Magnetoresistive-based real-time cell phagocytosis monitoring

Biosensors and Bioelectronics 36 (2012) 116–122

Contents lists available at SciVerse ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

Magnetoresistive-based real-time cell phagocytosis monitoring A. Shoshi a,n, J. Schotter a, P. Schroeder a, M. Milnera a, P. Ertl a, V. Charwat a, M. Purtscher a, R. Heer a, M. Eggeling a, G. Reiss b, H. Brueckl a a b

AIT Austrian Institute of Technology, Molecular Diagnostics, Donau-City-Strasse 1, 1220 Vienna, Austria Thin Films & Physics of Nanostructures, University of Bielefeld, D2, Universitaetsstrasse 25, 33615 Bielefeld, Germany

a r t i c l e i n f o

a b s t r a c t

Article history: Received 20 February 2012 Received in revised form 28 March 2012 Accepted 4 April 2012 Available online 20 April 2012

The uptake of large particles by cells (phagocytosis) is an important factor in cell biology and also plays a major role in biomedical applications. So far, most methods for determining the phagocytic properties rely on cell-culture incubation and end-point detection schemes. Here, we present a lab-on-a-chip system for real-time monitoring of magnetic particle uptake by human fibroblast (NHDF) cells. It is based on recording the time evolution of the average position and distribution of magnetic particles during phagocytosis by giant-magnetoresistive (GMR) type sensors. We employ particles with a mean diameter of 1.2 mm and characterize their phagocytosis-relevant properties. Our experiments at physiological conditions reveal a cellular uptake rate of 45 particles per hour and show that phagocytosis reaches saturation after an average uptake time of 27.7 h. Moreover, reference phagocytosis experiments at 4 1C are carried out to mimic environmental or disease related inhibition of the phagocytic behavior, and our measurements clearly show that we are able to distinguish between cell-membrane adherent and phagocytosed magnetic particles. Besides the demonstrated real-time monitoring of phagocytosis mechanisms, additional nano-biointerface studies can be realized, including on-chip cell adhesion/ spreading as well as cell migration, attachment and detachment dynamics. This versatility shows the potential of our approach for providing a multifunctional platform for on-chip cell analysis. & 2012 Elsevier B.V. All rights reserved.

1. Introduction The interaction of cells with their extracellular environment such as the cellular uptake of a variety of particles is of high biotechnological and biomedical interest. Phagocytosis is defined as the internalization of large particles or microorganisms by cells through plasma membrane derived vesicles (Conner and Schmid, 2003; Desjardins and Griffiths, 2003). This evolutionary conserved mechanism is associated with food uptake in unicellular organisms, while in metazoa it fulfills a variety of functions such as the clearance of infectious agents, apoptotic and senescent cells and, thus, participating in the immune and inflammatory response of the host defense (Chavrier, 2001; Greenberg and Grinstein, 2002). The phagocytic process is primarily restricted to professional phagocytes including macrophages, dendritic cells and neutrophils with a diversity of dedicated phagocytic receptors, albeit many mammalian cell types show phagocytic capacity (Chavrier, 2001; Conner and Schmid, 2003). Fibroblasts, classified as nonand paraprofessional phagocytes (Rabinovitch, 1995), are known to perform phagocytosis and contribute by remodeling of collagen

n

Corresponding author. Tel.: þ43 50550 4311; fax: þ 43 50550 4399. E-mail address: [email protected] (A. Shoshi).

0956-5663/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bios.2012.04.002

essentially to the maintenance of tissue structure and function (Abraham et al., 2007; Arlein et al., 1998; McCulloch and Knowles, 1993; Segal et al., 2000). Beyond the scope of biotechnological applications such as tissue engineering, biodegradation and deposition of collagen, fibroblasts are also of great importance to the progression of several diseases, amongst others wound healing, fibrosis, bone structure and the development of metastases in cancers and tumors (Kalluri and Zeisberg, 2006; McAnulty, 2007; McCulloch and Knowles, 1993). Since fibroblast can be found throughout the whole human body they are likely to be exposed to particles, which qualify them as suitable candidates for model cell systems investigating in vitro uptake. The most common techniques established for studying the engulfment process and uptake capacity are based on optical particle observation, e.g. flow cytometry (Arlein et al., 1998; Olivier et al., 2004; Segal et al., 2000; Semmling et al., 2008), fluorescence confocal microscopy imaging (Cox et al., 2000; Tollis et al., 2010; Zhang et al., 2009), confocal laser scanning microscopy (Lu et al., 2009), conventional fluorescence microscopy (Alberola and R¨adler, 2009; Berry et al., 2004) and phase contrast microscopy (Hasegawa et al., 2007; Herant et al., 2006; Korn and Weisman, 1967). Besides, also scanning and transmission electron microscopy are utilized especially for visualizing the initial events of particle engulfment (Aggeler and Werb, 1982) or (post-) phagocytosis cell

