Biomedical Microdevices (2019) 21: 29

https://doi.org/10.1007/s10544-019-0387-8

Influence of the static magnetic field on cell response

in a miniaturized optically accessible bioreactor for 3D cell culture
Luca Izzo 1 & Marta Tunesi 1 & Lucia Boeri 1 & Matteo Laganà 2 & Carmen Giordano 1 & Manuela Teresa Raimondi 1

Published online: 13 March 2019

# The Author(s) 2019

Abstract
Hydraulic sealing is a crucial condition for the maintenance of sterility during long term operation of microfluidic
bioreactors. We developed a miniaturized optically accessible bioreactor (MOAB) allowing perfused culture of 3D
cellularised constructs. In the MOAB, the culture chambers are sealed by magnets that generate a weak static
magnetic field (SMF). Here, we predicted computationally the exact level of SMF to which cells are subjected
during culture in the MOAB and we assessed its influence on the viability, metabolic activity and gene expression
of neuroblastoma-derived cells cultured up to seven days. The predicted SMF ranged from 0.32 to 0.57 T using an
axial-symmetric model of a single chamber, whereas it ranged from 0.35 to 0.62 T using a 3D model of the
complete device. Cell function was evaluated in SH-SY5Y neuroblastoma cells at 2 and 7 days of culture in the
MOAB, compared to 2D monolayer, 3D non-perfused constructs, and 3D perfused constructs cultured in a modified
MOAB with magnet-free sealing. We measured the cell metabolic activity normalized by the DNA content and the
expression levels of heat-shock protein 70 (Hsp-70), Bcl-2 and Bax. We found that the level of SMF applied to cells
in the MOAB did not influence their metabolic activity and exerted a stressful effect in 2D monolayer, not con-
firmed in 3D conditions, neither static not perfused. Instead, the magnets provided a significantly greater hydraulic
sealing in long-term culture, thus the MOAB might be potentially exploitable for the development of reliable in vitro
models of neurodegeneration.

Keywords MOAB . Static magnetic field . Perfusion bioreactors . Numerical characterization . 3D cell culture . Cell metabolic
activity . Gene expression

1 Introduction absence of a continuous flow of medium does not provide a

continuous nutrient supply and waste removal (Lv et al.
In vitro cell culture on two-dimensional (2D) glass or plas- 2017). Oppositely, in vitro cell modelling with miniaturized
tic substrates fails to model the complex physiological bioreactors shows great advantages. It requires small vol-
three-dimensional (3D) environment, enable cell differentia- umes of reagents and cells, both critical elements for valu-
tion and reproduce in vivo cell behavior; furthermore, the able samples and high-throughput screening. Portability, de-
sign versatility, potential for parallel operations and integra-
Luca Izzo and Marta Tunesi contributed equally to this work tion with existing devices or platforms are further advan-
tages (Lübberstedt et al. 2015). For instance, Lei and col-
Electronic supplementary material The online version of this article
leagues (Lei et al. 2014) provided an interesting example of
(https://doi.org/10.1007/s10544-019-0387-8) contains supplementary
material, which is available to authorized users. a microfluidic chip coupling both 3D microenvironment and
perfusion by perfusing human oral cancer cells embedded
* Luca Izzo in an agarose gel at a flow rate of 10 μL/h in a culture
luca.izzo@polimi.it chamber of 4x2x1 mm.
Optical accessibility represents a considerable improve-
1
Department of Chemistry, Materials and Chemical Engineering BG. ment, since it allows for the reduction of samples and the
Natta^, Politecnico di Milano, Piazza Leonardo da Vinci 32, analysis of the same constructs with time by non-destructive
20133 Milan, Italy techniques like viable staining and standard fluorescence mi-
2
Gemma Prototipi Studio, Erba, Italy croscopy. As an example, Kim and co-workers (Kim et al.
29 Page 2 of 12 Biomed Microdevices (2019) 21: 29

2012) applied a miniaturized optically accessible perfusion silicone gasket. It prevents the movements of the constructs by
bioreactor with the possibility to host 3D tissues for in vitro mechanical interference. By changing the shape of the gasket,
modelling of the human gut. it is possible to host differently shaped constructs (e.g. scaf-
In this context, Laganà and Raimondi (2012) developed a folds, hydrogels) in the culture chambers. The whole device
miniaturized optically accessible bioreactor (MOAB) for the may be sterilized with hydrogen peroxide gas plasma systems
interstitial perfusion of 3D cell constructs. To simplify the (e.g. STERRAD® 100S, ASP, Jonhson & Jonhson, Irvine,
assembly procedure while reducing the time required, we have CA, USA). The MOAB is a versatile and tunable platform,
optimized their initial prototype. The current device (Fig. 1, already validated for advanced in vitro cell modelling in sev-
top) is composed of three independent and magnetically lock- eral research fields, including neuroscience (Tunesi et al.
able chambers (9 mm3) assembled on the top surface of a 2016) and cancer (Marturano-Kruik et al. 2018). Thanks to
common main body (68 × 25 mm) made up of medical grade its optical accessibility, it was also applied for in vitro tracking
polystyrene. The main body has rigid edges to reduce the of exosomes in a model of gene therapy for muscular dystro-
optical path and enhance sample illumination during optical phy (Frattini et al. 2017).
transmission microscopy. The first magnet (NdFeB ring mag- The magnets sealing the culture chambers of the MOAB
net, 12 mm outer diameter, 9 mm internal diameter, 1.5 mm generate a static magnetic field (SMF) that might influence
thick) is located in the chamber, while the second one is in the cell functions. SMFs interact with biological systems by both
bioreactor body. Their magnetic coupling (closure force of electrodynamic and magneto-mechanical effects, meaning
14.7 N) ensures the hydraulic sealing during the perfusion of that moving ionic charges induce an electric potential because
3D constructs (6x3x0.4mm) and simplifies the assembly pro- of the Lorentz force and both diamagnetic and magnetic sub-
cedure, allowing for the self-centering and self-aligning of the stances are exposed to a torque and orient, respectively (Roth
locking system with respect to the feed channels. Each cham- 2011). SMFs are naturally present: the Earth itself generates a
ber has a 9 mm glass coverslip equipped with a medical grade magnetic field ranging between 25 and 65 μT and involved in

