Unveiling a novel macrophage-driven mechanism for obeticholic acid-induced liver injury in cholestatic: targeting NLRP12^Arg185 fuels Relb-TNFSF14-mediated ductular reaction

Abstract

For individuals with Primary Biliary Cholangitis (PBC) who are either intolerant or nonresponsive to ursodeoxycholic acid (UDCA), obeticholic acid (OCA) is authorized as a therapeutic treatment. Nevertheless, its clinical application has been limited due to the occurrence of several adverse effects. We employed the bile duct ligation (BDL) model to simulate cholestatic conditions observed in patients, revealing that high doses of OCA exacerbated hepatic inflammatory infiltration and fibrosis, accompanied by a marked augmentation in the extent of ductular reaction (DR) within BDL. Our findings mentioned that macrophages played a critical part in mediating OCA-induced aggravation of DR. OCA significantly upregulated the non-canonical NF-κB Relb signaling in macrophages, thereby contributing to proliferation of cholangiocytes. Cellular thermal shift assay (CETSA), surface plasma resonance (SPR) and liquid chromatography mass spectrometer (LC–MS) analysis showed that OCA directly bond to Nucleotide-binding oligomerization domain-like receptor (NLR) pyrin domain (PYD)-containing protein 12 (NLRP12), leading to a reduction in its protein level. Further, activity-based protein profiling (ABPP) method identified the Arg185 residue of NLRP12 as the binding site for OCA, elucidating the mechanism by which OCA downregulates NLRP12 protein expression. This interaction diminished the inhibitory effect of NLRP12 on the Relb pathway. Collectively, our study delineated the NLRP12^Arg185–Relb–TNFSF14 axis in macrophages as a key mechanism underlying OCA-induced hepatotoxicity.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10565-026-10176-1.

Introduction

Common biliary illnesses include primary biliary cholangitis (PBC), biliary atresia (BA), and primary sclerosing cholangitis (PSC) are associated with ductular reaction (DR), which is defined as the proliferation of reactive bile ducts after liver injury(Sato et al. 2019). DR is typically accompanied by inflammatory infiltration and portal fibrosis (Gadd et al. 2014), yet therapeutic options remain limited and pathogenesis is poorly understood (Sclair et al. 2015).

Obeticholic acid (OCA), a PBC-approved FXR agonist, has demonstrated efficacy but carries risks of adverse effects including gallstone formation and paradoxical exacerbation of biliary injury (Samur et al. 2017). Recent studies have reported that OCA may make susceptible persons more likely to develop gallstones (Al-Dury et al. 2019) and potentially boost the development of primary hepatic progenitor cells (HPC) towards the cholangiocyte lineage in vitro and encourage DR in healthy animals (An et al. 2020).

Nuclear factor kB (NF-kB) plays pivotal roles in liver disease progression (Luedde And Schwabe 2011). NF-kB signaling is divided in a canonical and a non-canonical pathway (Hayden And Ghosh 2008). While the canonical pathway mediates inflammatory responses (Sun 2017)., the non-canonical pathway, which activated by tumor necrosis factor receptor (TNFR) superfamily members including B-cell-activating factor belonging to TNF family (BAFF)-R (Claudio et al. 2002), CD40 (Coope, 2002), lymphotoxin β-receptor (LTβR)(Das et al. 2019), receptor activator for nuclear factor kB (RANK) (Novack et al. 2003), Fn14 (Short, et al., 2023), etc., triggers NIK-dependent processing of p100 to p52, enabling nuclear translocation of Relb-p52 dimers(Sun 2012). The biliary NIK directly increases cholangiocyte proliferation and promotes secretion of cholangiokines, augmenting liver inflammation and fibrosis (Zhang, et al., 2022). Additionally, the nuclear translocation of Relb, which was also necessary for the ductular reaction and the development of biliary fibrosis, is elevated in reactive bile ducts in individuals with chronic liver disorders (Elßner et al. 2019). Furthermore, many liver disorders, such as hepatocellular carcinoma (HCC) and liver fibrosis, have been causally related to dysregulated non-canonical NF-kB signaling (Cildir et al. 2016).

Liver macrophages are made up of both resident Kupffer cells (KCs) and monocyte-derived macrophages (MDMs) (Guillot And Tacke 2019). Macrophages from the bone marrow (BM) are drawn to engraft in the liver next to DRs during liver damage (Lorenzini, et al., 2010). Following acute injury to biliary epithelial cells (BECs), monocyte-derived CCR2 + macrophages gathered around damaged bile ducts, boosting BEC proliferation through the production of Itgβ6 (Guillot, et al., 2021). Bone marrow-derived macrophages dominate in cholestatic models (Li et al. 2023), where paracrine signals like TWEAK, another cytokine that directly stimulates the non-canonical NF-kB receptor Fn14, activate cholangiocyte NF-κB (Bird, et al., 2013).

In this study, we aimed at dissecting the contribution of the non-canonical NF-kB pathway in macrophages to OCA-mediated liver injury. For the first time, our research reveals how the rise in macrophage-derived non-canonical NF-kB brought on by OCA affects the DR's progression. By directly binding to Nucleotide-binding oligomerization domain-like receptor (NLR) pyrin domain (PYD)-containing protein 12 (NLRP12)^Arg185, OCA prevents the negative adjustment on non-canonical NF-kB signaling, encourages its breakdown, and increases the level of Relb. Together, our findings showed that OCA can bind to NLRP12 directly and raise the level of non-canonical NF-kB, which accelerates the DR.

Methods and materials

Reagents and antibodies

PUSH Biotechnology (Chengdu, China) supplied OCA. Phenol myristate acetate (PMA) (HY-18739), MG132 (HY-13259), and cycloheximide (CHX) (HY-12320) were acquired from MedChemExpress (New Jersey, USA). Cell Counting Kit-8 (CCK-8) (A311-02-AA, Vazyme, China) was acquired. The hydroxyproline (Hyp) kit (A030-2–1), aspartate transaminase (AST) (Cat No. C010-3–1), alkaline phosphatase (ALP) (Cat No. A059-2–2), and alanine aminotransferase (ALT) (Cat No. C009-3–1) assay kits were acquired from Nanjing Jiancheng Bioengineering Institute. The supplier of collagenase IV (C5138) was Sigma (USA). Clodronate liposomes (CLO^Lipo) (40337ES08) and control (PBS^Lipo) (40338ES08) were produced by Yeasen Biotechnology Co., Ltd. in Shanghai, China. Recombinant mouse macrophage colony-stimulating factor (M-CSF) protein (RP01216) was obtained from ABclonal Biotechnology Co., Ltd (Wuhan, China). Recombinant human TNFSF14 protein (664-LI-025) was purchased from R&D (USA). Recombinant human NLRP12 protein was synthesized by Wuhan Mabnus Biotech Co., Ltd (Wuhan, China). Adeno-associated virus (AAV)2/9-F4/80-Relb-ZsGreen was packaged by Hanbio (Shanghai, China). The antibodies were showed in Table 1.

Table 1.