A. Shoshi et al. / Biosensors and Bioelectronics 36 (2012) 116–122

morphology and for determining intracellular particle localization and/or quantity (Berry et al., 2004; Chithrani et al., 2006; Gupta and Curtis, 2004; Olivier et al., 2004). Long-term observations of the entire phagocytosis process with high time resolution are very challenging with the methods mentioned above. To elucidate the uptake kinetics/dynamics, generally several end-point measurements at different stages during phagocytosis are carried out. Especially for fluorescence based techniques, photobleaching is a limiting factor. Moreover, usually further cell analysis is not possible due to the fact, that in most cases the cells are not alive anymore (e.g. SEM and TEM imaging). For better understanding of the complexity of phagocytosis, real-time monitoring techniques are required to provide improved insights into this mechanism and, in a wider sense, to treatment strategies of phagocytosis related diseases (Abraham et al., 2007). To that end, Wang et al. (2010) introduced a label-free method based on plasmonic scattering which allows 3D mapping of internalized particles in living cells. Himmelhaus and Francois (2009) used fluorescent dye-doped polystyrene microspheres as optical sensors by applying whispering gallery mode excitations as transducer mechanism. Monitoring of the particle uptake was also realized by membrane capacitance measurements by means of a phase detection technique, which reflects plasma membrane changes during the phagocytic process (Holevinsky and Nelson, 1998). Lee et al. (2010) developed a capacitance-based sensor for real-time monitoring of cell polarization in an external AC electric field, and they were able to distinguish between receptormediated endocytosis and non-specific pinocytosis. In this study, we demonstrate a magnetic lab-on-a-chip (MAGLab) system to monitor phagocytosis in living cells in real-time and label-free. Our MAGLab platform combines magnetoresistive sensors, magnetic particles and microfluidics. A methodology is developed for analyzing the uptake dynamics of the dorsal cell membrane at different physiological conditions. Our method specifically focuses on phagocytosis of magnetic particles, which is especially relevant for a variety of biomedical applications such as magnetic resonance imaging, tissue repair, hyperthermia, targeted drug delivery (magnetofection) and magnetic cell separation (Gupta and Curtis, 2004; Plank et al., 2003).

117

magnetoresistive sensors, which changes when the particles are phagocytosed by cells. To that end, we follow the approach as depicted in Fig. 1. The cells are seeded on top of the sensor area and are grown to a confluent monolayer. Afterwards, the added beads reach the cell layer by sedimentation. The signal they produce in the underlying sensor depends on the strength of the externally applied magnetizing field during measurements as well as their density and mean vertical distance to the sensor layer. Since the stray field strength of the beads within the sensor region increases with decreasing distance while all other parameters are fixed, the sensor output increases proportionally to the progress of phagocytosis, thus allowing real-time monitoring. In contrast to fluorescent-based methods, this purely magnetic approach enables long-term monitoring without facing difficulties like photo-bleaching or background noise from tissue.

3. Materials and methods 3.1. MAGLab setup Our MAGLab setup is composed of two decoupled pairs of Helmholtz (HH) coils connected to two power supplies (Kepco BOP 50-20 MG) generating homogeneous magnetic fields parallel and perpendicular to the chip plane. The orientation of the out-ofplane field is adjustable relative to the chip plane. The in-plane field (max. 7285 Oe) enables magnetoresistive characterization of the sensors, while the out-of plane field (max. 7490 Oe) is required to magnetize the superparamagnetic particles during the measurement process. In the center of both HH-coils a chip holder with cooling and heating system (Jubalo F12) is positioned. Once the fluidic loaded chip is mounted on the chip holder, a connector lid placed on top of the chip establishes the fluidical and electrical connections. Additionally, our MAGLab setup is equipped with a long-range microscope (Leica MZ 16) and a CCD camera (Leica DFC 320) for on-chip optical observations. The magnetic field loops, read-out of the sensors and the peristaltic pump (Watson Marlow 205U) are computer controlled by a LabView (www.ni.com) program. 3.2. Sensor-chip fabrication, preparation and characterization

