In the realm of pharmaceutical research, Caco-2 cells have long been heralded as the gold standard for studying intestinal absorption and permeability.
These human epithelial colorectal adenocarcinoma cells are prized for their ability to differentiate into enterocyte-like cells, forming tight junctions that mimic the intestinal barrier.
However, as with any model system, Caco-2 cells come with their own set of limitations. This raises important questions: Do their limitations undermine their gold-standard status? Is it time to improve estimations with more physiologically relevant approaches? What new approaches are available and most importantly, do their benefits sufficiently outweigh the effort required to adopt a new approach?
The strengths of Caco-2 cells
The undeniable strengths and simplicity of Caco-2 cell models are testament to why they have remained a valuable tool in drug development for so long. So often we hear the phrase “If it isn’t broken, why fix it?” Therefore, innovations must be much more beneficial to drive change.
So, what’s the driver for change when Caco-2 cells are well known for their:
- Predictive Power: Their ability to form a monolayer with tight junctions on a Transwell® insert enables reliable prediction of drug permeability of passively diffused compounds.
- Reproducibility: The use of Caco-2 cells allows for consistent and reproducible results, which is crucial for comparative studies and regulatory submissions.
- Ease of Use: These cells are relatively easy to culture because they are derived from a human colon cancer, making them accessible to laboratories.
We believe the driver for change relates to drug metabolism. When a drug enters the body via the oral delivery route, its absorption and permeability through the gut is merely the first step in its journey. The gut is a metabolically active organ plus, before an orally administered drug has a chance to elicit its actual target effect, there’s the impact of a second organ to consider too – the liver.
Currently, we derive in vitro absorption & permeability data from Caco-2 models, then utilize in vitro liver clearance assays. The disconnect here is that this approach fails to account for the pharmacokinetics of a drug after it is absorbed and metabolized by the gut to estimate what proportion reaches the liver. Therefore, the use of isolated assays leaves a worrying gap in our knowledge when estimating a drug’s bioavailability.
Why is estimating human oral bioavailability a preclinical challenge?
Key to safe and efficacious dose setting ahead of clinical studies, and a regulatory requirement, is estimating a drug’s bioavailability.Data derived from isolated in vitro tests can be used to infer a drug’s bioavailability but only in extreme cases does the approach deliver actionable insights. Whilst these limitations can be countered to some extent by in vivo animal bioavailability studies and in silico models they too suffer limitations. In silico models are only as good as the data input, whilst the translatability of animal predictions is hindered by interspecies differences due to differences in physiology and metabolic capacity. A seminal study by Musther et al in 2014 investigated 184 compounds and found that bioavailability data from commonly used animal models weakly correlates with human data (mouse R2 = 0.25, rat =0.28, and dog = 0.37). Non-human primate (NHP) models provide an improved bioavailability correlation with humans (R2 =0.69). However, these models are not commonly used due to significant ethical considerations, stringent regulatory requirements, and high cost.
As a result, the preclinical estimation of human oral bioavailability is a complex process involving the “piecing together” of data derived from many sources. Improvements to the process via more humanized approaches represents an area of “quick win” potential.
Use of Caco-2 cells to mimic the process of first-pass metabolism in vitro
Fluidically interconnecting a gold-standard approach to measuring drug absorption/permeability with a gold-standard approach to measuring liver clearance would represent the obvious step towards improving the value of data derived from Caco-2 cell assays. At CN Bio, we explored this approach via the development of a gut-on-a-chip model by culturing a mixture of epithelial (Caco-2) and goblet (HT29-MTX) intestinal cells on the basolateral side of a Transwell® insert exposed to circulating media.
We fluidically connected the Caco-2 gut -on-a-chip with our highly characterized and validated primary human liver-on-a-chip model. The latter of which has been recognized by the US FDA (in the first publication between an Organ-on-a-chip provider and regulator) as delivering improved performance versus standard static liver culture approaches (Rubiano et al., 2021).
By fluidically linking these two models together to form a system, we combined previously isolated assays into one (using the PhysioMimix Multi-organ System & Multi-chip Dual-organ plates) to mimic the process of first pass metabolism in vitro. The advantages of this approach are that you can simulate IV and oral dosing to evaluate the individual and combined contributions of each model to derive the area under the curve, i.e., the pharmacokinetics of a drug after it is absorbed and metabolized by the gut, and estimate what proportion reaches the liver to provide the missing piece in the bioavailability estimation puzzle.
