Modeling alcohol-associated liver disease in humans using adipose stromal or stem cell-derived organoids

Summary

Alcohol-associated liver disease (ALD) is a prevalent liver disease, yet research is hampered by the lack of suitable and reliable human ALD models. Herein, we generated human adipose stromal/stem cell (hASC)-derived hepatocellular organoids (hAHOs) and hASC-derived liver organoids (hALOs) in a three-dimensional system using hASC-derived hepatocyte-like cells and endodermal progenitor cells, respectively. The hAHOs were composed of major hepatocytes and cholangiocytes. The hALOs contained hepatocytes and nonparenchymal cells and possessed a more mature liver function than hAHOs. Upon ethanol treatment, both steatosis and inflammation were present in hAHOs and hALOs. The incubation of hALOs with ethanol resulted in increases in the levels of oxidative stress, the endoplasmic reticulum protein thioredoxin domain-containing protein 5 (TXNDC5), the alcohol-metabolizing enzymes ADH1B and ALDH1B1, and extracellular matrix accumulation, similar to those of liver tissues from patients with ALD. These results present a useful approach for understanding the pathogenesis of ALD in humans, thus facilitating the discovery of effective treatments.

Graphical abstract

Highlights

• •Human adipose stromal/stem cells are used to generate hepatocellular and liver organoids
• •Ethanol exposure induces steatosis and inflammation in hepatocellular organoids
• •Ethanol-treated liver organoids show steatosis, inflammation, and fibrosis
• •These effects phenocopy the clinical manifestations of alcohol-associated liver disease

Motivation

Although the etiology and basic pathological processes of alcohol-associated liver disease (ALD) are well understood, the lack of suitable and reliable human disease models greatly limits the development of effective drugs. Here, we used human adipose stromal/stem cells to construct hepatic organoids and treated them with ethanol to simulate ALD in vitro. These organoids provide a disease model for exploring the mechanism of ALD and personalized therapy for patients with liver metabolic dysfunction.

Bi et al. present two approaches for generating human hepatic organoids using human adipose stromal or stem cells. The generated organoids, when exposed to ethanol, phenocopy several properties of alcohol-associated liver disease (ALD) in vitro. These models can potentially be useful in understanding the pathogenesis of ALD in humans and facilitate the discovery of effective treatments.

Introduction

Alcohol-associated liver disease (ALD) is one of the most prevalent types of chronic liver disease worldwide and encompasses steatosis, steatohepatitis, fibrosis, cirrhosis, and, ultimately, hepatocellular carcinoma.¹ Despite well-characterized disease progression and the mechanisms underlying its pathogenesis, effective drug therapies for any stage of ALD are lacking.² Due to this paucity, suitable and reliable human-relevant models can dependably be used for target discovery and compound screening.^3,4

Human hepatocyte organoids obtained from adult or fetal human hepatocytes,^5,6 human pluripotent stem cell (hPSC)-derived hepatocyte-like cells (HLCs),^7,8,9 and reprogrammed human hepatocytes¹⁰ are three-dimensional (3D) culture systems that recapitulate key morphological, functional, and gene expression features of mature in vivo hepatocytes. Human hepatocyte organoids may model steatosis, the first stage of nonalcoholic fatty liver disease (NAFLD). This stage includes three different triggers: free fatty acid loading, interindividual genetic variability (PNPLA3 I148M),^6,11 and monogenic lipid disorders (APOB and MTTP mutations).⁶ The 3D culture systems based on coculture approaches involving the mixing of hPSC-HLCs with fetal liver mesenchymal cells may simulate the next stage of NAFLD and ALD, steatohepatitis, which is characterized by inflammation and fibrosis.^1,9 However, these systems suffer from artifactual inflammation and fibrosis, in part due to the difficulty in choosing the culture medium and extracellular matrix (ECM) in which multiple cell lineages can be contained.

Recently, a multicellular human liver organoid was developed by codifferentiating epithelial and stromal lineages from hPSCs.¹² This model, coupled with free fatty acid treatment, recapitulates the progressive, stepwise nature of steatohepatitis-like pathology, including steatosis, inflammation, and fibrosis, and can potentially be leveraged for drug screening through the analysis of organoid stiffness.¹² Susceptibility to liver disease states such as steatohepatitis is highly variable. For example, not all individuals who are obese develop steatosis, and most cases of steatosis do not progress to steatohepatitis. Thus, in vitro personalized hepatic model systems are necessary for disease modeling, drug discovery, and drug toxicity studies. Alternatively, we developed a reproducible three-stage method to directly differentiate human adipose stromal/stem cells (hASCs) into functional hepatocytes.¹³

In this study, hASC-derived hepatocellular organoids (hAHOs) and multicellular hASC-derived liver organoids (hALOs) were developed in a 3D system using hASC-derived HLCs (hHLCs)¹³ and endodermal progenitor cells (hEPCs),¹⁴ respectively. hALOs coupled with ethanol treatment truly recapitulated the pathophysiological aspects of ALD. This study provides a powerful in vitro personalized model for understanding the pathogenesis of ALD in humans, thus facilitating the development of useful therapeutic targets for ALD.

Results

hAHOs exhibit the characteristics of expandable hepatocellular organoids

To investigate whether the spheroids of hHLCs derived from hASCs have the ability to support growth in suspension in Matrigel, an optimal expansion medium was designed for generating hAHOs.^5,7,15 During 3 days of culture, small organoids emerged from Matrigel-embedded hHLCs. The organoids expanded to a diameter of 150 μm within 10 days (Figure 1A).

Figure 1.

Establishment of the hAHO culture system

(A) Phase-contrast microscopy images of hAHOs at different stages (left). Scale bar: 20 μm. The line graph shows the sizes of the hAHOs (right). The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance: ∗p < 0.05 and ∗∗p < 0.01.

(B) Transmission electron microscopy image of hAHOs on day 10. Scale bar, 2.0 μm. Mt, mitochondrion; Mv, microvilli; N, nucleus; ER, endoplasmic reticulum; TJ, tight junction. G, Golgi complex; BC, bile capillaries.