List of antibodies

Antibodies	Manufacturer	Catalog number	Host	Reactivity
F4/80	Santa cruz	sc-377009	Mouse	Mouse, Rat, Human
p-IKKα/β	Santa cruz	sc-23470-R	Rabbit	Mouse, Rat, Human
IKKα/β	Santa cruz	sc-7607	Rabbit	Mouse, Rat, Human
p-IκB-α	Santa cruz	sc-7977	Goat	Mouse, Rat, Human
p-p65	Santa cruz	sc-33020	Rabbit	Mouse, Rat, Human
CK19	Developmental Studies Hybridoma Bank	TROMA-III	Rat	Mouse, Rat, Human
α-SMA	Cell Signaling Technology	19245	Rabbit	Mouse, Rat, Human, Hamster, Monkey
Relb	Cell Signaling Technology	10544	Rabbit	Mouse, Rat, Human
NIK	Cell Signaling Technology	4994	Rabbit	Mouse, Human
p100-52	Cell Signaling Technology	4882	Rabbit	Mouse, Rat, Human
Fibronectin 1	abcam	ab45688	Rabbit	Mouse, Rat, Human
Collagen I	abcam	ab260043	Rabbit	Mouse, Rat, Human
Ki67	abcam	ab16667	Rabbit	Mouse, Rat, Human
SOX9	abcam	ab185966	Rabbit	Mouse, Rat, Human
p65	abcam	ab16502	Rabbit	Mouse, Human
NLRP12	Bioss	bs-6864R	Rabbit	Human,  Rat, (predicted: Mouse, Rabbit, Pig, Sheep, Cow, Dog, Horse)
Lamin B1	Affinity	AF5161	Rabbit	Mouse, Rat, Human, Monkey, Fish
GAPDH	proteintech	60004-1-Ig	Mouse	Mouse, Rat, Human, Pig, Zebrafish, Yeast, Plant
F4/80	Biolegend	123116	Rat	Mouse
CD45	eBioscience	48–0451−80	Rat	Mouse
CD11b	eBioscience	48–0112−82	Rat	Mouse
CCR2	Biolegend	150609	Rat	Mouse
Goat anti-Rabbit IgG (H+L) Secondary Antibody, HRP	Invitrogen	31460	Goat	Rabbit
Goat anti-Mouse IgG (H+L) Secondary Antibody, HRP	Invitrogen	31430	Goat	Mouse
Alexa Fluor 488	Invitrogen	A-11001	Goat	Mouse
Alexa Fluor 568	Invitrogen	A-11077	Goat	Rat
Alexa Fluor 647	Invitrogen	A32728	Goat	Mouse

Studies in animals

Male mice were studied since their phenotypes varied less. Whether the results apply to female mice is uncertain. Male C57BL/6 J mice (6 to 8-week-old) used in this study were supplied by Shanghai SLAC Laboratory Animal Co. Ltd. The BDL surgery followed our earlier approaches (Wang et al. 2022). During a midline laparotomy, inhaled isoflurane was administered in addition to a 5 mg/kg intracutaneous injection of carprofen. It was subsequently given to the mice orally for seven days at a dosage of 40 mg/kg of OCA in 0.5% carboxymethyl cellulose sodium (CMC-Na). The guidelines of British Journal of Pharmacology (Lilley et al. 2020), ARRIVE guidelines (Percie Du Sert et al. 2020), and the Animal Ethics Committee of China Pharmaceutical University (2019 07-002) were followed in all animal procedures.

For macrophage depletion: On day 3 after BDL, mice were intraperitoneally injected with 200ul PBS^Lipo or CLO^Lipo, which was diluted at a ratio of 1:1 in PBS.

For AAV-mediated macrophage expression of Relb: AAV2/9 carrying ZsGreen, Relb-ZsGreen under the control of F4/80 promoter was injected into C57BL/6 J mice through tail vein at dose of 10¹¹ virus genome dissolved in saline before OCA treatment. After 3 weeks of tail vein injection, OCA was continuously administrated by oral gavage for another week.

Isolation and culture of liver macrophages and bone marrow-derived macrophages (BMDMs)

The primary liver cells were performed as described above (Seglen 1976). Briefly, after Ethylene glycol tetraacetic acid (EGTA) and collagenase solution perfusion, the cells were filtered and centrifuged at 50 g. The macrophages were collected and purified using an equal volume of 25%/50% Percoll density gradient centrifugation. The medium of macrophages were RPMI-1640 (12,633,012, Gbico, USA) supplemented with 10% FBS (26,170,043, Gbico, USA), 100 μg/ml streptomycin, and 100 μg/ml penicillin. BMDMs were extracted from the tibias and femurs of male C57BL/6 J mice aged 6–8 weeks. To generate single-cell suspensions, 10–15 ml of DMEM (12,491,015, Gibco, USA) were used, followed by filtering through 70 μm nylon mesh. The cells were centrifuged, lysed using red blood cell lysis solution, and then washed with PBS. The cell medium was DMEM with 10% FBS and 20 ng/ml of recombinant mouse CSF added. The medium supplemented with M-CSF was changed at day 3. BMDMs were obtained after 7 days and the purity was evaluated using flow cytometry analysis.

Cell culture

THP-1 cell line was cryopreservation by our own lab and was cultured in RPMI 1640 with 10% FBS, 0.05 mM β-Mercaptoethanol, streptomycin (100 μg/mL) and 1% penicillin (100 U/mL) (C0222, Beyotime, China). Prior to all tests, THP-1 monocytes were stimulated to develop into adherent macrophages for 48 h using 100 ng/mL PMA. Following 24 h treatment of DMSO or OCA, the supernatant of THP-1 was respectively collected as conditional medium (con CM or OCA CM). ScienCell (5100) supplied the human intrahepatic biliary epithelial cells (HiBECs), which were cultivated in the epithelial cell medium (4101, ScienCell, USA) supplemented with 2% FBS and epithelial growth factor. HEK293T were cryopreservation and was cultured in DMEM (10% FBS, 1% streptomycin (100 μg/mL) and penicillin (100 U/mL)) in our own laboratory at 37 °C with 5% CO₂.

Cell viability

The CCK-8 was utilized in accordance with the guidelines provided by the manufacturer. Each sample was run in triplicate.

RNA interference and cell overexpression

GenePharma (Shanghai, China) generated the plasmid vector and the siRNA. DNA and siRNA were employed in this investigation at concentrations of 2 μg and 50 nM, respectively. The random sequence of the non-conding (NC) siRNA had no biological impact. And empty plasmid was used as control (pcDNA3.1-NC). The plasmids or siRNA were transfected into THP-1 cells using Lipofectamine 3000 (L3000075, Invitrogen, USA). The sequences details were showed in Table 2. After 24 h of transfection, the efficacy was evaluated by RT-PCR or western blot. HEK293T cells were additionally transfected with the mutant plasmids in accordance with the methodology. Then HEK293T were administrated with OCA 20 μM for another 24 h after transfection of 24 h.

Table 2.

Sequence information of siRNA or plasmid

Product	Sense (5’−3’)	Antisense (5’−3’)
siLtbr (human)	GUGCCGAGAAUUCCUACAATT	UUGUAGGAAUUCUCGGCACTT
siRelb (human)	GGUGCAGAAAGAGGACAUATT	UAUGUCCUCUUUCUGCACCTT
siNlrp12 (human)−1615	GCCUUCCUCAACAUGAACATT	UGUUCAUGUUGAGGAAGGCTT
siNlrp12 (human)−179	CCACCUACUUGGAAGAACUTT	AGUUCUUCCAAGUAGGUGGTT
siNlrp12 (human)−1917	GUGGAUCCAAAGCAAAGCUTT	AGCUUUGCUUUGGAUCCACTT
siNlrp12 (human)−3094	GCUUGUGAGAAUCUUUACUTT	AGUAAAGAUUCUCACAAGCTT
PcDNA3.1(+)-NLRP12 (human)	> NM_144687.4:141–3326 Homo sapiens NLR family pyrin domain containing 12 (NLRP12), transcript variant 2, mRNA

Mutation overexpression plasmid construction

NLRP12 WT and mutation plasmids were generated by Wuhan Mabnus Biotech Co., Ltd (Wuhan, China). The sequence was inserted into pTT5-Flag vector. The corresponding glycerol stock bacteria were retrieved, plated onto LB agar, and incubated overnight. Then the monoclonal colony was selected and cultured in LB medium. The isolation of all plasmids was performed using endo-free plasmid DNA kit (D6950-01, Omega, USA). All plasmids were verified by sequencing. The primers were shown in the Table 3.

