49
Chapter 4
4.1 Models evaluation: ILT1
In the ILT1 design phase, theoretical models are needed in order to simulate fluid dynamics inside the bioreactor and for evaluation of solute passage through membrane. In order to reach these targets two models are studied, using COMSOL MULTIPHYSICS software. The first one, called Model A, has the aim to characterize the bioreactor from a fluid dynamic point of view, evaluating flow lines and maximum flow rates in order to keep laminar conditions and minimize shear stress. The second model, called Model B, has the aim to study solute passage through membrane pores, in order to characterize the porous insert. In both models it is necessary to reproduce the ILT1 wet volume, taking into consideration the porous membrane placed between the two chambers. COMSOL MULTIPHYSICS software uses a finite element approach in order to solve physical problems characterized by a system of differential equations, based on partial derivatives, such as Navier Stokes equations. Moreover several modules can be used in this software in order to simulate different phenomena of the same system. COMSOL MULTIPHYSICS is a suitable solution for its versatility and the need of parameters to initialize models that can be evaluated by simple experiments.
4.1.1 Model A
Model A has the aim to evaluate maximum flow rate in order to keep laminar conditions inside the chambers. In fact turbulence could cause uncontrolled localized shear stress on cells surface, damaging the culture, and unexpected resistances to solute passage through membrane. In COMSOL’s fluid dynamic module it is possible to initialize problems using liquid parameters such as viscosity and density, a temperature of 37 oC and setting several flow rates from 100 μL/min to 1000 μL/min. Moreover it is possible to simulate situations
50 in which the upper chamber flow rate to different to the bottom one. In table 4.1 initialization parameters are summarized.
Parameters Values Viscosity μ 1 cP Density ρ 1*103 kg/m3 Temperature T 310 K
Flow Up 100-1000 μL/min Flow Bottom 100-1000 μL/min
Table 4.1 parameters used to initialize the problem in model A. Viscosity and density are liquid parameters. Flow up and Flow bottom stand for flow in upper chamber and flow in
bottom one properly
In the post processing phase, COMSOL can show flow lines and evaluate the Reynolds Number. Figure 4.1 gives an example of flow lines evaluated at 100 μL/min inside both chambers. Laminar conditions are assured in this set up.
Figure 4.1 Flow lines evaluated for a flow of 100 μL/min inside both chambers
A range of flow between 100 μL/min and 350 μL/min was found suitable to maintain laminar flow conditions. A flow of 400 μL/min causes turbulence phenomena. Moreover it
51 is possible to evaluate the shear stress on the chamber walls (figure 4.2). In particular attention was focused where cells should be seeded. In the literature it is accepted that cells need suitable shear stress values in order to grow and develop their functions. The aim of this analysis is to verify that flow in bioreactors does not cause a higher shear stress than physiological values.
A B
Figure 4.2 Chromatic view of shear stress evaluation. Red colour indicates high value of shear stress. Blue colour is instead associated with low shear stress. A: Shear stress on
bottom side of ILT1. Maximum shear stress: 9.4*10-6 Pa B: Shear stress on membrane
surface. Maximum shear stress: 1.1*10-5Pa
It is possible to plot shear stress estimated by the model as a function of flow rate on a Cartesian graph. There is a range between 100 μL/min and 350 μL/min where shear stress is low, followed by a rapid increase from 400 μL/min (figure 4.3).
52 Figure 4.3 Plot of flow rate in relation to shear stress. A range between 100 μL/min and
350 μL/min assures low shear stress values. Flow rates higher than 400 μL/min show a rapid increase in shear stress value that could damage cells
This information and the study about flow conditions at different flow rates verify the presence of turbulence conditions caused by flow rate higher than 400 μL/min. Moreover the model permits to evaluate that the worst conditions from a shear stress point of view are reached when chambers inputs have the same flow rate.
4.1.2 Model B
Model B is used to study the passage of solutes dissolved in water from upper chamber to bottom chamber crossing membrane pores. It is coupled to model A through convection. In order to achieve this end point a chemical module has to be associated to the fluid dynamic module, in order to study diffusion and convection. Moreover to evaluate a solute passage, a kinetic module time variant setting is used. A time of analysis of 24 hours is performed in order to simulate a typical in-vitro experiment. Membrane characteristics are simulated by a porous layer where porosity and permeability are initialized by experimental evaluated data. The solute simulated by model B is Methylene Blue, an heterocyclic aromatic compound (C16H18N3SCI) used in a range of different fields such as chemistry or biology. Simulating solute passage using a solution, instead of particles dissolved in liquid, it is simpler than simulate nanoparticle passage. Model A is used to initialize solution of Model B, in order to base convection and diffusion evaluations on fluid. Parameters used
53 to initialize the problem in Model B are summarized in table 4.1. the solute concentration at the beginning of the simulation was 0.2 mg/ml. This value is constant in time in order to simulate the presence of a solution reservoir characterized by a volume bigger that the bioreactor chambers (mixing chamber volume is modelled as an infinite reservoir). Figure 4.4 shows results of the computational model after 24 hours. Basing on concept that shear stress is higher in presence of the same flow rates inside both chambers, simulation are performed imposing 200 μL/min in upper chamber and 100 μL/min in bottom one. After 1 day of simulated solute passage, an equilibrium point is not reached.