2. Concept of magnetoresistive-based real-time cell phagocytosis monitoring The basic principle of our approach is to monitor the signal induced by superparamagnetic particles (beads) in embedded

The biochip design and fabrication were previously presented by Schotter et al. (2009). Briefly, a continuous stack of ten Ni80Fe20/ Cu-double layers in the second antiferromagnetic coupling maximum (AFCM) is patterned into individual meander-shaped sensors

recognition engulfment

Cell

uptake APTES

r0(t0)

r1(t1)

… rn(tn) Sensor

Wafer

Fig. 1. (a) Sketch of the magnetoresistive-based phagocytosis monitoring methodology: (i) plain sensor, (ii) confluent cell growth, (iii) magnetic particle incubation. (b) Cross-section sketch of phagocytosis stages. After particle recognition, the cell starts engulfing and finally internalizing the particles in phagosomes, leading to a decrease of the particle-to-sensor distance.

A. Shoshi et al. / Biosensors and Bioelectronics 36 (2012) 116–122

(Supplementary Information (1)). The resistance of each sensor is in the range of 6–8 kO and its GMR amplitude is around 12%. On top of the sensors, except for the contact pads, a 230 nm thick Si3N4 passivation layer is chemical-vapor-deposited to prevent interactions with cells and cell medium. The biochip consists of 16 sensors arranged in two identical rows, representing separated reference and magnetic particle detection sensors (ref- and biosensors). For all cell experiments, the passivation layer is functionalized with 2% APTES (Aldrich, Inc.), dissolved in a buffer solution consisting of 99.2% (v/v) methanol and 0.8% (v/v) acetic acid, which ensures cell growth to the chip-surface. In addition, a fluidic fixture made of cross-linked polydimethylsiloxane (PDMS) with an integrated window of about 2.5 mm  15 mm is mounted on the chip and adjusted to the row of biosensors. The PDMS-fixture enables direct access to the biosensors for magnetic particle immobilization and cell seeding. The ref-sensors remain completely covered by the PDMS fixture throughout the whole experiment and are protected against interactions with fluids or cells. Prior to each experiment, the out-of-plane field is adjusted perpendicular to the chip plane. 3.3. Magnetic beads We use streptavidin-coated superparamagnetic magnetite (Fe3O4) particles of (nominally 0.5 mm) 0.9 mm and (nominally 1.0 mm) 1.2 mm in diameter, with a magnetization of 50 emu/g and a magnetic oxide content of 490% (MagSense Life Science, Inc.). Their physical and chemical properties are further characterized by electron-microscopy (ZEISS SEM, SUPRATM 40) as well as zeta potential and dynamic light scattering measurements (ZetaSizer Nano, MALVERN Instruments) in different buffer solutions (Supplementary Information (2)). pH measurements are carried out by a PB-11 Sartorius pH-meter. 3.4. Cell uptake capacity Different amounts of beads are immobilized onto sensor-like plain Si/SiO2/Si3N4/APTES surfaces. The samples are placed into microtiter plates (12 well), and 2 ml of bead-dH2O solution with different bead concentrations (mg/ml: 1.0, 2.5, 5.0, 10.0, 20.0, 40.0, 60.0, 120.0) are added. At the maximum concentration, the bead surface coverage almost reaches 100%. Afterwards, approx. 80,000 cells are seeded in each well and grown to a confluent monolayer. The number of internalized beads per cell is ascertained semiquantitatively by counting beads inside individual cell from images taken by optical- and electron-microscopy (EM). For this purpose, cells are fixed by a simple drying procedure at room temperature. In case of EM, an additional 100 nm thick Au-layer is sputter-deposited to ensure surface conductivity. Following optical and electron microscopy analysis, only cells with maximal bead loading are selected for bead counting at each bead immobilization concentration (Supplementary information (3)). 3.5. NHDF cell culture Normal Human Dermal Fibroblast (NHDF) cells from PromoCell are cultured at 37 1C under humidified atmosphere of 5% CO2 in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS). For on-chip cell experiments, additionally 10 mM HEPES buffer solution and 1% gentamicin are added (all components from PAA Laboratories GmbH). 3.6. Magnetoresistive-based real-time cell phagocytosis monitoring Approximately 5000–6000 cells are grown to a confluent cell monolayer within the area defined by the PDMS fluidic on top of