Table 3.

Primer pairs used for plasmids sequencing

Gene	Length	vector	Cloning sites	Sequencing primer (5’−3’)
Nlrp12(174A)	3276	pTT5	XbaI-HindIII	pTT5-seqF TGGCCATACACTTGAGTGAC pTT5-seqR CCATATGTCCTTCCGAGTGA seq 1 AAGTCTCTCTTGTCACTCCA seq2 CTCATCCGAGTTCCCGAGCG seq3GTCTGATGCAACCCAAGCCG seq 4AGCAGGGCTCCTTGGAGTTC seq 5 AGGATGGATCTCAGTGGCAA
Nlrp12(175A)	3276	pTT5	XbaI-HindIII
Nlrp12(177A)	3276	pTT5	XbaI-HindIII
Nlrp12(178A)	3276	pTT5	XbaI-HindIII
Nlrp12(185A)	3276	pTT5	XbaI-HindIII

Transcriptome sequencing: RNA-sequencing

BGI Technology (Shenzhen, China) extracted the RNA and examined its purity. The library was built and transcriptome sequencing was carried out on the machine once the RNA quality passed the test. In order to determine which genes were differentially expressed between the groups, DES eq2 assess differentially expressed genes (DEGs). The DrTOM received all of the raw data (https://genean-static-master.oss-cn-shenzhen.aliyuncs.com/Genean_win.zip) (http://report.bgi.com).

RNA isolation and real-time PCR

Using Trizol Reagent (R401-01, Vazyme, China) as directed by the manufacturer, total RNA was extracted. 1 μg of RNA was reverse-transcribed into cDNA using reagent HiScript Reverse Transcriptase (R123-01, Vazyme, China) and MyCyclerThermal Cycler (Bio-rad, USA). And then amplified in StepOnePlus (Applied Biosystems, USA) using oligo primers (Invitrogen, USA) and SYBR Green Master Mix (High ROX Premixed) (Q141-02, Vazyme, China) following the circumstances: pre-denaturation: 5 min at 95 °C; 40 cycles of 10 s at 95 ℃ and 30 s at 60 °C. Relative gene expression was normalized to Gapdh control and computed with the 2^−ΔΔCT technique. Tables 4 and 5 listed the primer sequences.

Table 4.

Mouse primer pairs for RT-PCR

Gene	Forward primer (5’−3’)	Reverse primer (5’−3’)
Gapdh	CTTTGGCATTGTGGAAGGGC	CAGGGATGATGTTCTGGGCA
Acta2	TGCTGACAGAGGCACCACTGAA	CAGTTGTACGTCCAGAGGCATAG
Timp1	TCTTGGTTCCCTGGCGTACTCT	GTGAGTGTCACTCTCCAGTTTGC
Timp2	AGCCAAAGCAGTGAGCGAGAAG	GCCGTGTAGATAAACTCGATGTC
Col1a1	CCTCAGGGTATTGCTGGACAAC	CAGAAGGACCTTGTTTGCCAGG
Il-1b	TGGACCTTCCAGGATGAGGACA	GTTCATCTCGGAGCCTGTAGTG
Tnf-a	GGTGCCTATGTCTCAGCCTCTT	GCCATAGAACTGATGAGAGGGAG
Cd68	GGCGGTGGAATACAATGTGTCC	AGCAGGTCAAGGTGAACAGCTG
Ccl2	GCTACAAGAGGATCACCAGCAG	GTCTGGACCCATTCCTTCTTGG
Ccr2	GCTGTGTTTGCCTCTCTACCAG	CAAGTAGAGGCAGGATCAGGCT
Ccl5	CCTGCTGCTTTGCCTACCTCTC	ACACACTTGGCGGTTCCTTCGA
Cxcl2	CATCCAGAGCTTGAGTGTGACG	GGCTTCAGGGTCAAGGCAAACT
Ccl20	GTGGGTTTCACAAGACAGATGGC	CCAGTTCTGCTTTGGATCAGCG
Trem2	CTACCAGTGTCAGAGTCTCCGA	CCTCGAAACTCGATGACTCCTC
Cd36	GGACATTGAGATTCTTTTCCTCTG	GCAAAGGCATTGGCTGGAAGAAC
Lpl	GCGTAGCAGGAAGTCTGACCAA	AGCGTCATCAGGAGAAAGGCGA
Il1rn	TGTGCCTGTCTTGTGCCAAGTC	GCCTTTCTCAGAGCGGATGAAG
Relb	GTTCTTGGACCACTTCCTGCCT	TAGGCAAAGCCATCGTCCAGGA
Tnfsf14	GGAGACATAGTAGCTCATCTGCC	CCACCAATACCTATCAAGCTGGC
Ltb	CCTGTTGTTGGCAGTGCCTATC	GACGGTTTGCTGTCATCCAGTC
Ltbr	TCCTTGGAGGAAGTGGTGCTAC	CGGTCACATGAATGCCATTCGC

Table 5.

Human primer pairs for RT-PCR

Gene	Forward primer (5’−3’)	Reverse primer (5’−3’)
Nlrp12	CAGGCATGATGCTGCTTTGCGA	AGCACAGAAGCCATCTCCTGAC
Relb	TGTGGTGAGGATCTGCTTCCAG	TCGGCAAATCCGCAGCTCTGAT
Tnfsf14	GGTCTCTTGCTGTTGCTGATGG	TTGACCTCGTGAGACCTTCGCT
Ltb	GGTTTCAGAAGCTGCCAGAGGA	CGTCAGAAACGCCTGTTCCTTC
Baff	ACCACGCGGAGAAGCTGCCAG	CTGCTGTTCTGACTGGAGTTGC
Tweak	AAAACACGGGCTCGAAGAGCGA	TGGAGCTGTTGATTCTGGCTTCC
Rankl	GCCTTTCAAGGAGCTGTGCAAAA	GAGCAAAAGGCTGAGCTTCAAGC
Cd40l	GCGGCACATGTCATAAGTGAGG	GTCCTTGTCTTTTAACGGTCAGC
Gapdh	GTCTCCTCTGACTTCAACAGCG	ACCACCCTGTTGCTGTAGCCAA

Western blot assay

The tissues and cells treated with lysis buffer RIPA (P0013C, Beyotime, China) provided the protein samples. The lysates were centrifuged for 15 min at 12000xg after being placed on ice for 15 min. Bradford Protein Aaasy Kit (P0006C, Beyotime, China) measured the concentration. 30 μg of protein each lane was deposited onto a polyvinylidene difluoride membrane (Millipore, USA) following separation on 10% SDS-PAGE. Following blocking the membranes with 5% milk, the membranes were exposed to several primary antibodies for an entire night at 4 °C. The membranes were rinsed and incubated with the relevant secondary antibodies for one hour before being identified using an ECL detection kit.

