This protocol details a monolayer, serum-free method to efficiently generate hepatocyte-like cells from human pluripotent stem cells (hPSCs) in 18 days. This entails six steps as hPSCs sequentially differentiate into intermediate cell-types such as the primitive streak, definitive endoderm, posterior foregut and liver bud progenitors before forming hepatocyte-like cells.
The liver detoxifies harmful substances, secretes vital proteins, and executes key metabolic activities, thus sustaining life. Consequently, liver failure—which can be caused by chronic alcohol intake, hepatitis, acute poisoning, or other insults—is a severe condition that can culminate in bleeding, jaundice, coma, and eventually death. However, approaches to treat liver failure, as well as studies of liver function and disease, have been stymied in part by the lack of a plentiful supply of human liver cells. To this end, this protocol details the efficient differentiation of human pluripotent stem cells (hPSCs) into hepatocyte-like cells, guided by a developmental roadmap that describes how liver fate is specified across six consecutive differentiation steps. By manipulating developmental signaling pathways to promote liver differentiation and to explicitly suppress the formation of unwanted cell fates, this method efficiently generates populations of human liver bud progenitors and hepatocyte-like cells by days 6 and 18 of PSC differentiation, respectively. This is achieved through the temporally-precise control of developmental signaling pathways, exerted by small molecules and growth factors in a serum-free culture medium. Differentiation in this system occurs in monolayers and yields hepatocyte-like cells that express characteristic hepatocyte enzymes and have the ability to engraft a mouse model of chronic liver failure. The ability to efficiently generate large numbers of human liver cells in vitro has ramifications for treatment of liver failure, for drug screening, and for mechanistic studies of liver disease.
The purpose of this protocol is to efficiently differentiate human pluripotent stem cells (hPSCs) into enriched populations of liver bud progenitors and hepatocyte-like cells2. Access to a ready supply of human liver progenitors and hepatocyte-like cells will accelerate efforts to investigate liver function and disease and could enable new cellular transplantation therapies for liver failure3,4,5. This has proven challenging in the past since hPSCs (which include embryonic and induced pluripotent stem cells) can differentiate into all the cell-types of the human body; consequently, it has been difficult to exclusively differentiate them into a pure population of a single cell-type, such as liver cells6.
To precisely differentiate hPSCs into liver cells, first it is critical to understand not only how liver cells are specified but also how non-liver cell-types develop. Knowledge of how non-liver cells develop is important to logically suppress the formation of non-liver lineages during differentiation, thereby exclusively guiding hPSCs towards a liver fate2. Second, it is essential to delineate the multiple developmental steps through which hPSCs differentiate towards a liver fate. It is known that hPSCs sequentially differentiate into multiple cell-types known as the primitive streak (APS), definitive endoderm (DE), posterior foregut (PFG) and liver bud progenitors (LB) before forming hepatocyte-like cells (HEP). Earlier work revealed the signals specifying liver fate and the signals that suppressed the formation of alternate non-liver cell-types (including stomach, pancreatic, and intestinal progenitors) at each developmental lineage choice2,7,8.
Collectively, these insights have given rise to a serum-free, monolayer method to differentiate hPSCs towards primitive streak, definitive endoderm, posterior foregut, liver bud progenitors and finally, hepatocyte-like cells2. Overall the method involves the seeding of hPSCs in a monolayer at an appropriate density, preparing six cocktails of differentiation media (containing growth factors and small molecules that regulate various developmental signaling pathways), and sequentially adding these media to induce differentiation over the course of 18 days. During the process, no passaging of cells is needed. Of note, because this method explicitly includes signals that suppress the formation of non-liver cell-types, this differentiation approach1 more efficiently generates liver progenitors and hepatocyte-like cells by comparison to extant differentiation methods2,9,10,11,12. Furthermore, the protocol described in this text enables the faster generation of hepatocytes that ultimately express higher levels of hepatic transcription factors and enzymes than those produced by other protocols9,10,11,12.
The protocol described here has certain advantages over current differentiation protocols. First, it entails monolayer differentiation of hPSCs, which is technically simpler compared to three-dimensional differentiation methods, such as those that rely on embryoid bodies13. Second, this method exploits a recent advance whereby definitive endoderm cells (an early precursor to liver cells) can be efficiently and rapidly generated within 2 days of hPSC differentiation2,7, thus enabling the subsequent production of hepatocytes with increased purity. Third, in side-by-side comparisons, the hepatocyte-like cells produced by this method2 produce more ALBUMIN and express higher levels of hepatic transcription factors and enzymes compared to hepatocytes produced in other methods10,11,12.
