Combinatorial microenvironmental regulation of liver progenitor differentiation by Notch ligands, TGFβ, and extracellular matrix

Abstract

The bipotential differentiation of liver progenitor cells underlies liver development and bile duct formation as well as liver regeneration and disease. TGFβ and Notch signaling are known to play important roles in the liver progenitor specification process and tissue morphogenesis. However, the complexity of these signaling pathways and their currently undefined interactions with other microenvironmental factors, including extracellular matrix (ECM), remain barriers to complete mechanistic understanding. Utilizing a series of strategies, including co-cultures and cellular microarrays, we identified distinct contributions of different Notch ligands and ECM proteins in the fate decisions of bipotential mouse embryonic liver (BMEL) progenitor cells. In particular, we demonstrated a cooperative influence of Jagged-1 and TGFβ1 on cholangiocytic differentiation. We established ECM-specific effects using cellular microarrays consisting of 32 distinct combinations of collagen I, collagen III, collagen IV, fibronectin, and laminin. In addition, we demonstrated that exogenous Jagged-1, Delta-like 1, and Delta-like 4 within the cellular microarray format was sufficient for enhancing cholangiocytic differentiation. Further, by combining Notch ligand microarrays with shRNA-based knockdown of Notch ligands, we systematically examined the effects of both cell-extrinsic and cell-intrinsic ligand. Our results highlight the importance of divergent Notch ligand function and combinatorial microenvironmental regulation in liver progenitor fate specification.

Figure 1. Liver progenitors differentiate into hepatocytes and cholangiocytes via TGFβ and Notch.

(A) Micrographs of BMEL progenitor cells cultured under differentiation conditions (TGFβ1±). BMEL cells in TGFβ1+ were cholangiocytic (ALB−/OPN+) while those in TGFβ1− were hepatocytic (ALB+/OPN−). Scale bars are 50 μm. (B) qRT-PCR analysis of Alb and Opn mRNA transcripts in BMEL cells treated with increasing doses of TGFβ1. Student’s t-tests were performed against 0 ng/ml for each concentration of TGFβ1 with P-values indicated for P < 0.05 (*). (C) qRT-PCR analysis of Alb, Opn, and Sox9 mRNA transcripts in BMEL cells treated with TGFβ1, γ-secretase inhibitor X (5 μM, GSI X), or SB-431542 (10 μM). For the DMSO treatment, Student’s t-tests were performed against 0 ng/ml for each concentration of TGFβ1 with P-values indicated for P < 0.05 (*). For the GSI X and SB-431542 treatments, Student’s t-tests were performed against equal TGFβ1 concentrations in the DMSO treatment with P-values indicated for P < 0.05 (^). Numeric callouts show y-axis values (not P-values). Data presented as mean ± s.e.m. with n = 3. log₂ errors are relative. See also Supplemental Figs S1 and S2.

Figure 2. Jag1 and TGFβ1 coordinate cholangiocytic fate specification.

(A) qRT-PCR analysis of Dll1, Dll4, Jag1, and Jag2 mRNA transcripts in BMEL cells under basal (growth) and differentiation (TGFβ1±) conditions. Student’s t-tests were performed against basal for TGFβ1±. (B) Representative immunoblot against JAG1 in BMEL cells under differentiation conditions (TGFβ1±). Cells were further treated with an equivalent volume of vehicle (DMSO), GSI X (5 μM, GSI), or SB-431542 (10 μM, SB). Molecular weight markers shown in kDa (left) and β-actin control at 45 kDa (bottom). (C) Quantification of JAG1 immunoblots described in (B). Student’s t-tests were performed against DMSO for TGFβ1±. (D) qRT-PCR analysis of Alb and Opn mRNA transcripts in BMEL cells infected with lentiviral shRNA constructs against a non-target sequence (control), Dll1 (shDll1), and Jag1 (shJag1). For shDll1 and shJag1, Student’s t-tests were performed against the same treatment condition (TGFβ1±) in control cells. Numeric callouts show y-axis values (not P-values). Data presented as mean ± s.e.m. with n ≥ 3. P-values indicated for P < 0.05 (*). See also Supplemental Figs S2 and S3.

Figure 3. GFP+/GFP− co-cultures confirm distinct roles for Jag1 and Dll1.