Cellular thermal shift assay (CETSA)

As previously reported (Jafari et al. 2014), the CETSA-WB experiment was conducted. In short, THP-1 cell soluble protein lysate was divided into PCR tubes and exposed to either DMSO or OCA (20 μM) for one hour at room temperature. And the solutions were heated at the indicated temperatures (37–65℃) for 3 min, followed by cooling at 4℃ for 3 min in a thermocycler. Use liquid nitrogen twice to freeze and defrost the cells. Centrifuge the cell lysate at 12000 g for 15 min at 4 °C after briefly vortexing the tubes to pellet cell debris along with precipitated and aggregated proteins. The supernatant containing the soluble protein fraction was ready for analysis by Western blot.

Nuclear protein extraction

The nuclear protein extraction kit was examined with the manufacturer's instructions (P0027, Beyotime, China).

Biochemical assay

AST, ALT, and ALP levels were determined using a biochemistry analyzer (Thermofisher, USA). Liver hyp level in mice was quantified using a detection kit following manufacturer’s instructions.

Immunohistochemistry (IHC) and immunofluorescence (IF)

The fresh liver tissues were embedded in paraffin and sectioned. The samples underwent hematoxylin and eosin (H&E) and Masson's trichrome staining after being divided into 3 μm slices. F4/80, CK19, and SOX9 antibodies were used for IHC. Fresh liver tissues were mounted with O.C.T, frozen on dry ice, and later preserved at −80℃. The cells were placed on glass coverslips in 12-well plates. After three PBS washes, 4% paraformaldehyde was used at room temperature for 15 min to fix cell samples and fresh-frozen liver slices (5 μm). After that, the liver slides and cell glass coverslips were cleaned with PBS and treated with 0.1% Triton X-100 for 10 min at room temperature. 10% goat serum in PBS was used to prevent non-specific binding after permeabilization for one hour at room temperature. The samples were incubated with the primary antibodies at 4℃ overnight. After washed with PBS, the samples were further incubated with DAPI and Alexa Fluor 488/647/568-conjugated secondary antibodies at 1:500 dilution in PBS for 1 h in the dark. The slides were visualized using FV3000 (Olympus, Japan).

Enzyme-linked immunosorbent assay (ELISA)

The relevant ELISA kit was used with the manufacturer's instructions to determine the amount of TNFSF14 present in mouse serum and cell supernatant (CSB-EL023991MO, CSB-EL023991HU, Cusabio, China).

Flow cytometry

After being separated, the liver macrophages underwent two PBS washes. The pellet was centrifuged at 500xg at 4 °C after the red blood cells were extracted by resuspending it in red blood lysis solution. After that, the pellet was reconstituted and treated with 10 μL of FcR blocking reagent per 90 μL FACS buffer (3% FBS in PBS) for 10 min at 4 °C. Fc receptor blocking was followed by the addition of fluorescein-conjugated primary antibodies (CD45, CD11b, F4/80, and CCR2) as instructed, and the mixture was incubated for 40 min at 4 °C under shadow. This was followed by surface staining. Following incubation, two 500 × g centrifugations were performed using FACS buffer for five minutes at 4 °C. The CytoFLEX (Beckman Coulter, Brea, USA) will be used to process the material for flow cytometry analysis. FlowJo software (TreeStar, Ashland, OR) was used to analyze the data.

Surface plasmon resonance (SPR) analysis

SPR was used to evaluate the binding kinetics of OCA with NLRP12 using a BiacoreT200 device and software supplied by the manufacturer (Cytiva, USA). The CM5 chip was activated using 75 mg/mL N-(3-dimethylaminopropyl)-N’ethylcarbodiimide hydrochloride and 10 mg/mL N-hydroxysuccinimide. Human recombinant NLRP12 protein was created and purified by Wuhan Mabnus Biotech Co., Ltd. Using 10 mM sodium acetate (pH 4.5) to dilute the protein to 20 μg/mL, it was then covalently immobilized on a CM5 chip by amino coupling. Approximately 8300 resonance units (RU) of immobilization were obtained by injecting the protein coupling solutions onto the surface of the active CM5 chip. Affinity or binding kinetics studies were carried out using PBST (0.05% Twain, 5% DMSO) buffer at 30 μL/min flow rate. The running parameters were 30 s of binding with the drug or DMSO solution. The Biacore T200 and evaluation software were used to process and analyze the sensor readings after the RU response values were recorded. The binding curves were fitted using the Steady-State calculating model for determining the equilibrium dissociation constant (KD).

Identification of OCA binding peptides on NLRP12

The synthesization, validation of photoaffinity probe (OCA-probe) and LC–MS/MS were performed by Genepioneer Biotechnologies Co., Ltd (Nanjing, China). NMR spectrometer (Bruker Biospin, AVIII-HD-400, USA), HPLC 2000 (Agilent, USA) and Mass spectrometer (AB, API 150EX, USA) were used to determine the details of chemical structure of OCA-probe. NLRP12 protein was incubated with OCA-probe (50 μM) or equal DMSO for 30 min on ice. The samples were exposed to 365 nm UV light on ice for 15 min, and heated to 95 °C for 10 min. After SDS-PAGE separation and removal of the major migrating protein band, the proteins were digested with trypsin and analyzed using LC–MS/MS. Specifically, liquid chromatography was used to dissolve the peptides. A 0.1% (v/v) formic acid water solution served as mobile phase A. The mobile phase B was an acetonitrile solution containing 0.1% formic acid. The following was the elution gradient: 3% ~ 5%B, 5 s; 5% ~ 15%B, 23 min55sec; 15% ~ 28%B, 21 min; 28% ~ 38%B, 7 min 30 s; 38%−100%B, 5 s; 100%B, 12 min 25 s. At 300 nL/min, the flow rate was kept constant. Thermo Q Exactive HF-X mass spectrometer analysis was performed after the peptide segments were eluted and ionized. Two kV was used as the ion source voltage. The parent ions of the peptide segments and their secondary pieces were identified and analyzed using high-resolution Orbitrap. The first mass spectrometry had a scan range of 350 to 1500 m/z and a scan resolution of 60,000; the secondary scan had a resolution of 15,000. Data-dependent acquisition (DDA) was the mode of data acquisition. The first stage's automatic gain control (AGC) was set to 3E6, and the second stage's to 1E5. Thirty seconds was chosen as the dynamic exclusion period for tandem mass spectrometry scanning. The MS data was searched using Proteome Discoverer (v2.4). The protein database was downloaded from Uniprot (AlphaFoldDB: P59046); the digestion method was set to Trypsin/P; the number of missed cleavage sites was set to 2; the minimum peptide length was set to 6 amino acid residues; the cysteine alkylation was set as a fixed modification, and the variable modifications were methionine oxidation, protein N-terminal acetylation, and OCA-probe.

Computational docking modeling

The 3D protein structure of human NLRP12 was obtained from UniProt Protein Data Bank (AlphaFoldDB: P59046). The ligand OCA molecule was plotted 2D structures by ChemDraw, saved as. MOL format and optimized conformations. The quick preparation of protein was all performed using Molecular Operating Environment (MOE) software. The sequence 160–185 from NLRP12 protein was selected to the following docking. The general docking parameters were all set as the default values.