Fig. 1 Views and technical sketches (AutoCAD® Software, (purple), the magnet (grey) and the glass coverslip (light blue). e Top view
Autodesk, San Rafael, California, USA) of the miniaturized of the MOAB without magnetic closure. f Technical sketch showing the
optically accessible bioreactor (MOAB) with magnetic (upper panel) MOAB without magnetic closure. g Top: top view of one culture chamber
or snap-fit closure (lower panel). a Top view of the MOAB with without magnetic closure; Bottom: side sketch of one culture chamber
magnetic closure. b Technical sketch showing the MOAB with without magnetic closure. h Technical sketch showing the lid (green), the
magnetic closure. c Top: top view of one culture chamber with gasket (purple) and the glass coverslip (light blue) in the MOAB without
magnetic closure; Bottom: side sketch of one culture chamber with magnetic closure. Scale bar 5 mm
magnetic closure. d Technical sketch showing the lid (green), the gasket
Biomed Microdevices (2019) 21: 29 Page 3 of 12 29

the orientation and migration of some animal species neurodegenerative disorders. MINERVA goal is to develop
(Feychting 2005). Industrial activities exploiting direct cur- a cutting-edge technological platform, based on organ on
rents also generate SMFs (e.g. in aluminum production or chip microfluidic device, to model the main players of the
chloralkali plants, workers are exposed to SMFs varying from microbiota-brain axis connection. In this context, with a view
4 to 50 mT (European Commission 1996)). Therapeutic de- to exploit the improved version of the MOAB also as the
vices (e.g. prosthetic cardiac pacemaker, defibrillator, standard basic functional unit for the development of reliable in vitro
orthodontic devices, Holter) also produce SMFs up to 10 mT. models of neurodegeneration in MINERVA project, in this
In particular, defibrillators and standard orthodontic devices work we focused on predicting the SMF generated by its
expose the patients to a SMF for very short periods, but the magnets and measuring its effects on viability, metabolic
operators for frequent times. For medical and research pur- activity and gene expression in SH-SY5Y neuroblastoma
poses, stronger SMFs may be applied. For instance, for mag- cells, a commonly used cell model in the research against
netic resonance imaging (MRI) the magnetic field can reach neurodegeneration. As a control, we used MOABs with a
intensities up to 3 T (Hartwig et al. 2018), while the operators non-magnetic lock (e.g. snap-fit closure, Fig. 1, bottom), a
are exposed to a magnetic field varying from 0.5 to 2 mT custom-made version of the device where the magnets are
(Zannella 1997). not present and the hydraulic sealing is assured by the me-
Valiron and colleagues (Valiron et al. 2005) observed chanical interference between the lids (showing a bulge) and
that exposure to high magnetic fields (over 10 T and 15 T the main body of the bioreactor.
for cycling cells and neurons, respectively) for 30–60 min
affects cell cytoskeleton, with deleterious effects on cell
viability (e.g. detachment from culture dishes), organiza- 2 Materials and methods
tion and differentiation. A reduction in cell viability was
also reported when decreasing the intensity of the SMF 2.1 Numerical prediction of the level of magnetic field
and increasing the exposure time. For example, Raylman
and co-workers (Raylman et al. 1996) obtained a reduction To study the distribution of the intensity of the SMF generated
in the number of viable HTB 63, HTB 77 IP3 and CCL 86 by the magnetic closures, we performed a numerical analysis
malignant human cell lines placed for 64 h in the isocenter with the software COMSOL Multiphysics® (Burlington, MA,
(~5 cm diameter) of a superconducting solenoid magnet USA), version 5.2. We exploited both an axial-symmetric and
generating a 7 T uniform SMF. Similarly, Ji and col- a 3D complete model and compared their results. In the first
leagues (Ji et al. 2009) observed a decrease in the number model, we considered only one chamber (thus hypothesizing
of E. coli colony-forming units after exposures up to that they are independent), while in the second one we includ-
60 min to two different permanent magnets (magnetic ed all the three chambers (thus taking into account their pos-
field induction between 45 and 450 mT and 0.45–3.5 T. sible reciprocal influences).
In the latter configuration, the maximum space for expo- The geometry of the 3D complete model (Fig. 2) was
sure was 17x17x10 cm). To study the effects of the inten- represented by three couples of permanent ring magnets
sity of the SMF that patients are exposed to during MRI, (Supermagnete, Gottmadingen, Germany) identical to
Zhang and co-workers (Zhang et al. 2017) plated 15 dif- those in the MOAB. We reported their properties in
ferent cell lines on the top surface of a magnet (1 T, 5x5x5 Table 1. We chose neodymium as the material for the
cm), reporting that the SMF affects cell proliferation in a domains of magnetic coupling. To simulate the external
cell type- and density-dependent manner. The dependence environment and visualize the distribution of the mag-
on cell type was also suggested by Aldinucci and col- netic field intensity around the chambers, we assumed
leagues (Aldinucci et al. 2003), showing that the combi- an air-made cylinder of 50 mm in diameter and 50 mm
nation of a static electromagnetic field (4.75 T) with a in height, with a magnetic insulation for the external
pulsed electromagnetic field (0.7 mT) generated by an surface. Its isocenter corresponded to the center of the
NMR apparatus had neither proliferative, nor activating central magnetic chamber; while the centers of the two
or proinflammatory effect on lymphocytes after 1 h-expo- lateral chambers were 16 mm far. This distance was in
sure, while it affected proliferation in Jurkat cells. accordance with the geometry of the MOAB. We chose
This work takes place inside two novel technological pro- the dimensions of the cylinder to observe a null value
jects, named MOAB (ID number 825159) and MINERVA of magnetic field intensity on its boundaries.
(ID number 724734) supported by the European Research The properties of neodymium and air were already
Council (ERC) Programme. In particular, MINERVA project present in COMSOL Multiphysics® database, except
aims, by using an innovative bioengineering approach, at for the relative permeability.
evaluating the potential impact of human gut microbial com- For both models, we inserted a value of 1.04 and 1 for
munity on central nervous system functionality also in neodymium and air, respectively. In addition, the Maxwell
29 Page 4 of 12 Biomed Microdevices (2019) 21: 29

