3.5.1. Next-Generation Sequencing

To obtain an unbiased, genome-wide mRNA expression profile of AML12 cells under routine culture conditions, we conducted next-generation bulk RNA sequencing (RNA-seq) on total RNA extracted from sub-confluent cultures. This was followed by standard quality control, library preparation, and sequencing. A complete overview of the RNA-seq dataset generated from AML12 cells, including all detected transcripts with raw read counts, normalized expression values (TPM), and associated gene annotations, is provided in Table S1. This table serves as a comprehensive reference for the global transcriptional landscape of AML12 under standard culture conditions and enables secondary analyses beyond the marker-focused evaluations presented in the main text. It also facilitates transparent re-use of the dataset for cross-study comparisons, pathway analyses, and the identification of additional lineage- or pathway-specific genes of interest.

For a more detailed analysis, we evaluated the expression of cell-type-enriched marker genes for hepatocytes, hepatic stellate cells, cholangiocytes, and Kupffer cells. The definitions and specificity annotations for these markers were obtained from the PanglaoDB database for the respective cell types. These marker panels are derived from curated lists in PanglaoDB [23] and the literature and are intended to reflect lineage enrichment rather than strict exclusivity, and some markers may appear in more than one cell type.

Expression of Hepatocyte-Specific Markers in AML12 Cells

RNA-seq analysis of AML12 cells reveals a distinct hepatocyte signature characterized by high expression of hepatocyte nuclear factor 4 alpha (Hnf4a, ~53 TPM) and forkhead box A2 (Foxa2, ~66 TPM), as well as detectable levels of prospero homeobox 1 (Prox1, ~9 TPM). Additionally, metabolic genes such as acyl-CoA dehydrogenase medium chain (Acadm, ~72 TPM) and ATP citrate lyase (Acly, ~146 TPM) are strongly expressed, indicating active hepatic lipid and energy metabolism (Table S2). The cells also exhibit secretory and transport functions, as evidenced by high expression of apolipoproteins like apolipoprotein B (Apob, ~112 TPM) and apolipoprotein A1 (Apoa1, ~2 TPM), as well as transporters such as the ATP binding cassette subfamily C member 3 (Abcc3, ~76 TPM) and ATP binding cassette subfamily D member 3 (Abcd3, 46 TPM). Acute-phase and iron-handling signals are prominent, with ceruloplasmin (CP, ~392 TPM) and GC and vitamin D binding protein (Gc, ~298 TPM) being highly expressed, while albumin (Alb) expression is modest (4.6 TPM) and the fetal marker Afp is detectable but low (0.901256 TPM), indicating an immortalized but hepatocyte-derived phenotype, and certain features of fully differentiated adult hepatocytes are diminished. These include urea cycle and gluconeogenic programs, with enzymes like carbamoyl-phosphate synthase 1 (Cps1, 0 TPM), argininosuccinate synthase 1 (Ass1, 6 TPM), ornithine carbamoyltransferase (Otc, ~0.26 TPM), phosphoenolpyruvate carboxykinase 1 (Pck1, 0 TPM), and glucose-6-phosphatase catalytic subunit (G6pc, 0 TPM) showing negligible expression levels. Key bile-acid-/drug-metabolizing cytochromes such as cytochrome P450 family 7 subfamily A member 1 (Cyp7a1, 0 TPM) and the Cyp3a family are either absent or expressed at very low levels. The cells also display stress-/progenitor-like traits, as indicated by high expression of transmembrane 4 L six family member 4 (Tm4sf4, 761 TPM) and phospholipid scramblase 1 (Plscr1, ~448 TPM), an interferon-responsive gene. Canonical serum proteins like serpin family A member 1 (Serpina1) are notably absent. Overall, these finding suggest that AML12 cells maintain a hepatocyte lineage program but show signs of partial de-differentiation and stress-related transcription typical of immortalized hepatocyte lines in 2D culture. However, it is important to note that our conclusions regarding the limited functionality of mature hepatocytes are based on these transcriptional patterns rather than direct measurements of metabolic fluxes or enzyme activities.

