Francesca
Fanizza
a,
Simone
Perottoni
a,
Lucia
Boeri
a,
Francesca
Donnaloja
a,
Francesca
Negro
a,
Francesca
Pugli
a,
Gianluigi
Forloni
b,
Carmen
Giordano‡
*a and
Diego
Albani‡
b
aDepartment of Chemistry, Materials and Chemical Engineering ‘Giulio Natta’, Politecnico di Milano, Milan, Italy. E-mail: carmen.giordano@polimi.it
bDepartment of Neuroscience, Istituto di Ricerche Farmacologiche Mario Negri IRCCS, Milan, Italy
First published on 12th February 2025
Current drug development pipelines are time-consuming and prone to a significant percentage of failure, partially due to the limited availability of advanced human preclinical models able to better replicate the in vivo complexity of our body. To contribute to an advancement in this field, we developed an in vitro multi-organ-on-a-chip system, that we named PEGASO platform, which enables the dynamic culturing of human cell-based models relevant for drug testing. The PEGASO platform is composed of five independent connected units, which are based on a previously developed millifluidic organ-on-a-chip device (MINERVA 2.0), hosting human primary cells and iPSC-derived cells recapitulating key biological features of the gut, immune system, liver, blood–brain-barrier and brain that were fluidically connected and challenged to model the physiological passage of donepezil, a drug prescribed for Alzheimer's disease. The nutrient medium flow rate of the connected units was tuned to obtain suitable oxygenation and shear stress values for the cells cultured in dynamic condition. A computational model was at first developed to simulate donepezil transport within the platform and to assess the drug amount reaching the last organ-on-a-chip. Then, we demonstrated that after 24 hours of donepezil administration, the drug was actually transported though the cell-based models of the platform which in turn were found viable and functional. Donepezil efficacy was confirmed by the decreased acetylcholinesterase activity at the brain model and by the increased expression of a donepezil-relevant multi-drug transporter (P-gp). Overall, the PEGASO platform is an innovative in vitro tool for drug screening and personalized medicine applications which holds the potential to be translated to preclinical research and improve new drug development pipelines.
One alternative lies in the development of human cell-based microphysiological systems as organ-on-a-chip (OOC). These devices can replicate in vitro the key microarchitectures of human organs and mimic human metabolism together with resembling physiopathological features. Moreover, OOCs can be linked together in a multi-OOC platform to analyze multi-organ crosstalk, drug–organ interactions and secondary drug toxicity.3,4
However, most OOC studies have largely been based on 2D culture of non-human-derived cell lines, or featuring immortalized or primary human cells, lacking 3D complexity and having just partial physiologic relevance. To contribute to the OOC field and offer a tool to address these shortcomings, we developed the MINERVA 2.0. device:5,6 a 3D-printed millifluidic OOC consisting of two hemi-chambers, apical and basal, interfaced through a porous membrane of a commercial Transwell-like insert. The peculiar design of the millifluidic MINERVA 2.0 allows the accommodation of millimetric 3D models as hydrogel-based scaffolds and the serial connection of single devices as plug-and-play units to build human OOC platforms. Here we present a multi-organ platform that we named PEGASO (PErsonalisation of induced pluripotent stem cell-based body-on-chips for in vitro testinG of therApeutic agentS against brain disOrders). It connects five MINERVA 2.0. devices under a common circulating medium, hosting primary or iPSC-derived human cells and recapitulating key features of organs such as the gut, immune system, liver, blood–brain barrier (BBB) and brain, to allow the resembling of the physiological passage of a drug approved for treating neurological diseases. Oral administration is the most preferred drug therapy route for donepezil since it has compliance for patients.7 The PEGASO platform aims at modeling the key passages of the orally-administered donepezil from the gut to the brain (Fig. 1). In the body, donepezil is first absorbed across the intestinal barrier and then the intestinal endothelium which represents the gateway to the body's systemic distribution, reaching peak plasma levels in a few hours.8 Along this path, the monocytes from the “blood circulation” have the potential to interact with the drug leading to its modification and/or the secretion of mediators which can freely diffuse back across the intestinal barrier. Donepezil is mainly excreted unchanged in the urine, but it also undergoes hepatic metabolism8 in the liver getting biotransformed before re-entering the blood circulation. The drug next reaches the BBB, a selective permeability barrier regulating fluid extravasation and drug and metabolite diffusion, finally targeting the brain parenchyma.9
![]() | ||
Fig. 1 Schematic representation of the main physical and biological features that are implemented in the PEGASO multi-OOC platform (created with Biorender®). |
We aimed at recreating the key features of this pathway by Transwell-based models of cell barriers, the immune system and a co-culture of brain parenchymatic cells (neurons and astrocytes) connected under the action of tuned fluid flow.