the biosensors. Then, 400 ml of particles dissolved in cell-medium (75 mg/ml) are added. The magnetoresistive responses of the ref- and biosensors are recorded frequently during each stage of the experiment and accompanied by optical microscopy imaging. Each experiment is terminated by complete cell removal using a trypsin-EDTA (0.25% trypsin and 0.02% EDTA) solution followed by several PBS and dH2O washing steps.

4. Results and discussions 4.1. Dependence of the sensor signal on bead coverage The beads employed here are superparamagnetic, which prevents their agglomeration in solution. Thus, a magnetizing field has to be applied in order to induce stray fields of the beads that can be measured by the embedded magnetoresistive sensors. Due to the large shape anisotropy of the nanometer-thin magnetic layers, the designed GMR-sensors are basically sensitive to inplane magnetic fields only. Thus, a large magnetic bead moment can be induced by applying a magnetizing field perpendicular to the sensor plane without directly affecting the sensor. However, the remaining minimal response to perpendicular fields represents the blank level. For bead-covered sensors, the in-plane components of the bead’s magnetic stray field locally affect the sensor resistance. To investigate the dependency of the sensor signal on the surface coverage of the MagSense particles, successively increasing concentrations of beads are immobilized on the same chip. The corresponding surface coverage denotes the x-axis in Fig. 2. After each immobilization step, the out-of-plane magnetizing field is applied and the sensor signal is recorded. The relative signal change obtained at zero and maximum magnetizing field denotes the y-axis in Fig. 2. This kind of sensor calibration is performed in two different manners. In one case, the complete chip is dried after each bead immobilization step (‘‘air-calibration’’), while for the other case (‘‘liquid-calibration’’) the sensors are read out without any drying step in between. The data points are fitted by an exponential growth function (solid line) according to y ¼ a2bnexpðx=dÞ: The fitting results are a ¼0.329% and 0.405% (saturation GMR amplitude), a–b ¼0.04% and 0.05% (blank level) and d ¼43% and 18% (bead coverage at which the saturation GMR amplitude decreases by b/e) for the liquid- and air-calibrations, respectively. As more and more of the total sensor area is affected by the local

0.4

GMR signal for beads in: air liquid exponential fits

0.3

GMR [%]

118

0.2

0.1 blank level 0.0 0

10

20

30

40

50

60

sensor surface coverage [%] Fig. 2. Dependency of the GMR sensor signal on the bead surface coverage.

A. Shoshi et al. / Biosensors and Bioelectronics 36 (2012) 116–122

stray fields of immobilized beads, the output signal initially increases linearly with the surface coverage. As the beads are getting closer to each other, their dipolar stray fields start to overlap and the sensor response saturates. Since the beads experience a buoyant force in liquid and can move within the constraint of their molecular binding length to the surface both by thermal agitation and mutual magnetic interaction, the average vertical distance of the beads to the sensor is larger in the liquid case, thus resulting in smaller sensor response for the same bead coverage. Within the window between depletion and saturation, the output signal of such magnetoresistive sensors is a direct measure of the density of magnetic particles bound on the sensor surface. 4.2. Micromagnetic simulations In order to model the change of the GMR sensor signal due to increasing bead distance induced by cellular uptake, we perform micromagnetic simulations (more details in Supplementary Information (5)) using the Object Oriented Micromagnetic Framework (OOMMF, http://math.nist.gov/oommf). The lateral dimensions of the modeled sensor region are 1.38 mm  1.38 mm, which directly resembles the average area per bead at the mean surface coverage of 60% encountered during the uptake experiments. To additionally incorporate the next neighbor interactions, the influences of the individual dipolar magnetic fields of an array of 9 beads arranged as depicted in Fig. 3 (inset) are considered. By successively increasing the vertical bead-array distance, the sensor signal decreases due to the diminishing effect of the beads’ stray fields, effectively vanishing for distances larger than 650 nm (Fig. 3). Here, the x-axis values describe the vertical distance between the passivation layer surface and the bead surface. Thus, in order to obtain the vertical sensor to bead-center separation, the passivation layer thickness (230 nm) and the bead radius must be added to the x-values in Fig. 3. This sensor model is used to determine the average bead distance decrease from the magnetoresistive measurements obtained during the uptake experiments. 4.3. Real-time monitoring of bead phagocytosis GMR-based real-time phagocytosis monitoring is carried out using MagSense particles with a mean diameter of 1.2 mm. We investigate the phagocytosis by the dorsal surface of fibroblast

