Endogenous phosphatidylcholine 16:0/20:4 increases to promote liver regeneration in hepatectomy

Summary

Liver regeneration after hepatectomy is a challenge in patients with metabolic dysfunction-associated steatohepatitis (MASH), as well as small-for-size syndrome and ex vivo organogenesis. Endogenous lipid metabolites promoting liver regeneration are poorly understood. A 70% partial hepatectomy (PHx) was performed in mice with healthy and MASH liver. Biomarkers promoting liver regeneration were explored using hepatic lipidomics followed by structural identification. The functional validation was performed in in vivo and in vitro models. Liver regeneration post-PHx peaked earlier in the MASH liver than that in the healthy liver. Phosphatidylcholine (PC) (16:0/20:4) increased in both MASH liver and non-MASH liver, and it promoted the proliferation of Hepa 1-6, Huh-7, human liver organoids, and mouse PHx models. In patients with metabolic dysfunction-associated steatotic liver disease (MASLD) and those undergoing hepatectomy, the serum PC (16:0/20:4) also increased, supporting the findings in this work. PC (16:0/20:4) is an endogenous lipid biomarker promoting liver regeneration which increased during hepatectomy in healthy and MASH liver.

Graphical abstract

Highlights

• •Liver regeneration post-PHx was advanced in the MASH liver
• •PCs increased after PHx in MASH liver, and TGs increased in healthy liver
• •Phosphatidylcholine (16:0/20:4) increased after PHx in both MASH and healthy livers
• •Phosphatidylcholine (16:0/20:4) promoted liver regeneration in vivo and in vitro

Health sciences; Hepatology; Internal medicine; Medical specialty; Medicine

Introduction

Liver regeneration restores hepatic mass, architecture, and function after substantial tissue loss.¹ Physiologically, hepatocytes remain dormant (G0 phase of the cell cycle) and exhibit minimal proliferative activity.² After partial hepatectomy (PHx), they are rapidly activated in response to hepatic tissue loss.² In rodents, normal hepatic histology and function are regained 7–10 days post-surgery, with peak DNA synthesis occurring during the initial 3 days.^3,4 In humans, liver regeneration attains complete mass recovery approximately 3 months post-PHx, with proliferative activity peaking between 7 and 10 days post-surgery.^3,4 This regeneration capacity is essential for recovery following hepatectomy and liver transplantation.⁵ Current research predominantly focuses on the downstream signaling molecules involved in liver regeneration, including those known as Wnt/β-catenin, hedgehog, Notch, and TGF-β. However, the essential functions of low-molecular-weight metabolites in liver regeneration are poorly known.

The global prevalence of metabolic dysfunction-associated steatotic liver disease (MASLD) is estimated to be 25%–35%.⁶ Approximately 20% of patients with MASLD develop cirrhosis, and 2% develop hepatocellular carcinoma.⁷ Metabolic dysfunction-associated steatohepatitis (MASH) is a progressive stage of MASLD development, characterized by steatosis, inflammation, liver damage, and varying degrees of fibrosis.⁸ MASH is currently the second most common indication for liver transplantation and is expected to become the leading indication in the coming years.^9,10 Patients with MASH and recipients of steatotic liver grafts experience higher mortality rates during liver transplantation due to reduced regenerative potential. However, the regeneration features of patients with steatosis are rarely investigated, and treatments aiming at stimulating liver regeneration to restore liver function are unknown.

Lipids, widely distributed biomolecules in living organisms, are primarily composed of fatty acids, alcohol derivatives, and metabolites associated with biosynthesis and biological functions.¹¹ Preliminary experimental observations indicated that acute liver steatosis is essential for normal liver regeneration.¹² Phosphatidylcholines (PCs) are a group of lipid components in eukaryotic cell membranes, accounting for 40%–50% of the total phospholipids in mammalian cells and their organelles.¹³ In all nucleated mammalian cells, PCs are synthesized by the cytidine diphosphate (CDP)-choline pathway, known as the Kennedy pathway. Upon entering cells, choline is phosphorylated by choline kinase to form phosphocholine, which is subsequently converted into CDP-choline by phosphocholine cytidylyltransferase. Finally, CDP-choline condenses with diacylglycerol to form PC.¹³ Given their crucial biological functions, PCs have been extensively investigated in many diseases, including MASLD,¹⁴ liver failure,¹⁵ and alcohol-related liver disease.¹⁶ However, the role and mechanism of PCs in liver regeneration remain unreported.

Based on the above mentioned knowledge, we hypothesized that there are some unknown endogenous lipid metabolites promoting liver regeneration after PHx. This work aimed to investigate lipid remodeling during liver regeneration using a 70% PHx model, identify regeneration-associated lipid biomarkers, and validate their function using in vitro and in vivo models.

Results

Liver regeneration was temporally advanced in the MASH PHx model

The chow diet (CD)-treated mice subgroup on day 1 exhibited a significant body weight loss of 14%, followed by a consistent increase in the subsequent 6 days (Figure 1A). The regenerated liver mass after PHx increased by 45.8% on day 3, 8.3% on day 5, and 21.3% on day 7 relative to measurements taken on day 1, 3, and 5, respectively (Figure 1B). On day 7, the liver weight in all mice reverted to its baseline, and their liver index in the CD-treated mice normalized after PHx (Figure 1C). Thus, in healthy mice, the remnant liver after PHx returned to its normal size within 7 days, and the peak regeneration occurred on day 3.

Figure 1.

Regeneration was advanced in the CDAHFD-treated PHx mouse model

(A) Body weight changes in the CD subgroups.

(B–C) Liver weight and liver index changes in the CD subgroups.

(D–G) Relative mRNA levels of Cyclin D1, Cyclin B1, Cyclin E1, and Ki 67 in the CD subgroups.

(H) Hepatic protein levels of pathways involved in regeneration.

