AUF1 helps prevent liver cell aging by keeping mitochondria balanced in a liver cell model

Mouse hepatocyte AML12 cells (American Type Culture Collection (ATCC), Manassas, VA, USA; CRL-2254) were purchased and cultured in Dulbecco's Modified Eagle Medium/Nutrient Mixture F12 (DMEM/F12; Capricorn Scientific, Ebsdorfergrund, Germany) supplemented with 10% fetal bovine serum (FBS), 1% antibiotics, 1× Insulin-Transferrin-Selenium Pyruvate Supplement (ITSP; Welgene, Gyeongsan, Republic of Korea), and 100 nM dexamethasone (Sigma-Aldrich, Burlington, MA, USA) at 37 °C. Small interfering RNAs (siRNAs; BIONEER, Daejeon, Republic of Korea) and plasmids were transfected using Lipofectamine™ 2000 (Invitrogen™, Waltham, MA, USA), according to the manufacturer's instructions. Sequences of siRNAs used in this study are provided in Supplementary Table S1. For adenoviral transduction, AML12 cells were infected with Ad-AUF1 virus (Ad-CMV-h-HNRNPD) or control adenovirus (Ad-CMV-Null) (Vector Biolabs, Malvern, PA, USA) at 2 × 10⁸ viral particles in serum-free medium for 4 h, after which the medium was replaced with complete culture medium.

To establish a hepatocyte senescence model, AML12 cells were continuously exposed to H₂O₂ for 7 days, following the protocols described in [23]. Briefly, on Day 1, cells were treated with 1 mM H₂O₂ (Sigma-Aldrich) in serum-free medium for 1 h at 37 °C. On Days 2–3, cells were treated with 750 μM H₂O₂ for 1 h each day at 37 °C and were subsequently sub-cultured on Day 3. On Days 4–6, cells were similarly treated with 750 μM H₂O₂ for 1 h daily at 37 °C, and cells were used for experiments on Day 7. Senescence induction was assessed by cell counting, BrdU incorporation assay (Merck, Taufkirchen, Germany), and TUNEL assays (Promega Corporation, Madison, WI, USA).

Cells were lysed using RIPA buffer (Biosesang, Inc., Seongnam, Republic of Korea) containing 1× protease inhibitor cocktail (Roche, Basel, Switzerland), and proteins were separated by SDS-PAGE. Proteins were transferred onto PVDF membrane (Millipore, Burlington, MA, USA) and incubated with primary antibodies against AUF1 (Merck; 07-260), p16 (Santa Cruz Biotechnology, Inc., Dallas, TX, USA; SC-1661), p21 (BD Bioscience, Franklin Lakes, NJ, USA; 556431), OPA1 (612606), DRP1 (611112), MFN2 (Abcam Plc., Cambridge, UK; ab56889), MFN1 (ab104274), MFF (ab81127), and β-actin (Genetex, Inc., Irvine, CA, USA; GTX629630), then sequentially incubated with HRP-conjugated secondary antibodies (Merck; AP124P and AP132P). Chemiluminescence was generated by Clarity Western ECL Substrate (Bio-Rad, Inc., Hercules, CA, USA) and detected by the ChemiDoc Imaging Systems (Bio-Rad, Inc.).

Total RNAs were extracted using RNAiso Plus (Takara Bio, Inc., Shiga, Japan) and reverse-transcribed into cDNA using the ReverTra^® Ace qPCR RT Kit (Toyobo Co., Ltd., Osaka, Japan). Relative RNA levels were analyzed by quantitative PCR (qPCR) using SensiFAST™ SYBR Hi-ROX kit (Meridian Bioscience, Inc., Cincinnati, OH, USA). Primer sequences used for qPCR are listed in Supplementary Table S1. Data were relatively quantified using the _ΔΔCT method, and Gapdh mRNA was used as the reference gene for normalization.

For RNP-IP, RNP complexes were immunoprecipitated using AUF1 (Merck; 07-260) or control IgG (Merck; 12-370) antibodies conjugated to Pierce™ Protein A Agarose (Thermo Fisher Scientific, Waltham, MA, USA). RNA was isolated from the RNP complexes, reverse-transcribed into cDNA, and analyzed using qPCR as described previously [24,25].

For synthesis of biotinylated RNA probes, PCR fragments containing T7 RNA polymerase binding sequence ([T7]:CCAAGCTTCTAATACGACTCACTATAGGGAGA) were amplified using Blend Taq^TM (Toyobo Co., Ltd.) and specific primers targeting the 3′UTR of mouse Opa1 (NM_001199177) and Mfn2 (NM_001285920) listed in Supplementary Table S1. The amplified PCR products were transcribed in vitro using MaxiScript T7 kit (Ambion, Waltham, MA, USA) and biotin-CTP (Enzo Life Sciences, Farmingdale, NY, USA). The biotinylated RNA probes were incubated with cell lysates and pulled down using Dynabeads™ Streptavidin Magnetic Beads (Invitrogen™), and the binding of AUF1 to RNAs was assessed by immunoblotting.