Figure 4.4 Solute passage through membrane. Red colour is associated with high Methylene Blue concentration areas. Blue colour is instead resented of low concentration
areas. Flow condition is: 200 μL/min in upper chamber and 100 μL/min in bottom one. Moreover red lines indicate the concentration gradient of Methylene blue due to flow
The model estimates that after 24 hours the percentage of solute passage is equal to 8% of the starting solute amount. these models are used in the bioreactor design phase, in order to evaluate the bioreactor designs. Once ILT1 is produced, Models A and B are validated by in-vitro experiments as described in paragraph 4.3.
54
4.2 Models evaluation and validation: ILT2
As described for ILT1 bioreactor, a computational fluidic model is implemented during the ILT2 design phase, in order to verify the efficiency of new designs. The aims are to study shear stress on the membrane surface and on the bottom chamber and velocity isobars to ensure the absence of turbulence. COMSOL MULTIPHYSICS is used as software to implement the simulations. Wet volume chamber geometry is used to create an appropriate mesh, in order to perform fluidic analysis. As boundary conditions, upper inlet is characterized by flow of 200 μL/min, in contrast to the bottom one where flow is imposed at 100 μL/min. Medium culture characteristics are implemented in order to initialize the liquid simulated by the CFD analysis, as described in paragraph 4.1. Figure 4.5 shows flow lines.
Figure 4.5 ILT2 wet volume and flow lines pathway
The flow rates imposed in the CFD model assure laminar flow conditions, as shown in figure 4.5. Figure 4.6 shows the shear stress on the membrane surface and at the bottom of the chamber.
55
A B
Figure 4.6 A: shear stress on membrane surface. B: shear stress on base of ILT2 bottom chamber
4.2.1 Problem solving: liquid passage
One of the main problems underlined by biologists during use of ILT1 was the passage of medium from the top chamber to the bottom one. In fact during the simulation phase it is impossible to detect: liquid passage from upper chamber to bottom one and consequently the reduction of liquid volume that flows inside the upper circuit, causing problems in duration of experiments. The emptying of one mixing chamber causes two problems:
When a sample is taken, it has to be related to a specific volume, in order to extrapolate mass of the analyte from a measured concentration value. Liquid passage causes a volume variation that it makes impossible to evaluate absolute quantities of measured analytes.
When one of the mixing chambers is empty or liquid level is too low, air bubbles enter in circuit and reach the bioreactors. Two phenomena are correlated to this point: local increase of pressure and obstruction of pores that causes a reduction in solute passage through membrane pores.
Analyzing the phenomena, a difference in pressure between the membrane is upper side and bottom side can be estimated. This disequilibrium causes passage of a small amount of
56 liquid from the upper chamber to the bottom one. A reduction in flow from the upper outlet is correlated with liquid passage through membrane. Following Bernoulli’s equation, reduction in flow is associated with an increase of pressure. As a consequence, increasing pressure in upper chamber causes an increase in pressure difference with the bottom chamber. This cascade of events ends when the upper circuit is emptied. Continuous variation of parameters makes it impossible to adjust the pressure disequilibrium by positioning mixing chambers on different levels, in order to increase the bottom circuit pressure in order to reduce the pressure difference on the membrane. A solution could be represented by a peristaltic pump controlled in pressure instead in flow. In that case the pump should compensate pressure differences by measuring flow rates. However flow increases can cause a shear stress peak that could damage cells. In order to solve the liquid passage problem it was decided to close the mixing chamber using an on/off valve. By this way the circuits become two closed systems in communication through the membrane. In a closed system, the pressure difference and then liquid passage is auto regulated. Flow is kept in a suitable range in order to have physiological value of shear stress on cells. So a stable equilibrium point avoiding liquid passage through membrane can be found. However in this condition it is not possible to control local increase of pressure due to gas bubbles that attach to bioreactor wall. Bubble traps are needed in order to avoid gas bubbles that flow inside chambers. Therefore a couple of them are assembled before the bioreactor, one per circuit, in order to capture air bubbles. ILT2 system, bubble traps and ILT0 bioreactors are analyzed carefully, simulating experimental conditions [Appendix B]. Using a complete fluid dynamic system, it is possible to evaluate watertightness and liquid passage between the upper and bottom circuits. Both mixing chambers are filled by deionised water. Meniscus levels are pointed on both mixing chambers. Upper and bottom flows are set up respectively at 200 μL/min and at 100 μL/min. Tests are repeated three times. After 24 hours no leakages are visualized and liquid levels inside the mixing chambers are constant.