(C) Relative mRNA levels of ALB, HNF4A, CYP3A4, and CK19 in hAHOs were determined by quantitative RT-PCR at different stages. The relative expression of each gene was normalized against that of 18S rRNA. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the hAHOs on day 0: ∗p < 0.05 and ∗∗p < 0.01. The hash (#) represents statistical significance compared with the hAHOs on day 5: #p < 0.05 and ##p < 0.01.

(D) Immunofluorescence staining for ALB, CEBPα, CYP2A6, GSTA2, and CK19 in hAHOs on day 10. Scale bar, 50 μm.

(E) PAS staining, LDL-DyLight uptake, and ICG uptake and release were determined in hAHOs on day 10. Scale bar, 20 μm.

(F) Albumin secretion, CYP3A4 activity, and urea production were analyzed in hASCs, hHLCs, and hAHOs on day 10. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the hASCs: ∗p < 0.05 and ∗∗p < 0.01. The hash (#) represents statistical significance compared with the hHLCs: #p < 0.05 and ##p < 0.01.

ALB, albumin; HNF4A, hepatocyte nuclear factor 4 alpha; CYP3A4, cytochrome P450 family 3 subfamily A member 4; CK19, cytokeratin 19; CEBPα, CCAAT enhancer binding protein alpha; CYP2A6, cytochrome P450 family 2 subfamily A member 6; GSTA2, glutathione S-transferase alpha 2; PAS, periodic acid-Schiff; LDL, low-density lipoprotein; ICG, indocyanine green.

To obtain a more comprehensive understanding of the hepatocellular structure, transmission electron microscopy was used for ultrastructural analysis. The typical hepatocellular structure, including abundant mitochondria, endoplasmic reticulum, Golgi apparatus, cell-cell junctions, microvilli, and bile canaliculus architecture, was present in the hAHOs (Figure 1B).

Gene expression over time was assessed by quantitative RT-PCR for three donors during the formation of hAHOs. The results revealed that the expression levels of albumin (ALB), hepatocyte nuclear factor 4 alpha (HNF4A), chrome P450 family 3 subfamily A member 4 (CYP3A4), and keratin 19 (CK19; a bile duct marker) in the organoids on days 5 and 10 were significantly greater than those in the hHLCs (day 0). The expression levels of CK19 in the organoids on day 10 were lower than those in the organoids on day 5 (Figure 1C). The protein levels in hAHOs were then analyzed by immunofluorescence staining. The results revealed that the expression of ALB, CCAAT enhancer binding protein alpha (CEBPα), CYP2A6, and glutathione S-transferase alpha 2 (GSTA2) was strongly positive in hAHOs but weak in CK19 (Figure 1D). They may maintain a stable phenotype during 1 week in culture. If the culture time is prolonged, the main hepatocyte markers will be significantly weakened after 1 week (Figure S1A).

Functionally, hAHOs showed strong positive periodic acid-Schiff staining, which is indicative of glycogen accumulation. Low-density lipoprotein (LDL) uptake was readily visualized by fluorescent probes. hAHOs showed indocyanine green (ICG) uptake and excretion, which are indicative of the interplay between sinusoidal uptake and biliary excretion in determining hepatic ICG clearance in hAHOs (Figure 1E). The levels of ALB secretion, CYP3A4 metabolic activity, and urea production in hAHOs were significantly greater than those in undifferentiated hASCs and hHLCs (Figure 1F). These data indicate that the hepatocellular organoids derived from hHLCs were expandable and acquired polarized epithelial functions.

hAHOs have the ability to engraft and maintain metabolic function in vivo

To address whether hAHOs are able to engraft to nude mouse liver tissue, enhanced green fluorescent protein (EGFP)-labeled hAHOs were produced from EGFP-hHLCs derived from hASCs stably transduced with EGFP, as described previously¹⁶ (Figure 2A). Subsequently, EGFP-hAHOs were implanted at the periphery of the superior right lobe of the damaged liver of athymic nude BALB/c male mice induced by carbon tetrachloride (CCl₄), as described in a previous study.¹⁷

Figure 2.

Transplantation and function of hAHOs in vivo

(A) Phase-contrast and fluorescence microscopy images of the hASCs, EGFP-hASCs, and EGFP-hAHOs. Scale bars, 200 μm and 50 μm (EGFP-hAHOs).

(B) IVIS spectrum imaging of EGFP-hAHOs in nude mice 2 weeks after implantation.

(C) Immunofluorescence analysis was performed on nude mouse liver sections transplanted with hAHOs to assess the expression of ALB, HNF4α, CK19, and EGFP. Scale bar, 50 μm.

(D) The drug metabolism activity of the hAHO graft in the nude mouse liver at 2 weeks post-implantation (right). Schematic representation of the experiments conducted on nude mice (left). Ketoprofen (15 mg/kg; Sigma-Aldrich) was administered intravenously to the mice post-transplantation of the medium (top) or EGFP-hAHOs (bottom) in the injured liver. Urine was collected 2 h after administration. Then, 100 μL urine was mixed with 100 μL of 0.5 M acetate buffer (pH 5.0), and 10 μL of 1 N KOH was added to the urine samples, The mixture was incubated at 80°C for 3 h, neutralized with 10 μL of 1 N HCl, and centrifuged (15,000 rpm, 4°C, 5 min). The supernatant was subjected to mass spectrometry (QuattromicroAPI; Waters). The ion spray voltage was −4,500 V, and the m/z transitions (Q1/Q3) for ketoprofen, 1-hydroxyketoprofen (OH-ketoprofen), and glucuronide-conjugated ketoprofen (G-ketoprofen) were 253, 269, and 429, respectively. The arrow shows the locations of the ketoprofen, OH-ketoprofen, and G-ketoprofen.

After 2 weeks of implantation, the distribution of EGFP-hAHOs in the murine liver was imaged by the IVIS spectrum imaging system (Figure 2B). Immunofluorescence analyses revealed that cells positive for ALB, HNF4α, and CK19 colocalized with EGFP-expressing cells in the graft (Figure 2C).