119

membranes both in absence and presence of metabolic inhibition. We first start with standard uptake experiments at 37 1C and then turn our focus to the metabolically inhibited uptake at 4 1C. Depending on the fluidic chamber dimensions, each experiment starts with the incubation of 5000–6000 cells, which in a confluent state cover the entire chip area. In contrast, the nonspecifically adsorbed cells on peripheral PDMS surfaces around the biosensors show a weak adhesion and are usually removed by the next cell medium exchange. On top of the confluent cell monolayer we add 400–450 ml of a 65 mg/ml bead-DMEM suspension to the fluidic chamber. The amount of beads added corresponds to a surface coverage of about 60%, which is slightly higher than the upper uptake limit of fibroblast cells (Supplementary information (3)). Subsequently, the beads suspended in cell medium start to settle down until all of them reach the cell layer on top of the chip surface. In order to determine the time required for all beads within the solution to reach the chip surface, bead sedimentation is first analyzed both with optical microscopy and sensor readout. For that, we use APTES functionalized sensor chips and apply the same amount and concentration of beads in DMEM to the PDMS-fluidic chamber. The resulting normalized sensor response and optically determined bead surface coverage are displayed in Fig. 4. As expected from the largely linear dependence of the sensor signal on bead coverage (Fig. 2), the two curves show good consistency and reveal a sedimentation time of about 80 min. In Fig. 5(a) the time-dependent GMR signal sequence of the entire top-down phagocytosis experiment is depicted for one sensor. In total, ten sensors from two separate experiments are included in the data analysis. Immediately after adding beads at t¼ 0 h, we observe an increase of the GMR response, which levels off at a certain value followed by an oscillatory signal behavior (Supplementary information (4)). In the early stage up to 2.470.8 h, we see a first fast signal increase up to a kink in the phagocytosis monitoring curve, which is common for all measurements (Fig. 5(b)). Beyond this kink, the signal increases gradually but much slower. The initial slope of the steep part is about 8.5 73.9 times larger than the later increase. If we compare the early stage increase to the bead sedimentation characteristic (Fig. 4), there is a good correspondence in their time dependence. Hence, we conclude that the main contribution to the initial signal is due to bead sedimentation to the top of the cell layer. In the time beyond the early stage the sensor signal increases further due to bead phagocytosis, which leads to a decrease of the average bead to sensor distance. At the end of the experiment

1.00

0.8

normalized data

norm. resistance [a.u.]

1.0

0.6 0.4 0.2 0.0

0.75

0.50

0.25 surface coverage GMR signal

0.00

0

300

600 900 1200 bead distance [nm]

1500

1800 0

20

40

60

80

100

120

140

160

time [min] Fig. 3. Simulated dependence of the relative sensor resistance on the vertical bead distance. Inset: Arrangement of the 9-bead array and the simulated sensor element with side length a ¼1.38 mm.

Fig. 4. Bead sedimentation characteristics analyzed by the GMR sensor signal and by optical bead surface coverage measurements.

120

A. Shoshi et al. / Biosensors and Bioelectronics 36 (2012) 116–122

0.16 0.30

0.14 0.12

GMR [%]

GMR [%]

0.25 0.20 0.15 0.10

0.10 0.08 0.06 0.04

0.05

0.02 0

20

40

60

80

100

120

-2

0.10

0.10

0.09

0.09

0.08

0.08

GMR [%]

GMR [%]

0.00 -20

0.07 0.06 0.05

0.03 5

10

15

20

25

30

6

8

10 12 14 16 18 20 22

0.05

0.03 0

4

0.06

0.04

-5

2

0.07

0.04

-30 -25

0

-30

-20

-10

0

10

20

30

40

50

Fig. 5. (a) Exemplary time-dependent GMR signal trace of the entire phagocytosis kinetics experiment. (b) Another measurement showing the early stage signal behavior. (c and d) Time-dependent GMR signal trace of metabolically inhibited phagocytosis. For the trace in (d), the phagocytosis is also monitored at physiological conditions (dashed line) following the temperature-induced inhibition up to 22 h.

