Genetic labeling does not detect epithelial-to-mesenchymal transition of cholangiocytes in liver fibrosis in mice

Abstract

Background & aims: Chronic injury changes the fate of certain cellular populations, inducing epithelial cells to generate fibroblasts by epithelial-to-mesenchymal transition (EMT) and mesenchymal cells to generate epithelial cells by mesenchymal-to-epithelial transition (MET). Although contribution of EMT/MET to embryogenesis, renal fibrosis, and lung fibrosis is well documented, role of EMT/MET in liver fibrosis is unclear. We determined whether cytokeratin-19 positive (K19(+)) cholangiocytes give rise to myofibroblasts (EMT) and/or whether glial fibrillary acidic protein positive (GFAP(+)) hepatic stellate cells (HSCs) can express epithelial markers (MET) in response to experimental liver injury.

Methods: EMT was studied with Cre-loxP system to map cell fate of K19(+) cholangiocytes in K19(YFP) or fibroblast-specific protein-1 (FSP-1)(YFP) mice, generated by crossing tamoxifen-inducible K19(CreERT) mice or FSP-1(Cre) mice with Rosa26(f/f-YFP) mice. MET of GFAP(+) HSCs was studied in GFAP(GFP) mice. Mice were subjected to bile duct ligation or CCl(4)-liver injury, and livers were analyzed for expression of mesodermal and epithelial markers.

Results: On Cre-loxP recombination, >40% of genetically labeled K19(+) cholangiocytes expressed yellow fluorescent protein (YFP). All mice developed liver fibrosis. However, specific immunostaining of K19(YFP) cholangiocytes showed no expression of EMT markers alpha-smooth muscle actin, desmin, or FSP-1. Moreover, cells genetically labeled by FSP-1(YFP) expression did not coexpress cholangiocyte markers K19 or E-cadherin. Genetically labeled GFAP(GFP) HSCs did not express epithelial or liver progenitor markers in response to liver injury.

Conclusion: EMT of cholangiocytes identified by genetic labeling does not contribute to hepatic fibrosis in mice. Likewise, GFAP(Cre)-labeled HSCs showed no coexpression of epithelial markers, providing no evidence for MET in HSCs in response to fibrogenic liver injury.

Figure 1. EMT and MET was studied using genetic cell fate mapping in mice in response to liver injury

A) Generation of K19^GFP mice to study EMT in K19⁺ cholangiocytes. Upon tamoxifen administration, genetic labeling of cholangiocytes was achieved in K19^GFP mice. B) Summary of genetic crosses in mice. EMT was studied in K19^GFP and FSP-1^YFP mice. MET was studied in GFAP^GFP and Col2(I)^YFP. C) Study design: K19^YFP mice were treated with tamoxifen (5 mg/mouse) prior and throughout the injury, BDL (3 w) or CCl₄- (0.5 µl/g/corn oil; 6 w). The regimen of tamoxifen administration (blue arrows) and liver injury induction (red arrows) is shown.

Figure 2. Induction of liver fibrosis in K19^YFP mice

A) Liver morphology and fibrosis is assessed K19^YFP mice by H&E, Sirius Red staining, immunohistochemistry for α-SMA and FSP-1 expression. Representative images are shown at 20 × and 40 × magnification. B). Quantification of collagen deposition by hydroxyproline, Sirius red and mRNA expression of collagen α1(I), α-SMA and FSP-1 in response to BDL. (* p<0.05, n=10) C) Collagen deposition and expression of fibrogenic genes were increased in response to CCl₄ treatment (* p<0.05, n=10).

Figure 3. Induction of liver injury in K19^YFP mice

A) Tamoxifen-induced Cre-loxP recombination in K19^Cre mice, as shown by immunohistochemistry for YFP (upper panel) and immunofluorescence for pancytokeratin (Pan-CK) and YFP (lower panel). Bile ducts (bd), hepatic artery (ha) and portal vein (pv) are indicated. B) Cre-loxP recombination was quantified in K19^Cre mice. The bars display the number of YFP labeled cholangiocytes in comparison with Pan-CK⁺ cholangiocytes (100%) (p<0,05, n=10). C) Efficiency of Cre-loxP recombination was estimated in tamoxifen-treated K19^YFP mice in comparison with total Pan-CK⁺ cholangiocytes (100%); in GFAP^GFP or Col2(I)^YFP mice was in isolated myofibroblast fractions (100%) by the number of genetically labeled GFP⁺ or YFP⁺ cells, respectively. D) Liver injury increases the number of K19⁺, FSP-1 and GFAP⁺ cells. The number of labeled cells is compared in BDL and sham-operated mice, or in CCl₄- and corn oil-treated mice, and expressed as the ratio calculated for each group.