(I–K) Gray value levels of hepatic proliferative proteins. Data were expressed as mean ± SD. One-way ANOVA between multiple groups (n = 5, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, compared with CD-D0 subgroup).

The mRNA levels of cell cycle regulatory genes, including Cyclin D1, Cyclin B1, Cyclin E1, and Ki 67, rapidly increased during the first 3 days, signifying the regeneration capacity of the mouse liver (Figures 1D–1G). The highest proliferation occurred on day 3, consistent with previously reported findings.¹⁷ ALT, AST, and TBA levels all increased except the ALP level (Figures S1A–S1D), which attributed to acute loss of the synthesis organ. The inflammatory factors Il-2, Il-10, Tnf-α, and Ccl2 all peaked on day 3 (Figures S1E–S1H). H&E staining revealed that the livers of the CD group on day 0 and 7 exhibited no signs of inflammation or necrosis. However, hepatic inflammatory cell infiltration was markedly evident on day 3, with an increased presence of binucleated cells identified on day 3 post surgery (Figures S1I and S2). The proliferation coincided with inflammation during liver regeneration in healthy mouse livers.

An examination of body weight, liver weight, and liver-to-body weight ratio in the choline-deficient, L-amino acid-defined, high-fat diet (CDAHFD)-treated PHx mice revealed that they did not fully recover within 7 days. This finding differed from the observation in the CD-treated mice, indicating that the liver mass recovery in MASH mice undergoing PHx was delayed (Figure S3). To further compare proliferation-associated regenerative signaling between the two models, YAP and ERK signaling were analyzed. Both pathways were reported to be involved in liver regeneration and hepatocyte proliferation.^18,19 The p-YAP/t-YAP in CDAHFD-treated mice was down-regulated by about 50% compared with that in CD-treated mice on day 1. In a similar comparison, both the p-ERK and PCNA in CDAHFD-treated mice were upregulated more than 100% on day 1 (Figures 1H–1K). The G1-phase progression following PHx peaked earlier in the CDAHFD-treated mice compared to the CD-treated mice.

Hepatic lipidome was different between healthy and MASH regeneration scenarios

In the orthogonal partial least squares discriminant analysis (OPLS-DA) analysis of the hepatic lipidome of the CD-treated mice, individuals from all groups clustered together but separated well from each other on various days (Figure 2A). In the permutation test, the R² value on the left of the intercept (R² = 0.630) was lower than the original point to the right, and the Q² regression line exhibited a negative intercept (Q² = −0.664) (Figure 2B). The present OPLS-DA model was valid without over-fitting. Paired comparisons revealed that 168 and 190 lipid metabolites on days 1 and 3 differed from the lipidome of day 0 (Figure 2C). Among them, 21 modified lipid metabolites were identical in the lipidomes of days 1 and 3 (Table S4). Their relative abundance was illustrated in a heatmap (Figure 2D), and the number of PCs (n = 3) was ranked second among eight lipid species (Figure 2E).

Figure 2.

The number of modified PCs ranked second in the regeneration-related lipidome in the CD-treated PHx mouse model

(A) OPLS-DA analysis in the CD group (R2X = 0.630, R2Y = 0.924, and Q2 = 0.664).

(B) Permutation test of the OPLS-DA models (repeated 200 times).

(C) Venn diagram of lipid components related to regeneration.

(D) Heatmap of differentially modified lipid metabolites in the CD-treated PHx mice.

(E) Classification of differentially modified lipid components in the CD-treated PHx mice.

In the MASH regeneration model, the distribution of CDAHFD subgroups exhibited a similar pattern as the CD subgroups, although they diverged in different directions (Figure 3A). The permuted R² value to the left of the intercept (R² = 0.686) and the intercept of the Q² regression (Q² = −0.767) indicated that the OPLS-DA model was valid (Figure 3B). In lipidomic analysis, 161 lipid metabolites in the CDAHFD-D1 subgroup and 189 in the CDAHFD-D3 subgroup exhibited significant differences when compared with the lipidome of the CDAHFD-D0 subgroup (Figure 3C). Similarly, 22 of them were identical in the lipidomes of days 1 and 3 (Table S5), and their relative abundance was illustrated in a heatmap (Figure 3D). The analyzed lipid species were sorted into nine lipid species, with PCs (n = 10) occupying the foremost position (Figure 3E).

Figure 3.

The number of modified PCs ranked first in the regeneration-related lipidome in the CDAHFD-treated PHx mouse model

(A) OPLS-DA analysis in the CD group (R²X = 0.686, R²Y = 0.972, and Q² = 0.767).

(B) Permutation test of the OPLS-DA models (repeated 200 times).

(C) Venn diagram of lipid components related to regeneration.

(D) Heatmap of differentially modified lipid metabolites in the CDAHFD group.

(E) Classification of differentially modified lipid components in the CDAHFD-treated PHx mice.

The hepatic lipidome during regeneration significantly differed between CDAHFD-treated and CD-treated mice. Therefore, a hypothesis was proposed that PCs significantly contributed to liver regeneration in MASH and healthy livers.

PC (16:0/20:4) was identified as a lipid biomarker for liver regeneration

The mRNA levels of proliferation genes Cyclin D1 and Cyclin E1 increased by 2.5- and 6.5-fold, respectively, on day 1 compared to those on day 0 in the CDAHFD-treated mice (Figures 4A and 4B). The mRNA levels of PC synthesis genes Pcyt1a, Pemt, Chka, and Cept1 increased by 2.6-, 2.6-, 2.0-, and 1.6-fold on day 1 relative to those on day 0 (Figures 4C–4F). Their modifications were higher than those in CD-treated mice, consistent with the number of PCs significantly modified in the two types of regeneration models (Figure S4). The correlation coefficients between Cyclin D1 and Pcyt1a, Chka, Cept1, and Pemt were 0.99, 0.95, 0.95, and 0.96, respectively. The correlation coefficients between Cyclin E1 and Pcyt1a, Chka, Cept1, and Pemt were 0.95, 0.88, 0.98, and 0.94, respectively (Figure 4G). These data indicated that the enhanced regeneration in MASH liver was probably attributed to the increase in PC components.