57
4.2.2 Problem solving: evaluation of NP concentration on the
membrane layer
Usually, when nanoparticles exposure is evaluated in-vitro doses are represented as surface densities (mass/area) because in static conditions nanoparticles are thought to sediment due to gravity. In actual fact, this situation is not representative of the in-vitro condition, because the physiological environment. In the InLiveTox project, on due one hand, it is desirable to simulate in-vitro conditions, while on the other hand, it is necessary to compare the results of InLiveTox system with the “gold standard” static controls. One of the objectives of this thesis was therefore to evaluate the surface density of nanoparticles on the membrane in the ILT circuit in the presence of flow, as a function of flow, and initial nanoparticle concentration (figure 4.7).
Figure 4.7 The model that is described in this section studies a correlation between nanoparticle concentration in the mixing chamber, expressed by a ratio between mass and
volume and nanoparticles concentration that is deposed on the membrane, expressed as ratio between mass and surface area
The aim of the model is to evaluate nanoparticle concentration (kg/m3) in dynamic conditions that permits to recreate the static in-vitro parameters (kg/m2), in order to compare the two experimental conditions. The bioreactor upper chamber is modelled as a tube as illustrated in figure 4.8
58
A B
Figure 4.8 A: cylindrical pipe is representation of ILT2. Parameters taken in consideration to evaluate are enounced in red. B: upper circuit of ILT2 is represented. The
correspondence between input and output parameters can be noted.
Using a fluid dynamic approach (equation 1) it is possible to evaluate mass transport in correspondence to the wall of a conduct where fluid flows :
(1)
Where:
: mass transport rate per unit surface area
C : concentration at infinite distance (in case of ILT2 it is the concentration of mixing chamber)
D : diffusion coefficient (estimated with equation 2), evaluated by stokes equation for Nanoparticles in a FBS 2% solution
W : width of the target area where compounds tends to be deposited (in case of ILT2 it is the membrane width)
L : length of the tube used to calculate the deposition (in case of ILT2 it is the membrane length)
γ : velocity field derivative on the y coordinate axis (in case of ILT2 it is the velocity field derivative on the thickness of the upper chamber)
59 (2)
Where:
R: Boltzman’s constant
T: Temperature in Kelvin degrees
N: Avogadro’s number
d: diameter of a nanoparticle
μ: dynamic viscosity
Velocity is a vector that has two components: fluid velocity imposed by peristaltic pump (vfluid) and deposition velocity (vdep) related to gravity force and to drug force that act on a nanoparticle (equation 3 and 4).
(3)
(4)
Where:
Q : flow rate imposed by peristaltic pump
r : inlet radius
g : gravity acceleration
60 ρs : density of the solute
ρf : density of fluid where solute is dispersed
Cd : constant coefficient calculated with equation (5):
(5)
Where:
Re : Reynolds number
The deposition velocity (equation 4) takes care of the difference between Nanoparticles density and fluid density. In recent studies focused on the evaluation of Nanoparticle hydrodynamic radius, it was discovered that nominal radius of compounds is slightly different from the measured one. Data evaluated in WP2 suggest that the real dimension of a nanoparticle is bigger than the nominal value, when it is dispersed in a water solution. The explanation is that fluid molecules adhere to Nanoparticle surfaces, creating nano spheres characterized by a bigger radius (rmeasured) than the nominal value indicated for a nanoparticle (rnominal). For example it is possible to measure a diameter of 120 nm instead a nominal one of 20 nm for Silver Nanoparticles dispersed in a FBS 2% solution. The density value of the sphere composed of the nanoparticle and the liquid molecules (ρs) used is in fact a weighted average value between fluid density (ρfluid) that cover the nanoparticle and the nano particle density (ρnanoparticle). Each contribution is related to the percentage of volume that represents if compared to the resulting sphere. In order to take care of the contributions to the ρs, density values of each component is correlated to the volume ratio that is occupied by water (1-perc) or nanoparticle (perc) (equation 6).
(6)
(7)
61 Once all parameters of the model are known (equation 1), it is possible to find a correlation between volumetric concentration kg/m3) in the upper chamber and surface concentration (kg/m2) on the membrane surface area. Taking for example, Ag NP parameters, figure 4.9 summarizes the model results.
