Cholangiocytes contribute to hepatocyte regeneration after partial liver injury during growth spurt in zebrafish

Abstract

The liver's regenerative ability depends on injury extent. Minor injuries are repaired by hepatocyte self-duplication, while severe damage triggers cholangiocyte involvement in hepatocyte recovery. This paradigm is well-documented for adult animals but is less explored during rapid growth. We design two partial liver injury models in zebrafish, which were investigated during growth spurts: 1) partial ablation, killing half the hepatocytes; and 2) partial hepatectomy, removing half a liver lobe. In both injuries, de novo hepatocytes emerged alongside existing ones. Single-cell transcriptomics and lineage tracing with Cre-driver lines generated by genome editing identified cholangiocytes as the source of de novo hepatocytes. We further identify active mTORC1 signalling in the uninjured liver of growing animal to be a regulator of the enhanced plasticity of cholangiocytes. Our study suggests cholangiocyte-to-hepatocyte transdifferentiation as the primary mechanism of liver regeneration during periods of rapid growth.

Fig. 1. CellCousin-enabled partial ablation of hepatocytes and subsequent tracking of hepatocyte regeneration.

A Schematic illustrating CellCousin construct for labeling hepatocytes by Cre/lox-mediated-recombination. B Strategy for the ablation of mCherry-NTR-expressing hepatocytes. C Schematic illustration showing the two potential modes of hepatocyte regeneration after partial ablation and the associated fluorescent labeling. D Schematic illustrating the experimental strategy. E Longitudinal in vivo confocal imaging of the same liver at 0, 1 and 2 dppa showing emergence of mTagBFP2+ hepatocytes in presence of spared H2B-mGL+ hepatocytes. Scale bar: 200 μm. F Barplots with mean ± SD of the percentage of hepatocyte labeling before (pre-Mtz, n = 5 animals) and after Mtz treatment (0,1 and 5 dppa, n = 10 animals each). G Min-to-max boxplot showing quantification of normalized number of mTagBFP2+ cells before and after Mtz treatment (pre-Mtz, n = 14 animals; after Mtz treatment (0,1 and 5 dppa), n = 10 animals each) (Kruskal Wallis test).

Fig. 2. Cholangiocytes proliferate in response to partial ablation of hepatocytes.

A Schematic illustrating the lineage tracing of cholangiocytes after partial ablation via the CellCousin method. Hepatocytes are labeled with H2B-mGL+ or mCherry-NTR+ after 4-OHT treatment in the Tg(fabp10a:CellCousin) line. Cholangiocytes, which are Notch-responsive, are marked with bright H2B-mCherry+ (nuclear red) using the Tg(tp1:H2B-mCherry) line. After MTZ administration, Specific ablation of mCherry-NTR+ hepatocytes leads to a liver with spared H2B-mGL+ hepatocytes and de novo nls-mTagBFP2+ hepatocytes. Early de novo hepatocytes derived from cholangiocytes are detected as nls-mTagBFP2 + /H2B-mCherry+ (nuclear magenta), while late de novo hepatocytes, due to dilution of H2B-mCherry label, are marked only with nls-mTagBFP2. B Confocal images of livers from EdU labeled uninjured and 6 hppa Tg(fabp10a:CellCousin); Tg(tp1:H2B-mCherry) transgenic animals. Arrows mark proliferating early de novo hepatocytes. Scale bar: 200 μm (left panels), 50 μm (insets). C Barplot with Mean ± SD showing quantification of the percentage of EdU+ hepatocytes, cholangiocytes, and early and late de novo hepatocytes in uninjured, 6 dppa, 24 dppa and 48 dppa (n = 6 animals). (ANOVA, followed by pair wise comparison using Two-Tailed Mann Whitney. For cholangiocytes and de novo hepatocytes, comparison with uninjured cholangiocytes is shown. ** p-value = 0.0022; * p-value < 0.05).

Fig. 3. Establishment of a partial hepatectomy (PHx) model of liver regeneration.

A Schematic illustration (above) and brightfield image (below) showing the surgical procedure for PHx of the left lobe in 10 dpf larvae. B Representative confocal images of the left and right liver lobe from uninjured or 0 hours-post-injury (hpi) PHx Tg(fabp10a:H2B-mGL) animals. Site of resection marked with white dashed line. C Min-to-max boxplot showing quantification of the total number of hepatocytes (nuclei) at 10 dpf in the livers subjected to PHx (n = 10 animals, each) (Two-tailed Welch’s t test). Each dot represents one animal; mean ± SEM is indicated above each box. D Confocal images of livers from EdU labeled Tg(fabp10a:H2B-mGL); Tg(tp1:H2B-mCherry) animals at 11 dpf (uninjured) and 1 dpi. Scale bar: 200 μm (a, b), 50 μm (a’, b’). E Schematic depicting the segregation of the liver lobes into a distal region of 50 μm thickness and a proximal region. F, G Quantification representing Mean ± SD of the percentage of EdU+ hepatocytes (F) and cholangiocytes (G) in uninjured, 1-, 2- and 3-dpi separated by distal and proximal regions of the left (injured) (n = 6 animals each) and right (uninjured) (n = 6, 6, 3, 5 animals, respectively) lobe. (ANOVA followed by Kruskal–Wallis test; *p-value < 0.05).