For immunofluorescence microscopy, cells were fixed using 4% FA (Biosesang, Inc.) and permeabilized with 0.4% Triton X-100 in PBS. After blocking, cells were incubated with primary antibody against NDUFV2 (Proteintech Group, Inc., Rosemont, IL, USA; 15301-1-AP) and further incubated with secondary antibody conjugated with Alexa Fluor^® 555 (Abcam Plc.; ab150074). Fluorescence signals from mitochondria were imaged using a ZEISS LSM 900 confocal microscope (Carl Zeiss, Oberkochen, Germany). Mitochondrial number, area, perimeter, and branch length were analyzed using the Mitochondrial Analyzer plugin (https://github.com/AhsenChaudhry/Mitochondria-Analyzer↗, accessed on 23 December 2025) for ImageJ/Fiji software version 1.54p (NIH, Bethesda, MD, USA), as previously described methods [26].

For analysis of mitochondrial Ca²⁺ levels and mitochondria-ER contact sites (MERCs), cells were transfected with pcDNA-4mtD3cpv (mitochondrial Ca²⁺ sensor; Addgene, Watertown, MA, USA; #36324) [27], SPLICS Mt-ER Short P2A (Addgene #164108), and SPLICS Mt-ER Long P2A (Addgene; #164107) [28,29]. Fluorescence signals from fixed cells were acquired using a ZEISS Axio Imager Z1 microscope (Carl Zeiss) and quantified using ImageJ/Fiji software (NIH) [30,31].

For transmission electron microscopy (TEM), cells were fixed with 2.5% glutaraldehyde and embedded in epoxy resin (Polysciences, Inc., Warrington, PA, USA). Ultrathin sections were prepared using an ultramicrotome (Leica Microsystems Ltd., Wetzlar, Germany), and mitochondrial morphology was examined using an HT7800 transmission electron microscope (Hitachi, Tokyo, Japan).

Mitochondrial ATP levels and membrane potential (Δψ_m) were analyzed using the Mitochondrial ToxGlo™ Assay (Promega Corporation; G8000) and the JC-1 Mitochondrial Membrane Potential Assay Kit (Abcam Plc.; ab113850), respectively, according to the manufacturer's instructions. Luminescence signals from ToxGlo™ Assay and fluorescence signals from JC-1 (Ex/Em = 530/590 nm) were measured using a BioTek Synergy H1 microplate reader (Agilent Technologies, Winooski, VT, USA).

Cell proliferation was evaluated by cell counting using a hemocytometer, and cell morphology was examined using a Leica DM IL LED microscope (Leica Microsystems Ltd.). For senescence-associated β-galactosidase (SA β-gal) staining, cells were incubated with an SA β-gal staining solution (40 mM citric acid/sodium phosphate buffer, pH 6.0; 150 mM NaCl; 2 mM MgCl₂; 5 mM potassium ferricyanide; 5 mM potassium ferrocyanide; and 1 mg/mL X-gal (BEAMS Biotechnology, Seongnam, Republic of Korea) at 37 °C for 16 h in the dark, and stained cells were subsequently examined using an Olympus IX70 microscope (Olympus Corp., Tokyo, Japan). SA β-gal-positive cells were quantified using ImageJ/Fiji software. The expression of senescence markers was assessed by RT-qPCR and immunoblotting analyses.

Intracellular and mitochondrial ROS levels were analyzed using 5 μM CM-H₂DCFDA (Invitrogen™) and 5 μM MitoSOX™ Red mitochondrial superoxide indicator (Invitrogen™), respectively, according to the manufacturer's instructions. CM-H₂DCFDA fluorescence intensity was evaluated using a FACS Canto™ flow cytometer (BD Biosciences). MitoSOX™ fluorescence signals were visualized using a ZEISS LSM 900 confocal microscope (Carl Zeiss) and quantified with ImageJ/Fiji software.

Data are represented as the mean ± SEM of three independent experiments. Statistical significance was determined using Student's t-test for comparisons between two groups and one-way ANOVA with Tukey's test for multiple comparisons. (* p < 0.05; ** p < 0.01; *** p < 0.001).