The nutrient medium flow rates within the OOC were adjusted to ensure optimal oxygenation and shear stress for the cultured cells. Additionally, a computational model was developed to analyze the drug transport and predict the drug amount reaching the final OOC.
After 24 hours from donepezil administration the cells were viable and able to express physiologic functionalities indicating that the multi-OOC can be used to test basic drug pharmacokinetics and biological targeting. These results present a new in vitro tool for drug screening and potentially personalized medicine which will be of benefit to the community in improving the drug development pipeline and reducing drug failure in the field of neurodegenerative disorders.
The platform is composed of five OOCs (based on MINERVA 2.0) representing the gut, immune, liver, BBB and brain, linked together through fluidic connections (Fig. 2).
The gut OOC mimicking drug absorption was linked to the immune OOC which was coupled to the liver OOC simulating metabolism that in turn was connected to the BBB OOC, which was linked to the brain OOC, the final target of the Alzheimer's disease (AD) drug donepezil.
The peculiar design of the MINERVA 2.0 device, composed of two culture chambers, enables physical separation of two cellular constructs which communicate through the OOC porous membrane.5
The first device of the platform, namely the gut OOC, contained primary human intestinal cells facing iPSC-derived endothelial cells; the immune OOC was populated with human PBMC CD14+ cells; the liver OOC hosted a collagen–poly(ethylene)glycol (COLL–PEG) hydrogel encapsulating iPSC-derived hepatocytes interconnected with iPSC-derived endothelial cells; the BBB OOC was based on iPSC-derived endothelial cells and astrocytes while the last device, the brain OOC, contained a co-culture of iPSC-derived astrocytes and neurons.
To ensure the survival and differentiation state of human primary cells and iPSC-derived cells, a vascular common culture medium was recirculated from the basal gut compartment to the apical BBB to nourish the endothelial cells and the monocytes while the BBB–brain common culture medium recirculated between the basal BBB compartment and brain compartment to perfuse astrocytes and neurons. Differently, the intestinal cells and the hepatocytes were maintained in specific cell media recirculated in a closed loop.
The integration of recirculation of culture media allowed the elimination of waste products and medium enrichment with secreting factors beyond simulating the blood circulation.
The glass windows present at the base of each OOC allow the user to monitor daily the morphology of the cultured cells (Fig. 2b and d).
To select the shared cell culture media to be circulated within the endothelial-lined vascular chambers, referred to as “vascular cell culture media” (VCCM) (Fig. 3a), and within the chambers hosting astrocytes and neurons, named “BBB–brain cell culture media” (BBCCM) (Fig. 3b), we conducted experiments with different single cell media.
In particular, we tested different media compositions onto cell-seeded inserts cultured in static condition before the insertion into the OOC.
The VCCM has to circulate through the basal chamber of the gut, the immune, the basal chamber of the liver and the apical chamber of the BBB. So, it was crucial for this medium to maintain the viability of both endothelial cells and PBMC CD14+ cells. Consequently, we cultured both cell types with different media compositions in static condition, and we found that the culture media composed of endothelial and monocyte media in a 1:
1 ratio provided optimal support for the viability of both cell types (Fig. 3a).
The same analysis was performed for the BBCCM that has to circulate through the basal chamber of the BBB and the brain and supports astrocytes alone and in co-culture with neurons. The viability assessment by the MTS assay onto cell-seeded inserts in static condition indicated that the brain medium ensured higher viability for both the astrocytes and neurons (Fig. 3b).
