Mfsd2a+ hepatocytes repopulate the liver during injury and regeneration

Abstract

Hepatocytes are functionally heterogeneous and are divided into two distinct populations based on their metabolic zonation: the periportal and pericentral hepatocytes. During liver injury and regeneration, the cellular dynamics of these two distinct populations remain largely elusive. Here we show that major facilitator super family domain containing 2a (Mfsd2a), previously known to maintain blood-brain barrier function, is a periportal zonation marker. By genetic lineage tracing of Mfsd2a⁺ periportal hepatocytes, we show that Mfsd2a⁺ population decreases during liver homeostasis. Nevertheless, liver regeneration induced by partial hepatectomy significantly stimulates expansion of the Mfsd2a⁺ periportal hepatocytes. Similarly, during chronic liver injury, the Mfsd2a⁺ hepatocyte population expands and completely replaces the pericentral hepatocyte population throughout the whole liver. After injury recovery, the adult liver re-establishes the metabolic zonation by reprogramming the Mfsd2a⁺-derived hepatocytes into pericentral hepatocytes. The evidence of entire zonation replacement during injury increases our understanding of liver biology and disease.

Figure 1. PP hepatocytes labelled by Mfsd2a-CreER.

Schematic figures showing (a) our knock-in strategy for Mfsd2a-CreER allele using CRISPR/Cas9 by homologous recombination and (b) genetic lineage tracing strategy for Mfsd2a⁺ hepatocytes by Cre-LoxP recombination in Mfsd2a⁺ hepatocytes. (c) Whole-mount fluorescence view of the adult liver from 6-week-old Mfsd2a-CreER;Rosa26-RFP mice. Tamoxifen was induced at 2 days before analysis. Scale bar, 1 mm. (d) Immunostaining for RFP, CK19 and HNF4a on liver sections shows Mfsd2a-expressing hepatocytes in PP zone (P). Scale bars, 100 μm. (e) Immunostaining for RFP, PECAM and 4,6-diamidino-2-phenylindole (DAPI) on liver sections shows Mfsd2a-expressing hepatocytes in the PP zone but not PC zone (*). Scale bars, 100 μm. (f) Isolation of RFP⁻ and RFP⁺ cells by flow cytometry followed by quantitative RT–PCR (qRT–PCR) analysis for expression of RFP and Mfsd2a. Expression level of genes in the RFP⁻ cells was set as 1. Error bars are s.e.m. of the mean for all the quantification in this study. (g) Expression of PP and PC genes detected by qRT–PCR. The x axis denotes RFP⁻ (black) and RFP⁺ (red) groups, and the y axis denotes fold induction.. *P<0.05; n=3, two-tailed unpaired t-test. Each immunostaining image is a representative of four individual samples.

Figure 2. PC hepatocytes proliferate faster than PP hepatocytes.

(a) Schematic figure showing time points for EdU injection. (b) Immunostaining for EdU, hepatocyte marker HNF4a and PP zonation marker CDH1 on liver sections of wild-type mice. Scale bars, 100 μm. (c) Z-stack images of PP and PC regions of liver sections. XZ and YZ indicate signals from dotted lines on Z-stack images. Inserts are magnified images. Quantification of percentage of EdU⁺ hepatocytes on PP and PC regions is shown in the middle panel. Scale bars, 100 μm. (d) Immunostaining for HNF4a, EdU and CDH1 on liver sections of Mfsd2a-CreER line. Quantification of percentage of EdU⁺ hepatocytes on PP and PC regions is shown in the middle panel. *P<0.05; n=6; two-tailed unpaired t-test; Scale bars, 100 μm. Error bars are s.e.m. of the mean. Each image is a representative of six individual samples.

Figure 3. PP hepatocytes reduced during liver homeostasis.

(a) Schematic figure showing experimental strategy for tamoxifen treatment (Tam) and tissue analysis at the indicated time points after Tam. (b) Whole-mount fluorescence view of Mfsd2a-CreER;Rosa26-RFP livers collected at the indicated time points. Scale bars, 1 mm. (c) Quantification of the percentage of RFP⁺ and RFP⁻ area. n=4. Error bars are s.e.m. of the mean. (d) Immunostaining for RFP, CK19 and HNF4a on liver sections shows that RFP⁺ hepatocytes remains in the PP regions during homeostasis. Scale bars, 100 μm. (e) Quantification of the percentage of RFP⁺ hepatocytes in the liver. (f) In situ hybridization of PP or PC genes on liver sections from mice at 6–20 weeks old. Scale bars, 500 μm. (g) Schematic figure showing PP hepatocytes reduced during liver homeostasis, whereas expression (Expr.) of zonation genes is maintained. Each image is a representative of four individual samples.

Figure 4. PP hepatocytes expand during liver regeneration.

(a) Schematic figure showing experimental strategy for tamoxifen induction and PH of Mfsd2a-CreER;Rosa26-RFP mice. (b) Whole-mount fluorescence view of sham-operated or PH liver. Inserts indicate the bright-field view of the same liver. Dotted lines demarcate liver lobule. Quantification of the liver lobule size based on sham-operated control (set as 1). *P<0.05; n=4; two-tailed unpaired t-test. Error bars are s.e.m. of the mean. Scale bars, 1 mm. (c) Immunostaining for RFP, CK19 and HNF4a on liver sections shows expansion of RFP+ hepatocytes during liver regeneration compared with the sham control. Inserts indicate quantification of the percentage of RFP⁺ hepatocytes in sham or PH livers. *P<0.05; n=4; two-tailed unpaired t-test. Scale bars, 500 μm. (d) In situ hybridization of PP and PC genes in sham or PH livers. Scale bars, 500 μm. (e) Schematic figure showing PP hepatocytes expand during liver regeneration, whereas expression (Expr.) of zonation genes is maintained. Each image is a representative of four individual samples.