Figure 4.9 correlation between chamber concentration (kg/m3) and surface concentration
on membrane surface area (kg/m2)
In this analysis, we hypothesize that the upper chamber concentration is the same in all the fluidic circuit. It has to be underlined that model is focused on the correlation between a surface concentration and a volumetric concentration. All the particles that are deposited on the membrane derive from deposition of nanoparticles that flow on the target surface area at steady state. It means that it is not necessary to consider all the forces presented in a typical situation such as the simulation of a nanoparticle movement inside a dynamic environment, referring in particular to Saffman lift. Since the aim is to find an index that represents the minimum value of nanoparticle volumetric concentration that permits a
62 particular surface deposition at steady state, only the forces that are related to the sedimentation are taken into account. Evaluating the slope of the curve shown in figure 4.9 it is possible to simply correlate volume concentration to surface concentration.
4.3 In-vitro characterization phase of the systems: materials and
methods
The characterization phase consists of experiments in order to characterize the systems with in-vitro data. The ILT circuits are set up as described in protocol [Appendix A, B]. The same procedures can be adopted also in case of MCmB 3.0 In all system that are analyzed the upper circuit is filled by solutions of solutes. In case of ILT system Methylene Blue or Polystyrene Nanoparticles labelled with FITC-PS are used. Methylene blue is an indicator of redox reactions. Its use is widely diffused in the biochemical lab. It is not a biocompatible molecule, so it can be used only in case of preliminary characterization of the circuits. The choice of these materials is related to the possibility to prepare, and analyze samples using spectrophotometer (Methylene blue maximum absorbance is at 610 nm). Moreover each sample can be analyzed step by step avoiding any problems of storage. FITC-PS of 200 nm in diameter are tested (Polysciences) because they are very easy to detect using a spectrofluoremeter. In order to prevent the possibility to form cluster in water, the Nanoparticles are dispersed in a Fetal Bovine Serum (FBS) 2% solution. Kinetic of solute passage through membrane is studied taking samples as indicated in table 4.2 from both bioreactor outlets.
Time [h] Samples 0 S1 1 S2 3 S3 7 S4 24 S5
63 Sample “S1” is taken when liquid starts to drop inside the mixing chambers. This sample is used as a base line to analyze data. Each experiment is replicated 3 times, in order to obtain statistic relevance. Methylene Blue dissolved in deionised water is used in order to validate model A and B. FITC-PS are used to study kinetics of nanoparticles. A spectrophotometer is used to analyze samples of both experiments. In next paragraphs Methylene Blue and Nanoparticles experiments will be explained.
In case of the MCmB 3.0 solutes used in order to characterize the passage through the system are: Methylene Blue, Urea And Diclofenac. Diclofenac (IUPAC: 2-(2-(2,6-dichlorophenylamino)phenyl)acetic acid) is an anti-inflammatory drug. Figure 4.10 shows the structure of this drug.
Figure 4.10 Diclofenac structure
Diclofenac was chosen because it is hydrophilic and absorbs UV light, so is easy to detect using a spectrophotometer. Moreover there are several studies that investigate cells exposure to this drug, so data that are obtained by In-vitro models can be correlated with published data. Urea is an organic compound that plays an important role in the metabolism of nitrogen containing substances. Figure 4.11 shows the structure of urea.
64 The role of urea is very important in metabolism, so testing the system with this molecule represents a key point in order to develop models that simulate human metabolism. A commercial kit was used to analyze urea samples with spectrophotometer. Commercial membranes that are placed gently in holder system are disc of 25 mm in diameter and 60 μm in thickness, made of Polycarbonate (PC) (Fisher Italy). They are cut in order to fit the holder. Pore diameter is 0.4 μm, and porosity is equal to 0.5. The data are described in the following paragraphs.
Curve fitting of each data sample is analyzed minimizing the mean squared error. The parametric equation used to evaluate the best fitting is (equation 8):
8)
Matlab Mathworks (7.0 R) calculates automatically “a” and “b” parameters in order to estimate the best fitting and characteristic time (τ*) of each case (equation 9).
9)
4.3.1 ILT1 characterization
In the following paragraphs the experiments performed in order to characterize ILT1 system are described. In-vitro experiments such as the kinetic studies of Methylene blue and nanoparticles passage through membrane pores are performed in order to validate the CFD models. Moreover a validation of TEER set up is presented.
4.3.1.1 Methylene Blue (ILT1)
The concentration of the Methylene blue solution used to characterise ILT1 system is 0.25 mg/ml in deionised water. The upper mixing chamber is filled by this solution before the experiment. Deionised water flows in the bottom circuit. Flow rates in both circuits are imposed at 100 μL/min in a first analysis. Samples of 100 μL are taken from both chambers following the timetable (table 4.2). Concentrations evaluated by the spectrophotometer are conversed in percentage of starting Methylene Blue concentration,
65 in order to have an immediate feedback in post processing phase. Comparing the upper circuit with bottom circuit data, it is clear that an equilibrium point is not reached after 24 hours as it would be expected (figure 4.12).