The functional maturation of hAHOs in the implants was evaluated by the ketoprofen metabolism activity assay, which is known to be metabolized differently in mice and humans.¹⁷ The results showed that 1-hydroxyketoprofen (a metabolite of mice) was detected in the urine samples from both the mice transplanted with hAHOs and the control mice (Figure 2D). However, glucuronide-conjugated ketoprofen (a human metabolite) was detected only in urea samples from mice transplanted with hAHOs (Figure 2D, bottom). These data show that hAHOs can be successfully engrafted to damaged livers and demonstrate metabolic function after transplantation.

Modeling hepatic steatosis and inflammatory responses in hAHOs by alcohol exposure

To assess whether alcohol exposure could trigger steatosis and result in an inflammatory response and fibrosis, hAHOs derived from three donors were treated with 100 mM ethanol for 72 h. Then, lipogenesis and lipid accumulation in hAHOs were analyzed using BODIPY 558/568 C₁₂ and HCS LipidTOX neutral lipid staining. The results showed that the levels of lipogenesis and lipid accumulation in ethanol-treated hAHOs were significantly greater than those in the control group (Figures S2A and S2B). The ethanol-treated hAHOs displayed a reduction in cell viability, as shown by an increase in the dead/live content ratio (Figure S2C).

Quantitative RT-PCR analysis revealed that the expression of the lipogenic-associated enzymes fatty acid synthase (FASN), stearoyl-CoA desaturase (SCD), and acyl-CoA oxidase 1 (ACOX1) in ethanol-treated hAHOs was significantly greater than that in the control group. Moreover, the mRNA levels of the inflammatory response factors interleukin (IL)-6 and tumor necrosis factor alpha (TNFA) were upregulated in ethanol-treated hAHOs and were significantly greater than those in the control group (Figure S2D).

However, the mRNA levels of the fibrosis markers collagen type I alpha 1 chain (COL1A1) and actin alpha 2 smooth muscle (ACTA2) did not change significantly (Figure S2D). To further verify whether hAHOs exhibit a fibrotic response to ethanol, we performed Sirius red staining, Masson staining, and immunohistochemical staining for COLLI and smooth muscle actin (α-SMA). The results showed that the expression levels of COLLI and α-SMA were similar in ethanol-treated hAHOs and control-treated hAHOs (Figure S2E). These data show that hAHOs can recapitulate the initial symptoms of ALD, including steatosis and inflammation, but not fibrosis.

Generation of liver organoids from hASCs

The major nonparenchymal cells, hepatic stellate cells, are lacking in the hAHO system, and thus this system has limited capacity to model fibrosis in ALD. Alternatively, we have developed methods to predominantly differentiate hASCs to hEPCs.^14,18 Then, another organoid culture method of codifferentiating hepatic parenchymal cells and nonepithelial cells from spheres of hEPCs derived from hASCs was developed as previously described.^12,19 After 5 days of retinoic acid (RA) treatment and hepatocyte maturation in differentiation medium for 10 days, human liver organoids from hASCs, named hALOs, were established (Figure 3A).

Figure 3.

Establishment of the hALO culture system

(A) Phase-contrast microscopy images of hALOs at different stages. Scale bar, 20 μm.

(B) Relative mRNA levels of ALB, HNF4A, CYP3A4, and CK19 in hALOs were determined by quantitative RT-PCR. The relative expression of each gene was normalized against that of 18S rRNA. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the hALOs on day 0: ∗p < 0.05 and ∗∗p < 0.01. The hash (#) represents statistically significant difference compared with the hALOs on day 5: #p < 0.05 and ##p < 0.01.

(C) Immunofluorescence staining for ALB, CK19, GSTA2, DES, CYP2A6, CD68, COLLI, and CEBPα in hALOs on day 10. Scale bar, 50 μm.

(D) The percentages of CD26-negative, CD68-negative, and CD166-positive populations in hALOs were determined by fluorescence-activated cell sorting. The results are presented as the means ± SEMs. n = 3.

The mRNA levels of hepatocyte markers (ALB, HNF4A, and CYP3A4) and a bile duct marker (CK19) in hALOs on day 10 were greater than those in hEPCs (day 0) and organoids on day 5 (Figure 3B). The types of cells in hALOs were characterized using immunofluorescence staining for the hepatocellular markers ALB, CEBPα, and GSTA2, the cholangiocyte marker CK19, the stellate cell markers desmin and COLLI, and the Kupffer cell marker CD68 (Figure 3C). They may maintain a stable phenotype during 2 weeks in culture. If the culture time is prolonged, then the main hepatocyte markers will be significantly weakened after 2 weeks (Figure S1B).

To determine the percentage of the different cell types in the hALOs, quantitative fluorescent antibody-based profiling was performed using flow cytometry analysis. The results showed that the frequency of CD26-positive cells (HLCs) was approximately 63.7% ± 0.8% in hALOs, whereas the frequencies of CD166-positive cells (fat-storing cells) and CD68-positive cells (Kupffer-like cells) were 8.1% and 0.5% ± 0.2% in hALOs, respectively (Figure 3D). Ultrastructural analysis showed that HLCs, Kupffer-like cells, fat-storing cells, and endothelial-like cells were found in the hALOs (Figure 4A, left). The typical liver ultrastructures included microvilli, cell junctional complexes, lipid droplets, endoplasmic reticulum, mitochondria, microvessels, and ECM (Figure 4A, right).

Figure 4.

The properties of the hALOs

(A) Transmission electron microscopy image of hALOs on day 10. Scale bars, 10 μm (left) and 2.0 μm. K, Kupffer-like cell; H, hepatocyte-like cells; F, fat-storing cell; E, endothelial-like cell; JC, junctional complexes; M, microvessel; Mv, microvilli; ECM, extracellular matrix; ER, endoplasmic reticulum; LD, lipid droplet; Mt, mitochondria.

(B) LDL-DyLight uptake, Rho-123 staining, and ICG uptake and release were determined in hALOs on day 10. Scale bar, 20 μm.