Fig. 2 COMSOL Multiphysics® modeling of the magnetic coupling air-made cylinder with two coupled neodymium-based ring magnets in
of the MOAB. Geometry of the COMSOL Multiphysics® models for the the isocenter (axial-symmetric model) or six coupled neodymium-based
three chambers of the MOAB. Left: Enlargement of a single couple of ring magnets (two in the isocenter and four 16 mm far, 3D complete
magnets and geometry of the exposure zone. Right: To visualize the model)
distribution of the intensity of the static magnetic field, we assumed an

equations modelling the physical phenomenon were already assumed were the temperature (293.15 K) and the absolute
present in the tool Magnetic Fields No Currents: pressure (1 atm). An axially magnetized magnetic couple
was present in the problem; therefore, we selected the correct
H ¼ −∇Vm ð1:1Þ domains to align the magnetized faces perpendicularly to the
where H is the magnetic field intensity and Vm is the mag- axial direction. We also assumed absence of radial magnetiza-
netic scalar potential; tion. To estimate the intensity of magnetic induction field as a
function of the distance from the symmetry axis of the mag-
∇  ðμ0 μr H þ Br Þ ¼ 0 ð1:2Þ net(s), for both models we chose the option very dense mesh
for both the magnet(s) and the air-made cylinder. We extracted
where μ0 is the vacuum permeability, μr is the relative perme- the results of both models in a colour map showing the distri-
ability of the material and Br is the residual magnetism (as- bution of the magnetic induction field around the magnet(s).
sumed 2.74 T, that is 1.37 T multiplied by two, since we
considered two coupled magnets). Other input parameters
2.2 Hydraulic characterization of magnetically
Table 1 Properties of NdFeB magnets
and non-magnetically lockable MOABs

Property Value 2.2.1 Static leakage tests

Material NdFeB
To assess the capability of both magnetically and snap-
Cross section shape Rectangular
fit-lockable MOABs to face a pressure increase, we per-
Outer diameter 12 mm
formed static leakage tests. We used a pressure regulator
Internal diameter 9 mm
to provide pressure steps of 0.02 MPa to a distilled
Height 1.5 mm
water reservoir connected to the inlet channel of each
Tolerance ± 0.1 mm
culture chamber by gas permeable tubes (internal diam-
Closure force 14.7 N
eter: 0.03″, outer diameter: 0.065″; Silastic® Tubing,
Direction of magnetization AXIAL (that is parallel to the height) Cole-Parmer, Vernon Hills, IL, USA). We tested the
Coating Ni-Cu-Ni three chambers one by one, while clamping the outputs
Magnetization N45 to allow for the pressure increase to the critical value.
Weight 0.56 g We maintained each pressure step for 30 s. We mea-
Residual magnetism 1.37 T sured the maximum pressure value when the first water
Max temperature 80 °C drop started coming out from the examined chamber.
List of the features of each ring magnet (Supermagnete, Gottmadingen,
We tested two magnetically lockable and two snap-fit
Germany) integrated in the microfluidic optically accessible bioreactor lockable MOABs and repeated the test three times for
(MOAB) with magnetic closure each chamber.
Biomed Microdevices (2019) 21: 29 Page 5 of 12 29

2.2.2 Hydraulic resistance Waltham, MA, USA) and split twice a week. We examined the
effects of the magnetic field generated by the NdFeB ring mag-
To characterize the hydraulic resistance of the fluid pathway, nets in both 2D and 3D (polystyrene scaffolds) conditions.
we used a precision pressure regulator to provide pressures
from 0.02 MPa to 0.1 MPa (with steps of 0.02 MPa) to a 2.2.4 Influence of the magnetic field on cells in 2D culture
distilled water reservoir connected to the inlet channels of
the culture chambers. We connected the output channels to a We studied the influence of the SMF on SH-SY5Y cells in 2D
reservoir placed on a precision electronic balance (AP250D, culture after both short (48 h) and longer exposures (7 days).
Ohaus, Greifensee, Switzerland) and tested the three chambers For the first condition, we plated 93,750 cells/cm2 in the cen-
one by one. We maintained each pressure step for 30 s, then ter of 35 mm glass bottom dishes with 7 mm diameter center
we clamped the inlet channel and calculated the flow rate from openings (MatTek Corporation, Ashland, MA, USA); while
the measured weight and the perfusion time. We tested two for the others, we plated 31,250 cells/cm2. These concentra-
magnetically lockable and two snap-fit lockable MOABs and tions allowed for having enough cells after 48 h and avoiding
repeated the test three times for each input pressure value. To over confluence after 7 days.
account for their hydraulic resistance, each chamber hosted a Since our predictions suggested that the SMF generated by
polystyrene scaffold (6x3x0.4 mm, 3D Biotek, Hillsborough, a single couple of NdFeB magnets is a reliable approximation
NJ, USA; Fig. 3, top) identical to those for cell experiments. It of the SMF to which the cells are exposed in the culture cham-
showed four layers of fibers (diameter: 100 μm, pore size: bers (regardless of their position in the MOAB), we performed
300 μm) shifted of 150 μm with respect to the adjacent one. experiments in 2D conditions by reproducing a single cham-
ber in a custom-made setup, also suitable for confocal micros-
2.2.3 Cell culture copy. To prevent the corrosion of the magnets when in contact
with medium and in humidified atmosphere, we coated the
We cultured SH-SY5Y human neuroblastoma cells (Izsler, code magnets with High Temp Resin (Formlabs, Somerville, MA,
BS TLC 232) at 37 °C and 5% CO2 in high-glucose Dulbecco’s USA). We disinfected the samples by 70% (v/v) ethanol,
modified Eagle’s medium supplemented with 10% (v/v) fetal washed extensively with distilled water and sterilized by UV
bovine serum, 2 mM L-glutamine, 100 U/mL penicillin, irradiation (254 nm) for 1 h. One day after seeding, we placed
0.1 mg/mL streptomycin sulfate (ThermoFisher Scientific, our setup in the center of the cell-seeded glass bottom dishes.