Lack of Hepatic Stellate Cell-Specific Markers in AML12 Cells

As a hepatocyte-derived line, AML12 cells are expected to show little-to-no expression of canonical activated hepatic stellate cell (HSC) cytoskeletal markers. Indeed, Acta2 (α-SMA), desmin (Des), transgelin (Tagln), myosin light chain 9 (Myl9), and glial fibrillary acidic protein (Gfap) are undetectable (Table S3). Low-level collagen transcripts are present: Col1a1 (1.7 TPM), Col1a2 (0.65 TPM), and Col3a1 (0.52 TPM), which are at levels far below the abundance seen in fully fibrogenic HSCs, as expected for hepatocytes in 2D culture. Unexpectedly for a pure hepatocyte line, several secreted extracellular matrix (ECM) modulators and inflammatory mediators are strongly expressed. These include C-C motif chemokine ligand 2 (Ccl2, 1718 TPM), insulin-like growth factor binding protein 7 (Igfbp7, 1536 TPM), cysteine-rich angiogenic inducer 61/cellular communication network factor 1 (Cyr61/Ccn1, 272 TPM), secreted protein acidic and cysteine rich (Sparc, 148 TPM), tissue inhibitor of metalloproteinase 1 (Timp1, 153 TPM), vascular endothelial growth factor A (Vegfa, 30 TPM), and insulin-like growth factor binding protein 3 (Igfbp3, 12 TPM). Hepatocyte identity remains evident with Alb (4.6 TPM) and the bile acid receptor nuclear receptor subfamily 1 group H member 4/farnesoid X-activated receptor (Nr1h4/Fxr, 30 TPM). A mixed “pre-activated” profile is further suggested by quiescence-associated retinol binding protein 1 (Rbp1, 70 TPM) alongside pro-inflammatory genes, and by very low platelet-derived growth factor receptor alpha (Pdgfra, 0.18 TPM) with absent fibroblast activation protein alpha (Fap) and ADAM metallopeptidase with thrombospondin type 1 motif 13 (Adamts13). Again, these data support the expected absence of the core contractile HSC program while revealing an unexpected secretory/pro-fibrotic and inflammatory signature that likely reflects stress or culture adaptation rather than true stellate cell contamination.

Lack of Cholangiocyte-Specific Markers in AML12 Cells

As a hepatocyte-derived cell line, AML12 would be expected to have minimal expression of cholangiocyte-specific genes. However, we have observed a strong expression of ductular/progenitor-associated genes, such as keratin 19 (Krt19, 162 TPM), keratin 7 (Krt7, 12 TPM), SRY-box 9 (Sox9, 55 TPM), HNF1 homeobox B (Hnf1b, 108 TPM), and the Notch ligand Jagged 1 (Jag1, 114 TPM). Additionally, secreted phosphoprotein 1 (Spp1)/osteopontin is exceptionally high (4642 TPM), which is unexpected for mature hepatocytes (Table S4). In contrast, key cholangiocyte markers like cystic fibrosis transmembrane conductance regulator (Cftr, 1.2 TPM), epithelial cell adhesion molecule (Epcam, 0 TPM), claudin 4 (Cldn4, 0 TPM), and aquaporin 4 (Aqp4, 0 TPM) remain low or absent. Trefoil factors are also largely undetectable, indicating an incomplete biliary differentiation program. However, hepatocyte characteristics are still present, as shown by detectable levels of injury/stress epithelial markers like alkaline phosphatase, liver/bone/kidney (Alpl, ~114 TPM), C-X-C motif chemokine ligand 1 (Cxcl1, ~68 TPM) and Lipocalin 2 (Lcn2, ~46 TPM). This confirms the hepatocytic origin rather than a cholangiocyte phenotype.