(C and D) Albumin secretion (C) and urea production (D) were analyzed in hHLCs, hAHOs, and hALOs. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the hHLCs: ∗p < 0.05 and ∗∗p < 0.01. The hash (#) represents statistical significance compared with the hAHOs: #p < 0.05 and ##p < 0.01. CD166, activated leukocyte cell adhesion molecule.

Functionally, hALOs possessed the hepatic-specific functions of LDL uptake, significantly high levels of ICG uptake and excretion, and rhodamine 123 transport (Figure 4B). The levels of ALB secretion and urea production in hALOs were significantly greater than those in hHLCs and hAHOs (Figures 4C and 4D). These data indicate that hALOs contain hepatocytes and stromal lineages of the liver and maintain typical liver function.

Modeling of ALD using hALOs by alcohol exposure

To investigate the differential gene expression in hALOs in response to ethanol exposure, quantitative RT-PCR was performed. The results indicated that the mRNA levels of FASN and SCD, which are related to lipid production; IL-6 and TNF-α, which are related to inflammation; and ACOX1 and carnitine palmitoyl transferase 1A, which are related to oxidative phosphorylation, were increased in ethanol-treated hALOs. This finding was similar to the hAHO response to ethanol exposure. Moreover, the mRNA levels of COL1A1 and ACTA2, which are related to fibrosis, were upregulated in ethanol-treated hALOs (Figure 5A).

Figure 5.

Modeling of ALD using hALOs by alcohol exposure

(A) Relative mRNA levels of FASN, SCD, COL1A1, ACTA2, IL-6, TNFA, ACOX1, and carnitine palmitoyl transferase 1A (CPT1A) in the control-treated and ethanol-treated hALOs were determined by quantitative RT-PCR. The ethanol-treated hALOs were exposed to 100 mM ethanol for 3 days. The relative expression of each gene was normalized against that of 18S rRNA. Significant differences were determined by unpaired two-tailed paired t tests. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control group: ∗p < 0.05.

(B) IL-6 secretion in the control-treated and ethanol-treated hALOs. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control group: ∗p < 0.05.

(C) The level of lipogenesis in the control-treated and ethanol-treated hALOs was analyzed using BODIPY 558/568 C12. Scale bar, 50 μm. The relative fluorescence intensity (548–568 nm) of the ethanol-treated group was normalized to that of the control group. The results are presented as the mean ± SEM. n = 3. The hash (#) represents a donor. The asterisk (∗) represents statistical significance compared with the control group: ∗p < 0.05.

(D) The level of lipid droplets in the control-treated and ethanol-treated hALOs was determined by HCS LipidTOX green neutral lipid staining. Scale bar, 50 μm. Relative fluorescence intensity (490–510 nm) of the ethanol-treated group was normalized to that of the control group. The results are presented as the mean ± SEM. n = 3. The hash (#) represents a donor. The asterisk (∗) represents statistical significance compared with the control group: ∗p < 0.05.

To verify whether pathological changes associated with ALD could be detected in hALOs exposed to alcohol, the inflammatory response and the level of steatosis were examined. The results showed that the levels of IL-6 secretion and lipid accumulation in ethanol-treated hALOs were significantly greater than those in the control group (Figures 5B–5D).

To assess whether hALOs exhibited a fibrotic response to ethanol exposure, alterations in the levels of ECM components were evaluated according to fibrosis clinical parameters. Sirius red and Masson staining analysis suggested that the levels of ECM components were significantly greater in ethanol-treated hALOs than in control hALOs. Immunohistochemical staining and immunofluorescence analyses demonstrated that the increased ECM deposition after ethanol treatment was type I collagen (Figures 6A and 6B). α-SMA, a commonly used marker of smooth muscle cells, influences ECM deposition. Immunofluorescence analyses revealed the increased colocalization of type I collagen and α-SMA in the hALOs after ethanol treatment (Figure 6B).

Figure 6.

The properties of hALO-ALD models

(A) Representative images of Sirius red staining, Masson staining, and COLLI staining in hALOs from the control group and the ethanol-treated group. The ethanol-treated hALOs were exposed to 100 mM ethanol for 3 days. Scale bar, 50 μm. Histogram showing the difference in the percentage of Sirius red-, Masson-, and COLLI-positive areas in the control group and the ethanol-treated group. Significant differences were determined by unpaired two-tailed paired t tests. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control group: ∗p < 0.05.

(B) Immunofluorescence staining for α-SMA and COLLI in the control-treated and ethanol-treated hALOs. The relative fluorescence intensities of α-SMA and COLLI were determined using ImageJ software. Scale bar, 50 μm. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control group: ∗p < 0.05.

(C) Transmission electron microscopy image of the control-treated and ethanol-treated hALOs. Scale bars, 1.0 μm and 5.0 μm. Mt, mitochondrion; N, nucleus; ER, endoplasmic reticulum; BC, bile capillaries; COLL, collagen; ECM, extracellular matrix.

(D) The levels of ECM deposition, Masson positivity, and α-SMA positivity were evaluated in liver tissues in control individuals and three different patients with ALD. The hash (#) represents a donor. There were 10 fields in each group. Scale bar, 50 μm.

More importantly, a large amount of fiber was found in the ethanol-treated hALOs according to ultrastructural analysis (Figure 6C). The mitochondria were significantly shorter and smaller, the endoplasmic reticulum became fragmented, and the lacuna became larger in the hALOs after ethanol treatment (Figure 6C). Certainly, the deposited matrix and the expression of α-SMA were further confirmed in the liver tissue from three donors with ALD (Figure 6D).

The endoplasmic reticulum protein thioredoxin domain-containing protein 5 (TXNDC5) is a member of the protein disulfide isomerase family that plays a critical role in fibrogenesis in various types of tissue.²⁰ Like those in the liver tissue of patients with ALD, the levels of TXNDC5 and α-SMA were significantly increased in ethanol-treated hALOs (Figures S3A and S3B).

Alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH) are the main ethanol-metabolizing ADHs.^1,21 The expression of ADH1B and ALDH1B1 was further determined in ethanol-treated hALOs and control hALOs. The results showed that the mRNA and protein levels of ADH1B and ALDH1B1 were greater in ethanol-treated hALOs than in control hALOs (Figures 7A, 7B, and S3C). Moreover, the expression of ADH1B and ALDH1B1 was further confirmed in the liver tissue from three donors with ALD (Figure 7C).

Figure 7.

The characterization of alcohol dehydrogenases in hALO-ALD models

(A) Relative mRNA levels of ADH1B and ALDH1B1 in the control-treated and ethanol-treated hALOs were determined. The relative expression of each gene was normalized against that of 18S rRNA. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control: ∗p < 0.05 and ∗∗p < 0.01.

(B) The expression of ADH1B and ALDH1B1 was evaluated in the control-treated and ethanol-treated hALOs. Scale bar, 50 μm. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control: ∗p < 0.05.

(C) The expression of ADH1B and ALDH1B1 was evaluated in liver tissues from control individuals and three different patients with ALD. The histogram shows the difference in the percentages of ADH1B and ALDH1B1 in each field between control individuals and patients with ALD. The hash (#) represents a donor. There were 10 fields in each group. Scale bar, 50 μm. The results are presented as the mean ± SEM. The asterisk (∗) represents statistical significance compared with the control: ∗p < 0.05 and ∗∗p < 0.01.

(D) The levels of cellular ROS in the control-treated and ethanol-treated hALOs were determined by a DCFDA/H2DCFDA-Cellular Reactive Oxygen Species Detection Assay Kit. The relative fluorescence intensity of ROS was determined using ImageJ software. Significant differences were determined by unpaired two-tailed paired t tests. Scale bar, 50 μm. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control: ∗p < 0.05.

(E) Cell viability in the control-treated and ethanol-treated hALOs was determined. The fluorescence intensity was determined using ImageJ software. Significant differences were determined by unpaired two-tailed paired t tests. Scale bar, 50 μm. The results are presented as the mean ± SEM. n = 3. The asterisk (∗) represents statistical significance compared with the control: ∗p < 0.05. ADH1B, alcohol dehydrogenase 1B; ALDH1B1, aldehyde dehydrogenase 1 family member B1.

ADH metabolizes ethanol into toxic acetaldehyde while producing reactive oxygen species (ROS), thus further inducing oxidative stress and steatosis in ALD.²¹ We examined the oxidative stress levels of hALOs. We found that ethanol treatment led to increased oxidative stress (Figure 7D). Further study revealed that hALOs exhibit damage, as indicated by a decrease in cell viability and an increase in the dead/alive content ratio (Figure 7E). These data confirm that hALOs may model the pathological process and partial mechanism of ALD.

Discussion

Although embryonic stem cell (ESC)-derived organoids may represent individualized responses to injury treatment, their limited availability is associated with ethical concerns.^9,22,23,24 Recently, organoids derived from irreversibly damaged livers of patients with nonalcoholic steatohepatitis have opened up experimental avenues for personalized disease modeling and drug development.²⁵ ALD comprises a spectrum of disorders and pathologic changes in individuals with acute and chronic alcohol consumption.²⁶ Genetic polymorphisms of alcohol-metabolizing enzymes are associated with heavy drinking habits and dependence.²⁷ The lack of registered drugs for ALD treatment is partly due to the paucity of human-relevant models that reliably simulate the pathophysiology of ALD in humans for target discovery.²⁸

Evidence indicates that hASCs and hASC-HLCs are attractive tools for establishing alternative therapies for liver dysfunction.^29,30,31 Recently, we found that hASC-HLCs from obese patients exhibit the characteristics of abnormal hepatic steatosis and altered mitochondrial structure and oxidative phosphorylation function.³² Therefore, we hypothesize that liver organoid-derived hASCs may be used for personalized ALD modeling.

In this study, we found that hAHOs exhibited more mature structures and metabolic functions than hHLCs.^13,17,18,33 Moreover, hAHOs could recapitulate the initial symptoms of alcohol-associated steatosis and inflammation. Considering that the progression of ALD involves complex cell-cell interactions, hALOs were developed using codifferentiating hepatic parenchymal cells and nonparenchymal cells from hEPC-derived spheres of hASCs.^12,19 The results confirmed that hALOs contain hepatocytes and stromal lineages in the liver and maintain typical liver function compared to hAHOs. But the gene expression of hALOs is still insufficient compared to that of human liver tissues (Figure S4).

Interestingly, the deposited matrix, along with the more α-SMA-positive cells, were present in ethanol-treated hALOs, which indicates that the increased ECM was secreted by the activated stellate cells in hALOs after exposure to ethanol. The increase in ROS production and damaged mitochondria suggested dysfunctional mitochondria in ethanol-treated hALOs. The expression of the endoplasmic reticulum protein TXNDC5 was significantly upregulated in hALOs after ethanol treatment, which may model the properties of endoplasmic reticulum stress in ALD.²⁰ ADH1B and ALDH1B1 exhibit the highest alcohol metabolism activity in the liver in humans.²¹ Ethanol-treated hALOs showed significantly upregulated expression of ADH1B and ALDH1B1, which was further confirmed in the liver tissue of patients with ALD. These results indicate that hALOs can be used as a reliable model to simulate the pathology of ALD in vitro. Overall, our strategy provides ideas for personalized treatment and research on ALD.