Figure 4.

PC metabolism correlated well with the regeneration peak in the CDAHFD-treated PHx mouse model

(A–B) The mRNA levels of Cyclin D1 and Cyclin E1.

(C–F) The mRNA levels of Pcyt1a, Pemt, Chka, and Cept1.

(G) Spearman’s correlation analysis between proliferation phenotype and PC synthesis in the CDAHFD-treated PHx mice.

Data were expressed as mean ± SD. One-way ANOVA between multiple groups. n = 5, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, compared with CDAHFD-D0 subgroup.

Since PC species ranked second in the CD subgroups and first in the CDAHFD subgroups, and the PC metabolism exhibited a strong correlation with the regeneration peak in the CDAHFD subgroups, it was hypothesized that PC components were involved in liver regeneration in both models. The Venn diagram illustrating key metabolites from the four paired comparisons (CD-D0 subgroup/CD-D1 subgroup and CD-D3 subgroup, CDAHFD-D0 subgroup/CDAHFD-D1 subgroup and CDAHFD-D3 subgroup) revealed that lipid component PC (16:0/20:4) increased in all regenerative groups (Figure 5A). Raw data analysis revealed that the relative abundance of PC (16:0/20:4) significantly increased during liver regeneration in CD and CDAHFD subgroups, and its level in the CDAHFD subgroups was higher than that in the CD subgroups (Figure 5B). These data indicated that PC (16:0/20:4) might function as a lipid biomarker related to liver regeneration in MASH livers and healthy livers.

Figure 5.

Biomarker promoting regeneration was identified as PC (16:0/20:4)

(A) Venn diagram of the potential biomarker that kept promoting regeneration in both CD- and CDAHFD-treated PHx mice.

(B) Relative abundance of potential biomarker PC (16:0/20:4) in CD and CDAHFD subgroups.

(C–D) Extracted chromatogram peak of the QC sample and authentic PC (16:0/20:4) in positive ion mode.

(E–F) Mass spectrum of the QC sample and the authentic PC (16:0/20:4) in positive ion mode.

(G–H) Extracted chromatogram peak of the QC sample and the authentic PC (16:0/20:4) in negative ion mode.

(I–J) Mass spectrum of the QC sample and the authentic PC (16:0/20:4) in negative ion mode.

(K) Supposed fragmentation pattern of PC (16:0/20:4) in positive ion mode.

(L) Supposed fragmentation pattern of PC (16:0/20:4) in negative ion mode.

In structural validation, the liver quality control (QC) sample and authentic PC (16:0/20:4) yielded a peak with a protonated m/z of 782.56 at the same retention time (RT) in the positive ion mode (Figures 5C and 5D). Figures 5E and 5F illustrate that the characteristic fragments of the liver QC sample and standard PC (16:0/20:4) in positive mode were m/z 184.07, representing the phosphocholine group in the structure of PC (16:0/20:4).²⁰ In the negative mode analysis conducted in a different analysis batch, the liver QC sample and authentic PC (16:0/20:4) yielded a peak with a protonated m/z of 826.56 at RT 13.6 min, which was anticipated to be the adduct ion [M + HCOO]⁻ (Figures 5G and 5H). This peak fragmented at m/z 766.56, 303.23, and 255.23 via losing one methyl group, a 20:4 fatty acid chain, and a 16:0 fatty acid chain (Figures 5I and 5J). Figures 5K and 5L depict the purported cleavage patterns in positive mode and negative mode. The biomarker suggested above in MASH livers and healthy livers was structurally identified as PC (16:0/20:4).

PC (16:0/20:4) facilitated rodent liver regeneration: in vitro and in vivo

After 24 and 48 h of challenge, PC (16:0/20:4) elicited a dose-dependent increase in Hepa1-6 proliferation rate at concentrations ranging from 2 to 50 μM (Figures 6A and 6B). A similar tendency was observed in H&E staining, where a significant increase in cell number was caused by 10 and 50 μM PC (16:0/20:4) (Figures 6C–6E). Biochemically, an increase in transcription of three out of four genes associated with proliferation (Cyclin D1, Ki67, and Cyclin B1) was detected in the 10 and 50 μM groups compared to those in the 0 μM group (Figures 6F–6I).

Figure 6.

PC (16:0/20:4) facilitated the proliferation of Hepa 1-6 and the mouse model

(A–B) CCK8 determination of Hepa 1-6 cells after 24 and 48 h of PC (16:0/20:4) challenge.

(C–D) Number of Hepa 1-6 cells after 24 and 48 h of PC (16:0/20:4) challenge.

(E) H&E staining of Hepa 1-6 cells after 24 and 48 h of PC (16:0/20:4) challenge. Scale bars, 50 μm.

(F–I) The mRNA levels of Cyclin D1, Ki 67, Cyclin B1, and Cyclin E1.

(J–L) The mRNA levels of Cyclin D1, E1, and Pcna determined using qPCR in the mouse liver challenged with PC (16:0/20:4).

(M) Hepatic protein levels of CYCLIN D1 and PCNA in the mouse liver challenged with PC (16:0/20:4).

Data were expressed as mean ± SD. One-way ANOVA between multiple groups. n = 5, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, compared with 0 μM group.

In vivo data supported the above in vitro results. In the mouse PHx model, PC (16:0/20:4) treatment for 24 h increased the transcription of the proliferation-related genes (2.3-fold, 1.3-fold, and 1.2-fold), including Cyclin D1, Cyclin E1, and Pcna, compared to the PHx group (Figures 6J and 6L). In protein analysis, Cyclin D1 and PCNA were significantly higher in the PHx + PC group than in the PHx group, indicating that the regeneration in the mouse PHx model was activated by PC (16:0/20:4) (Figure 6M).