Lack of Kupffer-Cell-Specific Markers in AML12 Cells

Finally, as a hepatocyte-derived line, AML12 is expected to lack lineage-defining Kupffer cell markers. Indeed, transcripts such as adhesion G protein-coupled receptor E1 (Adgre1/F4-80), adhesion G protein-coupled receptor E4 (Adgre4), C-type lectin domain family 4 member F (Clec4f), V-set and immunoglobulin domain containing 4 (Vsig4), Cd163, macrophage receptor with collagenous structure (Marco), colony-stimulating factor 1 receptor (Csf1r), toll-like receptor 4 (Tlr4), triggering receptor expressed on myeloid cells 1 (Trem1), T cell immunoglobulin and mucin domain containing 4 (Timd4), and Spi-C transcription factor (Spic) are absent or near-baseline (≈0–0.3 TPM) (Table S5). The one notable myeloid feature is robust Cd14 expression (56 TPM), while other macrophage receptors and scavenger/chemokine receptors remain undetectable or minimal (e.g., colony-stimulating factor 1 receptor (Csf1r, ~0.3 TPM), C-C motif chemokine receptor 5 (Ccr5, 0 TPM) deoxyribonuclease 1 like 3 (Dnase1l3, 0 TPM), and macrophage scavenger receptor 1 (Msr1, 0 TPM)). In contrast, iron-handling and metabolic genes characteristic of hepatocytes are prominent, including ferritin light chain (Ftl, 10,561 TPM) and solute carrier family 40 member 1 (Slc40a1/ferroportin, 232 TPM), with additional epithelial/metabolic factors such as peroxisome proliferator-activated receptor delta (Ppard, 17 TPM) and phospholipid transfer protein (Pltp, 3.5 TPM), present at modest levels. Complement C1q transcripts are low-to-absent (C1qa ~0.66 TPM; C1qb/c 0 TPM), and inflammatory mediators like Il1b and Tlr9 are not detected, with Mmp13 being only modest (2.7 TPM), arguing against meaningful Kupffer cell characteristics and instead reflecting hepatocytes with a mild innate–immune tone under culture conditions.

3.5.2. Verification of Next-Generation Sequencing Data

To validate the RNA-seq findings at the transcript and protein level, we next examined the expression of selected hepatocyte markers in AML12 cells using conventional RT-PCR (Figure 5A), quantitative RT-PCR (Figure 5B), and Western blotting (Figure 5C). RT-PCR confirmed the presence of mRNAs encoding canonical hepatocyte transcription factors and structural markers (e.g., Hnf4a, Foxa2, Krt18), as well as housekeeping genes such as Gapdh and peptidylprolyl isomerase A (Ppia), in line with the NGS-derived expression profile. However, Alb transcript levels were markedly lower than in primary hepatocytes or whole liver. Consistently, Western blot analysis detected robust expression of key hepatocyte proteins, including HNF4α, α-2-macroglobulin, Cyclophilin A, and GAPDH in AML12 lysates. However, it should be noted that the data established by conventional RT-PCR are qualitative and are meant only to corroborate the RNA-seq-based detection of the respective mRNAs. Albumin protein was only detectable in mouse liver extracts, indicating that low-level Alb transcripts do not translate into appreciable albumin protein expression under our culture conditions. Again, these data obtained in the Western blot analysis are intended to validate marker presence rather than to provide precise quantitative comparisons between the three sample types.

To further confirm the hepatocyte lineage identity of AML12 cells and verify the nuclear localization of HNF4α, we conducted immunofluorescence staining using a specific anti-HNF4α antibody and an isotype-matched goat IgG control (Figure 6). When cultured under standard conditions, almost all AML12 cells displayed strong HNF4α-specific fluorescence primarily localized in the nuclei, as indicated by co-localization with DAPI staining, supporting the presence of a robust hepatocyte transcription factor program in the majority of cells. In contrast, cells stained with the control IgG showed minimal background signal, again confirming the presence of HNF4α protein in AML12 cells and reinforcing the presence of a robust hepatocyte transcription factor program in AML12 cultures.