Limitations of the study

A limitation of this study is that compared to that of human liver tissues, the gene expression in hALOs is still relative lower. In addition, inter- or intrabatch variability can affect the phenotype of organoids. Moreover, the limitations of studying personalized treatments include small sample sizes and a lack of consideration of individual differences, such as the effect of genetic mutations on disease. Therefore, multiple organoids should be measured under each condition for quantification. Additionally, sample sizes need to be increased, and genetic mutations should be considered.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Collagen Type I Polyclonal antibody	Proteintech	Cat# 14695-1-AP; RRID: AB_2082037
Smooth Muscle Actin Polyclonal antibody	Proteintech	Cat# 14395-1-AP; RRID: AB_2223009
Anti-Albumin antibody	Sigma-Aldrich	Cat# A6684; RRID: AB_258309
Anti-CEBP Alpha/CEBPA antibody	Abcam	Cat# ab40761; RRID: AB_726792
Cytochrome P450 2A6 (CYP2A6) antibody	OriGene	Cat# TA503832; RRID: AB_11126761
Anti-GSTA2, (N-terminal) antibody	Sigma-Aldrich	Cat# SAB1402215; RRID: AB_10641865
Cytokeratin 19 antibody	Abcam	Cat# ab76539; RRID: AB_1523469
HNF4α (C11F12) Rabbit mAb	Cell Signaling	Cat# 3113; RRID: AB_2295208
Anti-Desmin antibody	Abcam	Cat# Ab15200; RRID: AB_301744
Anti-CD68 antibody	Abcam	Cat# Ab955; RRID: AB_307338
ADH1B Monoclonal antibody	Proteintech	Cat# 66939-1-Ig; RRID: AB_2882263
ALDH1B1 Polyclonal antibody	Proteintech	Cat# 15560-1-AP; RRID: AB_2224162
TXNDC5 Polyclonal antibody	Proteintech	Cat# 19834-1-AP; RRID: AB_10644285
BV421 Mouse Anti-Human CD26	BD Biosciences	Cat# 744448; RRID: AB_2742237
FITC Plus Anti-Human CD68	Proteintech	Cat# FITC-65187; RRID: AB_2924123
PE Mouse Anti-Human CD166	BD Biosciences	Cat# 560903; RRID: AB_397210
BV421 Mouse IgG1, κ Isotype Control	BD Biosciences	Cat# 562438; RRID: AB_11207319
FITC Plus Mouse IgG2b Isotype Control	Proteintech	Cat# FITC-65128 RRID: AB_2883824
PE Mouse IgG1, κ Isotype Control	BD Biosciences	Cat# 555749; RRID: AB_396091
Anti-mouse IgG (H + L), F(ab')2 Fragment (Alexa Fluor® 488 Conjugate) #4408	Cell Signaling Technology	Cat# 4408; RRID: AB_10694704
Anti-rabbit IgG (H + L), F(ab')2 Fragment (Alexa Fluor® 488 Conjugate) #4412	Cell Signaling Technology	Cat# 4412; RRID: AB_1904025
Alexa Fluor® 594 goat anti-mouse IgG(H + L)	Invitrogen	Cat# A11005; RRID: AB_2534073
Alexa Fluor® 594 goat anti-rabbit IgG(H + L)	Invitrogen	Cat# A11012; RRID: AB_2534079
Bacterial and virus strains
pLV[Exp]-EGFP:T2A:Puro-Null	VectorBuilder	Cat#LVS-VB160420-1011mqh
Biological samples
Human liver tissues	Beijing Friendship Hospital, Capital Medical University	N/A
Chemicals, peptides, and recombinant proteins
Matrigel	Corning	Cat# 356234
Recombinant Human/Murine/Rat Activin A	PeproTech	Cat# 120-14E
Retinoic acid	Sigma-Aldrich	Cat# R2625
B-27 Supplement	Gibco	Cat# 12587-010
N-2 Supplement	Gibco	Cat# 17502-048
Recombinant Human HGF	PeproTech	Cat# 100-39
Dexamethasone	Sigma-Aldrich	Cat# D4902
Recombinant Human Oncostatin M	PeproTech	Cat# 300-10T
Recombinant Human R-Spondin-1	PeproTech	Cat# 120-38
rhEGF	R&D Systems	Cat# 236-EG
N-acetylcysteine	Sigma-Aldrich	Cat# A7250
Gastrin	Sigma-Aldrich	Cat# T6515
FGF10	PeproTech	Cat# 100-26-25
Nicotinamide	Sigma-Aldrich	Cat# N3376
Forskolin	Tocris	Cat# 1099
LDL-DyLightTM 550	Cayman Chemical	Cat# 10011229
Cardiogreen	Sigma-Aldrich	Cat# I2633
Rhodamine 123	Sigma-Aldrich	Cat# R8004
Propidium iodide	Sigma-Aldrich	Cat# P4170
Schiff Reagent	Phygene	Cat# PH0643
Picrosirius red Staining Solution	Phygene	Cat# PH1099
Alcohol	Fisher Scientific	Cat# A995-4
HCS LipidTOX™ neutral lipid stain	Invitrogen	Cat# H34475
BODIPY 558/568 C12	Invitrogen	Cat#D3835
Advanced DMEM/F12	Invitrogen	Cat# 12634010
Critical commercial assays
P450-Glo CYP3A4 Assay with Luciferin-IPA	Promega	V9001
Human Albumin ELISA Kit	Alpha Diagnostic	Cat# 1190
Urea Assay Kit	Nanjing Jiancheng Bioengineering Institute	Cat# C013-2-1
RNeasy Plus Mini Kit	QIAGEN GmbH	Cat# 74134
DCFDA/H2DCFDA-Cellular Reactive Oxygen Species Detection Assay Kit	Abcam	Cat# ab113851
Human Interleukin 6 (IL-6) ELISA KIT	CUSABIO	Cat# CSB-E04638h
High-Capacity RNA-to-cDNA Kit	Fisher Scientific	Cat# 4387406
HCM Hepatocyte Culture Medium BulletKit	Lonza	Cat# CC-3198
Deposited data
Raw and analyzed data	This paper	https://doi.org/10.6084/m9.figshare.25728393.v1
Experimental models: Cell lines
Human adipose stem cells	Li et al.13	N/A
Experimental models: Organisms/strains
Athymic nude BALB/c male mice	Charles River	N/A
Oligonucleotides
Quantitative RT-PCR primers	This paper	Table S2
Software and algorithms
Primer Premier 5	Premier Biosoft	http://www.premierbiosoft.com/primerdesign/
ImageJ	NIH	www.ImageJ.nih.gov/ij/
EndNote		https://endnote.com/?language=en
Graphpad Prism	GraphPad Software	Version 7.0
CaseViewer		https://www.3dhistech.com/solutions/caseviewer/

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Haiyan Zhang (culture@ccmu.edu.cn).