PC (16:0/20:4) facilitated human liver regeneration: organoid and clinical evidences

The proliferation rate of human Huh-7 cells was enhanced in the 10 and 50 μM groups compared with those in the control group after 24 and 48 h of PC (16:0/20:4) challenge (Figures 7A and 7B). Similar to the mouse Hepa 1-6, the number of Huh-7 cells in the 10 and 50 μM groups increased significantly compared to the 0 μM group in H&E staining (Figures 7C–7E). Cyclin D1, Ki 67, Cyclin B1, and Cyclin E1 mRNA levels were increased by 10 and/or 50 μM PC (16:0/20:4) challenge for 24 h (Figures 7F–7I).

Figure 7.

PC (16:0/20:4) facilitated the proliferation of Huh-7 cells

(A–B) CCK8 determination of Huh-7 cells after 24 and 48 h of PC (16:0/20:4) challenge.

(C–D) The number of Huh-7 cells after 24 and 48 h of PC (16:0/20:4) challenge.

(E) H&E staining of Huh-7 cells after 24 and 48 h of PC (16:0/20:4) challenge. Scale bars, 50 μm.

(F–I) The relative mRNA levels of Cyclin D1, Ki 67, Cyclin B1, and Cyclin E1.

Data were expressed as mean ± SD. One-way ANOVA between multiple groups. n = 5, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, compared with 0 μM group.

In human liver organoids, PC (16:0/20:4) induced a significant increase in proliferation at 2–50 μM in the growing organoids by the 48 h challenge (Figures 8A and 8B). The expression level of proliferation genes (Ki 67, Cyclin B1, Cyclin D1, and Cyclin E1) in liver organoids was assessed. In contrast to the reaction in the differentiating organoids, the upregulation of gene transcription associated with proliferation (Cyclin D1, Ki67, and Cyclin B1) was observed in the 50 μM group compared to the 0 μM group in the growing organoids (Figures 8C–8E and S5).

Figure 8.

PC (16:0/20:4) facilitated the proliferation of human organoids and the clinical evidence

(A) Representative images of growing human liver organoids on day 1 and day 4 after PC (16:0/20:4) challenge. Scale bars, 250 μm.

(B) Proliferation of growing human liver organoids determined by CellTiter-Glo3D (n = 3/group).

(C–E) The mRNA levels of Cyclin D1, Ki 67, and Cyclin B1 after 3 days of PC (16:0/20:4) challenge (n = 3/group).

(F) Concentration of the serum PC (16:0/20:4) among patients with MASLD in comparison with that in healthy volunteers (n = 30–32/group).

(G) Concentration of the serum PC (16:0/20:4) increased in PHx patients whose samples on day 1 were not missing.

(H) Concentration of the serum PC (16:0/20:4) increased in PHx patients whose samples on day 1 were missing.

Data were expressed as mean ± SD. One-way ANOVA between multiple groups. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, compared with 0 μM group.

Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis revealed that compared with healthy individuals, the serum levels of PC (16:0/20:4) among patients with MASLD were significantly higher, consistent with the data in the mouse models shown above (Figure 8F). Among patients undergoing PHx, the serum levels of PC (16:0/20:4) were measured longitudinally. Additionally, 24 h after PHx, PC (16:0/20:4) levels decreased in most patients, attributed to the acute loss of liver mass. During the subsequent 10 days, a steady rebound of PC (16:0/20:4) abundance in the serum was observed in almost all of these patients, whether the samples on day 1 were missing or not (Figures 8G and 8H). This dynamic tendency of PC (16:0/20:4) abundance was temporally coupled with liver regeneration, but it went faster, considering the mass restoration in humans usually takes 1–3 months. This fast increase of PC (16:0/20:4) was supposed to be mobilized to promote regeneration by an unknown mechanism and not the effect of mass restoration.

Discussion

Liver regeneration is mediated by pre-existing hepatocytes, hepatic progenitor cells, and biliary epithelial cells, depending on the injury context.²¹ The classical 70% PHx model used in this study primarily represents hepatocyte-driven regeneration. Clinical hepatectomy in the MASLD/MASH context typically inhibits liver regeneration, leading to an increased incidence of complications and liver failures.¹³ Patients with MASLD-HCC exhibit larger tumors and a poorer prognosis than patients with HBV-HCC after hepatectomy.²² A study of 88 patients who underwent hepatectomy demonstrated impaired liver regeneration in moderate-to-severe steatosis cases, based on computed tomography volumetry.²³ This study indicated that liver regeneration outcome was delayed in the MASH mice as evidenced by the comparison of liver weight and liver index between CD- and CDAHFD-treated mice, corroborating the above clinical observation. However, when the proliferation potential was evaluated, the liver regeneration in the MASH PHx model was temporally advanced. This phenotype aligned with the reported function of the PC complex and did not contradict the adverse clinical outcome of PHx when the complicated regeneration mechanism was considered.

After PHx, the synthesis of mitochondrial oxides results in increased oxidative stress, thereby activating the expression of inflammatory factors, including TNF-α and IL-6. The elevated inflammation subsequently activates the expression of genes associated with proliferation.²⁴ In a mouse model of severe steatosis, the risk of performing PHx increased because the activation of lipid peroxidation and the pro-inflammatory response inhibit the regenerative capacity of the liver.²⁵ Herein, the mRNA levels of inflammatory factors Il-2, Il-10, Tnf-α, and Ccl2 were significantly increased after PHx in the CD-treated mice. Accordingly, the inflammation responses synchronized with regeneration dependency. These findings are consistent with reported observations and the known relationship between inflammation and regeneration.