the cells are trypsinized and completely removed by several washing steps (replacement). As a consequence the GMR signal drops down to the blank level (last data point in each trace). In order to attain the mean saturation level and bead incubation time, the data of each trace excluding the replacement value is fitted by an exponential growth function. Here, the fit value reached for t-N denotes the saturation level of 0.22970.032%, while the time required for reaching 95% of this level is defined as the saturation time of 27.7 714.5 h. The 5% difference to 100% signal level denotes the mean standard deviation of the GMR signals. The saturation time denotes the termination of the phagocytosis process. Taking into account the maximum uptake capacity of 1270 beads per cell (Supplementary information (3)), we calculate an upper limit of the phagocytic uptake rate of 45 beads per hour and cell. The mean saturation level corresponds to about 87% of the expected signal above blank level for beads bound directly to the sensor surface at 60% coverage (Fig. 2). By comparing to the simulated dependence of the sensor signal on bead distance (Fig. 3), the same relative signal decrease is obtained when lifting the beads by 20 nm, which implies a mean vertical position of internalized beads at the lowest possible limit of twice the cell membrane thickness (Alberts et al., 2008). In order to better distinguish bead sedimentation and phagocytosis, the experiments at 37 1C are supplemented by studies at 4 1C, where the cells are metabolically inhibited. That way, cell phagocytosis can be actively impaired (Berry et al., 2004; Olivier et al., 2004; Pratten and Lloyd, 2003). Two complementing types of investigations are performed: in the first experiment, the

temperature remains at 4 1C throughout the entire analysis time, whereas in the second, following the uptake monitoring time of 22 h at 4 1C we increase the temperature to 37 1C and continue monitoring the phagocytosis for another 22 h. Otherwise, the procedure of both experiment types corresponds to the specifications described above. To ensure appropriate cell growth, the cells are initially seeded at 37 1C onto the chip surface to form a confluent monolayer. Then, a smooth temperature transition from 37 1C to 4 1C with a cooling rate of 0.33 1C/min is applied. Once 4 1C is reached, we keep the cells for another 60 min at 4 1C before adding the particles (400–450 ml at 65 mg/ml), thus enabling comparability between all experiment types. Looking at the early stage GMR response of both experiment types (Fig. 5(c) and (d)), we again observe a rapid signal increase that is attributed to bead sedimentation. In the following monitoring time from  2 to 23 h, no further noticeable signal increase is observed for both experiment types. The saturation value is significantly lower compared to the results obtained at 37 1C (Fig. 5(a)) and provides evidence that phagocytic uptake is profoundly inhibited at 4 1C. Even within the early stage up to 2.5 h, the signal level reached for the experiments at 37 1C (0.10570.014%) exceeds the level obtained at 4 1C (0.071270.01%), which supports the assumption that bead phagocytosis and sedimentation take place in parallel at physiological temperatures. The mean saturation level for the thermally inhibited experiments corresponds to about 10% of the expected signal above blank level for beads bound directly to the sensor surface at 60% coverage (Fig. 2), which translates into a mean vertical bead distance to the sensor surface of about 400 nm (comparison to