Materials availability

This study did not generate any new unique reagents.

Data and code availability

• •All supporting raw and processed data along with the associated analysis scripts are publicly available on Mendeley and Figshare. The respective DOIs are https://doi.org/10.17632/zmnt867pc2.1 and https://doi.org/10.6084/m9.figshare.25728393.v1.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Human liver tissues and animals

Human liver samples were obtained with informed patient consent and under the approval of the Ethics Committee of Capital Medical University (Beijing, China). Age and gender of human patients are provided in the Table S1. Athymic nude BALB/c male mice aged 6–8 weeks were obtained and housed in the Laboratory Animal Center of Capital Medical University. To avoid changes in hormone levels, male mice were used in our study. The same reasoning was also extended to patients. All animal studies were conducted in accordance with the Animal Care and the Ethics Committee of Capital Medical University.

hASCs culture

hASCs were established by our laboratory. hASCs were cultured and differentiated into HLCs and hEPCs as described in our previous report.¹³

Method details

Generation of hASC-derived hepatocyte organoids (hAHOs)

hASCs were cultured and differentiated into HLCs as described in our previous report. HLCs were harvested and trypsinized by 0.05% Trypsin-EDTA (Invitrogen, Grand Island, NY) for 3–4 min at 37°C. The dissociated cells were washed with PBS and seeded into CellCarrier-96 Spheroid ULA/CS (PerkinElmer, Waltham, MA, USA) at ratios of 2500 cells per well with 100uL expansion medium (EM). EM consists of Advanced DMEM/F-12 (Invitrogen) plus 20 ng/mL R-SPO-1 conditioned medium (Peprotech), 2% B-27 (Gibco), 1% N-2(Gibco), 50 ng/mL EGF (Peprotech), 1.25 mM N-acetylcysteine (Sigma), 10 nM gastrin (Sigma), 25 ng/mL HGF (Peprotech), 100 ng/mL FGF10 (Peprotech), 10 mM Nicotinamide (Sigma), and 10 μM Forskolin (TOCRIS, Minneapolis, Minnesota, United States). Then they were centrifuged at 1000 rpm for 5 min and incubated at 37°C with 5% CO₂ for 3 days. Spheroids appeared on the plate at day 3 of culture. Then spheroids were mixed with 50uL Matrigel and 50uL EM per well. After Matrigel was solidified, another 100uL EM was added. Cultures were maintained at 37°C in 5% CO₂ with 95% air and the medium was refreshed every 3 days for 10 days. To analyze the properties of hAHOs, organoids were isolated from Matrigel by scratching and pipetting at Day0, Day5 and Day 10.

Generation of hASC-derived liver organoids (hALOs)

hASCs were cultured and differentiated into definitive endoderm progenitor cells (hEPCs) as described in our previous report.^13,18 hEPCs were harvested and trypsinized by 0.05% Trypsin-EDTA (Invitrogen) for 3–4 min at 37°C. The dissociated cells were washed with PBS and seeded onto CellCarrier-96 Spheroid ULA/CS (PerkinElmer) at ratios of 2500 cells per well with 100uL medium containing Advanced DMEM/F-12 (Invitrogen) plus 2 μM retinoic acid (RA; Sigma-Aldrich), 2% B-27 (Gibco), 1% N-2 (Gibco). Then they were centrifuged at 1000 rpm for 5 min and incubated at 37°C with 5% CO₂ for 5 days. Spheroids appeared on the plate at day 2 or 3 of culture. After 5 days, another medium was added. The medium consists of HCM (Lonza) plus 20 ng/mL HGF (Peprotech), 10 ng/mL OSM (Peprotech) and 1uM dexamethasone (Dex; Sigma-Aldrich). Cultures were maintained at 37°C in 5% CO2 with 95% air and the medium was refreshed every 3 days for another 5 days. The hALOs were generated. To analyze hALOs (Day0, Day5, and Day 10), organoids were isolated by scratching and pipetting.

Gene expression analysis

Quantitative RT-PCR was performed as previously described.³³ The relative expression of each gene was normalized against 18S rRNA. The data are presented as the means ± SEM.

For immunofluorescence analysis, the cells or organoids were fixed with 4% paraformaldehyde for 10 min at room temperature, followed by permeabilization with 0.3% Triton X-100 in PBS for 5 min. The cells were rinsed and blocked with 10% goat serum (Zsgb-Bio) for 60 min at room temperature. The cells or organoids were then incubated with the primary antibodies at 4°C overnight. Following three 5-min washes in PBS with gentle agitation, an Alexa Fluor-conjugated secondary antibody (Invitrogen) at 1:500 was added, and the samples were incubated for 1 h at 37°C. The nuclei were counter-stained with 4′, 6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich).³²

For histological preparation, parts of the liver were fixed with 10% formalin for 24 h, dehydrated with graded alcohols, and embedded in paraffin. The embedded tissues were then consecutively sliced into 5 mm thick sections and routinely stained with H&E. The stained areas visualized under a Leica light microscope (Leica, Wetzlar, Germany). The stained areas were analyzed using ImageJ software.¹⁶

Structure analysis

Structure of hAHOs and hALOs was analyzed using the transmission electron microscopy (TEM) as previously described.¹³ The samples were examined using an HT7700 transmission electron microscope (Hitachi).

Hepatic function assays

The ALB content of the culture supernatants was quantified using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Alpha Diagnostic Intl) according to the manufacturer’s protocol.¹⁷

For CYP3A4 enzyme activities were assessed by P450-Glo assay (V9001, Promega), according to the manufacturer’s instructions.³³

To detect urea synthesis, the cultures were incubated with 2 mM ammonium chloride (Sigma) for 48 h, the medium was collected and analyzed using the Urea assay Kit (Nanjing Jiancheng Bioengineering Institute) according the manufacturer’s instructions.