The performance of this novel Gut/Liver-on-a-chip model, also known as a Gut/Liver microphysiological system (MPS), was evaluated by Roche using the prodrug mycophenolate mofetil (Milani et al., 2022). However, in their publication Roche noted the limitations of the Caco-2 cell model when working with a drug of this type due to carboxylesterase 1 & 2 (CES1 and CES2) expression. In Caco-2 cells, expression of these enzymes is neither physiological nor human relevant. These enzymes are crucial for the hydrolysis of ester prodrugs and can affect the accurate prediction of the metabolism and absorption of ester-containing drugs.
But this is just one of many Caco-2 cell limitations that restrict the accuracy of estimations derived using the dual-organ model in our PhysioMimix Bioavailability assay.
The innovation leap: when two MPS providers are better than one
In January 2024 we announced a close strategic partnership with Altis Biosystems. This partnership enabled us to replace our perfused Caco-2 gut-on-a-chip with their primary human RepliGut®-Planar Jejunum model in our dual-organ system. The RepliGut model utilizes human intestinal stem cells to create an in vitro model that more closely mimics the human to overcome the additional limitations of Caco-2 cells including:
- Lack of Complexity: While Caco-2 cells mimic the intestinal barrier, they do not fully replicate the complexity of the human intestine. They lack the diversity of cell types found in the gut, which play significant roles in absorption and permeability.
- Mucus Secretion: Caco-2 cells do not secrete mucus. RepliGut® models include mucus-secreting goblet cells, better mimicking the in vivo environment.
- Barrier Integrity: The RepliGut model has a distinct cell expansion phase, with mostly stem and progenitor cells and a differentiation phase, mostly differentiated cells. As the cells mature, the barrier integrity increases monitored using TEER (Transepithelial Electrical Resistance). This more closely mimics the human intestine, which is highly regenerating.
- Overexpression of Efflux Transporters: Caco-2 lack key efflux transporters like P-glycoprotein, which can lead to an underestimation of drug absorption. Gene expression data from RepliGut models indicates transporters undergo differential regulation as cells mature with the Caco-2 cells looking more like immature cells.
- Metabolic enzyme expression: Caco-2 cells have limited expression of Phase 1 & 2 metabolic enzymes. This can lead to an incomplete understanding of a drug’s metabolic profile and potential bioavailability. RepliGut models express more physiologically relevant levels of metabolic enzymes, including cytochrome P450 (CYP), UGT, and CES enzymes, improving the accuracy of drug metabolism studies.
Improved ability to profile human bioavailability in vitro with a primary human RepliGut/Liver-on-a-chip versus Caco-2 cells in isolation
The performance of the fluidically linked RepliGut®-Planar Jejunum/Liver-on-a-chip (RepliGut®-Planar Jejunum/Liver MPS) was investigated using two compounds (temocapril and midazolam) who’s human ADME properties were not predicted by existing models. Midazolam is a drug that intestinal and liver clearance by the CYP3A4 enzyme which has negligible expression in Caco-2 cells. Temocapril is a pro-drug designed to be resistant to intestinal hydrolysis that is metabolized to temocaprilat by CES1. CES1 is more predominantly expressed in Caco-2 cells versus humans leading to an over estimation of drug clearance.
The results of this study demonstrated that the primary dual-organ model offers a viable alternative to circumvent the human-relevance limitations of the Caco-2 cell line for uniquely profiling human bioavailability in vitro.
More information can be found in our application note.
Conclusion
Despite their limitations, the ease of use of Caco-2 cells make them indispensable for initial screening and comparative studies. However, it is crucial to complement Caco-2 studies with other more physiologically relevant models. By doing so, you obtain a holistic understanding of the combined effects of the human gut and liver to more accurately profile oral drug bioavailability in vitro.
Generating more human-relevant data earlier in drug discovery means that observed issues can be flagged and addressed before costly preclinical in vivo studies. As the primary RepliGut/Liver-on-a-chip is entirely made up of primary human cells, there are no interspecies differences to account for, therefore, the model can be utilized to help overcome the poor correlation between animal model ADME predictions and human outcomes. Bridging the gap between in vitro assays and in vivo studies, the model enables researchers to confidently progress only the most promising drug candidates to support reductions in cost and the number of animals required. For this reason, we believe that the benefit of change significantly outweighs the benefit of maintaining the status quo and foresee that the gold standard will become redefined by the integration of dynamic models into the toolkit of ADME/DMPK scientists.
So, don’t delay tackling challenges in ADME drug development! The longer you wait the greater the risk of cascading problems down the line from miscalculated pre-clinical estimations.
Our primary human in vitro Gut/Liver model and Bioavailability assay can be accessed in two ways:
- Via our ADME Contract Research Services
- In your own laboratory using our recently launched PhysioMimix Bioavailability assay kit: Human 18 and PhysioMimix® Multi-organ System.