Figure 4. EMT in cholangiocytes does not contribute to the myofibroblast population in response to liver injury

A) In BDL- or CCl₄-injured K19^YFP mice, genetically labeled YFP⁺ cholangiocytes did not co-express myofibroblast markers (α-SMA, Desmin or GFAP), shown at 400 × magnification. B) Myofibroblast fraction isolated from BDL-operated K19^YFP mice lacked EMT-derived YFP⁺ cells, as detected by immonostaining with anti-GFP antibody. As a control, expression of YFP was detected in 97% of plated cells myofibroblasts isolated from CCl₄-treated Col2(I)^YFP mice (p<0.05). C). Liver tissues from K19^YFP mice were stained with anti-FSP-1 antibody. Liver sections from FSP-1-GFP reporter mice (n=7) were stained with anti-GFP and anti-Pan-CK antibodies. Similarly, FSP-1^YFP mice (n=8) were stained with anti-GFP and anti-Pan-CK antibodies. Representative images are shown at 40 × and 400 × magnification.

Figure 5. Genetically labeled quiescent or activated HSCs do not undergo MET in response to CCl₄ injury or during recovery

A) Morphology of liver tissues from GFAP^GFP mice is shown prior to injury (upper panel) and following CCl₄ administration (lower panel). Quiescent and activated HSCs are labeled by membrane-bound GFP expression (green arrows), and hepatocytes retain expression mTRed (white arrows; 200 × magnification). B) Genetically labeled GFAP⁺ quiescent HSCs do not express MET markers in response to CCl₄ or during recovery in GFAP^Cre mice (n=10). Images show morphology (upper panel) and co-staining (lower panel) of the same tissue section with anti-α-SMA, anti-Desmin, E-cadherin (E-cad) or Pan-CK antibodies and visualized using Alexa-Fluor-633-conjugated secondary antibodies for α-SMA, Desmin (shown in pseudo-red color), and E-cadherin and Pan-CK (pseudo-blue color; 600 × magnification). C) Serial liver sections from GFAP^GFP mice demonstrate differential localization of genetically labeled GFAP⁺ HSCs and Pan-CK⁺ cholangiocytes in response to CCl₄. D) Genetically labeled activated HSCs do not undergo MET in CCl₄-treated Col2(1)^YFP mice (n=10). Col2(1)^YFP mice upregulate YFP in all activated HSCs and co-express α-SMA and desmin, but lack Pan-CK expression (200 × magnification).

Figure 5. Genetically labeled quiescent or activated HSCs do not undergo MET in response to CCl₄ injury or during recovery

A) Morphology of liver tissues from GFAP^GFP mice is shown prior to injury (upper panel) and following CCl₄ administration (lower panel). Quiescent and activated HSCs are labeled by membrane-bound GFP expression (green arrows), and hepatocytes retain expression mTRed (white arrows; 200 × magnification). B) Genetically labeled GFAP⁺ quiescent HSCs do not express MET markers in response to CCl₄ or during recovery in GFAP^Cre mice (n=10). Images show morphology (upper panel) and co-staining (lower panel) of the same tissue section with anti-α-SMA, anti-Desmin, E-cadherin (E-cad) or Pan-CK antibodies and visualized using Alexa-Fluor-633-conjugated secondary antibodies for α-SMA, Desmin (shown in pseudo-red color), and E-cadherin and Pan-CK (pseudo-blue color; 600 × magnification). C) Serial liver sections from GFAP^GFP mice demonstrate differential localization of genetically labeled GFAP⁺ HSCs and Pan-CK⁺ cholangiocytes in response to CCl₄. D) Genetically labeled activated HSCs do not undergo MET in CCl₄-treated Col2(1)^YFP mice (n=10). Col2(1)^YFP mice upregulate YFP in all activated HSCs and co-express α-SMA and desmin, but lack Pan-CK expression (200 × magnification).

Figure 6. HSCs do not express markers of hepatic progenitor cells in response in CCl₄

Progenitor markers are not co-expressed in GFAP^GFP+ HSCs. Livers from CCl₄-treated GFAP^GFP mice were stained with antibodies that recognize hepatic progenitors in the ductal zone and periductular zone. Tissue structure (left) and co-staining (right) are shown for each immunostaining (pseudo-red color). Representative images are shown at 600 × magnification.