Model | Chamber | Numerical model | In vivo values | ||
---|---|---|---|---|---|
Shear stress (SS) (mPa) | Oxygen concentration (mol m−3) | Shear stress (SS) (mPa) | Oxygen concentration (mol m−3) | ||
Gut | Apical (0.5 mm): intestinal cells | 0.745 | 0.199 | >0.67 (ref. 10) | 0.004–0.1 (ref. 11) |
Basal (0.5 mm): endothelial cells | 0.913 | 300–9500 (ref. 12 and 13) | |||
Immune | Apical (2 mm): PBMC CD14+ monocytes | 0.062 | 0.177 | 300–9500 (ref. 12 and 13) | ∼0.053 (ref. 14) |
Liver | Apical (2 mm): hepatocytes | 0.022 | 0.169 | <0.2 (ref. 15 and 16) | >0.021 (ref. 17) |
Basal (0.5 mm): endothelial cells | 0.913 | 300–9500 (ref. 12 and 13) | |||
BBB | Apical (2 mm): endothelial cells | 0.062 | 0.151 | 0.5–2.3 (ref. 18) | ∼0.05 (ref. 19) |
Basal (0.5 mm): astrocytes | 0.913 | ||||
Brain | Apical (2 mm): astrocytes and neurons | 0.062 | 0.144 | 0.5–2.3 (ref. 18) | ∼0.05 (ref. 19) |
In detail, the intestinal cells experienced SS able to guarantee values higher than 0.67 mPa, which is the minimum value required to initiate epithelial cell differentiation.10
On the other hand, the PBMC CD14+ monocytes were subjected to negligible SS (average 0.062 mPa) which resembled the interstitial fluid flow conditions.
To minimize SS on hepatocytes and mimic the condition of the native liver sinusoid, a hydrogel encapsulating the hepatocytes reduced the average SS to 0.022 mPa, closer to the physiologic values,20 while for the common flow rate of 30 μL min−1, the SS reached average values in the BBB and brain of 0.913 mPa and 0.062 mPa, respectively.
However, with regard to the vascular compartment, the SS values ranging from 0.062 (in the apical chamber) to 0.913 mPa (in the basal chamber) were below the values experience by endothelial cells of the small veins.21
Fig. 4b shows the simulated oxygen concentration decaying from 0.2 mol m−3 in the gut OOC to 0.14 mol m−3 in the brain OOC. The theoretical oxygen levels in all the OOCs remained positive thus confirming that there was no oxygen deficiency.
We developed a computational model to predict donepezil diffusion though the platform to establish if an initial dose of 200 μM, injected into the gut, was sufficient to reach the final organ, namely the brain keeping a biologically active concentration.
We have considered neither drug uptake by cells or absorption by the device material nor drug metabolism by the liver OOC. The drug diffusion coefficients were measured experimentally for all the cell models and are shown in Table 3.
Fig. 5 shows the results of the computational model of donepezil diffusion through the platform.
After 24 hours from drug administration in the gut, a decay in donepezil concentration occurred from the basal chamber of the gut to the brain where it reached 141 μM. Counter-current flows for the apical and basal chambers chosen for the gut–immune–liver connection to increase mass transport allowed a slight increase in donepezil concentration.
Differently, the co-current flows of the liver–BBB–brain, chosen to reduce the tubing obstacle, determined a drop in concentration.
Moreover, a mean drug concentration of 20 μM reached the exit of the apical chamber of the liver after having crossed the basal chamber, the membrane and the hydrogel encapsulating the hepatocytes.
We measured through a three-way valve for media collection the donepezil concentration in the brain OOC 24 hours from drug injection (Fig. 6a). A discrepancy in drug levels between the experimental measurement and the numerical model was observed, as we found 45 μM donepezil in the brain compared to the 141 μM simulated.
To assess if donepezil induced unwanted toxic effects, we analyzed lactate dehydrogenase (LDH) levels in the cell culture media 24 hours from drug administration and by microscopic inspection which was feasible through the MINERVA 2.0 transparent bottom allowing live tissue imaging. Fig. 6b and c show the low levels of LDH (below 5%) in both untreated and treated samples indicating the absence of cytotoxicity.
The viability of all the cells in the platform after drug injection was evaluated with live/dead assay and confocal microscopy (Fig. 7). Cells were found viable 24 hours after donepezil administration as shown by the calcein green stained viable cells and no significant differences were detected.
Once the lack of evident cell toxicity was demonstrated, a biological correlation to donepezil passage through the platform was assessed by evaluating the P-glycoprotein (P-gp) expression, an ATP-dependent drug transport protein which can actively transport hydrophobic drugs such as donepezil.22
Since P-gp is predominantly found in the intestine, the liver and the BBB,23 we evaluated its expression by immunofluorescence in gut, liver and BBB OOC perfused samples treated and not-treated with donepezil (Fig. 8). An increased P-gp signal was observed in the donepezil-treated platform.