Figure 4.12 Percentage of Methylene Blue passage taken from upper and bottom chambers. Red triangles indicate data from upper chamber. Blue spheres indicate data
from bottom chamber. At 24 h percentages are not equal, so equilibrium point is not reached
Model A conclusions about flow lines can be verified by a visual analysis of bottom chamber. It can be noticed that dye water passes through the membrane following flow lines in the same direction as those ones predicted by A and B (figure 4.13). Model B evaluates concentration distributions following flow lines of model A, as it has to be expected since the 2 models are coupled. Moreover a picture taken during Methylene Blue experiment demonstrate that the dye water passes to the bottom chamber following flow lines similar to the models.
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
0
4
8
12
16
20
24
P er ce ntage [ % ] Time [h]66
L R
C
Figure 4.13 Comparison between “L” flow lines evaluated by Model A, “R” solute passage evaluated by Model B, and “C” dye water passage recorded by In-vitro
experiments
It is possible to compare passage estimated by model B with experimental data. The result is a good correspondence between model predictions and experiments. Despite the good correlation between the model and experimental data, the percentage of solute passage through the membrane after 24 h is too low if compared to diffusion theory in static conditions. In order to increase the passage, the upper circuit flow rate is imposed at 200 μL/min, maintaining the bottom one at 100 μL/min. By this way Methylene Blue is forced to pass through membrane pores by convection. In fact it is possible to register an increase of solute passage from upper chamber to bottom circuit (figure 4.14).
67 Figure 4.14 Comparison between bottom chamber kinetics evaluated for Methylene Blue
in conditions of 100 μL/min on both chambers (blue points) or upper flow set at 200 μL/min and bottom at 100 μL/min (red triangles)
4.3.1.2 FITC-PS Nanoparticles
Based on Methylene Blue data, working point is characterized by upper flow set at 200 μL/min and bottom flow at 100 μL/min. Nanoparticles concentration is 0.2 mg/ml. As described for Methylene Blue, before the experiment the upper mixing chamber is filled by a nanoparticles dispersion. In the bottom circuit a solution of FBS 2% flows. Once the peristaltic pump starts to run, samples are taken from both bioreactor outlets, following the time table (table 4.2). Kinetics evaluated by tests is similar to Methylene Blue. 24 h after the beginning of experiments only 12 % of starting Nanoparticles weight can be found in the bottom chamber (figure 4.15).
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
14.00%
0
4
8
12
16
20
24
P aer ce ntage [% ] Time [h]68 Figure 4.15 Data of Nanoparticles passage through membrane
Experiments described in preview paragraphs characterize ILT1 system without cells. They can be a base line to analyze data of experiments where cells are seeded in the chambers. Figure 4.16 shows the curve fitting and the characteristic time (τ*).
0% 2% 4% 6% 8% 10% 12% 0 5 10 15 20 P ass age [ % ] Time [h]
69 Figure 4.16 Curve fitting and τ*of FITC-PS data in ILT1
It remains to test the possibility to take TEER measurements and the system will be completely characterized.
4.3.1.3 TEER measurement tests
Once the ILT1 system is assembled, 4 electrodes are inserted in the bioreactor channels and then connected to EVOM system, in order to measure electrical impedance, as described in chapter 3. Several experiments are performed in order to evaluate membrane electrical impedance when saline solutions of different concentrations flow inside the circuits. NaCl is dissolved in deionised water in order to obtain several solutions characterized by different concentrations. A membrane is placed on its slot. The ILT1 is assembled. Each solution of NaCl is used to fill bioreactor ILT1 in both chambers. Once the bioreactor is completely filled up TEER measurements are performed in dynamic conditions (pump on) and static conditions (pump off), inserting the electrodes system in the channels and connecting wires to the EVOM. Each time, when a test on a particular concentration is completed, the system is empted and then it is washed with deionised water. Table 4.3 summarizes NaCl sample concentrations used in TEER experiments.
70 NaCl Conc [%] 0.05 0.0625 0.1 0.125 0.25 0.5 1 2 5 10
Conc-1 20 16 10 8 4 2 1 0.5 0.2 0.1
Table 4.3 NaCl samples concentrations used to test TEER system. First row summarizes NaCl concentrations in percentage. Second row shows normalized concentration values
The peristaltic pump is set up in order to have a flow of 200 μL/min in upper circuit and 100 μL/min in the bottom one. Each experiment is replicated 3 times, in order to obtain statistic relevance. Data are acquired in two conditions: pump on or off. The aim is to compare these situations, in order to evaluate flow influence on TEER measurements. In low concentration conditions, a low impedance value is expected. TEER data in relation to concentrations are represented in figure 4.17.