In this study, the CDAHFD PHx model was used to explore how a steatohepatitis background affects the lipidome after hepatectomy. CDAHFD feeding provides a rapid and reproducible non-obese model with steatohepatitis, allowing comparison of regenerative responses between healthy and MASH-like livers. This setting is also relevant to the clinical heterogeneity of fatty liver disease, as lean and non-obese NAFLD/MASLD are increasingly recognized. As reported, 19.2% of individuals with NAFLD were lean, and 40.8% were non-obese.²⁶ Thus, the CDAHFD PHx model in this work provided a useful experimental context for investigating liver regeneration with non-obese steatohepatitis background. Nevertheless, this model should be interpreted as an exploratory system rather than a model that fully recapitulates the natural progression of MASH.

The cell cycle is a crucial phase in cellular life activities, and cyclins are the key regulators. In the G1 phase, Cyclin C, Cyclin D, and Cyclin E expression are the highest, with Cyclin D widely regarded as the core regulator of the G1 phase.²⁷ In the mouse liver, three isoforms of Cyclin D exist, with Cyclin D1 being the most abundant.²⁸ Consequently, Cyclin D1 expression can serve as a direct indicator of liver regenerative capacity. This study found that the levels of proliferative protein YAP, p-ERK, PCNA, and G1-phase genes Cyclin D1 and Cyclin E1 mRNA were all elevated 24 h after PHx in the MASH mice. These modifications occurred earlier than those in healthy mice. Therefore, compared to the regeneration in a healthy liver, hepatocytes in steatosis entered the G1 phase earlier, which correlated with the higher PC (16:0/20:4) levels in the CDAHFD-treated subgroups.

Lipids are essential components of living organisms and are also involved in transport, storage, and signaling. mTORC2 promotes liver regeneration through sphingolipid-induced peroxisome proliferator-activated receptor-α/fatty acid oxidation pathway.²⁹ A significant increase in phosphatidylethanolamine after hepatectomy and its synthesis activation enhanced the proliferative capacity of hepatocytes in rats.³⁰ This study observed elevated TC and TG levels in the liver of the CD subgroups, coupled with decreased serum levels (Figure S6). A substantial quantity of TG components was elevated in the CD-treated PHx model. However, the quantity of PC components was ranked first in the CDAHFD-treated PHx model. This represented a significant difference in the lipidome of liver regeneration in these two contexts. Considering the different regeneration profiles in the two models, the PC components were deemed significant in regulating hepatocyte proliferation.

PC regulates the normal growth and development of organisms by influencing various stages of the cell cycle.³¹ In mice, the PC levels increased 2.6-fold 22 h after hepatectomy, and the activation of PC synthesis thereafter provided hepatic regeneration raw materials and promoted the proliferation of hepatocytes.¹³ PCs participate in cell proliferation by activating the intracellular nucleus PC-dependent phospholipase C.³² In this study, the quantity of modified PC ranked second in the CD-treated PHx model and first in the CDAHFD-treated PHx model. Significant correlations were observed between PC synthesis genes (Pcyt1a, Chka, Cept1, and Pemt) and proliferation markers (Cyclin D1 and Cyclin E1) in the MASH regeneration model. This strong association indicated that PC metabolism was associated with liver regeneration.

PC isoforms performed essential functions in mammalian cells, distinct from their structural role in biological membranes.³³ During the hepatic regeneration peak, the levels of monounsaturated PCs increase and polyunsaturated PCs decrease, indicating that monounsaturated PCs are closely related to the proliferation of hepatocytes.³⁴ However, no published study reported that any single identified PC compound promoted cell proliferation or liver regeneration. This study revealed that the endogenous lipid component PC (16:0/20:4), which facilitates liver regeneration, was undermined using a Venn diagram of regeneration in MASH livers and healthy liver. Notably, the structure of the potential biomarker was identified as PC (16:0/20:4) using the authentic compound. Its function was validated by promoting the proliferation of Hepa 1-6, Huh-7 cells, and human liver organoids. These data suggested that PC (16:0/20:4) was a lipid marker promoting liver regeneration.

In clinical practice, after hepatectomy or liver transplantation, patients may experience small-for-size syndrome (SFSS) or liver failure due to insufficient residual liver volume.³⁵ A systematic review revealed a mortality of 21% attributable to small-for-size syndrome after major hepatectomy.³⁶ External application of melatonin significantly enhanced survival in small-for-size liver transplant recipients by preventing ischemia-reperfusion injury, maintaining sinusoidal circulation, and restoring regenerative capacity.³⁷ Unfortunately, there is currently no effective medication available to resolve the SFSS problem. Herein, PC (16:0/20:4) was identified as an endogenous lipid component and demonstrated the potential to promote liver regeneration. It might be developed as a medicine promoting liver regeneration with no risk of toxicity-related failure in the future.

Considering the shortage of liver donors, ex vivo cultivating lesion-free liver segments represents a novel strategy to increase donor graft availability.³⁸ During ex situ dynamic organ preservation, cell-free therapeutics utilizing concentrated stem cell-derived extracellular vesicles demonstrated promising outcomes.³⁹ Organoid transplantation is gaining increasing interest in repairing damaged tissues. These 3D multicellular spheroids recapitulate in vivo conditions by modeling dynamic regulation of signaling pathways and paracrine signaling dependent on cell-cell and cell-matrix interactions. Indeed, primary hepatocyte- and cholangiocyte-derived organoids demonstrate functional engraftment in animal models of hepatobiliary injury.⁴⁰ In this study, a primary human hepatocyte-derived human liver organoid model was established to mimic hepatocytes after hepatectomy. The results showed that PC (16:0/20:4) promoted organoid proliferation in a dose-dependent manner. These findings necessitate more investigations into its capacity to enhance ex vivo liver regeneration.

In conclusion, differential liver regeneration characteristics in healthy and MASH scenarios closely relate to the distinct hepatic lipidome signatures. PC (16:0/20:4) is an endogenous lipid biomarker which increases to promote liver regeneration and holds a potential for development in ex vivo organogenesis and in vivo therapy.