Fig. 3 Axial-symmetric model of the magnetic field generated in a (diameter: 100 μm, with a pore size: 300 μm) shifted of 150 μm with
single culture chamber. a Intensity of magnetic induction field with the respect to the adjacent. A sketch of the positioning of the scaffolds in the
distance from the symmetry center of the magnets (blue line). The black culture chambers is also shown. b Color map showing the distribution of
rectangle highlights the dimensions of polystyrene scaffolds the magnetic induction field around the magnets (top view) and side view
(6x3x0.4 mm) identical to those for cell experiments. They are obtained of the arrow field. A sketch of the positioning of the scaffolds in the
by fused deposition modeling and composed of four layers of fibers culture chambers is reported
29 Page 6 of 12 Biomed Microdevices (2019) 21: 29

As a control, we cultured SH-SY5Y cells in the absence of 1. Presence of the SMF, dynamic culture: we cultured the
NdFeB magnets in the same glass bottom dishes. scaffolds for 48 h or 7 days in dynamic conditions
After 48 h and 7 days, we evaluated cell metabolic activity (MOAB with magnetic closing, flow rate: 0.5 μL/min;
by resazurin (Sigma-Aldrich, St. Louis, MO, USA) assay by PHD ULTRA programmable syringe pump, Harvard
measuring the fluorescence at 590 nm (excitation wavelength Apparatus, Holliston, MA, USA);
560 nm, Infinite M200PRO, Tecan, Männedorf, Switzerland). 2. Absence of the SMF, dynamic culture: we cultured the
At the end of this analysis, we assessed the number of viable scaffolds for 48 h or 7 days in dynamic conditions in the
SH-SY5Y cells in the samples. We detached the cells from absence of the magnetic field (bioreactor without magnet-
culture plates by trypsin, diluted with Trypan blue dye and ic closing, flow rate: 0.5 μL/min);
counted. For each condition, we counted at least four squares. 3. Presence of the SMF, static culture: we cultured the scaf-
Starting from a previous study reporting that every human folds for 48 h or 7 days in the presence of the magnetic
diploid cell contains about 6.4 pg DNA (Dolezel et al. field in static conditions (inside the custom-made setup
2003), we exploited these results to calculate the total content described in the previous paragraph);
of DNA and the specific metabolic activity (that is cell meta- 4. Absence of the SMF, static culture: we cultured the scaf-
bolic activity normalized to the DNA content of each sample). folds for 48 h or 7 days in the absence of the magnetic
On day 7, we performed the live/dead assay (ThermoFisher field in static conditions (inside a low-attachment culture
Scientific). Medium was removed, and cells were washed plate).
with phosphate-buffered saline before incubation with
0.5 μM ethidium homodimer-1 and 0.4 μM calcein AM. After 48 h and 7 days, we repeated the analyses described
Finally, samples were imaged by a confocal microscope for 2D cultures. We analyzed the mRNA expression levels of
(Fluoview FV10i, Olympus, Tokyo, Japan). Hsp-70, Bcl-2 and Bax for groups 1, 3 and 4.
On day 7, we also evaluated the effect of the SMF on these
target genes: Heat Shock Protein 70 (Hsp-70), Bcl-2 and Bax. 2.2.6 Statistical analysis
Hsp-70 is an anti-apoptotic molecule that regulates several
steps of the apoptotic cascade (Kong et al. 2016), while Bcl- We reported the results as mean ± standard deviation (SD). We
2 has a role in inhibiting cell death. Oppositely, Bax is a pro- performed the statistical analysis with GraphPad Prism® soft-
apoptotic molecule (Pawlowski and Kraft 2000). We per- ware (GraphPad Software, La Jolla, CA, USA). We used two-
formed the analyses on samples previously subjected to way analysis of variance (ANOVA) followed by Tukey’s mul-
resazurin assay. We extracted the total cellular RNA with tiple comparisons test for comparisons among the groups and
miRNeasy Mini Kit (Qiazol™, Qiagen, Hilden, Germany), time frames, while we used one-way ANOVA followed by
according to the manufacturer instructions. We evaluated the Dunnett’s multiple comparisons test for comparisons among
quality of the extracted RNA via spectrophotometry (ND- the groups. The significance level was set at p < 0.05.
1000; NanoDrop™, ThermoFisher Scientific). We synthe-
sized the cDNA starting from 500 ng of total RNA with the
High-Capacity cDNA Reverse Transcription Kit (Applied 3 Results
Biosystems™, ThermoFisher Scientific). We used the cDNA
to measure the relative mRNA expression levels of Hsp-70, 3.1 Numerical prediction of the level of magnetic field
Bcl-2 and Bax via quantitative real-time polymerase chain
reaction (qRT-PCR) with TaqMan™ Reagents (Applied Figure 3 and Fig. 4 show the results from the analysis with
Biosystems™). We used 18 s rRNA as a reference gene. For COMSOL Multiphysics® for the axial-symmetric and the 3D
each sample, we normalized the expression of target genes to complete model, respectively. The black rectangle highlights
the expression level of 18 s rRNA. Among the groups, we the dimensions of the polystyrene scaffold within the culture
normalized the expression ratio of magnetic samples to the chambers. For both models, according to Biot-Savart law, we
corresponding control. observed a quadratic decay for the intensity of magnetic in-
duction field in the area occupied by the scaffold. A magnetic
field is present outside the rings, but its values rapidly reach
2.2.5 Influence of the magnetic field on cells in 3D culture zero while increasing the distance from the symmetry center.
With the axial-symmetric model, considering only one of
We plated SH-SY5Y cells in polystyrene scaffolds, as reported the three chambers, we predicted that cell constructs are ex-
(Tunesi et al. 2016). After seeding, we kept the scaffolds in posed to a SMF ranging from 320 mT to 570 mT. The arrow
low-attachment culture plates for 2 days, and then we evaluated field lines showed the two magnetic poles of the coupling and
the effects of the SMF on SH-SY5Y cells both in dynamic and confirmed the non-zero value of the magnetic field in the
static conditions. We examined the following situations: center of the culture chamber.
Biomed Microdevices (2019) 21: 29 Page 7 of 12 29

