Building an NAFLD/NASH High-Throughput Amenable Human Liver Tissue Model

NAFLD/NASH

Nonalcoholic fatty liver disease (NAFLD) and its aggressive form non-alcoholic steatohepatitis (NASH) occurs in the liver often in association with obesity, diabetes, metabolic syndrome [1, 2]. With the increased prevalence of these diseases in developed countries, NASH is the most common cause of most chronic liver disease in these areas. While a significant number of candidate pharmaceuticals are entering clinical development aimed at treating NAFLD/NASH, these pharmaceuticals were developed using rodent models and disease pathways/drug treatment has been shown to differ between human and rodent species. As such, more accurate models that recapitulate the human disease progression and pathways are needed [2-4].


NAFLD first manifests as fat accumulation within the hepatocytes of liver tissue, termed steatosis. This phase of liver disease is generally thought to be clinically harmless to the patient. However, many patients will experience progression of NAFLD to a NASH phenotype without dietary/lifestyle changes. NASH also involves accumulation of fats/lipids within the tissue but is accompanied by inflammation and fibrosis of the liver tissue microenvironment. Left untreated, this NASH phenotype typically progresses to liver cirrhosis and liver cancer [5].



MVP™-Liver: Advanced 3D Human Liver Tissue Model

In previous work, we demonstrated the fabrication and characterization of a 3D thick (~1 cm) human liver tissue model compatible with standard multi-well plate formats that is fully automatable [6]. The high cell-dense liver tissue contains primary hepatocytes; non-parenchymal cells; and human, adipose-derived, microvessel fragments (Angiomics® haMVs) as a vascular source. The presence of the haMVs, which are comprised of multiple cell types normally residing in and on the microvessel wall, improves liver tissue function in the model over 14 days of culture, including increased drug transporter expression, P450 responsiveness, and increased non-parenchymal cell presence [6]. Importantly, the tissue is structured such that, despite the relative thickness, all cells are within 300 mm of a surface available for diffusion, regardless of the plate format used in the application.


Here we use the same tissue modeling approach to fabricate this high cell-dense, 3D human liver tissue in a high throughput compatible format (standard 384 well plate) and explored its utility as a model for NAFLD/NASH. In this application, a single column of liver tissue was formed around a central space producing a cylinder approximately 0.6 cm tall within a single well (Figure 1A). The lumen provides good surface area for nutrient and gas exchange between the tissue space and the medium. Over the 14 days of culture examined, the tissue height and structure, including the central lumen persisted. In this application, the fabricated tissues consisted of a mix of 60 million primary hepatocytes (PH) per ml, 6 million non-parenchymal cells (NPCs) per ml (a 10:1 PH to NPC ratio), and 200K haMVs/ml.


To begin exploring the utility of the model for NAFLD/NASH, liver tissues were exposed to a high fat medium (HFM) comprised of a base medium supplemented with glucose (25mM), palmitic acid (45μM), and sodium oleate (65μM). Additionally, dynamic tissue responses were explored by adding TGFβ1 (10ng/ml) alone or in combination with an ALK5 inhibitor (ALK5i; 5μM) to the high fat media supplements, where media was exchanged daily throughout the culture period. Findings were then compared to the appropriate vehicle control groups. After the experiment, tissues were fixed, end bloc stained with BODIPY, and imaged to measure lipid uptake, then processed for histology. Histologic sections and media supernatants were used to measure matrix deposition, urea, and inflammatory cytokine expression.


Here, we observed increased lipid accumulation in hepatocytes and increased human collagen type I deposition of the liver tissues cultured in high fat medium, the two hallmarks of the disease. Additionally, we observed dynamic response of the NASH tissue to TGFβ1, which was abrogated by ALK5i, a TGFβ1 signaling inhibitor.