(A) Schematic of GFP+/GFP− co-culture experiment. GFP+ cells were generated using a GFP adenovirus and co-cultured at a 1:50 ratio with GFP− cells under differentiation conditions (TGFβ1±). GFP+ cells were collected after 72 h of culture by flow sorting for downstream RNA isolation and qRT-PCR analysis. (B) qRT-PCR analysis of Alb, Opn, and Sox9 mRNA transcripts in GFP+ cells from co-cultures of every GFP+/GFP− combination of control-, shDll1-, or shJag1-infected BMEL cells. Results were normalized to expression in cultures grown under basal conditions in parallel with co-cultures. For each gene, Student’s t-tests were performed against Control^GFP+ (Control^GFP−) for every combination of GFP− and GFP+ cells. P < 0.05 indicated separately for TGFβ1− (*) and TGFβ1+(^). Numeric callouts show y-axis values (not P-values). Data presented as mean ± s.e.m. with n ≥ 3. See also Supplemental Fig. S4.

Figure 4. Cellular microarrays enable studies of combinatorial microenvironmental regulation.

(A) Schematic of a cellular microarray experiment. Biomolecules and ECM proteins are patterned on a polyacrylamide hydrogel substrate using contact printing. Cells seeded on arrays adhere only to the patterned regions and are exposed to the deposited biomolecules and any experiment-specific soluble factors, fixed at endpoint, immunolabeled, imaged, and analyzed. (B) Analytical pipeline for cellular microarrays. Individual cells on islands are automatically identified by nuclear stain (DAPI) and associated with intensities in other channels, resulting in both single-cell and summary quantifications (e.g., percentage of cells positive for a marker) of results by deposited biomolecule and soluble factor treatment. Scale bars are 100 μm. (C) Experimental pipeline for cellular microarrays. BMEL cells are seeded for 2 h on arrays (sufficient to populate each patterned region), cultured under differentiation conditions (TGFβ1±) for 72 h, fixed, and labeled for nuclei, ALB, and OPN. Arrays shown are 18 × 4.5 mm (40 × 8 spots).

Figure 5. Microenvironmental regulation of liver progenitor differentiation by ECM proteins.

(A) Scatter plot of ALB intensity against OPN intensity by TGFβ1 treatment. Each point represents a single arrayed ECM protein combination. (B) Immunofluorescence micrographs of selected ECM conditions. Scale bar is 50 μm. (C) Single-cell histograms of ALB and OPN label intensity for selected ECM proteins by TGFβ1 treatment. Data presented as mean ± s.e.m. with n = 3. Abbreviations: 1 = collagen I, 3 = collagen III, 4 = collagen IV, F = fibronectin, L = laminin. Combinations denoted by “•”, e.g., “1•3•4” denotes an ECM combination containing collagen I, III, and IV. See also Supplemental Figs S5, S6 and S7.

Figure 6. Arrayed Notch ligands drive cholangiocytic fate specification.

(A) Immunolabeling of arrayed Fc-recombinant JAG1 and DLL1. Scale bar is 50 μm. (B) Immunofluorescence micrograph showing BMEL cells in TGFβ1−. Arrowhead shows spatial specificity of cholangiocytic (OPN+) differentiation at the edge of the island surrounding an OPN—core. Scale bar is 150 μm. (C) ALB quantification of shRNA-infected BMEL cells in TGFβ1− on five ECM proteins. (D) OPN quantification of shRNA-infected BMEL cells in TGFβ1− on five ECM proteins. Break in y-axis applies only to bar for JAG1/laminin/shDll1 condition. Data presented as mean ± s.e.m. with n ≥ 3. Hypothesis testing in (C,D) was performed as follows: For control cells, Student’s t-tests were performed against IgG for each arrayed Notch ligand within each ECM condition with P-values indicated for P < 0.05 (*). For shDll1 and shJag1 cells, Student’s t-tests were performed against the corresponding arrayed Notch ligand for control cells, again within each ECM condition with P-values indicated for P < 0.05 (^). See also Supplemental Figs S8 and S9.

Figure 7. Imaging cytometry of Notch ligand arrays.

(A) Contour maps showing imaging cytometry of shRNA-infected BMEL cells responding to Notch ligands on collagen III. Dotted lines show cutoffs determining cell positivity for both ALB (x-axis) and OPN (y-axis). (B) Immunofluorescence micrographs showing varying response to Notch ligand by shRNA-infected BMEL cells. Scale bar is 150 μm.

Figure 8. Schematic summary of approach and findings.

Cellular microarrays enable controlled studies of the combined effects of microenvironmental signals, including TGFβ, Notch, and ECM. Analysis of BMEL cell differentiation within cellular microarrays and complementary co-culture formats is further suggestive of the following roles for distinct Notch ligands: TGFβ1 and cell-extrinsic Notch ligands (JAG1 and DLL1) cooperate to induce cholangiocytic fate; cell-intrinsic DLL1 plays a role in the suppression of hepatocytic fate in response to TGFβ1; and cell-intrinsic DLL1 inhibits the generation of double-positive (ALB+/OPN+) cells during differentiation.