1. Preparation of Differentiation Media
NOTE: Refer to the Table of Materials for manufacturer information regarding the materials and reagents used.
2. Seed hPSCs onto Plates at Defined Densities for Differentiation
3. Differentiation of hPSCs into Endodermal Cells and Liver Progenitors
4. Characterization of Endodermal Cells and Liver Progenitors by Immunostaining
5. Characterization of Liver Progenitors by Fluorescence Activated Cell-sorting (FACS) Analysis
NOTE: Use FACS to precisely quantify the percentage of AFP+ differentiated LB cells that emerge by day 6 of differentiation. Follow the same steps to quantify the percentage of ALB+ differentiated hepatocytes by day 18 of differentiation.
After 24 h of APS differentiation, colonies will generally adopt a different morphology than undifferentiated colonies concomitant with a loss of the bright border that typically circumscribes hPSC colonies. Morphologically, primitive streak cells generally have ragged borders and are more spread and less compact than hPSCs-this is evocative of an epithelial-to-mesenchymal transition as pluripotent epiblast cells differentiate and ingress into the primitive streak in vivo. If the colony size of the hPSCs prior to differentiation was too large, there will be an overtly-raised center that comprises undifferentiated cells. Cultures containing such large clumps should be discarded, as these undifferentiated colony centers will confound subsequent differentiation. If clumps of the correct size and density were plated, APS differentiation is highly reproducible and uniform, generating a 99.3±0.1% MIXL1-Gfp+ primitive streak population (MIXL1 is a primitive streak marker)7. Note that some cell death is observed after 24 h of APS differentiation.
By day 2 of differentiation, primitive streak cells have differentiated into day 2 definitive endoderm cells, the vast majority of which express SOX17 (Figure 2) and FOXA2. In dozens of independent experiments with 14 hESC and hiPSC lines (including H1, H7, H9, HES2, HES3, BJC1, BJC3, HUF1C4 and HUF58C4), this approach scalably, consistently and efficiently generated pure definitive endoderm populations (94.0% ± 3.1%)7. This method to generate definitive endoderm from hPSCs yields higher percentages of CXCR4+PDGFRA– endodermal populations compared to other endoderm-forming approaches2,14.
By day 3 of differentiation, endoderm has differentiated into foregut progenitors that appear polygonal in shape (Figure 1). Later, by 6 days of differentiation, the foregut progenitors differentiate into liver bud progenitors expressing AFP, TBX3, and HNF4A (Figure 3). Across three hESC lines (H1, H7, H9 hESCs), this method generates day 6 AFP+ liver progenitors form at an efficiency of 89.0±3.1% (Figure 2). Finally, after 18 days of liver differentiation from hPSCs, ALBUMIN+ hepatocyte-like cells appear. Morphologically, they appear epithelial, forming bright borders reminiscent of bile canaliculi (Figure 1). At this stage, the cytoplasm of the day 18 hPSC-derived hepatocytes appears darker than the nucleus (Figure 1).
Figure 1: Schematic of differentiation strategy and morphology of undifferentiated hPSCs, differentiating endoderm and hepatocyte-like cells. Differentiation process and timeline were shown. Abbreviations: d = day, hPSC = human pluripotent stem cells, APS = anterior primitive streak, DE = definitive endoderm, PFG = posterior foregut, LB = liver bud progenitors, HP = hepatocyte progenitors, HEP = hepatocyte like cells. Scale bar = 400 μm. Please click here to view a larger version of this figure.
Figure 2: Percentage of day 1 MIXL1+ primitive streak, day 2 SOX17+ definitive (def) endoderm, day 6 AFP+ liver bud progenitors and ALB+ hepatocyte populations as shown by FACS. Please click here to view a larger version of this figure.
Figure 3: Immunostaining analysis of day 6 liver progenitors. hPSC were first differentiated into liver bud progenitors, which were immunostained for liver bud transcription factors HNF4A (red), TBX3 (green) as well as cytoplasmic liver bud marker AFP (red). Nuclei were counterstained with DAPI (blue) to assess total cell number. Scale bar = 50 μm. Please click here to view a larger version of this figure.
Figure 4: Cell seeding density of hPSCs. Low (left) and appropriate (middle and right) density of cells seeded. Scale bar = 1,000 μm. Arrow points to "bridges" between colonies of cells. Please click here to view a larger version of this figure.