Lipid Accumulation

To induce a NAFLD/NASH phenotype, we supplemented the hepatocyte base media with glucose, sodium oleate, and palmitic acid to simulate dietary conditions most often associated with NAFLD/NASH. Furthermore, it has been shown by others to be an effective media to induce the relevant phenotype in 2D and 3D monocultures [7]. In both the high fat media (HFM) and vehicle control groups, urea production in the liver tissues dropped after the first day, likely reflecting cell dysfunction secondary to primary cell thawing and some energetic constraints (Figure 1B) [6]. However, no significant difference was observed in urea production between the treatment groups (Figure 1B). While urea cycle dysfunction can occur in NASH, it is not a typical hallmark of the disease and is often prevalent in only later manifestations of the disease [8]. Similarly, bile acid production was not significantly different between treatment groups (Figure 1C). While increased plasma bile acid levels are associated with NAFLD, it appears to occur in patients with insulin resistance co-morbidity [9], a situation likely not present in our liver tissues. To visualize accumulated lipids, tissues were en bloc stained with BODIPY and imaged via confocal microscopy. As expected, hepatocyte lipid accumulation was significantly increased in HFM-treated groups (Figure 1D), compared to the control group. This is an exciting result as it demonstrates the uptake of fatty acids by the hepatocytes, which is a clinical hallmark of NAFLD.


Figure 1: A) Side view of 3D human liver tissue in a single well of a 384 well plate. Urea (B) or bile acid (C) production in culture supernatants collected from control or high fat medium (HFM) treated human liver tissues. D) Assessment of lipid accumulation in control- and HFM-treated liver tissues via confocal imaging and corresponding measurements of BODIPY-positive staining.
Figure 1: A) Side view of 3D human liver tissue in a single well of a 384 well plate. Urea (B) or bile acid (C) production in culture supernatants collected from control or high fat medium (HFM) treated human liver tissues. D) Assessment of lipid accumulation in control- and HFM-treated liver tissues via confocal imaging and corresponding measurements of BODIPY-positive staining.

Tissue Fibrosis

Progression from an NAFLD phenotype towards a NASH phenotype is marked by inflammation and fibrosis within the tissue. Immunostaining for human collagen type I, an extracellular matrix (ECM) protein commonly associated with liver fibrosis, in histology sections revealed a significant increase in collagen I deposition in liver tissues of the HFM group as compared to control tissues (Figure 2A). In contrast, the deposition of collagen type III, another ECM protein associated with fibrosis, was not found to be different (Figure 2B). Interestingly, while the numbers of a-actin positive and CD45-positive cells tended to be higher in the HFM groups, a concomitant marker with fibrosis, no statistically significant difference was found (Figures 2C & 2D). Additionally, neither TNFα nor IL-6, the two major inflammatory cytokines associated with NASH, were detected in tissue supernatants via ELISA (data not shown). Therefore, the presence of increased collagen I deposition in the absence of inflammatory cytokine production suggests that fibrosis is initiated in the liver model with HFM treatment in the absence of an inflamed tissue environment.


Figure 2: Characterization of in vitro, fabricated 3D human liver tissues following a NAFLD challenge. Histology sections of tissues were immunostained for A) Collagen I, B) Collagen III, C) SM actin, and D) CD45 positive following exposure to either the vehicle control or a high fat medium (HFM). In all cases, measurements from stained sections were normalized to the # of hepatocytes in each section.
Figure 2: Characterization of in vitro, fabricated 3D human liver tissues following a NAFLD challenge. Histology sections of tissues were immunostained for A) Collagen I, B) Collagen III, C) SM actin, and D) CD45 positive following exposure to either the vehicle control or a high fat medium (HFM). In all cases, measurements from stained sections were normalized to the # of hepatocytes in each section.