The integrity of the intestinal and BBB barrier was assessed with TEER measurements. No difference in TEER values was observed for both the intestinal–endothelial barrier and the BBB, thus indicating that the drug presence did not affect the barrier functionality (Fig. 9a and d).
Hepatocyte functionality was assessed by measuring albumin and urea production (Fig. 9b and c). Albumin concentration in the cell culture media was comparable between the untreated and donepezil-treated conditions, while urea secretion showed a significant increase after drug exposure.
Finally, we evaluated the inhibitory effect of donepezil on acetylcholinesterase (AChE) activity of neurons in the brain OOC (Fig. 9d). As expected, the AChE activity greatly diminished in the treated platform.
In the end, we evaluated the viability of endothelial cells present in the basal chamber of the gut and liver and in the apical chamber of the BBB after donepezil administration under continuous recirculating medium (Fig. 10). The endothelial cells in the gut, liver and BBB OOCs seemed to align along the direction of fluid flow, as evidenced by zonulin-1 (ZO-1) staining, and were viable as demonstrated by the MTS assay.
Here we have described the design and biological characterization of the PEGASO platform, a novel human multi-OOC solution for drug testing and pharmacokinetic studies.
The PEGASO platform relies on five compartments based on millifluidic devices, hydraulically connected in series and loaded with human cell models, representing most of the key biological systems and pathways involved in pharmacokinetics of drugs for neurological disease.
Drug absorption, metabolism and distribution were introduced in the multi-OOC platform by fluidically coupling relevant OOC models of the gut,6 immune system, liver,5 BBB and brain through vascular endothelium-lined channels.
The presence of these channels enabled the multi-OOC system to be perfused with a common vascular cell culture medium (VCCM) and to mimic the systemic circulation, a desirable feature to measure drug concentration and facilitate extrapolation to human systems.
Others have already reported fluidic linking of individual OOC leading to multiple microfluidic devices, but focusing on a single type of parenchymatic organ (such as liver, lung, or kidney epithelium) and not including a separate vascular compartment.25–28
It is worth underscoring that the modular design of the platform enables the addition or removal of an OOC and isolation of organ subsystems which would be more difficult in animal models. In addition, the placing into the device of the seeded Transwell-like insert after their maturation in static condition avoids using enzymatic dissociation for cell harvesting from the cell culture support and re-plating inside the device, which could impair for example iPSC differentiation and survival, particularly for neuronal cell types.
The administration of the drug into the gut OOC takes place through a 3-way valve and can be changed to other modes of delivery, for instance, intravenous by dosing to the immune OOC or to the interconnected vascular medium.
To optimize the flow rate for the multi-OOC perfusion we set up a numerical model to evaluate the oxygen supply and shear stress (SS) profile for the cells in the OOCs serially connected.
With a flow rate of 30 μL min−1 oxygen concentrations decayed from the gut (20%) to the brain (14%) but kept positive indicating that there was no oxygen deficiency.
However, these values are above the physiological oxygen levels of human organs and tissues and were near the ambient oxygen levels. More investigation about this point is warranted, for instance by measuring the local level of oxidative stress, even if the lack of cell toxicity does not support a major issue to this respect.
Differently, the SS generated by fluidic flow was already demonstrated to be able to induce intestinal epithelial differentiation,29 mimic the physiological conditions of the human sinusoid,30 and resemble interstitial fluid flow conditions together with avoiding monocyte and astrocyte damage and detachment.31–33
With regard to the vascular compartment, the physiological SS found in cerebral capillaries is believed to be approximately 0.3–2 Pa.34 However, the range of SS endothelial cells are generally subjected to in OOCs is wide, ranging from a few mPa to values greater than 1 Pa.35–37 The values experienced by endothelial cells in our platform were below the in vivo values of the capillary beds21 due to the need of selecting a common input flowrate for multiple organ models. Nevertheless, our tests showed endothelial cells viable and functional and apparently able to elongate along the direction of the fluid flow similar to native tissue. To have a clearer picture of endothelial cell functionality more investigation should be performed, for instance by assessing the transcriptomic profile of these cells under different flow rates in our device. Moreover, future engineering of the platform would be possible in order to apply larger input flow rates in the vascular compartment, which, for example, were necessary for Harding et al.38 to reach 1.2 Pa SS on endothelial cells cultured in a Transwell-based OOC device, as our platform is.