To investigate hepatocyte-intrinsic mechanisms of cellular senescence, we established a senescent AML12 model by continuously exposing the cells to H₂O₂ following the STAR Protocols method [23]. Over the time course, cells developed characteristic morphological changes (Figure 1A), accompanied by a progressive decrease in cell number (Figure 1B), which appeared to reflect reduced proliferation rather than increased cell death (Figure 1C,D). By Day 7, senescence was confirmed by an increased portion of SA β-gal-positive cells (Figure 1E) and elevated expression of canonical markers, including p16, p21, and p53 (Figure 1F). Levels of p21 and p16 were increased in senescent cells (Sen) compared with proliferating control cells (Pro) (Figure 1G). These results indicate that cellular senescence was successfully induced in AML12 cells. Notably, AUF1 expression was significantly reduced at both mRNA and protein levels in senescent AML12 cells compared to proliferating controls (Figure 1F,G), suggesting that AUF1 may act as a regulator of hepatocyte senescence.

Mitochondrial dysfunction is a hallmark of cellular senescence [3,32,33]. To investigate mitochondrial architecture and function in the senescent AML12 model, we assessed mitochondrial networks by quantitative morphometry following immunofluorescence staining for the complex I subunit NDUFV2 [26]. Relative to proliferating control cells, senescent AML12 cells exhibited a reduced mitochondrial number, accompanied by an increased mean area, perimeter, and branch length, consistent with an elongated, fusion-dominant morphology (Figure 2A). This structural remodeling was corroborated by TEM, which revealed more elongated mitochondria in senescent AML12 cells (Figure 2B). In line with chronic oxidative stress, cellular ROS (CM-H₂DCFDA) and mitochondrial ROS (mitoSOX) were elevated in senescent cells (Figure 2C,D). Functionally, senescent AML12 cells showed reduced mitochondrial ATP synthesis and diminished mitochondrial membrane potential (Δψ_m) as measured by JC-1 staining, indicating a bioenergetic compromise (Figure 2E,F).

Given the established function of mitochondria-endoplasmic reticulum contact sites (MERCs) in regulating Ca²⁺ exchange and thereby influencing mitochondrial dynamics and bioenergetics [34,35,36], we next asked whether MERCs are remodeled in senescent AML12 cells. Using the split-GFP contact-site sensor SPLICS [28,29], we monitored short-range (SPLICS-S) and long-range (SPLICS-L) mitochondria-ER appositions. As shown in Figure 2G, senescent AML12 cells exhibited reduced SPLICS-S fluorescence with a concomitant increase in SPLICS-L signal compared to proliferating control cells, indicating a loss of tight MERCs and a shift toward more distant contacts. Consistent with the reduced contacts, mitochondrial Ca²⁺ uptake, assessed with the mitochondria-targeted FRET probe 4mtD3cpv [27], was diminished in senescent cells (Figure 2H). Collectively, these data demonstrate that senescent AML12 cells undergo mitochondrial elongation accompanied by MERCs reorganization and impaired Ca²⁺ transfer, providing a mechanistic link to the observed bioenergetic decline.

To determine whether AUF1 is a functional regulator of hepatocyte senescence, we downregulated AUF1 in AML12 cells by transient transfection of siRNA and confirmed AUF1 downregulation at the protein level (Figure 3A). AUF1-knockdown (AUF1-KD) AML12 cells developed the characteristic senescent morphology, becoming larger and more flattened in shape (Figure 3B). Furthermore, AUF1-KD decreased cell number and increased the number of SA β-gal-positive cells in AML12 cells (Figure 3C,D). In addition, AUF1 downregulation increased the expression of senescence markers at both the mRNA and protein levels (Figure 3E,F). These results indicate that AUF1 reduction is sufficient to drive AML12 hepatocytes into a senescent state.

To determine whether AUF1 downregulation is sufficient to induce the mitochondrial abnormalities observed in senescent AML12 cells, we analyzed mitochondrial morphology, function, and MERCs organization under AUF1-KD. Analysis of NDUFV2 immunofluorescence and TEM revealed enhanced mitochondrial elongation in AUF1-KD cells (Figure 4A,B). ROS levels increased at both the whole-cell and mitochondrial levels relative to controls (Figure 4C,D). AUF1-KD also reduced mitochondrial ATP levels and decreased mitochondrial membrane potential (Figure 4E,F). Furthermore, short-range contacts (SPLICS-S) decreased, while long-range contacts (SPLICS-L) increased under AUF1-KD (Figure 4G). In accordance with this looser coupling between mitochondria and ER, mitochondrial Ca²⁺ uptake, as measured by 4mtD3cpv probe, was reduced in AUF1-KD cells (Figure 4H). Collectively, AUF1-KD resulted in mitochondrial remodeling and functional deficits as seen in senescent AML12 cells. These results suggest that AUF1 is an upstream regulator in the mitochondria-senescence axis in AML12 hepatocytes.