Figure 4.17 TEER data taken in static (blue points )or dynamic conditions (red crosses) in relation to NaCl samples concentrations
As explained, there is a linear dependence between impedance and NaCl concentrations. Moreover about impedance values in relation to concentrations are correct. The data underline a good system stability in static and dynamic conditions. Data acquired in static conditions are not so different from data in dynamic condition, as can be observed in figure 4.17. This observation reveals the possibility to take TEER measurements in real time during the experiments, so as to monitor the status of the cells in real time.
71
4.3.2 ILT2 validation
As the wet geometry of ILT2 was identical to ILT1, it was not necessary perform tests to optimize flow rates. The standard flow of to 200 μl/min in the upper circuit and 100 μl/min in the bottom chamber is used. Methylene blue and Polystyrene Nanoparticles are used to characterize the system. In next paragraphs the results of each test will be described. The new membranes used are realized by CSEM and have pores of 1 μm in diameter and a fill factor of 15%. That means less open surface to have an exchange of the solute between upper and bottom circuit, but, on the other hand is less fragile than the membrane described in paragraph 3.2. This makes it easier to manage. The tests were carried out over 24 hours because this is the duration of the InLiveTox cell cultivate experiments. Pressure measurements and TEER system validation contribute to a complete characterization of ILT2 system.
4.3.2.1 Methylene Blue
The upper circuit is filled by a solution of Methylene blue and deionised water in concentration of 0.25 mg/ml. In the bottom circuit flows deionised water. It is possible to monitor passage of dye water through membrane pores by sampling from the outlets of the chamber. Figure 4.18 shows data sampled from the bottom chamber outlet.
72 Figure 4.18 Percentage of Methylene blue that flows through membrane pores. Sampling
the outlet of ILT2 bottom chamber
If compared to figure 4.18, the data shown in figure 4.12 follows the same trend and reaches the same percentage of solute passage through the membrane after 24 hours. Figure 4.19 shows the fitted data.
0% 2% 4% 6% 8% 10% 12% 14% 0 5 10 15 20 Pe rc e n tage [% ] Time [h]
73 Figure 4.19 Curve fitting and τ* of Methylene blue data in ILT2
Figure 4.20 shows a comparison between data evaluated for the bottom circuit and data for the upper one.
Figure 4.20 Passage of Methylene blue through membrane. Comparison between samples of the upper chamber (red triangles) and bottom chamber (Blue balls)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 5 10 15 20 Per ce n tage[% ] Time [h]
74 Figure 4.20 clearly shows that the system does not reach an equilibrium point characterized by an equal distribution of dye water between chambers. To reach equilibrium will probably require a longer period
4.3.2.2 FITC-PS Nanoparticles
Experiments with ILT2 based on Nanoparticles were identical to those described for Methylene blue. The difference is related to presence of FBS 2% in the solvent which reduces nanoparticles aggregation. Figure 4.21 shows data analysed by fluorescence method using a spectrophotometer.
Figure 4.21 Percentage of Nanoparticles that flow through membrane pores. Sampling is focused at the outlet of ILT2 bottom chamber
As described for Methylene blue, it is possible to compare figure 4.21 with figure 4.15. ILT2 data follow a similar curve to ILT1 even for the Nanoparticles. Moreover after 24 hours it is possible to evaluate a small percentage of passage through the membrane pores identical to ILT1 tests. However the characteristic time of Methylene blue passage in ILT1 (figure 4.15) is smaller than the ILT2 one, as shown in figure 4.22.
0.00% 2.00% 4.00% 6.00% 8.00% 10.00% 0 4 8 12 16 20 24 Percent age [%] Time [h]
75 Figure 4.22 Curve fitting and τ* of FITC-PS in ILT2
Figure 4.23 shows a comparison between data evaluated from samples of the upper chamber outlet and data related to bottom chamber outlet.
Figure 4.23 Passage of Nanoparticles through the membrane. Comparison between samples of the upper chamber (red triangles) and bottom chamber (Blue balls)
0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00% 90.00% 100.00% 0 4 8 12 16 20 24 Percent age[ %] Time [h]
76 As before after 24 hours the equilibrium point is not reached as it is evident from the unequal distribution of solute between circuits.
4.3.2.3 TEER measurements
As described for ILT1 system, different solutions of NaCl are used to feel the ILT2 system and perform TEER measurements. The goal is to demonstrate a correlation between ILT1 data and ILT2, verifying the possibility to monitor cells in real time using the TEER set up in both cases. The experimental procedure is the same adopted and described for ILT1. Figure 4.24 describes the comparisons between ILT2 data performed in static conditions and dynamic conditions and with ILT1 data.