A. Shoshi et al. / Biosensors and Bioelectronics 36 (2012) 116–122

simulated distance dependence in Fig. 3). According to SEM crosssection imaging of fibroblast-like endothelial cell monolayers (Gimbrone et al., 1974), the smallest cell thickness is reached at the cell–cell borders and is found to be around 500 nm, which agrees well to the distance derived by our sensor signal analysis. When increasing the temperature with a heat-up rate of 5.5 1C/ min from 4 1C to 37 1C for the second experiment type (indicated by dashed line in Fig. 5(d)), (heat-up rate of 5.5 1C/min), the GMR signal remains constant for another 22 h, thus indicating that the cells do not recover their ability for phagocytosis during this period. To assess whether the beads are partly internalized, superficially bound or simply lie on top of the cell surface monolayer, a small cylindrical NbFeB permanent magnet (diameter 3 mm, length 15 mm) is drawn through the medium at close vicinity to the sensors ( 2 mm vertical distance). Reference experiments show that this procedure reliably removes beads from an un-functionalized sensor surface (data not shown). In case of the cell monolayer the respective sensor signals drop to 0.06070.007% (data point denoted by ‘‘magnetic bead replacement’’). Thus, we conclude that about one half of the beads on top of the cell monolayer are weakly bound or simply unbound, which further proves that bead uptake of the cells is inhibited under these conditions. At the end of the experiment, all cells are replaced and the signal drops down to the blank level. If we compare the uptake rate of NHDF cells at physiological conditions with Acanthamoeba or professional phagocytic cells such as macrophages and neutrophils, we observe a much lower uptake rate. An ameba is capable to internalize about 7.9 beads of 1.3 mm (Weisman and Korn, 1967) and up to 10 beads of 1.1 mm (Wetzel and Korn, 1969) in diameter per cell and minute, respectively, which corresponds to 10–13 fold higher uptake rates. Cox et al. (2000) showed that murine thio-macrophages are capable of ingesting the equivalent of 48–145% of their macroscopic plasma surface area (1367 mm2) within 30 min. This area corresponds to the surface area of 145–438 beads of 1.2 mm in diameter and, thus, an uptake rate of 4.8–14.6 beads per minute, which is 6–19 times higher than observed here for human fibroblasts. Comparable results are observed by neutrophils of about 9 mm in diameter, which are incubated with polystyrene beads in the range of 2–10 mm in diameter (Herant et al., 2006). The relatively low uptake rates of fibroblast in comparison to professional phagocytes are consistent with literature results (McCulloch and Knowles, 1993; Segal et al., 2000).

5. Conclusion In vitro phagocytosis of magnetic particles by human fibroblast cells is monitored in real-time by means of a magnetic cellchip based on magnetoresistive sensors. Crucial bead parameters influencing phagocytosis, i.e. size, surface chemistry and charge, are characterized. In preliminary experiments, phagocytosis-relevant cell specific properties are determined. The maximum uptake capacity of NHDF cells for beads of 1.2 mm and 0.9 mm in diameter is 1270 and 2265, respectively. The stability of beads in phagolysosomes is assessed by long-term monitoring measurements of internalized beads showing no noticeable bead degradation. Moreover, the magnetoresistive properties of our GMR sensor are fitted by a downhill simplex method and analyzed by micromagnetic simulations, which serve as a model for evaluating our GMR-based detection concept. For the real-time phagocytosis monitoring we have established a magnetoresistive based method for measuring the uptake kinetics of the dorsal plasma membrane. The results show an overall uptake rate of 45 beads of 1.2 mm in diameter per cell and hour. In general, the uptake rate is not a linear function with time. It is higher at early stages of

121

phagocytosis and diminishes until it reaches saturation after an average time of 27.7 h. The investigations of the uptake at 4 1C clearly demonstrate that we are able to distinguish between cellmembrane adherent and phagocytosed beads. In a wider sense, our results show that the described method is well suited to detect possible disease-related or environmental impacts leading to a profound inhibition or malfunction in phagocytosis. From the biomedical point of view collagen-coated beads can be utilized to simulate actual receptor-mediated cellular phagocytosis processes during collagen degradation by fibroblasts. Besides the demonstrated real-time monitoring of phagocytosis mechanisms, also on-chip cell adhesion/spreading as well as cell migration, attachment and detachment dynamics can be analyzed, which shows the potential of our MAGLab system for providing a multifunctional platform for on-chip analysis of a multitude of different cell model systems.

Acknowledgments We gratefully acknowledge T. Uhrmann for LabView programming. The research leading to these results has received financial ¨ support by the ‘‘Oesterreichische Forschungsforderungs-gesellschaft (FFG)’’ under Grant no. 810985.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2012.04.002.