In the view of exploiting the PEGASO platform for drug testing studies, we developed a computational model to predict the pharmacokinetic profile of the Alzheimer's disease prescribed drug donepezil through the whole multi-OOC platform.
We experimentally obtained parameters such as drug diffusion coefficients and partition coefficients for all the models, and we investigated the drug diffusion mechanism occurring though the hydrogel in the liver.
After 24 hours, starting from 200 μM donepezil concentration in the gut OOC, a decay turned out from the gut to the brain OOC model, where a value of 141 μM was predicted. To validate the numerical model, we performed experimental tests to measure the donepezil concentration in the brain OOC cell culture media. The drug concentration found was 45 μM, a biologically relevant dose but lower than the simulated numerical one.39–41 We hypothesize that the reduction of the donepezil concentration in the experimental tests was due to hepatic metabolism or other biological interactions, for instance with cell media components, not considered into the numerical model.
We recognize the importance of measuring the active metabolite concentrations in the system as they contribute to the overall therapeutic and adverse effects of a drug. Therefore, it will be relevant to quantify not only donepezil but also the levels of donepezil principal active metabolite, 6-OH donepezil, using liquid chromatography to distinguish between the biological effects caused by donepezil itself and those induced by its metabolite.42
As for the biological effect of donepezil exposure in the platform, the live/dead and LDH assays showed the absence of cytotoxicity after drug injection and a high viability of over 80% in all the OOCs of the platform.
This finding was in agreement with several studies indicating that donepezil at a dose of 200 μM is safe.43
To get a clearer picture of the donepezil effect on our platform, we evaluated hepatocyte functionality by measuring albumin and urea levels. Albumin concentration in the cell culture media was comparable between the untreated and donepezil-treated conditions, while the urea secretion increased after drug injection. A hypothesis being considered is that the presence of the drug stimulated liver activity with increased protein turnover, which could lead to an increase in protein breakdown and subsequently result in higher urea production.44
We also demonstrated that the TEER levels of gut and BBB models were unaffected by donepezil.
Moreover, a key point was to show that our platform was able to predict not only donepezil transport but also its effect on brain cells. To this respect, the enzymatic inhibitory effect of donepezil was assessed by investigating the acetylcholinesterase (AChE) activity in the brain OOC. AChE enzymatic activity greatly diminished in the multi-OOC treated platform with respect to the control untreated samples. This result demonstrated both the effective transport of donepezil or its metabolites through all the OOCs and the suitability of the platform to recapitulate donepezil efficacy as an AChE inhibitor.
The passage of donepezil through the cell barriers and the circulating system was further confirmed by the correlated expression of P-glycoprotein (P-gp), a multidrug efflux pump, in the intestinal, hepatic and BBB layers of the multi-OOC.45
To the best of our knowledge, the PEGASO platform presented in this work is the first multi-OOC platform developed to test drugs against neurodegenerative diseases.4 In fact, few studies focused on developing tools to study neurological diseases such as Alzheimer's and Parkinson's diseases but with single OOC such as the BBB or the brain and not with multi-OOC platforms.46,47 The results described in this work contribute to multi-OOC field development through the integration of multiple organ systems in mutual communication, thus being potentially suitable as a support of available animal models to investigate systemic scenarios, such as drug metabolism, PD/PK, and drug toxicity. Regarding the latter, our model represents a technological step forward starting from reported two-organ liver–brain OOCs for neurotoxicity evaluation.3 Moreover, the integration into the platform of human cells, such as primary and iPSC-derived cells, increases the physiological relevance of the model. In a scenario where only a few iPSC-based multi-OOCs have been reported to host fully differentiated and functional models48 we were able to connect five functional iPSC-based and/or primary cell-based OOCs. In this study we have implemented cell and barrier models of representative players potentially involved in orally administered drug routes. In particular, we carried out a first characterization of the immune cells hosted in the platform in the view of obtaining an immunocompetent multi-OOCs. In fact, it is believed that incorporating immune cells into OOC systems is a necessary step to get platforms capable of being efficient in detecting drug-induced toxicity, also taking into account the action of the immune system.49 This would be relevant in the light of literature evidence on donepezil modulatory effects on the immune system.50,51
We acknowledge the limitations of the multi-OOC PEGASO platform lacking other important organs, such as kidney or heart, on which donepezil or drugs in general may cause undesired effects.52–54 The inclusion of a kidney OOC would be essential for excretion modelling and evaluation of optimal drug dosages. However, the versatile platform design will allow us to implement the hosted in vitro models according to the upcoming literature and further application needs. Additionally, we have to implement our protocols for long-term drug exposure assays.