Limitations of the study

There were some limitations in this work which should be mentioned. After structural identification and function validation of PC (16:0/20:4), it is interesting to explore more about its downstream signaling and molecular targets mediating the regeneration. Since the longitudinal sham-operated controls were not included, the effect of persistent surgical stress and baseline MASH pathology on the lipid dynamics could not be excluded. Additionally, the CDAHFD model did not fully recapitulate the complete natural history of obesity-associated human MASH. So, studies using additional MASH-associated hepatectomy model and larger clinical cohort will strengthen the translational relevance of PC (16:0/20:4), where the effects of hepatocyte enlargement and hypertrophy are suggested to be considered.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Caide Lu (lucaide@nbu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The raw LC-MS data analyzed in this work were deposited in the MetaboLights (identifier: MTBLS13644; URL: https://www.ebi.ac.uk/metabolights/reviewer963724bf-6bb1-4295-a180-ddfc14c48c8f).
• •All data reported in this article will be shared by the lead contact upon request.
• •This article does not report the original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

This work was supported by the Ningbo Major Research and Development Plan Project (2024Z179), Ningbo Top Medical and Health Research Program (2024020818), the National Natural Science Foundation of China (82304126), the Natural Science Foundation of Zhejiang Province (Y24H030001 and LQ24H280002), the Natural Science Foundation of Ningbo (2023J072, 2024J464, and 2023J069), the Health Fund of Translational Biomedicine (H2023000588), the Zhejiang Provincial Medical and Health Science and Technology Projects (2025KY1588), and the K.C. Wong Magna Fund in Ningbo University.

Author contributions

L.C., conceptualization, methodology, investigation, data curation, and writing – original draft; Y.W., formal analysis, validation, and visualization; J.Y., methodology and investigation; W.S., formal analysis and software; P.L., formal analysis; W.Z., software; X.X., investigation; X.G., resources; J.H., investigation; J.F., investigation; X.Y., conceptualization and supervision; M.D., methodology, validation, and visualization; A.L., project administration and writing – review and editing; C.L., conceptualization, funding acquisition, and writing – review and editing.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
GAPDH Rabbit pAb	ABclonal	Cat# AC001, RRID: AB_2619673
ERK1/2 Rabbit mAb	ABclonal	Cat# A4782, RRID: AB_2863347
Phospho-ERK1-T202/Y204 + ERK2-T185/Y187 Rabbit pAb	ABclonal	Cat# AP0472, RRID: AB_2756833
YAP1 Rabbit mAb	ABclonal	Cat# A21216
YAP1 (Phospho-Ser127) Rabbit mAb	Signalway antibody	Cat# 14199
PCNA Polyclonal antibody	Proteintech	Cat# 10205-2-AP, RRID: AB_2160330
Cyclin D1 Rabbit mAb	ABclonal	Cat# A19038, RRID: AB_2862530
Biological samples
Human serum samples from clinical cohort 1, see Table S1	the First Hospital of Jiaxing	N/A
Human serum samples from clinical cohort 2, see Table S2	the Affiliated Lihuili Hospital of Ningbo University	N/A
Chemicals, peptides, and recombinant proteins
LPC (12:0)	Sigma-Aldrich	Cat# 855475P, CAS: 20559-18-6
PC (15:0)	Sigma-Aldrich	Cat# 850350P, CAS: 3355-27-9
PC (16:0/20:4)	Santa Cruz	Cat# sc-213759, CAS: 35418-58-7
Critical commercial assays
Alanine aminotransferase Assay Kit	Meikang Biotechnology	N/A
Aspartate aminotransferase Assay Kit	Meikang Biotechnology	N/A
Alkaline phosphatase Assay Kit	Meikang Biotechnology	N/A
Total bile acid Assay Kit	Meikang Biotechnology	N/A
Total cholesterol Assay Kit	Meikang Biotechnology	N/A
Triglyceride Assay Kit	Meikang Biotechnology	N/A
TransScript Uni&RT qPCR Kit	TransGen	Cat# AUQ-01
HepatiCult™ Organoid Kit (Human)	Stemcell	Cat# 100-0386
Deposited data
The raw data of lipidomics generated	This paper	Metabolights: MTBLS13644
Uncropped western blot images, see Data S1	This paper	N/A
Raw microscopy images, see Data S2	This paper	N/A
Experimental models: Cell lines
Huh7 cell	Ningbo University	N/A
Hepa1-6 cell	Ningbo University	N/A
Experimental models: Organisms/strains
ICR mice	Slac Laboratory	N/A
Oligonucleotides
Primer List, see Table S3	This paper	N/A
Software and algorithms
Graph pad PRISM v10	GraphPad software	https://www.graphpad.com/, RRID: SCR_002798
Masslynx	Waters Corp.	https://www.waters.com/waters/zh_CN/MassLynx-MS-Software/nav.htm? locale = zh_CN&cid = 513662; RRID: SCR_014271
ImageJ	National Institutes of Health	https://ImageJ.net/; RRID: SCR_003070
Analyst software (version 1.6)	AB SCIEX	N/A
Other
Choline-deficient, L-amino acid-defined, high-fat diet (CDAHFD)	SYSE Bio-tech Corp.	Cat# A16092003

Experimental model and study participant details

Human samples

The studies were approved by the ethics committee of the Affiliated Lihuili Hospital of Ningbo University (Approval Number KY2024SL094-01) and the First Hospital of Jiaxing (Approval Number 2024-LY-646). All research procedures were in compliance with guidelines of the 1975 Declaration of Helsinki. Written informed consent was obtained from each individual before participation.