Statistics

Every value for the three separate experiments was captured as mean ± standard error of mean (SEM). The Student's t-test and the Student–Newman–Keuls test were employed for statistical significance assessments comparing the two groups. The Turkey test and a one-way ANOVA were used to analyze data from three or more groups. On the graph, the p values were displayed. GraphPad Prism 8.0 was used for all statistical analyses.

Results

Increased accumulation of monocytes/macrophages in the liver of BDL mice administrated with OCA

Our previous study used the BDL model, which confirmed the toxicology of OCA at the dose of 40 mg/kg (Wang et al.2022). OCA (40 mg/kg) exacerbated bile duct proliferation and extracellular matrix (ECM) deposition in BDL mice (Fig S1). IHC and IF staining of F4/80 revealed an increased F4/80-positive area around injured bile ducts and necrotic regions in OCA-treated BDL liver sections (Fig. 1A, 1). The study has demonstrated that following cholestatic or toxic liver damage, MDMs are replenished and dominate the KC pool (Li et al. 2023). To further characterize the population recruited in the liver, we examined CD45, CD11b and F4/80 using flow cytometry. In OCA-treated BDL mice, KCs (CD45 + F4/80 + CD11b^inter) were reduced whereas infiltrating MDMs (CD45 + F4/80 + CD11b^hi) were markedly elevated (Fig. 1C). Moreover, study has shown that infiltrating CCR2⁺ monocytes played an important role in promoting cholangiocytes proliferation in vivo (Guillot, et al., 2021). The proportion of CCR2⁺ macrophages were also elevated following OCA treatment (Fig. 1D). Consistently, inflammatory factor mRNA expression including Il-1b, Tnf-a, Cd68, Ccl2, Cxcl2, etc., was much elevated in the BDL OCA group in contrast to BDL (Fig. 1E). All of these results suggested that recruited MDMs were a significant source of inflammatory mediators and might be crucial in accelerating the course of the disease in the BDL mice received OCA.

Fig. 1.

Increased accumulation of monocytes/macrophages in the liver of BDL mice administrated with OCA. A Images and quantification of F4/80 immunohistochemical staining in liver slides (n > 5). 250 μm scale bars. B Immunofluorescence staining for F4/80 (red) and CK19 (green) in liver tissues, with nuclei counterstained by DAPI (blue), and corresponding quantification of F4/80-positive area (n > 5). 100 μm scale bars. C Flow cytometry of Kupffer cells (KCs) and monocyte-derived macrophages (MDMs) frequencies among CD45⁺ leukocytes. D Proportion of CCR2⁺ cells within total liver macrophages (n = 6). E Heatmap and RT-PCR analysis of inflammatory gene expression in liver tissues (n = 5). Each row represents a gene, and significance versus BDL group is indicated (*P < 0.05). Data: mean ± SEM. Two-way ANOVA was used (*P < 0.05 vs. BDL group)

Macrophage depletion attenuates OCA-aggravated liver injury and ductular reaction in BDL mice

To explore the role of macrophages in the exacerbation of DR mediated by OCA, we injected the mice with clodronate-loaded liposomes (CLO^Lipo) to deplete monocytes and macrophages and evaluated the degree of liver injury. PBS^Lipo or CLO^Lipo injections, either with or without OCA therapy, had no effect on sham mice (Fig S2). Following macrophage reduction, the OCA-treated BDL animals showed decreased serum ALT, AST, ALP, and liver Hyp content (Fig. 2A, B). CLO^Lipo-mediated macrophage depletion led to significantly attenuated necrosis, cholangiocyte proliferation, and fibrosis as demonstrated by the H&E staining, Masson 3 staining and IF of CK19-α-SMA (Fig. 2C, D), which was further confirmed by CK19 IHC (Fig. 2E, F). Consistently, the inflammatory-related (Il-1β, Tnf-α) and fibrogenesis-related gene (Timp-1, Timp-2, Col1a1 and Acta2) and protein (CK19, Collagen I) were markedly reduced in the CLO^Lipo-treated BDL OCA mice (Fig. 2G, H). These data indicated that macrophages participated the process of exacerbation of liver DR mediated by OCA.

Fig. 2.

Macrophage depletion attenuates OCA-aggravated liver injury and ductular reaction in BDL mice. A Levels of serum ALT, AST, and ALP in mice from BDL, BDL OCA group with PBS^Lipo or CLO^Lipo treatment (n = 6–7). B Hepatic hydroxyproline (Hyp) content (ng/mg) (n = 6–7). C Images of H&E- and Masson 3-stained liver sections. Scale bars: 250 μm. D Liver segment immunofluorescence pictures of α-SMA (yellow) and CK19 (purple). 100 μm scale bars. E, F Immunohistochemical and quantitative analysis of CK19 in liver slices. Scale bars: 100 μm. G Heatmap of Il-1β, Tnf-α, Timp-1, Timp-2, Col1a1 and Acta2 mRNA levels in liver tissues was analyzed by RT-PCR. *P < 0.05 revealed significant differences between BDL OCA CLO^Lipo group and BDL OCA group in each row. H Western blot examination of liver lysates for the expression of the proteins CK19 and Collagen I. The mean ± SEM is used to illustrate the data. Two-way ANOVA, analysis of variance, was applied. *P < 0.05 represents significance different in BDL OCA CLO^Lipo group vs BDL OCA PBS^Lipo group, BDL OCA PBS^Lipo group vs BDL PBS^Lipo group

RNA-seq analysis revealed OCA influenced non-canonical NF-κB signaling in macrophages

Given the pivotal role of MDMs involved in regulated ECM and exacerbation of DR(Li et al. 2023), RNA-seq was performed to furtherly explore the mechanism of macrophages affected by OCA on the THP-1 cells (Fig. 3A). A total of 293 DEGs (186 upregulated, 107 downregulated) were identified (|log2Fold|> 1, P < 0.05) (Fig. 3B). GO and KEGG enrichment analyses revealed enrichment in immune-related processes, including “T cell differentiation”, “Positive regulation of interleukin-1 beta production”, “Immune system process”, “Positive regulation of NIK/NF-κB signaling” (Fig. 3C, D). Combining the result of KEGG enrichment analysis, the non-canonical NF-κB pathway components Relb and p52 were significantly upregulated (showed in red) by OCA (Fig S3), whereas canonical NF-κB pathway components were not affected (Fig S3, S4). So, we primarily concentrated on investigating the impact of OCA on the non-canonical NF-κB pathway.

Fig. 3.