Figure 5. Lineage tracing of Mfsd2a-derived hepatocytes during liver regeneration.

(a) Schematic figure showing experimental strategy for tamoxifen induction and PH of Mfsd2a-CreER;Rosa26-RFP mice. (b) Whole-mount fluorescence view of sham-operated or PH liver. Inserts indicate the bright-field view of the same liver. Dotted lines define liver lobule. Scale bars, 1 mm. (c) Quantification of the liver lobule size based on sham-operated control (set as 1). *P<0.05; n=4; two-tailed unpaired t-test. (d) Immunostaining for RFP, CK19 and HNF4a on liver sections showing expansion of RFP⁺ hepatocytes during liver regeneration compared with the sham control. Scale bars, 100 μm. (e) Quantification of the percentage of RFP⁺ hepatocytes in sham or PH livers. *P<0.05; n=4; two-tailed unpaired t-test. Error bars are s.e.m. of the mean. Each image is a representative of four individual samples.

Figure 6. PP hepatocytes expand and replace almost all hepatocytes in the liver lobule after injury.

(a) Schematic figure showing experimental strategy for tamoxifen induction and CCl₄ treatment. (b) Sirius red staining images showing robust fibrotic responses in CCl₄-treated group, compared with oil-treated group (control). Scale bars, 1 mm. (c) Whole-mount fluorescence view of Mfsd2a-CreER;Rosa26-RFP livers from control (left) and CCl₄ (right)-treated groups. n=4. Error bars are s.e.m. of the mean. Scale bars, 1 mm. (d) Immunostaining for RFP, CK19 and HNF4a on Mfsd2a-CreER;Rosa26-RFP liver sections of control and CCl₄-treated mice. Scale bars, 100 μm. (e) Sequential whole-mount fluorescence view of the same liver from individual mouse before injury and at week 4 after chronic injury. Scale bars, 1 mm. (f) Schematic figure showing expansion of PP hepatocytes after injury. CV, central vein; PV, portal vein. Each image is a representative of four individual samples.

Figure 7. Lineage tracing of Mfsd2a⁺ hepatocytes after single injection of CCl₄.

(a) Schematic figure showing strategy for tamoxifen induction (Tam), CCl₄ treatment and time points for tissue analysis. (b) Haematoxylin and eosin staining of liver sections collected at day 2 or week 2 after after single CCl₄ injection. C, central vein; P, portal vein. Scale bars, 100 μm. (c,d) Immunostaining for RFP, CK19 and HNF4a on liver sections collected at day 2 (c) and week 2 (d) after single CCl₄ injection. Scale bars, 100 μm. Each image is a representative of four individual samples.

Figure 8. Restoration of PC gene expression during recovery.

(a) Schematic figure showing experimental strategy at the indicated time points. (b) Images of liver sections stained with Sirius red. Scale bars, 500 μm. (c) Whole-mount fluorescence view of liver of CCl₄-treated mice during recovery. Scale bars, 500 μm. (d) Immunostaining for RFP, CK19 and HNF4a showing RFP⁺ PC and PP hepatocytes in the recovered liver. Scale bars, 500 μm. (e) Images of in situ hybridization showing recovered PC gene expression during recovery. Scale bars, 500 μm. (f) A cartoon figure showing expansion of PP hepatocytes throughout the entire liver after injury and recovery, whereas some PC genes (grey) were reduced after injury and restored during recovery. Each image is a representative of four individual samples.

Figure 9. Mfsd2a-derived hepatocytes remain throughout the entire liver lobule after recovery from liver injury.

(a) Schematic figure showing strategy for tamoxifen induction (Tam), CCl₄ treatment and analysis after recovery. (b) Sirius red staining on liver sections collected from CCl₄-treated mice after recovery and control mice (no CCl₄ treatment). Scale bars, 100 μm. (c) Whole-mount bright-field and fluorescence view of liver after recovery from CCl₄ treatment. Scale bars, 1 mm. (d) Immunostaining for RFP on liver sections showing RFP⁺ cells retained throughout entire liver lobules after recovery. Scale bars, 100 μm. (e–g) Immunostaining for RFP, CK19 and HNF4a on liver sections showing most hepatocytes in PP zone (f) and PC zone (g) as RFP⁺; f,g are magnified images of boxed regions in e. Scale bars, 100 μm. Each image is a representative of four individual samples.

Figure 10. Reduction of pre-labelled PP hepatocytes after injury induced by BDL.

(a) Schematic figure showing strategies for PP hepatocytes labelling and injury model induced by BDL. (b) Whole-mount bright view of sham and BDL livers. Scale bars, 2 mm. (c) Sirius red staining of liver sections. Scale bars, 200 μm. (d) Whole-mount fluorescence view of sham and BDL livers. Inserts are bright-field images of the same liver. Scale bars, 500 μm. (e) Immunostaining for RFP, CK19 and HNF4a on sham or BDL liver sections. Boxed regions are magnified in lower panels. Insertions indicate the quantification of the percentage of RFP⁺ hepatocytes in sham or BDL livers. n=4. Error bars are s.e.m. of the mean. Scale bars, 500 μm. Each image is a representative of four individual samples.