Base medium | Composition of base medium |
CDM2 | 1:1 IMDM/F12, 0.1% m/v PVA, 1% concentrated lipids, 0.7 μg/mL human recombinant Insulin, 15 μg/mL Transferrin, 18 nM 1-thioglycerol |
CDM3 | 1:1 IMDM/F12, 0.1% m/v PVA, 1% concentrated lipids, 10% KOSR |
CDM4 | 1:1 IMDM/F12, 1% concentrated lipids, 15 μg/mL Transferrin |
CDM5 | CMRL, 10% KOSR, 1% Glutamax |
Table 1: Composition of chemically defined media CDM2, CDM3, CDM4, and CDM5.
Differentiation Stages | Duration | Factors | Doses | Base medium |
Day 1 Primitive Streak | 1 day | Activin | 100 ng/mL | |
CHIR99201 | 3 μM | CDM2 | ||
PI103 | 50 nM | |||
FGF2 | 10 ng/mL | |||
Day 2 Definitive Endoderm | 1 day | Activin | 100 ng/mL | |
DM3189 | 250 nM | CDM2 | ||
PI103 | 50 nM | |||
Day 3 Posterior Foregut | 1 day | A83-01 | 1 μM | |
TTNPB | 75 nM | CDM3 | ||
BMP4 | 30 ng/mL | |||
FGF2 | 10 ng/mL | |||
Day 6 Liver Bud progenitors | 2 days | Activin | 10 ng/mL | |
C59 | 1 μM | CDM3 | ||
BMP4 | 30 ng/mL | |||
Forskolin | 1 μM | |||
1 day | Activin | 10 ng/mL | ||
CHIR99201 | 1 μM | CDM3 | ||
BMP4 | 30 ng/mL | |||
Forskolin | 1 μM | |||
Day 12 Hepatic progenitors | 6 days | BMP4 | 10 μg/mL | |
OSM | 10 ng/mL | |||
Dexamethasone | 10 μM | |||
Forskolin | 10 μM | CDM4 | ||
Ro4929097 | 2 μM | |||
AA2P | 200 μg/mL | |||
Insulin | 10 μg/mL | |||
day 18 Hepatocytes | 6 days | Dexamethasone | 10 μM | |
Forskolin | 10 μM | CDM4 or | ||
Ro4929097 | 2 μM | CDM5 | ||
AA2P | 200 μg/mL | |||
Insulin | 10 μg/mL |
Table 2: Composition of differentiation media.
Problem | Possible Reason | Proffered Solution |
Cells in center of colony do not differentiate | i) Colony size was excessively large, cells in middle of large colonies were not accessible to differentiation signals | i) Check cell counting technqiue. |
ii) Uneven distribution of clumps which merged together in the center of the well (or its periphery), forming very large colonies | ii) Shake the plate in a cross fashion to evenly distribute clumps during plating, and check under a microscope before placing it into the incubator | |
iii) Cells received insufficient differentiation signals | iii) Add adequate volumes of differentiation media: add 1 mL of medium per well in a 12-well plate and 3 mL per well in a 6-well plate | |
Poor efficiency of differentiation | i) Starting cell culture partially differentiated | i) Use a new, high-quality batch of undifferentiated hPSCs |
ii) Colonies were too densely seeded, forming a confluent sheet of cells | ii) Seed cells and count cells accurately for differentiation, and shake evenly to distribute them | |
iii) Residual media and unwanted signals were not washed off due to inadequate washing | iii) Residual mTeSR1 or induction media from the previous stage of differentiation will block differentiation; wash cells with IMDM prior to adding differentiation medium | |
iv) Washing too harsh; inappropriate washing conditions will severely disrupt cell morphology | iv) Prior to adding either differentiation medium, wash briefly with IMDM. Use of DPBS (or media with different osmolarity or cold media) or extended washing, will compromise cell morphology and viability | |
v) Prolonged or shortened period of differentiation stages, longer or shorter than recommended. | v) Adhere to the recommended timings for each stage of differentiation. |
Table 3: Potential problems and their possible causes and solutions.
This method enables the generation of enriched populations of liver bud progenitors, and subsequently hepatocyte-like cells, from hPSCs. The ability to generate enriched populations of human liver cells is important for the practical utilization of such cells. Previous methods to generate hepatocytes from hPSCs yielded impure cell populations containing both liver and non-liver cells that, upon transplantation into rodents, yielded bone and cartilage in addition to liver tissue15. Hence the explicit suppression of non-liver differentiation is critical to generate enriched liver populations that might be suitable for a variety of applications.