RNA-seq analysis revealed in vitro OCA impaired non-canonical NF-κB signaling. A The schematic diagram of cell experiment for RNA-sequencing. B Volcano plot depicting the differential enrichment of genes in OCA vs DMSO groups. Gray dots show genes that have not changed significantly, whereas significantly elevated and downregulated genes are indicated by red and blue dots, respectively. Screening metrics were defined as | log2fc |≥ 1 and P < 0.05. C Analysis of GO enrichment in biological processes. D KEGG enrichment of DEGs

The upregulation of macrophages Relb induced by OCA promoted the proliferation of cholangiocytes

We first examined whether OCA could affect Relb signaling and whether this phenomenon could contribute to the DR, we performed cell experiments firstly. OCA significantly increased Relb mRNA levels (Fig. 4A) and nuclear translocation of Relb (Fig. 4B) in THP-1. Among these cytokines involved in the non-canonical NF-κB pathway, Tnfsf14, Ltb, Baff and Tweak mRNA levels were upregulated, with Tnfsf14 showing the most pronounced increase (Fig. 4D), representing the non-canonical NF-κB pathway being activated. TNFSF14 protein was also elevated in the supernatant of OCA-treated THP-1 (Fig. 4C). Consistently, OCA upregulated Relb signaling in primary mouse BMDMs and isolated liver macrophages (Fig. 4E-H, Fig S5). To maintain the consistency of species, we choses THP-1 and human intrahepatic biliary epithelial cells (HiBECs) to investigate the co-culture experiments. Conditioned medium (CM) from OCA-treated THP-1 significantly enhanced the viability and Ki67 positivity of HiBECs, an effect mimicked by recombinant human TNFSF14 (hTNFSF14) (Fig. 4I-K). Among these released cytokines, TNFSF14 and Ltβ were at a relatively higher level, we then knockdown of LTβR, the receptor of TNFSF14 and Ltβ (Ware 2008), to assess whether the promotion of proliferation was TNFSF14/Ltβ-LTβR dependent. Knockdown of LTβR in HiBECs abolished the proliferative effect of OCA-treated THP-1 CM (Fig. 4L). These data indicated that macrophages influenced by OCA through upregulation of Relb signaling and the subsequent release of cytokines, particularly TNFSF14 and LTβ, further promoted the cholangiocytes proliferation mediated by LTβR.

Fig. 4.

The upregulation of Relb induced by OCA promoted the proliferation of cholangiocytes. A Relb mRNA expression treated with OCA (20 μM) or DMSO control in THP-1 (n > 6). B Immunofluorescence and quantitative analysis of Relb in THP-1 cells after OCA (20 μM) or DMSO treatment. Scale bars: 100 μm. C TNFSF14 protein concentration in supernatant of THP-1 cells treated with OCA (20 μM) or DMSO, measured by ELISA (n = 6). D Heatmap of log2Fold change for genes related to the non- classic NF-κB pathway in THP-1. E Relb mRNA expression in BMDMs treated with OCA (20 μM) or DMSO control (n = 6). F Heatmap of log2Fold change for indicated genes in BMDMs. G Immunofluorescence and quantitative analysis of Relb in BMDMs after OCA (20 μM) or DMSO treatment. 100 μm scale bars. H Western blot of Relb expression in BMDMs treated with increasing concentrations of OCA. I Viability of HiBECs treated with recombinant human TNFSF14 (400 ng/mL) or vehicle control for 24 h, assessed by CCK-8 assay (n = 6). J Viability of HiBECs treated with conditioned medium (CM) from OCA- or DMSO-treated THP-1 cells (n = 6). K Proliferation of HiBECs treated with control CM or OCA CM from THP-1 cells, assessed by Ki67 immunofluorescence. Scale bars: 100 μm. L Viability of Ltbr-knockdown HiBECs treated with control CM or OCA CM from THP-1 cells (n ≥ 6). Data: mean ± SEM. Analysis of Student’s t-test and variance (two-way ANOVA) was employed

Macrophage-specific Relb overexpression potentiates OCA-induced ductular reaction in vivo

To further confirm the effect of OCA-induced macrophages on the DR was attributed to the high level of Relb, adeno-associated virus serotype2/9 (AAV2/9)-F4/80-Relb was administered via tail vein injection to achieve macrophage-specific Relb overexpression, with efficiency validated by IF (Fig S6A). Mice were administrated with OCA 40 mg/kg or 0.5% CMC-Na for 14 days. In the absence of liver injury, Relb overexpression alone did not alter serum ALT, AST, ALP levels or liver histology, regardless of OCA treatment (Fig S6B, Fig. 5A). However, upon macrophage-specific Relb overexpression, OCA treatment led to low-grade ductular proliferation, indicated by IHC and IF staining of CK19 and SOX9 (Fig. 5A, B), the enlargement lumen and numbers of the bile ducts (Fig. 5E, F), and the positive areas (Fig. 5C, D) in the absence of liver injury. These data suggested that OCA can further promoted the progress of DR in the high level of Relb in macrophages.

Fig. 5.

Macrophage-specific Relb overexpression potentiates OCA-induced ductular reaction in vivo. Mice were transduced with AAV-F4/80-NC (negative control) or AAV-F4/80-Relb via tail-vein injection, followed by OCA (40 mg/kg) or vehicle treatment for two weeks after 4 weeks of virus incubation. A Representative H&E staining of liver sections. 250 μm scale bars. B Representative immunofluorescence staining of CK19 (green) with DAPI (blue) nuclear counterstain. 50 μm scale bars. C, D Representative immunohistochemical staining of CK19 (C) and SOX9 (D). Scale bars: 250 μm (overview), 50 μm (enlarged insets). E, F Quantification of CK19-positive (E) and SOX9-positive (F) areas from IHC images (n > 6). The data are displayed as mean ± SEM

NLRP12 negatively regulates OCA-induced upregulation of Relb signaling in macrophages in vitro

While the above results confirm OCA's impact on the non-canonical NF-κB pathway, the question of how it achieves this regulation remains. Therefore, we focused on elucidating the precise mechanism. The heatmap (Fig. 6A) displayed the DEGs involved in NF-κB pathway regulation. Some of the genes that have changed most significantly were Nlrp12, Cd86, Trim44 (upregulated), and Myo1e, Trem2, Tlr4 (downregulated). Considering the previous results, we then pay more attention to the influence affected by OCA between NLRP12 and non-canonical NF-κB. Given the reported feature of NLRP12 on regulated non-canonical NF-κB (Tuladhar And Kanneganti 2020), whether NLRP12 was involved in the upregulated Relb signaling caused by OCA, NLRP12-overexpression and -knockdown experiments were performed in vitro (Fig. 6B, C). As shown in Fig. 6B, NLRP12 siRNA significantly reduced Nlrp12 mRNA expression in THP-1 cells. The results showed that small-interfering RNA (siNlrp12)-mediated NLRP12 knockdown significantly increased the mRNA level (Fig. 6B) and the IF intensity of Relb induced by OCA (Fig. 6D), while the overexpression of NLRP12 inhibited this effect (Fig. 6C, D). These data indicated that OCA negatively modulated the Relb signaling pathway via NLRP12.

Fig. 6.

NLRP12 negatively regulates OCA-induced upregulation of Relb signaling in macrophages. THP-1 cells were treated with OCA (20 μM) or DMSO for 8 h after being transfected with siRNAs or Nlrp12 plasmids for 24 h. A A heatmap displays the expression of genes that control the NF-κB signaling pathway. B, C RT-PCR measurements of Nlrp12 (B) and Relb (C) mRNA levels (n ≥ 6). D, E Representative immunofluorescence images (D) and Relb expression quantification (E). Scale bars: 50 μm. Data: mean ± SEM. *P < 0.05 represents significance different from scRNA group or pcDNA3.1-NC group

OCA promotes proteasomal degradation of NLRP12 in macrophages

In sham mice, the expression of NLRP12 was low and OCA treatment caused a little effect on its level (Fig. 7A). After BDL, the protein NLRP12 level was increased probably because of the increased numbers of infiltrated macrophages in the cholestatic liver. And the expression was further declined following OCA therapy (Fig. 7B). Experiments conducted in vitro were used to examine the mechanism involved in the decreased NLRP12 protein by OCA. Interestingly, OCA significantly upregulated the Nlrp12 mRNA (Fig. 7C) but decreased NLRP12 protein in a time-dependent manner (Fig. 7D). Therefore, we treated THP-1 cells with either cycloheximide (CHX), a protein synthesis inhibitor, or MG132, a proteasome inhibitor, to ascertain if OCA directly encouraged NLRP12 breakdown. The OCA group treated with CHX had a much shorter NLRP12 half-life than the control group (Fig. 7E). And after MG132 (50 μM) treatment, NLRP12 reached higher level (Fig. 7F). Consistently, OCA reduced NLRP12 protein level in isolated BMDMs (Fig. 7G) and primary liver macrophages of whole groups (Fig. 7H). These data suggested that OCA administration influenced the stability of NLRP12, resulting the decreased level of NLRP12 protein.