For clinical cohort 1, a total of 62 subjects were separated into two groups: health control (n = 30) and MASLD (n = 32). Serum triglycerides, serum cholesterol and abdominal ultrasound were performed to detect MASLD. Individuals with abnormalities in laboratory test and abdominal ultrasound were excluded from the discovery cohort. The clinical characteristics of the 62 subjects were listed in Table S1. Serum samples from cohort 1 were used for targeted serum PC (16:0/20:4) quantification. Clinical cohort 2 included 15 individuals undergoing PHx to validate PC (16:0/20:4) changes during liver regeneration. The clinical characteristics of the 15 subjects were summarized in Table S2. Serum samples from cohort 2 were used for PC (16:0/20:4) quantification too.

Mice

Male Crl:CD1 (ICR, Institute for Cancer Research) mice (7-week old, 26 ± 2 g) were procured from Shanghai Slac Laboratory Animal Co., Ltd. (Shanghai, China). All mice were acclimatized for at least 7 days to laboratory conditions (room temperature 21°C–25°C, humidity 60–70%, and 12/12 h light–dark cycle) and allowed ad libitum water and CD. Animal procedures adhered to the Institute of Laboratory Animal Resources guidelines. All protocols were approved by the Animal Care and Use Committee of Ningbo University (Approval number: NBU20220122). Male mice were used in this study to avoid potential confounding effects from the estrous cycle in females.

Cell lines and cell culture

Hepa1-6 and Huh-7 cell lines used in this study are listed in the key resources table. Both cells were cultured in Dulbecco’s modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin.

Method details

Animal models

The mice were randomly stratified into CD (n = 25) and CDAHFD groups (n = 25) and fed with Chow diet (CD) and Choline-deficient, L-amino acid-defined, high-fat diet (CDAHFD) (45 kcal% fat with 0.1% Methionine, no added choline and 1% cholesterol) for 1 week. In both dietary groups, the mice in the D0 subgroup underwent midventral laparotomy without manipulation of the liver (sham surgery). The mice in the D1, D3, D5, and D7 subgroups underwent a standard 70% PHx as detailed by Nevzorova et al.⁴¹ All surgical, bleeding and killing procedures were performed under isoflurane inhalation anesthesia. After surgery, the mice were treated with the above two diets until euthanasia. Five mice from each subgroup were euthanized postoperatively on day 0, 1, 3, 5, and 7. Terminal blood collection was performed under anesthesia. After euthanasia, the liver tissues were promptly excised and scaled. Half of the largest liver lobe from each mouse was fixed in 10% formalin solution. The residual liver tissues were stored at −80°C pending analysis.

To verify the role of PC (16:0/20:4) during liver regeneration, a second experiment was performed using 10 male ICR mice (7-week old, 26 ± 2 g) subject to a standard 70% PHx. The PHx+PC group was treated with 500 mg/kg PC (16:0/20:4) dissolved in corn oil (2% dimethyl sulfoxide) by single intraperitoneal injection before peritoneal closure (n = 5), whereas the PHx group was administered an equivalent volume of the vehicle. 500 mg/kg was selected as an exploratory dose based on the endogenous serum PC (16:0/20:4) abundance, mouse circulation volume, tissue distribution, and pharmacological estimation. All the mice were euthanized 24 h post-surgery to assess early post-hepatectomy regenerative responses, particularly hepatocyte cell-cycle entry and proliferation-associated signaling.

Cellular experiments

Hepa1-6 and Huh-7 cells were inoculated, cultured for 24 h, and subsequently exposed to a series of concentrations of PC (16:0/20:4) for 24 and 48 h at approximately 50% confluence. The cells underwent transcription analysis and were assessed for proliferative capacity using the cell counting kit-8 (CCK-8) assay. At designated time points, the cells were incubated with 10 μL CCK-8 solution for 2 h at 37°C. Absorbance was subsequently measured at 450 nm. Cellular proliferation phenotype was captured using a microscope (Olympus BX41) after hematoxylin and eosin (H&E) staining.

Regeneration assay using human liver organoids

Human liver tissues were acquired from the non-tumor margin of hepatoblastoma with full informed consent from the patients. Approximately 2 cm³ fresh liver tissue was washed in ice-cold DMEM/F-12 containing 1% FBS and minced into 0.5–1 mm³ pieces. The isolation and culture followed the manufacturer’s guidelines provided in the commercial kits (Stemcell Technologies, 100–0386). Organoids were passaged every 5–7 days at a split ratio of 1:4–6, and passages 3–7 were used in this study. Liver organoids of the growth stage (3 days post passage) and differentiation stage (7 days post passage) were treated in twelve replicates with PC 16:0/20:4 (2, 10, and 50 μM) for 24 h. One RNA sample from each triplicate was pooled for subsequent mRNA analysis. The pooled sample was reverse-transcribed in triplicate for cDNA synthesis and subsequent transcription analysis.

A CellTiter-Glo 3D assay (Promega, Wisconsin, USA) was performed to evaluate the cell viability of the organoid growth stage. The culture medium of the organoids was discarded, and 50 μL of pre-warmed detection reagent was added according to the procedure. Finally, the organoids were incubated for 30 min at 25°C and the luminescence was measured.

Biochemical analysis

The TC and TG in serum were determined using the endpoint method recommended in the manufacturer’s instructions. To measure the hepatic TC and TG, 20 mg of liver tissue was prepared, to which 180 μL of saline was added. The samples were subsequently homogenized using Magna Lyser (Roche, USA), and the supernatants were extracted after centrifugation. The analysis procedures were performed according to the kit’s descriptions. Serum ALT, AST, TBA, and ALP levels were measured using the dynamic method recommended in the manufacturer’s instructions.

Histopathological assessment

Paraffin-embedded liver tissue sections (4 μm) were dewaxed and rehydrated in alcohol gradients for subsequent H&E staining. Histopathological changes were observed under light microscopy (Olympus BX41). The sections were examined and the images were independently analyzed by a pathologist who was anonymized to the study design.