References Abraham, L.C., Dice, J.F., Lee, K., Kaplan, D.L., 2007. Experimental Cell Research 313 (5), 1045–1055. Aggeler, J., Werb, Z., 1982. Journal of Cell Biology 94, 613–623. ¨ Alberola, A.P., Radler, 2009. Biomaterials 30, 3766–3770. Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., Walter, P., 2008. Molecular Biology of the Cell, 5th ed. New York. Arlein, W.J., Shearer, J.D., Caldwell, M.D., 1998. American Journal of Physiology— Regulatory, Integrative and Comparative Physiology 275, 1041–1048. Berry, C.C., Charles, S., Wells, S., Dalby, M.J., Curtis, A.S.G., 2004. International Journal of Pharmaceutics 269, 211–225. Chavrier, P., 2001. Seminars in Immunology 13, 337–338. Chithrani, B.D., Ghazani, A.A., Chan, W.C.W., 2006. Nano Letters 6 (4), 662–668. Conner, S.D., Schmid, S.L., 2003. Nature 422, 37–44. Cox, D., Lee, D.J., Dale, B.M., Calafat, J., Greenberg, S., 2000. Proceedings of the National Academy of Sciences of the USA 97 (2), 680–685. Desjardins, M., Griffiths, G., 2003. Current Opinion in Cell Biology 15, 498–503. Gimbrone, M.A., Cotran Jr., R.S, Folkman, J., 1974. Journal of Cell Biology 60, 673–684. Greenberg, S., Grinstein, S., 2002. Current Opinion in Immunology 14, 136–145. Gupta, A.K., Curtis, A.S.G., 2004. Journal of Materials Science Materials in Medicine 15, 493–496. Hasegawa, T., Hirota, K., Tomoda, K., Ito, F., Inagawa, H., Kochi, C., Soma, G.-I., Makino, K., Terada, H., 2007. Colloids and Surfaces B: Biointerfaces 60, 221–228. Herant, M., Heinrich, V., Dembo, M., 2006. Journal of Cell Science 119 (9), 1903–1913. Himmelhaus, M., Francois, A., 2009. Biosensors & Bioelectronics 25, 418–427. Holevinsky, K.O., Nelson, D.J., 1998. Biophysical Journal 75, 2577–2586. Kalluri, R., Zeisberg, M., 2006. Nature Reviews Cancer 6, 392–401. Korn, E.D., Weisman, R.A., 1967. Journal of Cell Biology 34, 219–227. Lee, R., Kima, P.H., Choi, J.-W., Oh-Joon, K., Kim, K., Kim, D., Yun, C.-O., Yoo, K.-H., 2010. Biosensors & Bioelectronics 25, 1325–1332. Lu, F., Wu, S.-H., Hung, Y., Mou, C.-Y., 2009. Small 5 (12), 1408–1413. McCulloch, C.A.G., Knowles, G.C., 1993. Journal of Cellular Physiology 155 (3), 461–471. McAnulty, R.J., 2007. International Journal of Biochemistry & Cell Biology 39, 666–671. Olivier, V., Rivie re, C., Hindie´, M., Duval, J.-L., Bomila-Koradjim, G., Nagel, M.-D., 2004. Colloids and Surfaces B: Biointerfaces 33, 23–31. ¨ Plank, C., Schillinger, U., Scherer, F., Bergemann, C., Re´my, J.S., Krotz, F., Anton, M., Lausier, J., Rosenecker, J., 2003. Biological Chemistry 384 (5), 737–747.

122

A. Shoshi et al. / Biosensors and Bioelectronics 36 (2012) 116–122

Pratten, M.K., Lloyd, J.B., 2003. Biochimica et Biophysica Acta—General Subjects 881 (3), 307–313. Rabinovitch, M., 1995. Trends in Cell Biology 5, 85–87. Segal, G., Lee, W., Arora, P.D., McKee, M., Downey, G., McCulloch, C.A.G., 2000. Journal of Cell Science 114, 119–129. ¨ J., Parak, W.J., 2008. Semmling, M., Kreft, O., Javier, A.M., Sukhorukov, G.B., Kas, Small 4 (10), 1763–1768. Schotter, J., Shoshi, A., Brueckl, H., 2009. Journal of Magnetism and Magnetic Materials 321, 1671–1675.

Tollis, S., Dart, A.E., Tzircotis, G., Endres, R.G., 2010. BMC Systems Biology 4, 149. Wang, S.-H., Lee, C.-W., Chiou, A., Wei, P.-K., 2010, Journal of Nanobiotechnology 8, 33. Weisman, R.A., Korn, E.D., 1967. Biochemistry 6 (2), 485–497. Wetzel, M.G., Korn, E.D., 1969. Journal of Cell Biology 43, 219–227. Zhang, S., Li, J., Lykotrafitis, G., Bao, G., Suresh, S., 2009. Advanced Materials 21, 419–424.