Figure 4.24 comparisons between TEER measurements performed with ILT2 in dynamic conditions (blue points), static conditions (red points) and ILT1 (green points)
From figure 4.24 it is evident that TEER measurements performed with different devices in different conditions show good correlation. The conclusion is that it is possible to perform a stable TEER measurement with ILT2 in static or dynamic conditions. This is the validation of the correct design realized in order to maintain the electrodes in dry conditions.
77
4.3.2.4 Pressure measurements
In order to characterize ILT2 in all its aspects pressure measurements are performed. Bubbles traps are placed between the peristaltic pump and ILT2. Two ILT0 are joined to the bottom outlet, in order to simulate chambers where endothelial and hepatic cells are going to be seeded. Each mixing chamber is filled with 15 ml of deionised water, and placed between bioreactors and the pump. 3 ways valves are placed at the inlets and outlets of ILT2. Protocol [Appendix B] summarizes the passages required to set up the ILT2 system. In order to measure pressure inside the bioreactor a sensor can be attached to each valve by a tube of 2 mm in diameter (i. d.) filled with deionised water. It is important to avoid air inside pipe that connects the sensor to the hydraulic circuit, in order to obtain high sensitivity to small pressure variations. Pressure is measured at the inlets and at the outlets of the upper and bottom chambers, in order to characterize the pressure drop inside the bioreactor. A problem is represented by the mutual influence of fluid conditions inside the circuits: opening the three way valve of the upper or bottom chamber in order to monitor pressure imposes a change in conditions of the other bioreactors. In order to avoid this effect a silicon chip is used instead of a porous membrane. The aim is to separate the upper chamber environment from the bottom one, in order to perform pressure measures in an environment influenced only by the section of circuit that is taken in consideration. The peristaltic pump imposes a flow of 100 μL/min in the bottom circuit and 200 μL/min in the upper one. Once the sensor is attached to a 3 way valve, the stopcock is opened and the sensor is able to record pressure values. A time delay of 30 s after the connection of the sensor permits to avoid transients. Pressure evaluation takes only 120 s per test. Figure 4.25 shows the fall in pressure between the inlet and outlet of the upper chamber.
78 Figure 4.25 Pressure drop between inlet and outlet of the upper chamber. Different
colours are different data sets
Table 4.4 summarizes the complete data and statistical analysis on the upper and bottom circuits. Inlet Outlet Average Up 424.99 307.50 STD dev 40.20 15.43 Dev % 9.46% 5.01% Average Bottom 626.85 549.21 STD dev 20.38 17.04 Dev% 3.25% 3.10%
Table 4.4 Summary of pressure measurements at the inlet and outlet of ILT2 in the upper and bottom chambers in the complete InLiveTox circuit
30 80 130 180 0 20000 40000 60000 80000 100000 120000 Pr e ssur e [Pa] Time [ms]
79 Comparing mean pressure values evaluated for the upper circuit with the bottom one, it is possible to notice that the bottom circuit shows higher pressure values than the upper one. This phenomena could be correlated to the hydraulic resistance of the two circuits: in contrast to upper circuit, the bottom one has two ILT0 bioreactors joined in series to the bottom ILT2 chamber. This increases the resistance to flow, a higher pressure is required to obtain the same flow rate in comparison to the upper circuit. In order to verify this hypothesis ILT2 bioreactors should be characterized in the absence of ILT0 chambers. In this way the upper circuit should have the same hydraulic resistance as the bottom one. Table 4.5 shows pressure measurements of ILT2 on its own.
Inlet Outlet
Average top units 394.59 386.38
Average Bottom 384.80 374.44
Table 4.5 Summary of pressure measurements of the inlet and outlet of the upper and bottom chamber of ILT2
From the table it is apparent that in the presence of ILT0 (in the complete circtuit), the bottom pressure is higher than the top pressure. This is because the pressure of 2 ILT0s impose a higher hydraulic resistance. Pressure is then a parameter correlated to the topology of the hydraulic circuit.
80
4.3.3 MCmB 3.0 validation
As described in paragraph 4.3, tests of solute passage through membrane pores are performed in order to characterize also the MCmB 3.0 system. The strategy is always the same: the upper circuit is filled with a solution of known concentration, in the bottom circuit only the solvent flows. Experiments are performed imposing the same flow rate, 250 μL/min, in both circuits.
4.3.3.1 Methylene Blue
Once the membrane is wetted in deionised water in order to avoid any trapped air bubbles in the pores, it is placed gently on the bottom holder. The upper part is used to keep it in a fixed position, then it is possible to join the bioreactor upper and bottom chamber, using the clamp system to avoid any leakage. The hydraulic circuit set up and the operation to fill the bioreactor are the same as ILT2, described in [Appendix B]. Figure 4.26 shows data from the outlet of the bottom chamber.