3.6. Karyotype Based on Molecular Cytogenetic Analyses

Molecular cytogenetic analysis revealed that AML12 is highly aneuploid with a near-tetraploid chromosome set of 81<4n> and a female sex chromosome complement (XX), despite its reported male origin, suggesting extensive sex chromosome instability. The composite karyotype is 81<4n>,XX,del(X)(A3)x2,-Y,-Y,+1,-2[11/20],der(3)t(2;3)(A2;H4)[11/20],−4,+6,−12,+15,−16,del(17)(D1)x2,dic(X;17)(:XA3->XA1::17A1->17qter)x2,−18,+19,+19,+19,+19[cp20]. This means that the cell line consists of two clones: one with four normal chromosomes 2 and 3 and one with three chromosomes 2 and a derivative chromosome 3 der(3)t(2;3)(A2;H4). The breakpoint architecture of the X;17 dicentric and 2;3 derivative chromosomes was only resolvable after mcb (Figure 7 and Figure 8).

Segmental imbalances across mouse chromosomes were determined through in silico comparative genomic hybridization-style copy number mapping, as resolved by mcb and summarized in the accompanying table. These include gains of 1A1-qter, 2A1-A2, 6A1-qter, 15A1-qter, and 19A1-qter and losses of 3H4-qter, 4A1-qter, 12A1-qter, 16A1-qter, 17D1-qter, 18A1-qter, XA3-qter, and YA1-qter. Human cytoband projections are provided for comparison.

Segmental copy number imbalances mapped from mouse to human cytogenomic coordinates highlight recurrent gains such as 1q23-qter, 2q13-q22/2q32.3-qter, 6p13-q13, 7p22-p14, 7q21-q36, 8q22-qter, and 12p13-p11, as well as losses including 1p, 3p12-q21/3q27-qter, 16p13, 18p11.2-q21.3, Xp21.1-q24, and Yp-qter. These imbalances define the complex genomic landscape of the AML12 cell line (Table 4).

Notably, 57.5–68% of these alterations overlap with recurrent copy number changes reported in hepatocellular carcinoma, and 34–41% overlap with those in cholangiocarcinoma, highlighting a transformation-like profile (Table 5) [30,31].

3.7. In Silico Comparative Genomic Hybridization-Style Copy Number Mapping

To visualize the global distribution of chromosomal imbalances in AML12, we used an in silico comparative genomic hybridization-style copy number mapping approach. This approach integrates segmental gains and losses inferred from mFISH/mcb with their positions on the mouse reference genome. This strategy enabled us to project murine copy number changes onto orthologous human cytobands, creating genome-wide profiles that summarize the complex pattern of recurrent amplifications and deletions mentioned earlier. The resulting virtual aCGH plots for the mouse and mapped human genomes (Figure 9 and Figure 10) highlight broad gains affecting chromosomes 1, 2, 6, 15, and 19, along with corresponding losses on chromosomes 3, 4, 12, 16, 17, 18, X, and Y. This provides an integrated overview of the AML12 karyotype that complements metaphase-based analyses. These virtual aCGH plots are in silico reconstructions that integrate the segmental gains and losses inferred from mFISH/mcb. No independent array-based or low-pass whole-genome sequencing CNV assay was performed.

4. Discussion

This study provides an integrated cytogenetic and transcriptomic analysis of AML12 cells, revealing a cell line that maintains a clear hepatocyte lineage program while displaying significant karyotypic complexity and transcriptional features indicative of partial de-differentiation and stress adaptation in standard 2D culture. To our knowledge, this is the first report to combine multicolor cytogenetics, STR-based authentication, and RNA-seq-driven lineage profiling into a unified reference dataset for AML12.

Multicolor FISH and mcb banding reveal a near-tetraploid background with two related clones and multiple recurrent structural and numerical aberrations; these include del(X)(A3)×2, dic(X;17)(:XA3→XA1::17A1→17qter)×2, der(3)t(2;3)(A2;H4), and whole-chromosome gains and losses (+1, −2, −4, +6, −12, +15, −16, −18, and multiple +19), with copy numbers ranging from two to seven per chromosome in individual cells. The composite karyotype (81<4n>, XX) is consistent with the loss of both Y chromosomes from a line initially described as male, highlighting ongoing sex-chromosome instability in long-term culture. When mapped to human cytobands, segmental gains (e.g., 1q23-qter, 2q13-q22 and 2q32.3-qter, 6p13-q13, 7p22-p14 and 7q21-q36, 8q22-qter, 12p13-p11) and losses (e.g., 1p, 3p12-q21 and 3q27-qter, 16p13, 18p11.2-q21.3, Xp21.1-q24, Yp-qter) show significant overlap with recurrent copy number changes reported in hepatocellular carcinoma (57.5–68%) and a smaller but notable overlap with cholangiocarcinoma (34–41%), highlighting transformation-like features that can arise in immortalized lines despite their non-tumorigenic origins.