Transcription analysis

Table S3 presents the primer sequences utilized in this study. Total RNA extraction, reverse transcription, and amplification protocols were similar to those published.⁴² A 5 μL PCR system was designed for the 384 plate, comprising 1 μL cDNA, 2.5 μL 2 × Perfectstart Green qPCR SuperMix, 0.2 μL forward and reverse primers, and 1.1 μL nuclease-free water. The qPCR was performed using a LightCycler 480 II (Roche, USA). The 2^–ΔΔCt method was employed to quantify transcript abundance. The transcription of 18S or GAPDH was used as an internal standard. The expression of target genes was expressed as relative abundance compared with the internal standards.

Western blot analysis

The liver proteins were isolated and quantified as previously described.⁴³ The proteins were separated on a 10% SDS-PAGE and subsequently transferred to 0.2 μM polyvinylidene difluoride (PVDF) membranes following standard preparation and quantification. The PVDF membranes were subsequently blocked in 5% skimmed milk for 3.5 h at room temperature, followed by overnight incubation with primary antibodies at 4°C. The blotted PVDF membranes were incubated with secondary antibodies for 2 h at room temperature, followed by three washes with TBST. The images were collected using ChemiScope 6000 Touch (Shanghai, China). Western blot results were analyzed using ImageJ (version 1.53 K; National Institutes of Health, Bethesda, USA).

Lipidomic analysis

Liver samples for LC-MS analysis were prepared according to a previous report with minor modifications.⁴⁴ Briefly, 350 μL methanol-water (4:3, v/v) and 400 μL cold chloroform containing LPC 12:0 as an internal standard were added to 50 mg of liver tissue. Samples were centrifuged (12000 rpm for 5 min at 4°C), and the lower phase was evaporated using a nitrogen stream. Isopropyl alcohol-acetonitrile-water 2:1:1 (v/v/v) served as the reconstitution solvent. Subsequently, 200 μL supernatant was transferred to a glass vial for LC-MS analysis. Equal volumes of the supernatant form each sample were combined to create a quality control (QC) sample.

The lipidomic analysis conducted using UPLC-ESI-QTOFMS adhered to the previous methods.⁴⁵ For all the samples, an aliquot 5 μL were injected and separated using reverse-phase chromatography with an Acquity UPLC C18-RP column (ACQUITY UPLC BEH C18 1.7 μM, Waters). The mobile phase A comprised 10 mM ammonium formate, acetonitrile-water (60:40) (10 mM formic acid). The mobile phase B comprised 10 mM ammonium formate and isopropanol-acetonitrile (90:10) (10 mM formic acid), utilized in the positive ion mode. In the negative ion mode, 10 mM ammonium acetate was employed with the same composition of solvents A and B as in the positive ion mode. The elution gradient was 40–50% B from 0 to 5 min, 50–99% B from 5 to 18 min, 40% B from 18 to 20 min. The flow rate was 0.3 mL/min. The QC sample was injected after every seven injections to validate stability.

Biomarker undermining and structure validation

The raw LC-MS files were analyzed using MS-Dial (http://prime.psc.riken.jp/) to provide peak tables of m/z and retention time (RT) for the samples. Lipid metabolites were provisionally identified using MS-Dial with the Lipid Blast database and reconfirmed through LIPID MAPS (www.lipidmaps.org) and HMDB (https://hmdb.ca/) to confirm the m/z of the parent compound and its fragments. The abundance data were imported into SIMCA13 (Umetrics, Umea, Sweden) for principal component analysis, orthogonal partial least squares discriminant analysis (OPLS-DA), and S-plot analysis. The relative abundance of these lipids was normalized by comparing their peak areas to those of the internal standard. Variable importance projection (VIP) values exceeding 1.0 and p < 0.05 were selected as differential metabolites.

Authentic compound PC (16:0/20:4) was solubilized in isopropyl alcohol-acetonitrile-water (2:1:1, v/v/v). The QC sample was generated in accordance with the lipidomic analysis. MS and MS/MS data were acquired to obtain the accurate mass measurement of precursor ions and their product ions. The biomarker authenticity was validated by comparing the RT and mass spectrum of the QC sample with those of the authentic standard in positive and negative modes.

Quantification of PC (16:0/20:4) in the human samples

We added 190 μL methanol with 100 ng/mL internal standard PC 15:0 to 10 μL serum, followed by vigorous vortexing for 10 min. After centrifugation at 12,000 g for 20 min, 200 μL supernatant was diluted by adding methanol with internal standard PC 15:0 and subsequently analyzed using LC-MS/MS.

The LC-MS/MS analysis was conducted using a QTRAP 4500 mass spectrometer (AB SCIEX, Foster City, CA). Chromatographic separation was conducted using an EVO C18 column (50 × 2.1 mm, 2.6 μm, Phenomenex, Tianjin, China) at 40°C. The mobile phase (water containing 0.1% formic acid and 5 mM ammonium acetate (A) and acetonitrile (B)) was maintained at a flow rate of 0.4 mL/min. The elution gradient comprised 40% B from 0 to 0.5 min, 40–98% B from 0.5 to 4 min, 98% B from 4 to 5 min, 98–40% B from 5 to 5.1 min, and 40% B from 5.1 to 6 min. The mass spectrometric detection was performed with ion pairs 782.6/184.1 and 782.6/599.2 in positive mode. Data acquisition and processing were performed using the Analyst software (version 1.6) from AB SCIEX.

Quantification and statistical analysis

Data are expressed as the mean ± standard deviation (SD). Graphs and statistical tests were performed using GraphPad Prism software (version 10; San Diego, California, USA). After confirming normal distribution of data, data normality was assessed using the Shapiro–Wilk test. One-way analysis of variance with the post hoc Bonferroni test was utilized for group comparison. Differences were considered significant at p < 0.05 and were further stratified to p < 0.01 and p < 0.001.