Fig. 4 Complete 3D model of the magnetic field generated by the symmetry center of the magnets for the central chamber (blue line). The
three chambers. a Color map showing the distribution of the magnetic maximum estimated value was 620 mT. c Intensity of magnetic induction
induction field around the magnets (top view). The black rectangles field as a function of the distance from the symmetry center of the
highlights the dimensions of polystyrene scaffolds. b Intensity of magnets for the lateral chambers (blue line). The maximum estimated
magnetic induction field as a function of the distance from the value was 620 mT

With the 3D complete model, considering all the chambers In addition, for both MOABs we did not observe breaking
and their possible reciprocal influences, we predicted that the while increasing the pressure.
magnetic field intensity ranges from 350 mT to 620 mT in the
lateral chambers and from 370 mT to 620 mT in the central 3.2.2 Hydraulic resistance
one. Therefore, when the two other chambers are included in
the model, in each lateral chamber the intensity of the SMF Figure 5b shows the results related to the hydraulic resistance
increases by less than 10%. For the central chamber, the min- of the fluid pathway. For both magnetically and snap-fit lock-
imum intensity of the SMF increases with respect to the lateral able MOABs, the flow rate increased while raising the pres-
ones, but the maximum value is unchanged, suggesting that sure. For a pressure of 0.02 MPa we observed a greater flow
the influence of adjacent chambers is negligible. Therefore, a rate (mean ± SD) for the snap-fit bioreactors (1.8 ± 0.08 vs 3.5
single couple of NdFeB magnets and an axial-symmetric ± 0.01 mL/min, *, p < 0.05); while for a pressure of 0.04 MPa
model is suitable to estimate the intensity of the SMF in the we did not find differences between the two configurations
scaffolds in the magnetically lockable MOABs with a reliable (ns, p > 0.05). At this pressure, we recorded a flow rate of
approximation. 6.0 ± 0.01 and 6.0 ± 0.04 mL/min for the magnetic and snap-
Since our predictions have suggested that in the culture fit MOAB, respectively. When we further increased the pres-
chambers cells are exposed to non-zero values of the SMF, sure, the magnetically lockable bioreactor offered a greater
effects on their viability, metabolic activity and gene expres- flow rate (**, p < 0.01 when we set pressure at 0.06 MPa;
sion might be observed. ****, p < 0.0001 when we set pressure at 0.08 and
0.1 MPa). At these pressures, we recorded a flow rate of
3.2 Hydraulic characterization of magnetically 10.0 ± 0.82 and 8.0 ± 0.41 mL/min; 14.0 ± 0.82 and 9.5 ±
and non-magnetically lockable MOABs 0.41 mL/min; 18.0 ± 0.82 and 14.5 ± 0.41 mL/min for the
magnetic and snap-fit MOAB, respectively. These values are
3.2.1 Static leakage tests greater than those applied for cell perfusion.

Figure 5a shows the results from static leakage tests. We did 3.2.3 Influence of the magnetic field on cells in 2D culture
not find any difference (ns, p > 0.05) between MOABs with
magnetic and snap-fit closing, with values (mean ± SD) of Figure 6 shows the experimental set-up and the results related to
(0.110 ± 0.017) MPa and (0.122 ± 0.035) MPa, respectively. the effect of the magnetic field on the viability, metabolic activity
29 Page 8 of 12 Biomed Microdevices (2019) 21: 29

Fig. 5 Hydraulic characterization of MOABs with magnetic and ± SD (6 replicates/group). We performed the statistical analysis with one-
snap-fit closing. a Static leakage tests. We reported the results as mean way ANOVA followed by Tukey’s multiple comparisons test. ns,
± SD (6 replicates/group). We performed the statistical analysis with t- p > 0.05; *, p < 0.05; **, p < 0.01, ****, p < 0.0001
test. ns, p > 0.05. b Hydraulic resistance. We reported the results as mean

and gene expression of SH-SY5Y cells in 2D monolayer. We (but also on standard tissue culture-treated dishes, TCP) cul-
exposed the cells to the SMF one day after seeding because we tured in the absence of the magnets (ns, p > 0.05), suggesting
noticed that its presence impaired cell adhesion, especially in the no effects of the magnetic field generated by the NdFeB mag-
case of the lowest cell density (31,250 cells/cm2). nets in the examined time window. Live/dead assay and con-
For both metabolic activity and specific metabolic activity, focal microscopy confirmed these results. Fig. 6c reports an
we did not find differences with controls on glass substrates example image: the great majority of SH-SY5Y cells cultured

Fig. 6 Influence of the magnetic field on SH-SY5Y cells in 2D cul- magnetic field. Scale bar: 20 μm. d Specific metabolic activity (that is
ture: specific metabolic activity. a Experimental set-up to evaluate the metabolic activity normalized with respect to the DNA content) of SH-
effect of the magnetic field generated by the ring magnets of the MOAB SY5Y cells with time. For the shortest exposure time (2 days), we plated
on SH-SY5Y cells. Scale bar: 5 mm. b Metabolic activity of SH-SY5Y 93,750 cells/cm2; while for the other exposure times, we plated 31,250
cells with time. For the short exposure time (2 days), we plated 93,750 cells/cm2. In all the graphs, we showed the results from resazurin assay (3
cells/cm2; while for the long exposure time (7 days), we plated 31,250 replicates/group). We reported the results as mean ± SD. We performed
cells/cm2. TCP indicates SH-SY5Y cells plated on standard tissue the statistical analysis with two-way ANOVA followed by Tukey’s mul-
culture-treated dishes. c Example of a confocal microscopy image show- tiple comparisons test. ns, p > 0.05; *, p < 0.05; **, p < 0.01; ****, p <
ing live (stained green by calcein AM) and dead (stained red by ethidium 0.0001
homodimer-1) SH-SY5Y cells after 48 h culturing in the absence of the
Biomed Microdevices (2019) 21: 29 Page 9 of 12 29

for 7 days in glass bottom dishes was alive (stained green by conditions; MOAB-dynamic conditions) and control samples
calcein AM). Only a few dead cells (stained red by ethidium (No Mag, static conditions) with respect to SH-SY5Y cells in
homodimer-1) were visible. 2D conditions. The expression levels of the anti-apoptotic
Figure 7 reports the mRNA expression levels of the Hsp70 did not decrease (ns, p > 0.05) and the expression
anti-apoptotic Hsp70 and Bcl-2 and the pro-apoptotic levels of the pro-apoptotic Bax did not increase (ns, p >
Bax after 7-days exposure. It shows a lower expression 0.05) in magnetic samples with respect to controls.
of Hsp70 (*, p < 0.05) and a higher expression of Bax However, the expression levels of Bcl-2 were significantly
(**, p < 0.01) in SH-SY5Y cells exposed to the magnetic lower in the presence of the SMF (**, p < 0.01 for Mag sam-
field with respect to controls on glass substrates. We did ples; ****, p < 0.0001 for Mag samples), suggesting a stress-
not detect significant differences in the mRNA expres- ful effect that does not reduce cell-specific metabolic activity.
sion levels of Bcl-2 (ns, p > 0.05).