At the transcriptome level, RNA-seq confirms a hepatocyte identity with the expression of lineage-defining transcription factors (such as Hnf4a, Foxa2, Prox1) and core metabolic modules (including mitochondrial β-oxidation and lipid metabolism), alongside apolipoproteins and transporters consistent with a secretory, xenobiotic-responsive phenotype. However, several hallmarks of fully differentiated adult hepatocytes are diminished: albumin expression is modest, urea cycle and gluconeogenic enzymes are low-to-absent, and key drug-/bile-acid-metabolizing cytochromes (such as Cyp7a1 and major Cyp3a family members) are barely detectable. This collectively indicates partial de-differentiation typical of immortalized hepatocytes in 2D culture. Concurrently, the cells exhibit a stress- or progenitor-like transcriptional profile, including high expression of interferon-/stress-associated genes (like Plscr1) and epithelial/progenitor markers. This suggests culture adaptation rather than a loss of lineage.

Two unexpected but informative aspects of the transcriptome refine how AML12 should be interpreted in vitro. Firstly, despite lacking the cytoskeletal/contractile program that defines activated hepatic stellate cells (Acta2, Des, Tagln, Myl9, Gfap all undetectable), AML12 expresses high levels of secreted extracellular matrix (ECM) modulators and inflammatory mediators (e.g., Ccl2, Igfbp7, Cyr61/Ccn1, Sparc, Timp1), together with quiescence-associated Rbp1, yielding a mixed “secretory/pro-fibrotic” profile without a bona fide HSC identity. Secondly, AML12 shows robust induction of ductular/progenitor-associated cholangiocyte markers (Krt19, Krt7, Sox9, Hnf1b, Jag1) and very high Spp1/osteopontin, yet it lacks several hallmark transporters of mature cholangiocytes (e.g., Cftr is low; Epcam and Cldn4 are absent), indicating a partial ductular program superimposed on a hepatocyte backbone rather than full cholangiocytic transdifferentiation. Importantly, lineage-defining Kupffer cell markers are broadly absent (e.g., Adgre1/F4-80, Clec4f, Vsig4, Csf1r), arguing against meaningful macrophage contamination; the presence of Cd14 with strong iron-handling signatures (very high Ftl and elevated Slc40a1) is compatible with hepatocytes in a mild innate–immune state rather than a mixed hepatocyte–macrophage culture. Together, these data support a model in which AML12 retains hepatocyte identity but adopts a progenitor-like, injury-responsive transcriptional program under routine conditions, a state that is common in immortalized epithelial lines.

Compared to other commonly used hepatocyte-derived cell lines such as mouse Hepa1-6 [32], human cell line HepG2 [33], HuH-7 [22,34], and Hep3B [33], the integrated molecular profile we describe positions AML12 in a distinct niche along the spectrum from primary hepatocytes to fully transformed hepatocellular carcinoma lines. On one hand, AML12 retains lineage-defining transcription factors (Hnf4a, Foxa2, Prox1), a broad set of mitochondrial and peroxisomal metabolic genes, and ultrastructural features such as abundant rough endoplasmic reticulum, peroxisomes, and bile canaliculi-like structures that are characteristic of hepatocytes [2,35,36]. On the other hand, the near-tetraploid karyotype with complex structural rearrangements, the marked overlap of its copy number changes with human HCC and CCA, and the partial loss of adult hepatocyte functions (including urea cycle, gluconeogenic, and key Cyp transcripts) argue that AML12 should not be regarded as a surrogate for fully differentiated, genomically stable liver parenchyma [4,37]. Instead, our data underscore the notion that AML12 occupies an intermediate position, combining many experimentally convenient features of immortalized lines with a lineage program that still reflects its hepatocyte origin but with important constraints on the interpretation of metabolic- and xenobiotic-handling endpoints [4].