NASH promotion: TGFb1 treatment

Given that tissue inflammation is associated with NASH, we included TGFβ1 alone or in combination with ALK5i in the high fat media supplements as treatment groups to assess the dynamic response of the tissue to TGFβ1, an inflammatory mediator. Neither the presence of TGFβ1 alone nor in combination with ALK5i had a significant effect on urea or bile acid production (Figures 3A & 3B). Interestingly, TGFβ1 promoted increased lipid accumulation by hepatocytes as compared to the vehicle control group, an increase of which was abrogated by ALK5i (Figures 3C - 3F). Interestingly, given the generalized role of TGFβ1 in tissue inflammation, there was no evidence of additional fibrosis (collagen I or III deposition), inflammatory cytokine expression (IL-1 and IL-6), or changes in NPC cell populations in the presence of TGFβ1 with or without ALK5i (Figures 4A - 4H) above that induced by HFM + vehicle, suggesting that in our model, treatment with TGFβ1 alone is insufficient to induce a NASH phenotype. Additionally, we observed vehicle effects when comparing values for the HFM + vehicle (EtOH + citric acid) in this set of TGFβ1-related experiments to the HFM + vehicle (EtOH) in the previous set of experiments, suggesting that the EtOH + citric acid combination had a suppressive effect on lipid accumulation (see Figures 1 vs 3).


Figure 3: Liver tissue responses to TGFβ1 w/ and w/o ALK5i in high fat medium. Urea (A) and bile acids (B) produced in supernatants from tissues in high hat medium + vehicle (H+VC), with TGFβ1 (H+T), or with TGFβ1 and ALK5i (H+T+A). C-F) Assessment of lipid accumulation in liver tissues from the 3 different treatment groups via confocal imaging and corresponding measurements of BODIPY-positive staining.
Figure 3: Liver tissue responses to TGFβ1 w/ and w/o ALK5i in high fat medium. Urea (A) and bile acids (B) produced in supernatants from tissues in high hat medium + vehicle (H+VC), with TGFβ1 (H+T), or with TGFβ1 and ALK5i (H+T+A). C-F) Assessment of lipid accumulation in liver tissues from the 3 different treatment groups via confocal imaging and corresponding measurements of BODIPY-positive staining.

Figure 4: Characterization of in vitro, fabricated 3D human liver tissues following a NASH challenge. Histology sections of tissues were immunostained for A) Collagen I, B) Collagen III, C) SM actin, and D) CD45 following exposure to the HFM/vehicle control (H + VC), HFM + TGF-β1 (H + T), or HFM + TGF-β1 + ALK5i (H + T + A). In all cases, measurements from stained sections were normalized to the # of hepatocytes in each section.
Figure 4: Characterization of in vitro, fabricated 3D human liver tissues following a NASH challenge. Histology sections of tissues were immunostained for A) Collagen I, B) Collagen III, C) SM actin, and D) CD45 following exposure to the HFM/vehicle control (H + VC), HFM + TGF-β1 (H + T), or HFM + TGF-β1 + ALK5i (H + T + A). In all cases, measurements from stained sections were normalized to the # of hepatocytes in each section.

Summary and Discussion

In this pilot study, we demonstrate the reconfiguration of our 3D thick, human liver tissue model into a standard 384 well format towards improving assay throughput. Indeed, we were able to fabricate, culture and assay the high cell-dense liver tissues (60 million hepatocytes/ml of starting tissue) over a 2-week period similar to what we have reported earlier for larger sized tissue formats [6]. While not explored here, we have previously demonstrated that this human vascularized liver tissue model is amenable to laboratory automation using standard well-plate formats [6], making this 384 well plate format an enticing option for high throughput screening. Furthermore, we demonstrate that the small tissue volumes, which make the model more economical given the high cell density, continue to provide sufficient biology for assay endpoints.


Related to this, we demonstrate clear differences in liver tissue responses towards NAFLD- and NASH-like in vitro challenges related to lipid accumulation and fibrosis. Other indicators such as non-parenchymal cell activation and inflammatory cytokine production did not show a significant difference, although there were trends in these assay endpoints consistent with an NAFLD/NASH phenotype, particularly with the TGFβ1 treatment consistent with a possible NAFLD/NASH-like response.