Fig. 7.

OCA promotes proteasomal degradation of NLRP12 in macrophages. A, B NLRP12 protein expression in liver tissues from sham (A) and BDL (B) mice treated with OCA (40 mg/kg, 7 days) was measured using Western blot analysis (n > 6). C Nlrp12 mRNA expression in OCA-treated THP-1 for the indicated durations (n = 6). D NLRP12 protein expression in THP-1 treated with OCA for the indicated durations. E, F NLRP12 protein levels in THP-1 treated with OCA in combination with cycloheximide (CHX, 20 μg/mL; E) or MG132 (50 μM; F). G NLRP12 expression in primary mouse BMDMs treated with increasing concentrations of OCA. (H) NLRP12 expression in isolated liver macrophages from the four groups. Data: mean ± SEM. *P < 0.05 represents significance different from control group or BDL group

NLRP12^Arg185 was a key residue in OCA-induced protein degradation

To clarify how OCA affects the protein degradation of NLRP12, we carried out CETSA. OCA can bind to NLRP12 and the bond was stronger at different temperatures (Fig. 8A). This effect was verified by SPR experiment, indicating the direct binding between OCA and NLRP12 protein (KD = 1.851e-5) (Fig. 8B). According to the results above, OCA directly targeted the NLRP12 protein. Then, by adding a difunctional side chain with diazirine and alkyne, we selected a molecular probe (OCA-p) based on OCA to precisely identify the binding sites on the NLRP12 protein (Jiang, et al., 2023) (Li, et al., 2024) (Fig S7A, S8 and S9). The result indicated that OCA-p showed similar bioactivity and down-regulated the protein level of NLRP12 (Fig S7). After 30 min of room temperature incubation with pure human NLRP12 protein, the OCA-p was exposed to UV light on ice (Fig. 8C). Therefore, photoaffinity diazirine chemically cross-linked OCA-p with NLRP12 at the binding sites. Labelling ability of OCA-p was confirmed in gel fluorescence (Fig. 8D). After the OCA-p-NLRP12 complex was digested by trypsin, an LC–MS/MS analysis was performed. The photoactivated cleaved fragment of the NLRP12 protein, 160-LLLVKEHSNPMQVQQQLLDTGRGHAR-185, was found to have a molecular weight increase of 512.387 Da (C₃₃H₅₂O₄) (Fig. 8E, Fig S10C), indicating that the pocket where this peptide is located may contain potential OCA binding sites. To map this binding reaction between OCA and NLRP12, we docked OCA into the structure of human NLRP12 protein, created by MOE software (Fig. 8F). The pocket of NLRP12 around the residues 160–185 aa was specifically selected. To further verify which residues contributed to the NLRP12 degradation, these residues were all mutated to alanine A, respectively. Interestingly, mutation of these residues didn’t show effect on NLRP12 protein in normal condition in HEK 293 T cells (Fig S10A). Only mutating Arg185 greatly destroyed the degradation of NLRP12 after OCA treatment (Fig. 8G), and residue Arg185 exhibited binding with OCA in different conformations and locations in computational docking (Fig S10B). Therefore, these results further proved that key residue Arg185 was responsible for the interaction and protein degradation of NLRP12.

Fig. 8.

NLRP12Arg185 was a key residue in OCA-induced protein degradation. A CETSA measuring the interaction between OCA (20 μM) and NLRP12 protein in THP-1. B Surface plasmon resonance (SPR) analysis of OCA binding to NLRP12. Equilibrium dissociation constant (KD) = 18.51 μM. C Schematic workflow for identifying OCA binding sites on NLRP12 using photoaffinity probe OCA-p. D In-gel fluorescence detection of OCA-p binding to NLRP12 protein. E LC–MS/MS analysis of OCA-p-crosslinked peptides from NLRP12. F Structural model of OCA binding within the NLRP12 protein pocket. G NLRP12 protein levels in HEK293T cells expressing wild-type (WT) or Arg185Ala (R185A) mutant NLRP12 after treatment with OCA (20 μM) were determined by Western blot analysis

Discussion

The pathophysiology of chronic cholestatic liver illnesses is largely dependent on cholangiocytes, the specialized epithelial cells that line the intrahepatic and extrahepatic bile ducts (Jalan-Sakrikar et al. 2024). Despite making up only 3% to 5% of the liver's total cell mass, cholangiocytes execute a number of vital tasks carried out by a number of significant molecules (Yoo et al. 2016). In order to limit liver damage, cholangiocytes must be able to react to pathogenic insults by triggering an inflammatory response (Gouw et al. 2011). Cholangiocytes provide a crucial barrier role that shields the liver parenchyma from harm caused by biliary components. Additionally, intrahepatic cholangiocytes multiply in the portal tract and into the hepatic parenchyma to overcome cholestasis conditions, possibly in an effort to compensate by increasing bile drainage. The DR, or increase in intrahepatic cholangiocyte proliferation, also causes fibrogenesis and an inflammatory wound-healing response (Banales et al. 2019). Activated cholangiocytes have a unique secretory phenotype that influences immune cell recruitment, as well as mesenchymal cell migration and activation (Boyer And Soroka 2021). Dysregulation and aggravation of the repair mechanism may result in cirrhosis, fibrosis, and chronic inflammation; in the end, it may act as a breeding ground for liver cancer and organ failure.

According to a recent study, in a chronic (Mdr2^−/−) animal model of sclerosing cholangitis, macrophages are essential for triggering the DR, portal region fibrosis, and monocyte-driven inflammation (Guicciardi et al. 2018). Additionally, our results showed that OCA in BDL dramatically enhanced the amount of macrophages in the liver. The mononuclear phagocyte system includes liver macrophages. Macrophages after liver injury are diverse and can come from a variety of sources, including peritoneal macrophages for the liver's subcapsular areas, KCs, and two patrolling populations of MDMs (Guillot And Tacke 2019). It became unclear how the KCs and MDMs clustered after OCA was administered for seven days in a row. The liver's KC count decreased and its MDMs count significantly increased. At the same time, it’s also suggested that the complex liver pathological state caused by OCA in BDL can’t be defined by the traditional cell population M1 or M2 marker. Therefore, we didn’t define hepatic macrophages in terms of M1 or M2.