The developed platform is modular making it cost-affordable and versatile, especially with iPSC-derived cells. At present, our model features commercial iPSC-derived cells and human primary cells but may be improved by inserting patient-specific iPSC-derived cells to develop a “patient-on-a-chip” for testing personalized therapies. This advancement holds potential for further exploitation in the neurodegenerative disease field studded by a high drug discovery failure rate.
Briefly, the millifluidic device consists of two nylon 3D-printed components assembled with a snap-fit closure system and a 12-well Transwell-like insert having a PET membrane with a pore diameter of 0.4 μm and density of 2 × 106 pores per cm2. Inside the device there are two hemi-chambers, apical and basal, interfaced through a porous membrane of the Transwell-like insert. The apical chamber can be 0.5 mm or 2 mm high, while the basal chamber is 0.5 mm. To implement the PEGASO multi-OOC platform using MINERVA 2.0, our device hosts Luer-lock connectors coupled to millifluidic channels with diameters of 0.5–1 mm.
MINERVA 2.0. devices were sterilized by UV rays (SafeMate cabinet) for 10 h or with hydrogen peroxide (V-PRO® 60 Low Temperature Sterilization System).
The gut OOC had a 0.5 mm high apical chamber and a basal chamber of 0.5 mm, while the immune OOC, liver OOC, BBB OOC and brain OOC had a 2 mm high apical chamber and a basal chamber of 0.5 mm.
Endothelial cells were maintained in endothelial medium composed of VascuLife VEGF Medium (Lifeline Cell Technologies, Frederick, MD) supplemented with the complete growth factors.
Intestinal cells were maintained in Epithelial Cell Growth Medium® (iXCells Biotechnologies).
The cell-seeded inserts were maintained 7 days in static conditions before being assembled once inserted in the MINERVA 2.0.
The cell-seeded inserts were maintained for 4 hours in static conditions before being assembled once inserted in the MINERVA 2.0.
The cell-seeded inserts were maintained for 5 days in static conditions before being assembled once inserted in the MINERVA 2.0.
The membrane sides of endothelial cells and astrocytes were pre-coated with 10 μg ml−1 laminin (Sigma-Aldrich) and 30 μg ml−1 fibronectin (PromoCell), respectively. All the culture inserts were incubated at 37 °C and 5% CO2 for 7 days with medium renewal every 2 days.
Endothelial cells were maintained in endothelial medium composed of VascuLife VEGF Medium (Lifeline Cell Technologies, Frederick, MD) supplemented with the complete growth factors included in the kits. Here, only 10 mL of the glutamine solution was added per 500 mL of media and 50 mL of the CDI-provided supplement replaced the VascuLife FBS component.
Astrocytes were maintained in DMEM/F12, HEPES (ThermoFisher), FBS hyclone (GE Healthcare Life Sciences), and N2 supplement 10× (ThermoFisher).
The cell-seeded inserts were maintained for 7 days in static conditions before being assembled once inserted in the MINERVA 2.0.
Astrocytes and neurons were maintained in BrainPhys Neuronal Medium (STEMCELL), iCell Neural Supplement B (CDI), iCell Nervous System Supplement (CDI), N-2 Supplement 100× (ThermoFisher), and penicillin–streptomycin 100× (ThermoFisher).
The cell-seeded inserts were maintained for 7 days in static conditions before being assembled once inserted in the MINERVA 2.0.
OOC | |||||
---|---|---|---|---|---|
Gut | Immune | Liver | BBB | Brain | |
Apical chamber | Intestinal medium | Vascular common medium | Hepatocytes medium | Vascular common medium | BBB–brain common medium |
Basal chamber | Vascular common medium | — | Vascular common medium | BBB–brain common medium | — |
Two hours after platform assembly, donepezil hydrochloride 200 μM was dissolved into the apical compartment of the gut OOC. The PEGASO platform treated with DMSO, the vehicle of donepezil, was used as the control. After 24 hours of dynamic perfusion and donepezil administration the platform was disassembled, and biological tests were performed.