Fig. 4. Contribution of de novo hepatocytes to regeneration following partial hepatectomy (PHx).

A Schematic showing the potential modes of regeneration after PHx and the associated fluorescent labeling of hepatocytes. B Experimental strategy for 4-OHT-pulse labeling and PHx in the Tg(fabp10a:CellCousin) background. C Representative confocal images of the left lobe from live imaging of the Tg(fabp10a:CellCousin) animals that were uninjured or subjected to PHx. D Min-to-max boxplot showing quantification of the normalized number of mTagBFP2+ cells in uninjured (n = 29 animals) and PHx animals (n = 30 animals) at 4 dpi (Two-tailed Mann–Whitney test). Each dot represents one animal; mean ± SEM is indicated above each box. Scale bar: 200 μm (B). E Barplot with Mean ± SEM showing quantification of the percentage of mTagBFP2+ hepatocytes using FACS in uninjured (n = 4 biological replicates) and 4 dpi (n = 3 biological replicates) animals (Two-tailed Welch’s t-test). Each dot represents one biological replicate; mean ± SEM is indicated above each box.

Fig. 5. De novo hepatocytes display transcriptional convergence with uninjured hepatocytes.

A Schematic representation of the single-cell RNA-seq approach to investigate the cellular and transcriptional landscape in basal condition (control-uninjured, N = 2) and during regeneration after partial ablation (0,1,9 dppa) and partial hepatectomy (4, 11 dpi). B UMAP visualization of cell types detected in zebrafish larval liver, combining data from injured and uninjured liver conditions. C Heatmap displaying the average scaled expression of selected markers across different cell types. D UMAP visualization focusing on specific injured conditions, highlighting mTagBFP2+ cells and spared hepatocytes (0,1,9 dppa and 4,11 dpi), control hepatocytes, control cholangiocytes, and cholangiocytes in injured conditions (cholangiocytes-rest). The remaining clusters are shown in gray. E Correlation matrix of hierarchically clustered mTagBFP2+ and spared hepatocytes compared to control hepatocytes and cholangiocytes. F Heatmap showing the normalized expression of hepatocyte and cholangiocyte-specific marker genes. Expression of selected genes that are detected in mTagBFP2+ cells at 0 dppa, 1 dppa and 4 dpi are outlined (dashed rectangles).

Fig. 6. her2+ cells generate hepatocytes after PHx.

A The transgenic lines used to lineage trace her2+ cells. B Schematic illustrating the tracing of her2 + -derived hepatocytes following PHx. C Experimental timeline for lineage tracing. D, F Live confocal images of left lobe in uninjured and PHx condition at 4 dpi (D) and at 8 dpi (F). Representative cholangiocyte-derived de novo hepatocytes are marked with arrowheads. Dashed line marks the border of the regenerate. E, G Min-to-max box plots showing the normalized number of H2B-mGL⁺/mCherry⁺ hepatocytes per unit volume at 4 dpi (E) and 8 dpi (G). Each dot represents one animal; mean ± SEM is indicated above each box. Statistical significance was assessed using the Two-tailed Mann–Whitney test. Sample sizes are indicated below the respective groups. Scale bars: 100 μm (left panels), 50 μm (insets) (D, F).

Fig. 7. mTORC1 activity in uninjured liver regulates cholangiocyte plasticity during liver regeneration.

A UMAP visualization showing the integration of uninjured zebrafish larval and adult liver cells. B Unsupervised clustering of cell types presents in larval and adult datasets. The dashed area indicates cholangiocytes present in the two datasets. C Heatmap of mTORC1 signaling pathway-related gene expression in larval and adult cholangiocytes (top). Fold changes and p-values from differential gene expression (DGE) analysis displayed in dot plot (bottom). D Immunostaining for pS6 in uninjured larval and adult zebrafish liver sections. Scale bars: 10 μm. E Min-to-max boxplot showing quantification of pS6 intensity in cholangiocytes from livers of uninjured larval and adult zebrafish (n = 50 cells from 10 animals) (ANOVA followed by Kruskal Wallis test). F Western blot against pS6 and GAPDH for whole livers from 10 dpf larvae and adult zebrafish. G Schematic of the experimental approach for inhibiting the mTORC1 signaling pathway in Tg(fabp10a:CellCousin) zebrafish larvae. H Confocal images of the liver from 3 dpi Tg(fabp10a:CellCousin) larvae treated with DMSO or 10 μM Rapamycin. Dashed line marks the border of the regenerate. Scale bars: 200 μm. I Min-to-max boxplot showing quantification of normalized mTagBFP2+ de novo hepatocytes following treatment with Rapamycin treatment (n = 17 animals) compared to DMSO controls (n = 17 animals). Each dot represents one animal; mean ± SEM is indicated above each box. Statistical significance was assessed using the Two-tailed Mann–Whitney test.