Furthermore, we used clodronate liposomes to verify whether the increased macrophage infiltration contributed to the DR promoted by OCA in BDL mice. However, it would be misleading to consider macrophages as foes during liver injury (Duffield, et al., 2005). Considering that in the acute phase of BDL, macrophages may resist external stimuli by phagocytosis and promoting tissue repair response, we chose to inject liposomes on day 4 after BDL surgery. The outcome also illustrated the role of macrophages on DR in OCA-induced liver damage in BDL. The possible mechanism of macrophages affected by OCA in vitro was investigated by RNA-seq. The non-classical NF-κB pathway was found to have aberrant overexpression by OCA, according to the KEGG and GO pathway enrichment analyses. Among the DEGs mentioned above, NLRP12 has been reported to negatively regulate non-canonical pathways by interacting with proteins such as NIK and TRAF3 within the non-canonical NF-κB signaling cascade (Lich, et al., 2007); Tuladhar And Kanneganti 2020). The results of our in vitro and in vivo experiments also showed that NLRP12 was crucial in mediating the non-canonical pathway's negative regulation, where Relb served as a key transcription factor, leading to an aberrant upregulation in macrophages. The elevated level of Relb expression in macrophages facilitated the progression of DR aggravated by OCA.

Importantly, we provide first-ever evidence that OCA directly binds NLRP12 at molecular resolution. CETSA and SPR assays confirmed direct binding between OCA and NLRP12, providing us with a foundation to explore the specific binding sites in detail. To pinpoint the interaction site, we employed activity-based protein profiling (ABPP) (Cravatt 2008). In order to better clarify the mechanism of action and adverse consequences of tiny bioactive compounds, ABPP combines proteomics and activity-based probe technologies to discover the protein targets of these molecules (Wang, et al., 2018). Based on this, we designed an OCA-probe with diacridine as the photoaffinity group and a clickable alkyne tag (Fig S7, S8, S9), which can detect the non-covalent interaction between OCA and NLRP12 protein followed by click chemistry and target enrichment (Chen, et al., 2020). Subsequent LC–MS/MS analysis identified the peptide region (160-LLLVKEHSNPMQVQQQLLDTGRGHAR-185) within NLRP12 that directly binds OCA. Mutagenesis studies, together with computational docking analyses establishes Arg185 as a key determinant of OCA-NLRP12 interaction. However, the precise molecular mechanism by which Arg185 facilitates OCA-induced degradation—whether through enhanced ubiquitination, recruitment of specific E3 ligases, or allosteric modulation of NLRP12 conformation—remains to be fully elucidated. While the complete mechanistic details remain to be uncovered, the identification of Arg185 as a critical residue for OCA-induced NLRP12 degradation represents a significant advance in our understanding of OCA's hepatotoxicity.

To examine whether bile acids mediated OCA’s effect on NLRP12, we treated THP-1 cells with OCA and other bile acids at equal concentrations. Notably, only OCA upregulated NLRP12 mRNA while reducing its protein levels (Fig S11), suggesting post-transcriptional regulation. This may explain why hepatic NLRP12 protein in BDL mice did not differ significantly from sham controls. The results on macrophages indicated that OCA specifically influenced stability of NLRP12 protein, which accelerating the progress of proteasome-dependent degradation.

While previous studies implicated cholangiocyte-specific non-canonical NF-κB signaling (NIK, Relb) in DR progression (Elßner et al. 2019;  (Zhang, et al., 2022), our results demonstrated that Relb was expressed both in macrophages and cholangiocytes. Whole-liver tests further confirmed that OCA in BDL mice activated this pathway (Fig S12). And the in vitro experiments on HiBECs also confirmed that OCA also directly promoted the cholangiocyte proliferation via Relb-TNFSF14-LTβR axis (Fig S13). Therefore, both macrophages and cholangiocytes may contribute to OCA-aggravated DR through non-canonical NF-κB activation, which would explain the high levels of non-canonical NF-κB signaling throughout the liver. However, whether OCA also regulate Relb signaling through decreasing NLRP12 protein need further investigation in cholangiocytes.

An intriguing observation from our previous studies is that OCA significantly downregulated both FXR protein and mRNA levels in BDL mice, yet OCA-induced hepatotoxicity persisted (Wang et al. 2025). In this context, FXR-independent pathways—particularly the NLRP12-Relb axis identified in the present study—may become the predominant drivers of OCA-mediated liver injury.

In summary, we confirmed the critical role of macrophages in the contribution of DR aggravated by OCA in cholestatic BDL model. We provide mechanistic insight into macrophage–cholangiocyte communication and identify a specific binding site underlying OCA-induced liver injury. OCA promoted the release of cytokines, such as TNFSF14 which were involved in mediating the DR and ECM deposition, accelerating the liver fibrosis, by upregulating the non-canonical NF-κB pathway in macrophages. And this regulation was achieved by the direct binding of OCA to the residues Arg185 of NLRP12 protein, resulting in its decrease of protein levels and abnormally elevated Relb signaling (Fig.9). These findings underscored the therapeutic potential of targeting the macrophage NLRP12^Arg185-Relb-TNFSF14 axis to ameliorate OCA-induced hepatotoxicity. Moreover, they highlighted the complexity of drug actions under pathological conditions and reinforce the importance of evaluating drug safety in disease-relevant models, rather than relying solely on target engagement in healthy systems.

Fig. 9.

Graphical abstract

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

OCA

Obeticholic acid

PBC

Primary biliary cholangitis

PSC

Primary sclerosing cholangitis

BA

Biliary atresia

UDCA

Ursodeoxycholic acid

BDL

Bile duct ligation

DR

Ductular reaction

HPC

Hepatic progenitor cell

TNFR

Tumor necrosis factor receptor

(BAFF)-R

B-cell-activating factor belonging to the TNF family

LTβR

Lymphotoxin β-receptor

RANK

Receptor activator for nuclear factor kB

NIK

NF-κB-inducing kinase

HCC

Hepatocellular carcinoma

TWEAK

Tumor necrosis factor-like weak inducer of apoptosis

PMA

Phorbol myristate acetate

ALT

Alanine aminotransferase

AST

Aspartate transaminase

ALP

Alkaline phosphatase

CETSA

Cellular thermal shift assay

IHC

Immunohistochemistry

IF

Immunofluorescence

ELISA

Enzyme-linked immunosorbent assay

SPR

Surface plasmon resonance

Hyp

Hydroxyproline

ECM

Extracellular matrix

MDMs

Monocyte-derived macrophages

KCs

Kupffer cells

HiBECs

Human intrahepatic biliary epithelial cells

BEC

Biliary epithelial cell

NLRP12

Nucleotide-binding oligomerization domain-like receptor (NLR) pyrin domain (PYD)-containing protein 12

OCA-p

Obeticholic acid-probe

CMC-Na

Carboxymethyl cellulose sodium

AAV

Adeno-associated virus

EGTA

Ethylene glycol tetraacetic acid

M-CSF

Macrophage colony-stimulating factor

CCK-8

Cell Counting Kit-8

DEGs

Differentially expressed genes

LC-MS

Liquid chromatography mass spectrometer

ABPP

Activity-based protein profiling

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by J.W., Z.Y., H.Z., Q.T., and Y.M. The first draft of the manuscript was written by J.W., with contributions from Z.Y., and all authors commented on previous versions. Q.Y. performed data validation. L.Z. and Z.J. provided resources, supervision, and funding acquisition. All authors read and approved the final manuscript.

Funding

This work was supported by Natural Science Foundation of Jiangsu Province (BK20221526 to ZJ); National Natural Science Foundation of China (82274200, 82074114 to ZJ); and "Double First-Class" University project (CPU2018GY33 to ZJ).

Data availability

The data are available upon request from the authors.

Declarations

Ethical approval

All animal procedures were performed as approved by the Animal Ethics Committee of China Pharmaceutical University (Approval number: 2019–07-002) and in accordance with the ARRIVE guidelines.

Conflicts of interests

The authors declare no conflicts of interest.