For analyzed the uptake and release of indocyanine green (ICG), the cultures were incubated with 1 mg/mL ICG (Sigma-Aldrich) for 30 min at 37°C, the medium containing ICG was discarded and the cells washed with PBS. The uptake of ICG was examined under the microscopy. Organoids were then returned to the culture medium and incubated for a further 8 h to determine the release of cellular ICG. Following completion of each experiment, images were taken using an inverted fluorescence microscope (Olympus).⁷

For analyzed the Rhodamine 123 transport, the organoids were incubated with 1ug/mL rhodamine 123 (Sigma-Aldrich) for 5 min at 37°C, and then washed 3 times with PBS. Following completion of each experiment, images were taken using an inverted fluorescence microscope (Olympus).

For low-density lipoprotein (LDL) uptake cell-based assay, the organoids were incubated LDL-DyLightTM 550 (Cayman chemical, Ann Arbor, Michigan, USA) working solution in serum-free medium. Incubate the cells at 37°C with 5% CO₂ for an additional 24 h. At the end of LDL uptake incubation, the culture medium was replaced with fresh culture medium. The degree of LDL uptake was examined under a Leica TCS SP8 confocal microscope (Leica) with filters capable of measuring excitation and emission wavelengths 540 and 570 nm, respectively.¹⁵

Intracellular glycogen was analyzed by Periodic Acid-Schiff (PAS) Staining as previously described.³⁴ The PAS-positive cells and whole cells were examined under a Leica light microscope (Leica).

Determination and analysis of lipid intensity

Lipogenesis in cells were labeled with 10μM BODIPY 558/568 C₁₂ (Red C12, D3835, Invitrogen) for 12 h at 37°C. Lipid droplets in cells were staining with HCS LipidTOX neutral lipid stain (H34475, Invitrogen) diluted 1:1000 for at least 30 min at room temperature and examined using a Leica TCS SP8 STED confocal laser scanning microscope (Leica) as previously described.³² Quantification of the lipid droplets number per cell was performed using ImageJ software (National Institutes of Health). The data are presented as the means ± SEM.³²

The lipid droplets fluorescence intensity in organoids was measured with a High-performance multimode detection (EnSpire).

Sirius red staining

Organoids were fixed with 4% PFA, dehydrated in ethanol and xylene, and embedded in paraffin. The 5μm sections were deparaffinized, then stained with Sirius red (LEAGENE) for 60 min at room temperature, washed in the running water. They were dehydrated with two changes of 100% ethanol, and cleared in xylene. The stained areas visualized under a Leica light microscope (Leica, Wetzlar). The stained areas were analyzed using ImageJ software.⁹

Cell viability assay

The organoids were incubated for 30 min with Propidium iodide (PI; Sigma) at 37°C. The experiment was carried out according to the manufacturer’s instructions and was analyzed with a Leica TCS SP8 STED confocal laser scanning microscope (Leica).

Inflammatory factor (IL-6) assay

The IL-6 content of the culture supernatants was quantified using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (CUSABIO) according to the manufacturer’s protocol.

ROS assay

The organoids were incubated for 45 min with DCFDA/H2DCFDA (Abcam) at 37°C. The experiment was carried out according to the manufacturer’s instructions and was analyzed with a Leica TCS SP8 STED confocal laser scanning microscope (Leica).

Flow cytometry

For flow cytometric detection of surface antigens, hALOs (3×10⁵ cells) were digested into individual cells using enzymes and resuspended in stain buffer (FBS) (BD Biosciences, San Jose, CA) containing saturating concentrations (1:100 dilution) of the following conjugated mouse monoclonal antibodies against human antigens (BD Biosciences) on ice for 30 min in the dark: CD166-PE, CD68-FITC and CD26-BV421. FITC labeled mouse IgG2b Isotype control, PE labeled mouse IgG 1, κ Isotype control and BV421 labeled mouse IgG1, κ Isotype control were also included. The cell suspensions were washed twice and resuspended in FBS for flow cytometer (BD Accuri C6; BD Biosciences) using FLOWJOTM software (TreeStar, Inc., Ashland, OR). The antibodies used are listed in supplementary materials Table.

Acute liver injury induction in mice and organoids transplantation

Athymic nude BALB/c male mice aged 6–8 weeks received care according to the Capital Medical University guidelines. All protocols were approved by the Committee for Animal Care. The mice were injected with a single intraperitoneal dose of CCl₄ solution in olive oil (5.0 mL/kg body weight as 1%; Sigma-Aldrich) to induce acute liver injury.¹⁷ Vehicle (olive oil) injected mice (n = 3) were used as controls. At 24 h after CCl₄ treatment, organoids that contained 3 × 10⁴ cells were implanted at the edge of the superior right lobe of the liver with a 0.5 cm long incision under isoflurane anesthesia. Matrigel without cells were used as controls. The mice were sacrificed at day 14 post implantation. In each group, there are three to six mice for transplantation. Histological analysis of liver tissues was conducted by serial tissue sectioning and staining with hematoxylin & eosin (H&E). ALB, HNF4α and CK19 expression were examined by immunofluorescence.

For lentiviral-stable Transfection of hASCs, the cells at 80%–90% confluence was cultured with concentrated lentiviral vectors harboring pLV (Exp)-EGFP: T2A: Puro-Null (Cyagen Biosciences) with 5 mg/mL polybrene (Solarbio) for 24 h. Then, the cells were seeded into a new plate for selection using 1.0 mg/mL puromycin (Solarbio). After 7 days of screening, the medium was replaced with complete medium without puromycin, and cultivation was continued. The transduction results were evaluated using an inverted fluorescence microscope (Olympus).¹⁶

The fluorescence Imaging of transplantation in vivo was analyzed with an IVIS Spectrum In Vivo Imaging System (PerkinElmer) according to the manufacturer’s instructions. The excitation/emission wavelengths for measuring transplantation in vivo are 480/500 nm.

The ketoprofen metabolism activity of transplantation was analyzed as described previously.¹⁷

Quantification and statistical analysis

At least three independent determinations of each parameter were compared among the treatment groups by unpaired two-tailed Student’s t test using Graphpad prism 8 software. Data presented as the mean ± SEM. The differences were considered significant if p < 0.05.

Published: May 14, 2024