3D liver cell co-cultures (hepatocytes plus other liver tissue cell types) have been repeatably shown to be more functional and representative of the in vivo microenvironment, as signaling between NPC cells, vascular cells, and hepatocytes is important in healthy and pathological dynamics. Interestingly, in contrast to 2D mono- and co-cultures and 3D monocultures which exhibit simple markers of NASH due to elevated oleic and palmitic fatty acid levels [13], 3D liver co-culture models required an additional insult, such as with cyclosporine [10], methotrexate [11], or fluid flow-related cues [12] to induce the NASH phenotype. As others have suggested, a more complex cellular environment in a tissue model likely plays a modulating role in tissue disease progression, requiring a more aggressive and/or chronic insult to progress towards disease in vitro. This may explain why the inclusion of TGFβ1 in our studies, intended to provide a more natural co-stimulus with the HFM, did not push our model fully towards a frank NASH phenotype. Regardless, the early indicators of NAFLD/NASH in our 3D human liver tissue model, which has a more dynamic cellular environment configured in a throughput-friendly format, may prove useful in gaining more insight into the relevant disease processes, which existing in vitro models have not been able to fully address.


 

References

  1. Kořínková, L., et al., Pathophysiology of NAFLD and NASH in Experimental Models: The Role of Food Intake Regulating Peptides. Front Endocrinol (Lausanne), 2020. 11: p. 597583.

  2. Ramos, M.J., et al., In vitro models for non-alcoholic fatty liver disease: Emerging platforms and their applications. iScience, 2022. 25(1): p. 103549.

  3. Ströbel, S., et al., A 3D primary human cell-based in vitro model of non-alcoholic steatohepatitis for efficacy testing of clinical drug candidates. Sci Rep, 2021. 11(1): p. 22765.

  4. Boeckmans, J., et al., Human-based systems: Mechanistic NASH modelling just around the corner? Pharmacol Res, 2018. 134: p. 257-267.

  5. Fernando, D.H., et al., Development and Progression of Non-Alcoholic Fatty Liver Disease: The Role of Advanced Glycation End Products. Int J Mol Sci, 2019. 20(20).

  6. Moss, S.M., et al., Point-of-use, automated fabrication of a 3D human liver model supplemented with human adipose microvessels. SLAS Discov, 2022.

  7. Müller, F.A. and S.J. Sturla, Human in vitro models of nonalcoholic fatty liver disease. Current Opinion in Toxicology, 2019. 16: p. 9-16.

  8. De Chiara, F., et al., Urea cycle dysregulation in non-alcoholic fatty liver disease. J Hepatol, 2018. 69(4): p. 905-915.

  9. Grzych, G., et al., NASH-related increases in plasma bile acid levels depend on insulin resistance. JHEP Rep, 2021. 3(2): p. 100222.

  10. Bell, C.C., et al., Characterization of primary human hepatocyte spheroids as a model system for drug-induced liver injury, liver function and disease. Sci Rep, 2016. 6: p. 25187.

  11. Leite, S.B., et al., Novel human hepatic organoid model enables testing of drug-induced liver fibrosis in vitro. Biomaterials, 2016. 78: p. 1-10.

  12. Feaver, R.E., et al., Development of an in vitro human liver system for interrogating nonalcoholic steatohepatitis. JCI Insight, 2016. 1(20): p. e90954.

  13. Green, C.J., et al., Studying non-alcoholic fatty liver disease: the ins and outs of in vivo, ex vivo and in vitro human models. Horm Mol Biol Clin Investig, 2018. 41(1).

  14. Strobel, H.A., T. Gerton, and J.B. Hoying, Vascularized adipocyte organoid model using isolated human microvessel fragments. Biofabrication, 2021.

 

Angiomics Human Adipose Microvessels by Advanced Solutions are for RESEARCH USE ONLY and not for use in humans under any circumstances. Advanced Solutions Life Sciences, LLC. and Advanced Solutions, Inc. are not responsible or liable for how this product is used. The RESEARCH USE ONLY limitation supersedes any written, oral, or implied understanding between the parties. View Human Adipose Microvessels Safety Data here.