These dual features suggest that AML12 may be particularly informative for modeling early or pre-malignant stages of hepatocellular transformation, in which hepatocytes have acquired substantial genomic instability and stress-responsive transcriptional programs but have not yet progressed to overt carcinoma [1]. In this context, the strong induction of ductular/progenitor markers (Krt19, Krt7, Sox9, Hnf1b, Jag1, Spp1) on a hepatocyte backbone, together with the secretion of extracellular matrix modulators and chemokines (e.g., Ccl2, Sparc, Timp1, Igfbp7), mirrors the mixed hepatocyte–ductular, pro-fibrotic, and inflammatory signatures observed at the interface of chronic injury, ductular reaction, and early neoplasia in vivo [38]. Interestingly, others have identified significantly dysregulated genes associated with cell death and survival that are essential for early HCC transformation in humans, including TGFBR1, RBM5, THOC2, USP11, MDM2, TPP2, EIF4E, VRK1, CCNA2, ZMAT3, AGPAT2, CREB1, FIGNL1, and GSK3B [39]. Our NGS data indicates that all these are also highly expressed in AML12 cells. AML12 therefore offers a tractable platform for cross-species transcriptomic analysis of gene networks to dissect mechanisms of epithelial plasticity, hepatocyte-intrinsic inflammatory signaling, and matrix remodeling while still allowing for alignment with in vivo murine models that share the same genetic background. Such applications may be particularly powerful when AML12 is used in co-culture or conditioned-medium experiments with hepatic stellate cells, cholangiocytes, or immune cells, where the secretory profile we define can be explicitly integrated into experimental design and interpretation.

In summary, under basal conditions, AML12 cells integrate hepatocyte [23,40,41,42,43,44,45,46,47,48], ductular/progenitor [23,49,50], and cholangiocyte signatures [23,51,52,53], highlighting the coexistence of robust hepatocyte programs with pronounced ductular/progenitor features. This is consistent with a mixed hepatocyte–ductular, injury-adapted phenotype (Figure 11).

While Figure 11 summarizes lineage-associated gene expression in AML12 cells, it is important to emphasize that many of these markers are lineage-enriched rather than truly lineage-specific. Additionally, several markers are known to be shared across transitional or injury-associated states. The TPM-based cutoffs used to define “high”, “medium”, or “low” expression provide a practical visualization, but they do not constitute a formal sensitivity threshold. Low-level expression may still be biologically meaningful or reflect technical background noise. Therefore, the marker patterns in Figure 11 should be interpreted as indicative of dominant lineage programs rather than as definitive, highly specific or fully sensitive classifiers of cell identity.

Several implications follow for experimental design and interpretation. The extensive aneuploidy and clonal heterogeneity documented here can shift gene dosage and pathway activity, potentially contributing to the observed bias toward stress, ductular/progenitor, and ECM-modulating programs. Such dosage effects are also consistent with the substantial overlap between AML12 copy number imbalances and those recurrently observed in human HCC and, to a lesser extent, CCA. Functionally, the reduced expression of gluconeogenic, urea cycle, and key Cyp genes cautions against using AML12 as a stand-alone surrogate for adult hepatocyte physiology in studies requiring high-fidelity xenobiotic metabolism, nitrogen handling, or bile-acid synthesis. Pre-conditioning with appropriate nuclear receptor agonists, adaptation to 3D or matrix-supported cultures, or complementary validation in primary hepatocytes may be warranted depending on the endpoint.