Because AUF1-KD induced a fusion-dominant mitochondrial morphology, we investigated whether AUF1 modulates the expression of genes governing mitochondrial dynamics. In AML12 hepatocytes, AUF1-KD increased expression of Opa1 and Mfn2, consistent with mitochondrial elongation; although Drp1 and Mff exhibited modest statistically significant changes, we focused subsequent analyses on Opa1 and Mfn2 to test a fusion-biased remodeling mechanism (Figure 5A). RT-qPCR confirmed increased levels of Opa1 and Mfn2 mRNAs in AUF1-KD AML12 cells (Figure 5B). Since AUF1 is an RNA-binding protein mainly linked to mRNA decay [19,22], we investigated whether Opa1 and Mfn2 mRNAs physically associate with AUF1. RNP-IP revealed enrichment of Opa1 and Mfn2 mRNAs in AUF1-containing ribonucleoprotein complexes compared with IgG control (Figure 5C). In silico analysis identified GU/UG-rich or AU-rich RNA sequences within the 3′UTRs of Opa1 and Mfn2 mRNAs that are predicted to be accessible for AUF1 (Supplementary Figure S1) [17,37]. A pulldown assay using biotinylated RNAs harboring defined 3′UTR fragments revealed direct interaction of AUF1 (Figure 5D). Taken together, these data suggest that AUF1 binds the mRNAs of Opa1 and Mfn2, and limits their abundance. Loss of AUF1 therefore promotes expression of mitochondrial fusion factors, including Opa1 and Mfn2, shifting dynamics toward elongation and providing a mechanistic link between AUF1 deficiency and the mitochondrial phenotype-associated senescence.

To investigate whether loss of AUF1 sensitizes hepatocytes to senescence-associated remodeling, we induced senescence using chronic oxidative stress (Figure 1) after AUF1-KD and then assessed mitochondrial dysfunction and expression of senescence markers (as in Figure 1 and Figure 2). Senescence induction further increased Opa1 and Mfn2 transcripts in AUF1-KD AML12 cells relative to control senescent cells (Sen-siCtrl), with parallel increases in Opa1 and Mfn2 proteins (Figure 6A,B). Quantitative morphometry likewise showed that senescence induction in AUF1-KD cells significantly increased mitochondrial perimeter, area, and branch length, consistent with accentuated network elongation (Figure 6C). Concomitantly, total ROS and mitochondrial ROS increased further, while mitochondrial ATP levels decreased and membrane potential became more depolarized compared to the control senescent AML12 cells (Figure 6D–G). At the cellular level, senescence markers, including p16, p21, and p53, exhibited higher levels after senescence induction (Figure 6H). Levels of p21 and p16 in senescent cells were further increased by AUF1-KD, and the population of SA β-gal-positive cells increased accordingly (Figure 6I,J). Together, these results suggest that AUF1 reduction primes hepatocytes for an amplified mitochondrial dysfunction and cellular senescence in response to pro-senescent stimuli, supporting a model in which AUF1 restrains the mitochondria–senescence axis.

To evaluate whether AUF1 upregulation can counteract the mitochondrial remodeling and senescence-like phenotypes, we ectopically expressed AUF1 in AML12 cells, subjected the cells to senescence induction, and compared the effects of AUF1 rescue on mitochondrial dysfunction and cellular senescence, with proliferating control AML12 cells included as baseline controls. Induction of Opa1 and Mfn2 expression in senescent AML12 cells was attenuated by AUF1 expression (Figure 7A,B). Morphometric analysis of mitochondria showed that AUF1 expression retrieved the shift towards elongated networks upon senescence induction relative to vector-transduced senescent cells, consistent with maintenance of a more balanced mitochondrial architecture (Figure 7C). In addition, total ROS and mitochondrial ROS were lower, whereas mitochondrial ATP levels were higher and membrane potential better preserved in AUF1-expressing cells than in vector controls after senescence induction (Figure 7D–G). Levels of senescence markers were partially reduced compared with induced vector controls (Figure 7H,I), and the population of SA β-gal-positive cells was decreased by AUF1 expression (Figure 7J). Taken together with the loss-of-function data (Figure 3, Figure 4 and Figure 6), these gain-of-function results demonstrate that increasing AUF1 levels prevents fusion-dominant remodeling, bioenergetic decline, and elevation of senescence markers in response to pro-senescent stimuli. These findings support the notion that AUF1 is an essential regulator of the mitochondria-senescence axis in AML12 hepatocytes by maintaining mitochondrial homeostasis.