3.2.4 Influence of the magnetic field on cells in 3D culture 4 Discussion

Figure 8 shows the results related to the effect of the magnetic This study aimed at assessing the effects of the SMF generated
field on the metabolic activity and specific metabolic activity by the magnetic closure of the optimized prototype of a recently
of SH-SY5Y when in 3D scaffolds both in dynamic or static developed miniaturized bioreactor (Laganà and Raimondi 2012)
conditions. on the metabolic activity, viability and gene expression of target
After 2 days in static conditions, we did not observe effects genes (Hsp-70, Bcl-2 and Bax) in human SH-SY5Y neuroblas-
on metabolic activity and specific metabolic activity. After toma cells. Our results indicated that: 1) in the MOAB with
7 days, we measured a significant difference in metabolic magnetic closing, cells are exposed to non-zero values of SMF;
activity (***, p < 0.001), but it was recovered when calculat- 2) the maximum value of the SMF estimated with the 3D com-
ing the specific metabolic activity (ns, p > 0.05). plete model (620 mT) is independent on the chamber position
In dynamic conditions, again we observed a statistical dif- (lateral or central); 3) the maximum value of the SMF estimated
ference (***, p < 0.001) in both metabolic activity and specif- by the axial-symmetric model (570 mT) is reduced by less than
ic metabolic activity after 7 days of exposure to the SMF. 10% with respect to the SMF estimated by the 3D complete
However, the situation was reversed with respect to the static model (620 mT); 4) the SMF generated by the NdFeB ring
condition, with the cells exposed to magnetic field exhibiting magnets does not reduce cell-specific metabolic activity after
greater values than controls. 48 h and 7 days in 2D and 3D conditions (both static and dy-
Live/dead assay and confocal microscopy supported these namic); 5) the SMF exerts a stressful effect in 2D conditions, but
results. Fig. 8e shows an example image: the great majority of it decreases in 3D conditions (both static and dynamic).
SH-SY5Y cells cultured for 7 days in static conditions was Due to the very limited experimental accessibility of our
alive (stained green by calcein AM). Only a few dead cells miniaturized system to measurement of the magnetic field, we
(stained red by ethidium homodimer-1) may be detected. used a numerical analysis to predict the magnetic field gener-
Finally, confocal microscopy suggested a good degree of scaf- ated by the magnetic rings. In fact, the culture chambers are
fold colonization, reporting the presence of living cells at least very small (9 mm3) and it was not possible to find a suitable
on three of the four layers. magnetometer able to avoid any influence on the read-outs
Gene expression analyses (Fig. 9) showed different Hsp- while measuring. Moreover, a magnetometer is not able to
70, Bcl-2 and Bax profiles between magnetic (Mag-static show the distribution of the magnetic field. Peng and co-

Fig. 7 Influence of the magnetic field on SH-SY5Y cells in 2D cul- expression levels of Bax. We reported the results as mean ± SD, 3 repli-
ture: mRNA expression levels. a mRNA expression levels of heat shock cates/group. We performed the statistical analysis with t-test. ns, p > 0.05;
protein-70 (Hsp70). b a) mRNA expression levels of Bcl-2. c mRNA *, p < 0.05; **, p < 0.01
29 Page 10 of 12 Biomed Microdevices (2019) 21: 29

Fig. 8 Influence of the magnetic field on SH-SY5Y cells in 3D cul- the results as mean ± SD. We performed the statistical analysis with two-
ture: specific metabolic activity. a Metabolic activity of SH-SY5Y cells way ANOVA followed by Tukey’s multiple comparisons test. ns,
with time (in static conditions). b Metabolic activity of SH-SY5Y cells p > 0.05; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. e
with time (in dynamic conditions). c Specific metabolic activity (that is Example of confocal microscopy images showing live (stained green by
metabolic activity normalized with respect to the DNA content) of SH- calcein AM) and dead (stained red by ethidium homodimer-1) SH-SY5Y
SY5Y cells with time (in static conditions). d Specific metabolic activity cells after 48 h culturing in static conditions in the absence of the mag-
of SH-SY5Y cells with time (in dynamic conditions). In all the graphs, we netic field. Results from a z-stack acquisition. Scale bar 20 μm
showed the results from resazurin assay (3 replicates/group). We reported

workers described an analytical method to calculate the mag- Taking as a reference the intensity of Earth magnetic
netic field only for the central axis of an axially magnetized field (25–65 μT), we estimated greater values (320–620
ring (Peng et al. 2004), but to our knowledge no analytical mT) in the geometrical center of the rings. Therefore,
methods calculating the distribution of the magnetic field the assessment of a possible influence of the magnetic
around a magnetic ring exist. Instead of NdFeB, we consid- field on cell constructs was necessary.
ered neodymium as the material for the domains of magnetic In the last few years, several studies have focused on the
coupling because it is the strongest magnetic material of the influence of SMFs at the cellular level. They have reported a
alloy. In fact, neodymium magnets have a coercivity of about dependence on magnetic field intensity, exposure time, cell
four orders of magnitude greater than that of a material with type and density, but results are contradictory and a compari-
only a ferrous component. son is difficult because of the differences in experimental