An important practical consequence of the karyotypic and transcriptional heterogeneity we observe is that different laboratories, or even different freeze–thaw cycles within the same laboratory, may work with AML12 populations that occupy slightly different positions along this phenotypic spectrum. Because the two major cytogenetic clones share a common set of structural abnormalities yet differ in the presence of der(3)t(2;3) and specific whole-chromosome losses, it is conceivable that relative clonal proportions will drift over time as a function of passage number, culture conditions, and selection imposed by particular experimental manipulations [54]. In practice, this argues for establishing low-passage master stocks, limiting the total passage range used in critical studies, and periodically re-evaluating key phenotypic readouts (for example, the expression of hepatocyte, ductular, and inflammatory markers) to ensure that conclusions are not driven by unrecognized clonal shifts [54]. It also underpins the advice for cell storage and banking that, in general, immortalized cells should be frozen at an early passage number to provide a reputable source [55]. When marked changes in functional behavior are observed after long-term culture or strong selective pressure, re-assessment of karyotype features and comparison with the reference profile presented here can help distinguish true biological effects of the experimental perturbation from culture adaptation artifacts [40]. Conversely, the significant high expression of genes such as Spp1 and Ccl2 suggests that AML12 can serve as a practical model for investigating hepatocyte-intrinsic responses to stress, inflammation, and epithelial plasticity. However, it is important to consider the underlying progenitor-like state when drawing conclusions.

This work also emphasizes best practices for cell line quality control. Our STR profile shows strong concordance with the Cellosaurus AML12 reference at multiple loci, with only one discrepancy and several loci lacking public reference calls. All alleles fall within known ranges, confirming the authenticity of our stock and highlighting the importance of routine authentication in long-term projects. Given the observed karyotypic dynamics and clonal diversity, we suggest monitoring passage windows, periodically re-evaluating karyotype features in critical studies, and verifying key functional readouts (e.g., marker panels for hepatic, ductular, and inflammatory states) before and after significant perturbations.

5. Limitation of This Study

Our study has several limitations that need to be addressed in future studies. The pronounced chromosomal instability observed in AML12 raises the question of which underlying mechanisms drive the accumulation of structural and numerical aberrations. It is not known if the identified mouse CNVs in AML12 cells fall within heterochromatic regions or segmental duplication. Recently, it was proposed that segmental copy number gains might result from a rearrangement process termed breakage–replication/fusion [56]. However, without further molecular data on the base pair level of the breakpoint region, we cannot speculate on the mechanisms underlying the observed chromosomal aberrations in AML12 cells. Moreover, we do not know if the observed pattern of CNVs in AML12 cells is consistent with chromoanasynthesis, chromothripsis, or chromoplexy, which are mechanisms [57].

In our dataset, breakpoints of the major rearrangements can be delineated at the level of cytogenetic bands by mFISH/mcb (e.g., XA3, 2A2, 3H4, 17D1), and these are summarized in Table 4, but we do not currently have base-pair-resolved breakpoint information. Accordingly, we cannot reliably distinguish between mechanisms such as breakage–fusion–bridge cycles, chromothripsis, chromoanasynthesis, or chromoplexy at the sequence level. Given the presence of two related clones and ongoing aneuploidy, together with the absence of a single catastrophic rearrangement pattern, we consider it more likely that the observed karyotype reflects stepwise, culture-driven adaptation with accumulation of mutations over time, a scenario frequently encountered in long-term propagated cell lines rather than a one-off chromothriptic event.

Additional limitations of our study include relying primarily on mRNA (and selected protein) measurements to infer functional capacities, without performing dedicated assays of urea cycle activity, gluconeogenic flux, or cytochrome-P450-mediated xenobiotic metabolism. We also focused on cells in standard 2D culture without testing condition-dependent plasticity (e.g., sandwich cultures, 3D spheroids, or differentiation paradigms), and we used cross-species mapping to human cytobands to contextualize murine CNAs.

Because our AML12 cultures were maintained in DMEM + 10% FBS without the insulin, transferrin, selenium, and dexamethasone supplements recommended by ATCC, we cannot exclude the possibility that certain functional and transcriptional features, especially those related to albumin synthesis, urea production, and specific aspects of xenobiotic metabolism, may be quantitatively different under ATCC-recommended conditions. However, the cytogenetic and STR profiles are expected to be generally applicable across different media formulations.

