Non-alcoholic steatohepatitis (NASH) in vitro

We use primary human hepatocytes, Kupffer cells and stellate cells with the ratio of 10:1:1.We source the primary cells from vendors including LifeNetHealth and Lonza, subject each lot to stringent validation testing to ensure assay performance in 3D tri-culture. More information about our validation process can be found in the following blog. To fast-track your route to assay adoption, we have developed a NASH-in-a-box kit for use with our PhysioMimix^® OOC Systems. The kit contains everything that you need to recreate our human NASH in vitro assay in your laboratory.

Spheroids are a very good system to study fibrosis, better than a 2D immortalised line model, however, spheroids models are still limited in terms of the number of end point measurements that can be detected, and their longevity.

The 3D tissue is totally perfused throughout an experiment, providing nutrients and O₂ to all cells in the 3D tissue, thus enabling the culture of stable 3D tissue for longer periods of time (14 days for our standard NASH-in-a-box assay). Internally, we have run the assay for longer and offer flexibility in terms of assay duration through our NASH Contract Research Services. Contact us for more info.

Compared to static spheroid cultures, our human NASH in vitro model enables many more endpoints to be measured per well, due to the larger scale microtissue that we generate (over 480,000 cells per well) and large volume of media (~1 mL) from which to sample. Scaffolds containing the 3D liver microtissues can be cut in half; one half can be used for microscopy while the other can be used for -omics analysis e.g. RNAseq for transcriptomics. The large volume of media available for sampling allows for the analysis of various soluble biomarkers that include cell health (LDH, albumin, ALT/ AST), cell function (CYP activity), inflammation (IL-6, TNF-α, etc.) and fibrosis (TIMP1, Fibronectin)

Read the following publication for more info Kostrzewski et al, 2019.

Additionally, use of flow perfusion enables healthy (and diseased) liver tissue to remain functionally and metabolically active for over 4 weeks, much longer than is possible when culturing under static conditions. A co-publication with the FDA demonstrates the assay performance benefits of culturing using flow perfusion versus gold standard spheroid and liver sandwich approaches (Rubiano et al, 2020).

We have used and adapted various genetic modulation methods in our system. We have looked at key single nucleotide polymorphisms (SNPs), for example, knocked-out patatin-like phospholipase domain containing 3 (PNPLA3) in Human Stellate cells (HSCs). In this study, we confirmed that mutated SNPs on PNPLA3 gene in HSC enhanced the overall NASH disease state (Kostrzewski et al, 2019). We have also successfully used small interfering RNA (siRNA) and antisense oligonucleotides (ASOs), both of which were shown to be effective in our model.

Fibrosis is key to the disease progression into severe NASH, cirrhosis and, if left untreated, liver cancer. Developing in vitro models that mimic the more advanced clinical phenotypes of the disease is therefore crucial, and it is something that we have recently started to investigate. For example, by adding cues such as tumor growth factor-beta (TGF-β), our 3D in vitro model can already progress to an advanced and more severe NASH stage with a quantifiable fibrosis phenotype (Kostrzewski et al, 2021), but we think we can go even further to reach a critical point where we generate, or trigger, hepatocellular cancer.

Have you tried iPS-derived hepatocytes instead of PHH in your human NASH in vitro model?

Our primary focus has always been to use primary cells to be as close to the human liver’s physiology as possible. However, there is a growing interest in developing iPS-derived hepatocytes models, especially for researchers who cannot easily access primary cells. We have successfully developed a monoculture liver model with iPS-derived hepatocytes, however this model has yet to be used to replicate a disease state or be used in a co-culture with other nonparenchymal cells (NPCs)

We typically examine expression of various disease markers for steatosis (Nile Red), inflammation (IL-6, IL-8, TNF-α), and fibrosis (TIMP-1, Pro-collagen, Fibronectin, smooth muscle actin). Also, we evaluate cell function and health by measuring Urea, Albumin, LDH and Aspartate Transferase (AST)/Alanine amino transferase (ALT)

For deeper insights, periodic media sampling over time enables secreted biomarker assessments and post assay, RNA isolation for transcriptomic analysis.

As a rule, reliability is good between different batches. One of the advantages of CN Bio’s system is the ability to run a lot of experiments at the same time. We have 12 wells available on a single Multi-chip Liver-12 plate and we can run up to 6 plates at the same time. Generally, we get great reproducibility using the same donor within an experiment. When you start using different donors you do start to see some variability, but outcomes tend to trend the same. You would, however, expect to see donor to donor variations in the same way as variations are observed patient to patient. All cells used within the model are subjected to significant validation checks prior to experiments to ensure they give the best quality data. The following blog gives more information about our thorough lot validation process.