Fig. 9 Influence of the magnetic field on SH-SY5Y cells in 3D cul- replicates/group. We performed the statistical analysis with one-way
ture: mRNA expression levels. a mRNA expression levels of heat shock ANOVA followed by Dunnett’s multiple comparisons test. ns, p > 0.05;
protein-70 (Hsp70). b a) mRNA expression levels of Bcl-2. c mRNA ***, p < 0.001; ****, p < 0.0001
expression levels of Bax. We reported the results as mean ± SD, 3
Biomed Microdevices (2019) 21: 29 Page 11 of 12 29

parameters and read-outs (e.g. cell number, cell metabolic reduction of this stressful effect. The significant lower levels of
activity, orientation of intracellular components, DNA tran- Bcl-2 in SH-SY5Y cells subjected to the SMF suggested a pos-
scription). Since changes in cell cycle or at intracellular level sible weak stressed condition that should be further investigated.
(e.g. growth factor signaling, DNA transcription) have a direct The hydraulic characterization performed in the present
impact on cell viability and functions (Albuquerque et al. study indicated that both the magnetically and snap-fit lock-
2016), in this work we focused on cell metabolic activity, able MOABs may stand pressure values higher than those
number and gene expression. For 2D studies, we preferred usually involved in cell perfusion experiments, where maxi-
dishes with glass bottom to standard polystyrene microplates mum flow rates are in the order of μL/min. However, at higher
because the latter does not allow for cell observation by con- pressures the flow rate offered by the bioreactor with magnetic
focal microscopy. Since the multi-layer nickel-copper-nickel closing is significantly greater. This is probably due to the
plating may not be sufficient to prevent the corrosion of neo- better hydraulic sealing of the magnetic coupling with respect
dymium magnets for all applications, we covered NdFeB to the simple snap-fit closure for mechanical interference. Cell
magnets with a protective layer of High Temp Resin after proliferation inside the scaffolds increases the mechanical re-
assessing its biocompatibility with SH-SY5Y cells sistance to the fluid flow within the chambers. To avoid leak-
(Online Resource 1). age and guarantee a suitable perfusion to cell constructs dur-
In both 2D and 3D conditions, we observed that the SMF ing experiments up to 7 days, we inserted screws and cable
does not reduce cell-specific metabolic activity after 48 h and ties on non-magnetically bioreactors. Despite these precau-
7 days of exposure. For cells in 2D conditions, Miyakoshi tions, the specific metabolic activity was significantly lower
(2005) published that most of studies reported no significant than in magnetic dynamic samples (p < 0.0001), suggesting
effects of SMF on cell viability. Similarly, Romeo and co- that our snap-fit bioreactors are not suitable for perfusion ex-
workers (Romeo et al. 2016) reported lack of effects on cell periments longer than 48 h. For these reasons, we did not
viability when exposing MRC-5 human lung fibroblasts to a perform genetic assays on these samples.
SMF of 370 mT. Oppositely, biological effects were observed However, the MOAB with snap-fit closure would remain a
when the magnetic field was coupled to stimuli like x-irradiation forced choice for magnetic particle tracking, apart from their
and in the literature there is evidence that SMFs can affect several dimensions and charge.
endpoints when in the mT range. For SH-SY5Y cells exposed
for 24 h to 2.2 mT SMF, Calabrò and colleagues (Calabrò et al.
2013) observed a decrease of membrane mitochondrial potential 5 Conclusions
up to 30% and an increase in the production of reactive oxygen
species. On the other end, Stolfa and co-workers (Stolfa et al. We predicted the values and distribution of the SMF gen-
2007) reported an increase in viability for human chondrocytes erated by the NdFeB ring magnets in the culture chambers
exposed for 72 h to a SMF of 600 mT. To our knowledge, this is of the MOAB by a numerical analysis, indicating that the
the first study examining the effects of the SMF on SH-SY5Y intensity of the magnetic field has a quadratic decay trend
cells in 3D conditions, therefore no direct comparison with data along the radial direction of the culture chamber, with
in the literature is possible. very low values affecting all the central areas of the cul-
We deepened our investigations with genetic assays and tured scaffold. Furthermore, the maximum intensity of the
analysed the expression profile of Hsp70, Bcl-2 and Bax. magnetic field generated in each chamber of the bioreactor
Hsp70 can be activated when cells are exposed to particular is not influenced by the adjacent chamber(s). The hydrau-
stressors and it represents one of the actors of the apoptosis lic characterization recorded a significantly higher leakage
regulation. Bcl-2 is an anti-apoptotic protein that inhibits cell for non-magnetically lockable MOABs compared to the
death by regulating the Bcl-2-regulated apoptotic pathway, while standard magnetic ones. We found that after 2 and 7 days
Bax has a pro-apoptotic role in the events triggering programmed of cell exposure to the SMF generated by the magnetic
cell death. We selected these genes from a previous work closure of the MOAB, the cell-specific metabolic activity
(Tenuzzo et al. 2009). Here, the authors evaluated the biological of SH-SY5Y cells is not reduced, neither in 2D monolay-
effects of a SMF of 6 mT on aged human lymphocytes. Similarly er, nor in 3D static culture, nor in the MOAB. Instead, the
to Tenuzzo and colleagues, we demonstrated that the expression expression levels of Hsp70, Bcl-2 and Bax measured in
of Hsp70 decreases and the expression of Bax increases after 7- cells at seven culture days shows that this level of SMF
day exposure to our SMF in 2D conditions. Together with the exerts a stressful effect in 2D monolayer, which decreases
results from cell metabolic activity, these results suggest that the to a negligible level in 3D static culture, and in the
SMF generated by the NdFeB ring magnets is not stressful MOAB. Taken together, these results suggest that
enough to reduce cell-specific metabolic activity. In 3D dynamic MOAB device with magnetic closure might be potentially
conditions (MOAB samples), Hsp70 and Bax expression was used also as the basic functional unit of the ERC
comparable with controls (No Mag samples), suggesting a BMINERVA^ multi-organ bioengineered platform.
29 Page 12 of 12 Biomed Microdevices (2019) 21: 29