For each experiment, three platforms treated with donepezil and three with DMSO (control) were used.
Fluid velocity vector u was determined using the Navier–Stokes equation in the stationary condition:
ρ(u·∇)u = ∇·(−pI + K) + F |
ρ ∇·u = 0 |
K = μ(∇u + (∇u)T) |
Shear stresses were calculated as:
We estimated oxygen distribution with the equation of mass transport of diluted species:
∇·(−Di·∇ci) + u·∇ci = Ri |
For the 2D cell models (gut, immune, BBB and brain) we set the reaction term as null, and we assumed a homogeneous cell distribution through the membranes. The oxygen consumption rate was calculated as: , where Vmax is the oxygen cell consumption rate, d is the cell density and Vol is the volume.
For the 3D cell model (liver), it was assumed a reaction term function of the local oxygen concentration according to the Michaelis–Menten kinetics: , where Vmax is the maximum molar consumption rate, c is the local oxygen concentration and Km is the Michaelis–Menten constant.
We estimated donepezil distribution with the equation of mass transport of diluted species (∇·(−Di·∇ci) + u·∇ci = Ri) in which the reaction term was set as 0.
We assumed an initial AD drug donepezil equal to 0.2 mol m−3 at the inlet of the apical chamber of the gut OOC.
The drug diffusion coefficient for all the organs of the platform was estimated experimentally and calculated with the Fick's law equation for a diffusion cell:
To run the simulations, we chose the “fine” element size for mesh building.
All the characteristic parameters are summarized in Table 3.
Parameter | Value |
---|---|
Inlet oxygen concentration | 0.195 mol m−3 |
Outlet pressure | 0 Pa |
Oxygen diffusion coefficient | 2 × 10−9 m2 s−1 |
Condition at the wall | No slip condition |
Density of the culture media at 37 °C | 1000 kg m−3 |
Dynamic viscosity of the culture media | 8.1 × 10−4 Pa s |
COLL–PEG gel permeability | 6.45 × 10−17 m2 |
COLL–PEG gel porosity | 0.93 |
Donepezil diffusion coefficient in the GOC | 1.11 × 10−9 m2 s−1 |
Donepezil diffusion coefficient in the IMMUNE | 1.87 × 10−9 m2 s−1 |
Donepezil diffusion coefficient in the LOC | 4.12 × 10−11 m2 s−1 |
Donepezil diffusion coefficient in the BBBOC | 1.05 × 10−9 m2 s−1 |
Donepezil diffusion coefficient in the BOC | 0.98 × 10−9 m2 s−1 |
Intestinal cell oxygen consumption | J c = 1.14 × 10−8 mol m−2 s−1 (ref. 56) |
Endothelial cell oxygen consumption | J c = 1.46 × 10−9 mol m−2 s−1 (ref. 57) |
PBMC CD14+ oxygen consumption | J c = 3 × 10−8 mol m−2 s−1 (ref. 58) |
Hepatocyte oxygen consumption | R c = 5.53 × 10−4 mol m−3 s−1 (ref. 17) |
Astrocyte oxygen consumption | J c = 1.56 × 10−13 mol m−2 s−1 |
Astrocyte + neuron oxygen consumption | J c = 9.37 × 10−13 mol m−2 s−1 |
The live/dead images were postprocessed in ImageJ software, and viable and dead cells were manually counted. The viability was calculated as follows:
The following equation was used to calculate the percentual toxicity of donepezil treated samples with respect to control samples:
After 24 h of static culture, 80 μL of CellTiter 96 Aqueous One Solution Reagent was pipetted into the Transwell-like insert containing the sample in 500 μL of medium.
After 2 h of incubation at 37 °C in a humidified 5% CO2 atmosphere, the amount of soluble formazan produced by cellular reduction of MTS was measured with a plate reader (490 nm absorbance).
TEER = (Rmeasured − Rblank) × Membrane Area |
In more detail, the quantification of the P-gp fluorescence signal was based on the recognition of the region of interest within the acquisition area. From the segmentation of the ROI it was possible to obtain parameters like integrated density, mean gray value and area. The total fluorescence of an ROI within the acquisition area was calculated with the following formula:
TF = integrated density − (total area × mean gray background value). |
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4lc00273c |
‡ These authors are co-last authors. |
This journal is © The Royal Society of Chemistry 2025 |