Figure 4.26 Percentage of Methylene blue that flows through a commercial polycarbonate membrane. Sampling phase is at the outlet of MCmB 3.0 bottom chamber
0.00% 2.00% 4.00% 6.00% 8.00% 10.00% 12.00% 14.00% 16.00% 0 1 2 3 4 5 6 7 8 Pas sag e t h ro u gh m e m n b ran e [% ] Time [h]
81 As expected the percentage of solute that crosses the membrane is equal to that evaluated for ILT1 and ILT2 but in a shorter time. In fact in order to reach a percentage of passage between 10% and 15% 8 hours are necessary in MCmB 3.0, instead of 24 hours required with ILT bioreactors. Moreover it is not necessary to use a flow rate in the top chamber double that one of the bottom one. The characteristic time is correlated to a faster passage through the commercial membrane in MCmB 3.0 if compared to ILT bioreactors (figure 4.27). These differences are related to the membrane characteristics.
Figure 4.27 Curve fitting and characteristic time of Methylene blue that flows through a commercial polycarbonate membrane
Figure 4.28 shows a comparison between samples that derive from upper outlet (red triangle) and bottom outlet (blue spheres).
82 Figure 4.28 Passage of Methylene blue through a commercial membrane. Comparison between samples of the upper chamber (red triangles) and bottom chamber (Blue balls)
Figure 4.28 underlines that the situation after 8 h in MCmB 3.0 is the same as that one at 24 h in ILT bioreactors. It can be hypothesised that in 24 h in MCMB 3.0 it will be possible to find an equilibrium point between upper and bottom chambers, characterised by percentage of passage closed to 50 %. In order to increase the velocity of solute exchange between upper and bottom chambers membrane porosity or surface area exposed to flows should be increased.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0
2
4
6
8
P ass age through m em brane [% ] Time [h[83
4.3.3.2 Diclofenac
It is possible to estimate the kinetics of diclofenac passage by the membrane pores, sampling from the outlets of the MCmB 3.0. Figure 4.29 shows data sampled from the bottom chamber outlet.
Figure 4.29 Percentage of Diclofenac range through the membrane. Sampling isat the outlet of MCmB 3.0 bottom chamber
As shown in figure 4.29 diclofenac flows through membrane pores following similar kinetics to Methylene blue. Diclofenac passage is significantly higher and faster than the Methylene blue, as outlined by characteristic time shown in figure 4.30.
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
14.00%
16.00%
0
2
4
6
8
P assage thro ug h mem bra ne[% ] Time[h]84 Figure 4.30 Curve fitting and characteristic time
Figure 4.31 Passage of Diclofenac through the membrane. Comparison between samples of the upper chamber (red triangles) and bottom chamber (Blue balls)
Over the 8 hours duration of the experiment the system had not reach an equilibrium point represented by equal distribution of solute between chambers (figure 4.31).
0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00% 90.00% 100.00% 0 1 2 3 4 5 6 7 8 Pas sag e t h ro u gh m e m b ran e [% ] Time[h]
85
4.3.3.3 Urea
Experiments with Urea are performed in order to increase the characterization of the system, using a physiological product of digestion phase. Figure 4.32 describes data sampled from the bottom chamber outlet.
Figure 4.32 Percentage of urea passage through the membrane pores. The sampling phase is at the outlet of MCmB 3.0 bottom chamber
Urea flows through the membrane pores following a similar curve as compared to other type of solutes analyzed with MCmB 3.0. At the end of the experiment (8h) the percentage of passage through the membrane is 20 % of the starting concentration. This high value is due to the low molecular weight of urea compared with the other 2 solutes. Figure 4.33 shows the fitted data.
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
0
2
4
6
8
P ass age through m em brane[ % ] Time[h]86 Figure 4.33 Fitted data characteristic time of urea passage through the membrane pores
Passage of urea through the membrane is a rapid phenomena as it is outlined by the small characteristic time (figure 4.33).
Figure 4.34 Passage of Urea through the membrane. Comparison between samples of the upper chamber (red triangles) and bottom chamber (blue balls)
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
0
2
4
6
8
P ass age through m em brane[ % ] Time[h]87 Over the 8 hours duration of the experiment the system had not reach an equilibrium point of equal distribution of solute between chambers (figure 4.34). Characteristic times evaluated in each in-vitro test are summarized in table 4.6, in order to compare micro fabricated membrane in ILT1, ILT2 with commercial membrane in MCmB 3.0 and validate the previous discussions.
Micro fabricated membrane Commercial membrane
Methylene blue 9h (ILT2) 2h 57'
FITC-PS 5h (ILT1), 7h (ILT2) _
Urea _ 1h 49'
Diclofenac _ 2h 21'