Please refer to the work of Vacca et al., 2020 who also explored TGFβ signalling and BMP signalling for modeling NASH. For BMP, their investigations confirmed that SMAD phosphorylation occurs. After 5 hours the target genes of SMAD are upregulated as you would expect. Vacca et al we were able to inhibit both the branches of the family using chemical inhibitors thus affecting the transduction of the signal. Their results concluded that the assay represents an effective in vitro tool for modeling NASH in preclinical science.

Additionally, our co-publication with AstraZeneca (Kostrzewski et al, 2021) explores transcriptional profile of NASH liver MPS model demonstrating that it aligns with profile from human clinical samples and the model’s potential for elucidating mechanism of action data. The scale of the NASH in vitro model is ideally suited to the analysis of multiple factors, both from the media and from the cells, allowing for transcriptome and proteomic analysis from every replicate.

This topic is covered in more detail in the following blog.

As part of our R&D work, we have run NASH cultures for up to one month, providing plenty of time to include fat loading, genetic manipulation, and compound dosing. However, our NASH-in-a-box kit, which contains everything required to recreate our industry proven assay in your laboratory using a PhysioMimix^® OOC System, is validated for use up to 14 days.

Our model has been designed to be highly adaptable allowing for the culture of a range of primary cells. Our standard model, as utilized in our NASH-in-a-box kit, uses a tri-culture of hepatocytes, Kupffer cells and stellate cells with a modified media to support all three cell types. We can run simpler versions without non-parenchymal cells, however, there is the potential to develop more complex models too.

One PhysioMimix^® OOC Controller unit can operate up to six Multi-chip Liver-12 plates at once with 12 wells in each plate. This results in a total throughput of 72 individual samples/run. As standard we run experiments in triplicate providing the ability to investigate 24 independent conditions. More complex studies can be scaled up using more than one PhysioMimix^® Controller unit.

When using our NASH-in-a-box kit, two plates are supplied which means up to 6 conditions in triplicate can be run per kit.

Compared “like for like” versus technologies that use microfluidic flow (pneumatics that permit organ-specific tuning) instead of gravity-driven flow, we provide a higher throughput. Increasing the throughput is part of our roadmap now that we have the PhysioMimix^® Higher Throughput (HT) System and Multi-chip Liver-48 plate, however, a commercialized solution is pending until a thorough evaluation of assay performance (in this miniaturized format) has been performed by our Science and Technology Team.

When developing in vitro models, each cell lot/donor will react differently to the environment/conditions used. Some may do really well in 2D, under static conditions (e.g.: monolayer, liver sandwich or spheroid) but perform poorly under flow in a microphysiological system (MPS), or vice-versa. To develop the best model for your research, it is therefore important to source and test a variety of donors. A recent blog outlines the process that we go through to validate tri-cultures for NASH assay use.

We source our cells from a wide range of large cell suppliers such as Lonza, and LifeNet Health and thoroughly test each cell lot/donor to identify the best donors for our disease models. To save time and effort, customers can access CN Bio validated Hepatocyte and Kupffer cell lots (i.e. those that have been internally tested and proven to generate healthy, functional 3D liver tissues) through our 3D validated cell portfolio or directly from LifeNet Health. However, our NASH-in-a-box kit offers the fastest and easiest route to assay adoption, containing everything you need to recreate our model in your laboratory using PhysioMimix OOC.

Although using the same donor each time to generate data with consistency is ideal, it is impossible to guarantee as stocks of donors will run out at some point. It is also important to observe donor-donor variability to gain a better understanding of variations in metabolism between patients. When starting a project, we tend to use the same donor for a specific set of experiments to ensure consistency and build confidence in the data, then, switch to a different donor(s), to assess donor-donor variability. Donor variability can often be observed in our model; however, we generally see a consistent trend between donors. For example, when inducing severe fibrosis (by adding cues such as TGF-β), the severity of fibrosis may vary between donors, however, an increase in fibrosis is always observed.

When comparing cues/components, we look at expression profile using transcriptomics/RNA seq and compare the data to that of NASH patients to define the overlap. Although all cues are very good and show great NASH-like phenotype, TGF-β drives the NASH phenotype a little bit further than the other components. Depending on what researchers want to do with the model, adding different cues to subtly change the profile is possible. For example, if an inflammatory response is needed, lipopolysaccharide (LPS) is a very good component to add. Which cue to add and at what dose greatly depends on the research question being asked.

This topic is covered in detail in the following blog.