Moreover, we did not generate matched bulk or single-cell RNA-seq datasets from primary hepatocytes or intact liver for direct transcriptome-wide comparison, and we did not perform an unbiased global proteomic analysis of AML12. As a result, our conclusions about limited mature hepatocyte functionality and overlap with in vivo liver cell states should be viewed as expression-based inferences that require confirmation by functional assays and integrative multi-omics studies in primary cells and animal models. Nonetheless, the combination of mFISH/mcb cytogenetics, STR authentication, and cell-type-informed RNA-seq provides a clear picture of AML12 as a genomically unstable yet lineage-faithful hepatocyte line that exists in a progenitor-like, injury-responsive state in vitro.

Future work could build on our reference map in several directions. Sophisticated strategies that combine single-cell multi-omics approaches with karyotyping or copy-number profiling, as well as transcriptomics [58] and integrative single-cell multi-omics subclone mapping [59], would help disentangle how specific aneuploidies and structural variants in subclones shape the transcriptional states we observed at the bulk level. This could reveal whether the ductular/progenitor-like and pro-fibrotic signatures are limited to particular subclones or are widely shared across the population. Other useful methods might enable the study of cell state diversity or help identify the mechanisms by which the genome and transcriptome interact in the production of cellular subclones observed in our AML12 cultures [60].

Similarly, comparing AML12 cells cultured under 3D, matrix-supported, or hormonal differentiation conditions with the 2D baseline reported here could clarify the extent to which their reduced urea cycle, gluconeogenic, and Cyp expression is reversible. It could also determine whether more physiological configurations shift the balance between hepatocyte and ductular programs. Finally, targeted genome editing or comparative analyses with additional murine hepatocyte lines lacking specific CNAs might help identify which of the recurrent gains and losses are functionally most relevant for the distinctive stress-responsive, progenitor-like phenotype of AML12.

In summary, AML12 remains a valuable and manageable murine hepatocyte model, especially for mechanistic studies of hepatocyte signaling, stress responses, and epithelial plasticity. However, caution should be taken when addressing questions that require fully mature hepatocyte functions. The cytogenetic and transcriptomic reference provided here serves as a practical framework to assist with experimental design, benchmarking culture adaptations, and interpreting results within the context of the capabilities and limitations of AML12 cells.

6. Conclusions

In this work, we present an integrated cytogenetic and transcriptomic reference for AML12, clarifying its strengths and limitations as a murine hepatocyte model. Cytogenetically, AML12 cells have a near-tetraploid karyotype with extensive numerical and structural aberrations across two related clones. Cross-species mapping of copy number imbalances reveals significant overlap with recurrent alterations in human hepatocellular carcinoma and some overlap with cholangiocarcinoma, indicating transformation-like genomic features acquired in culture. At the RNA level, AML12 cells maintain a clear hepatocyte lineage program but show partial de-differentiation with decreased expression of urea cycle, gluconeogenic, and key xenobiotic-metabolizing transcripts. This is consistent with an immortalized, 2D-adapted state. Additionally, the cells up-regulate ductular/progenitor markers (such as Krt19, Sox9, Hnf1b, Jag1, Spp1) while lacking several hallmark features of fully differentiated cholangiocytes, suggesting an incomplete biliary-like program on a hepatocyte backbone. They also express high levels of secreted ECM modulators and inflammatory mediators without the contractile cytoskeletal signature of activated hepatic stellate cells, indicating a stress-responsive, pro-fibrotic hepatocyte state. Kupffer cell identity is similarly unsupported, with core macrophage-restricted markers largely absent. Isolated Cd14 expression and strong iron-handling transcripts align with hepatocyte biology under mild inflammatory conditions. STR profiling confirms close concordance with the Cellosaurus AML12 reference, supporting the authenticity of the analyzed stock. Overall, AML12 remains a valuable platform for mechanistic studies of hepatocyte signaling, stress responses, and epithelial plasticity. However, for applications requiring fully mature metabolic functions or precise bile acid/xenobiotic handling, complementary models or condition-dependent optimization should be considered.