Notably, controlled plating of hPSCs at the very first step is essential for efficient downstream differentiation. It is important to accurately plate hPSCs at a certain density for differentiation, and to evenly distribute these hPSCs across the well or plate during the process of plating. For example, for each well of a 12-well plate, seed 160,000 to 250,000 hPSCs per well; overall, it is imperative to titrate the cell seeding density and to ultimately test the cell density that is appropriate for each cell line (Figure 4). If too many hPSCs are seeded per well, they will form large colonies, which will adversely affect differentiation efficiencies, as cells in the middle of large colonies are less accessible to differentiation signals compared to cells near the periphery; colonies of heterogeneous sizes also will present similar issues. If cell densities are too low (e.g., below 200,000 cells/well of a 12-well plate), then there may not be sufficient material for differentiation and extensive cell death may be observed. The passaging and seeding method described above consistently generates hPSC clumps of suitable size for downstream differentiation.
One limitation of this method is that the hepatocyte-like cells generated are not identical to adult hepatocytes, as the hPSC-derived cells still express immature liver marker AFP. Moreover, CYP3A4 enzymatic activity is approximately 55 times lower in these hPSC-derived hepatocyte-like cells compared to primary adult human hepatocytes. A coming challenge will be to mature these hPSC-derived hepatocyte-like cells into fully-fledged, adult-like cells. A second limitation is that efficient differentiation is extremely dependent on the starting density of cells and therefore, it is very important to seed at the recommended density and to evenly disperse them across the plate (Table 3).
Overall, this protocol produces liver bud progenitors and hepatocyte-like cells at 89.0±3.1% purity in at least 3 hPSC lines. Second, hepatocyte-like cells expressed liver enzymes, secreted human ALBUMIN and expressed higher levels of liver genes than cells generated using extant differentiation approaches. Finally, the resultant hepatocyte-like cells not only exhibit certain hepatocyte functions in vitro, but most importantly they can function to some extent in vivo, as they can engraft a mouse model of chronic liver injury and improve short-term survival2.
The authors have nothing to disclose.
We thank Bing Lim for discussions and the Stanford Institute for Stem Cell Biology & Regenerative Medicine for infrastructure support. This work was supported by the California Institute for Regenerative Medicine (DISC2-10679) and the Stanford-UC Berkeley Siebel Stem Cell Institute (to L.T.A. and K.M.L.) and the Stanford Beckman Center for Molecular and Genetic Medicine as well as the Anonymous, Baxter and DiGenova families (to K.M.L.).
Geltrex | Thermofisher Scientific | A1569601 | |
1:1 DMEM/F12 | Gibco | 11320033 | |
0.2 μm pore membrane filter | Millipore | GTTP02500 | |
mTeSR1 | Stem Cell Technologies | 5850 | |
Thiazovivin | Tocris Bioscience | 3845 | |
Accutase | Gibco or Millipore | Gibco A11105, Millipore SCR005 | |
IMDM, GlutaMAX™ Supplement | Thermofisher Scientific | 31980030 | |
Ham's F-12 Nutrient Mix, GlutaMAX™ Supplement | Thermofisher Scientific | 31765035 | |
KOSR, Knockout serum replacement | Thermofisher Scientific | 10828028 | |
Poly(vinyl alcohol) | Sigma-Aldrich | P8136 | |
Transferrin | Sigma-Aldrich | 10652202001 | |
Chemically Defined Lipid Concentrate | Thermofisher Scientific | 11905031 | |
human Activin | R&D | 338-AC | |
CHIR99201 | Tocris | 4423 | |
PI103 | Tocris | 2930/1 | |
human FGF2 | R&D | 233-FB | |
DM3189 | Tocris | 6053/10 | |
A83-01 | Tocris | 2939/10 | |
Human BMP4 | R&D | 314-BP | |
C59 | Tocris | 5148 | |
TTNPB | Tocris | 0761/10 | |
Forskolin | Tocris | 1099/10 | |
Oncostatin M | R&D | 295-OM | |
Dexamethasone | Tocris | 1126 | |
Ro4929097 | Selleck Chem | S1575 | |
AA2P | Cayman chemicals | 16457 | |
human recombinant Insulin | Sigma-Aldrich | 11061-68-0 |