Acknowledgments This study was funded by the European Research F. Kong, H. Wang, J. Guo, M. Peng, H. Ji, H. Yang, B. Liu, J. Wang, X.
Council (ERC) under the European Union’s Horizon 2020 research and Zhang, S. Li, In Vitro Cell Dev Biol Anim. (2016). https://doi.org/
innovation program (Grant agreements No. 724734-MINERVA to C.G. 10.1007/s11626-016-0005-5
and 825159-MOAB to M.T.R.). The results reflect only the authors’ M. Laganà, M.T. Raimondi, Biomed. Microdevices 14, 225 (2012).
views and the Agency is not responsible for any use that may be made https://doi.org/10.1007/s10544-011-9600-0
of the information contained. We thank Dr. Alessandro Marturano-Kruik K.F. Lei, M.H. Wu, C.W. Hsu, Y.D. Chen, Biosens. Bioelectron. 51, 16
for his technical support, Prof. Giambattista Gruosso, Politecnico di (2014). https://doi.org/10.1016/j.bios.2013.07.031
Milano, for his suggestions on the numerical models and Dr. Diego M. Lübberstedt, U. Müller-Vieira, K.M. Biemel, M. Darnell, S.A.
Albani, Istituto di Ricerche Farmacologiche Mario Negri-IRCCS, for Hoffmann, F. Knöspel, E.C. Wönne, D. Knobeloch, A.K. Nüssler,
critically reviewing the manuscript. J.C. Gerlach, T.B. Andersson, K. Zeilinger, J. Tissue Eng. Regen.
Med. (2015). https://doi.org/10.1002/term.1652
Compliance with ethical standards D. Lv, Z. Hu, L. Lu, H. Lu, X. Xu, Oncol. Lett. (2017). https://
doi.org/10.3892/ol.2017.7134
Conflict of interest The authors declare that they have no conflict of A. Marturano-Kruik, M.M. Nava, K. Yeager, A. Chramiec, L. Hao, S.
interest. Robinson, E. Guo, M.T. Raimondi, G. Vunjak-Novakovic, Proc.
Natl. Acad. Sci. U. S. A. 115, 1256 (2018). https://doi.org/10.
Open Access This article is distributed under the terms of the Creative 1073/pnas.1714282115
Commons Attribution 4.0 International License (http:// J. Miyakoshi, Prog. Biophys. Mol. Biol. 87, 2–3 (2005)
creativecommons.org/licenses/by/4.0/), which permits unrestricted use, J. Pawlowski, A.S. Kraft, Proc Natl Acad Sci U S A. 97, 2 (2000)
distribution, and reproduction in any medium, provided you give Q.L. Peng, S.M. Mc Murry, J.M.D. Coey, J. Magn. Magn. Mater. 268, 1–
appropriate credit to the original author(s) and the source, provide a link 2 (2004)
to the Creative Commons license, and indicate if changes were made. R.R. Raylman, A.C. Clavo, R.L. Wahl, Bioelectromagnetics. 17, 5 (1996)
S. Romeo, A. Sannino, M.R. Scarfì, R. Massa, R. d'Angelo, O. Zeni, Sci.
Rep. 6 (2016). https://doi.org/10.1038/srep19398
B.J. Roth, Exp. Biol. Med. (Maywood). (2011). https://doi.org/10.1258/
ebm.2010.010236
References S. Stolfa, M. Skorvánek, P. Stolfa, J. Rosocha, G. Vasko, J. Sabo, Physiol
Res. 56(Suppl 1) (2007)
W.W. Albuquerque, R.M. Costa, S. Fernandes Tde, A.L. Porto, B. Tenuzzo, C. Vergallo, L. Dini, Tissue Cell. 41, 169 (2009). https://doi.
Prog. Biophys. Mol. Biol. 121, 16 (2016). https://doi.org/10. org/10.1016/j.tice.2008.09.004
1016/j.pbiomolbio.2016.03.003 M. Tunesi, F. Fusco, F. Fiordaliso, A. Corbelli, G. Biella, M.T.
C. Aldinucci, J.B. Garcia, M. Palmi, G. Sgaragli, A. Benocci, A. Meini, F. Raimondi, Front. Aging Neurosci. 8 (2016). https://doi.org/
Pessina, C. Rossi, C. Bonechi, G.P. Pessina, Bioelectromagnetics. 10.3389/fnagi.2016.00146
24, 6 (2003) O. Valiron, L. Peris, G. Rikken, A. Schweitzer, Y. Saoudi, C. Remy, D.
E. Calabrò, S. Condello, M. Currò, N. Ferlazzo, D. Caccamo, S. Magazù, Job, J. Magn. Reson. Imaging 22, 334 (2005). https://doi.org/10.
R. Ientile, Bioelectromagnetics. 34, 618 (2013). https://doi.org/10. 1002/jmri.20398
1002/bem.21815 S. Zannella, in CAS-CERN Accelerator School: Measurement and
J. Dolezel, J. Bartos, H. Voglmayr, J. Greilhuber, Cytometry A. 51, 2 (2003)
Alignment of Accelerator and Detector Magnets. (1997) https://
European Commission (EC). Non-ionizing radiation: sources, exposure doi.org/10.5170/CERN-1998-005.375
and health effects. B-1049, Brussels 1996
L. Zhang, X. Ji, X. Yang, X. Zhang, Oncotarget. (2017). https://doi.org/
M. Feychting, Prog. Biophys. Mol. Biol. 87, 2–3 (2005)
10.18632/oncotarget.14480
P. Frattini, C. Villa, F.D. Santis, M. Meregalli, M. Belicchi, S. Erratico, P.
Bella, M.T. Raimondi, Q. Lu, Y. Torrente, Hum. Mol. Genet. 26,
3682 (2017). https://doi.org/10.1093/hmg/ddx252
Publisher’s note Springer Nature remains neutral with regard to
V. Hartwig, S. Romeo, O. Zeni, Med. Biol. Eng. Comput. 56, 531 (2018).
jurisdictional claims in published maps and institutional affiliations.
https://doi.org/10.1007/s11517-017-1779-7
W. Ji, H. Huang, A. Deng, C. Pan, Micron. 40, 894 (2009). https://doi.
org/10.1016/j.micron.2009.05.010
H.J. Kim, D. Huh, G. Hamilton, D.E. Ingber, Lab Chip 12, 2165 (2012).
https://doi.org/10.1039/c2lc40074j