A novel environmental chamber for neuronal network multisite recordings

ARTICLE A Novel Environmental Chamber for Neuronal Network Multisite Recordings E. Biffi,1,2 G. Regalia,1 D. Ghezzi,3 R. De Ceglia,4 A. Menegon,2 G. Ferrigno,1 G.B. Fiore,5 A. Pedrocchi1 1

Politecnico di Milano, Bioengineering Department, Neuroengineering and Medical Robotics Laboratory, p.zza Leonardo da Vinci 32, 20133 Milano, Italy; telephone: þ39 02 2399 9509; fax: þ39 02 2399 3360; e-mail: emilia.biffi@mail.polimi.it 2 Advanced Light and Electron Microscopy Bio-Imaging Centre, Experimental Imaging Centre, San Raffaele Scientific Institute, Milano, Italy 3 Department of Neuroscience and Brain Technologies, Istituto Italiano di Tecnologia, Genova, Italy 4 Neuroimmunology Unit-DIBIT2, INSpe, Department of Neuroscience, San Raffaele Scientific Institute, Milan, Italy 5 Politecnico di Milano, Bioengineering Department, Laboratory of Micro and Biofluid Dynamics, Milano, Italy

Introduction ABSTRACT: Environmental stability is a critical issue for neuronal networks in vitro. Hence, the ability to control the physical and chemical environment of cell cultures during electrophysiological measurements is an important requirement in the experimental design. In this work, we describe the development and the experimental verification of a closed chamber for multisite electrophysiology and optical monitoring. The chamber provides stable temperature, pH and humidity and guarantees cell viability comparable to standard incubators. Besides, it integrates the electronics for long-term neuronal activity recording. The system is portable and adaptable for multiple network housings, which allows performing parallel experiments in the same environment. Our results show that this device can be a solution for long-term electrophysiology, for dual network experiments and for coupled optical and electrical measurements. Biotechnol. Bioeng. 2012;109: 2553–2566. ß 2012 Wiley Periodicals, Inc. KEYWORDS: Micro Electrode Arrays; neuronal cell culture; long-term recording; temperature control; hyperosmolarity

Correspondence to: E. Biffi Contract grant sponsor: Fondazione Confalonieri Additional supporting information may be found in the online version of this article. Received 6 December 2011; Revision received 29 February 2012; Accepted 2 April 2012 Accepted manuscript online 17 April 2012; Article first published online 24 April 2012 in Wiley Online Library (http://onlinelibrary.wiley.com/doi/10.1002/bit.24526/abstract) DOI 10.1002/bit.24526

ß 2012 Wiley Periodicals, Inc.

Neuronal networks show a high degree of sensitivity to slight changes in the physical and chemical composition of the cell culture environment (Gross and Schwalm, 1994). For this reason, the ability to control and maintain the environmental parameters during optical and electrophysiological measurements is an important requirement in the experimental design. Particularly, applications involving Micro Electrode Arrays (MEAs), planar matrices of non-invasive microelectrodes (Johnstone et al., 2010; Morin et al., 2005), suffer from some limitations imposed by the standard experimental equipment. This is usually composed of an incubator, where cell cultures are grown, and of the freestanding electronics for neuronal activity recording. Thus, during experimental sessions, the MEA chip needs to be moved from the incubator to the recording stage and this may induce changes in the cell culture temperature, contamination of the cellular environment and mechanical disturbances. Moreover, the content of CO2 in the environmental air, lower than the peculiar percentage of an incubator, rapidly causes elevation of the cell culture medium pH. This limits electrophysiological sessions in room air to approximately 120 min (Potter and DeMarse, 2001). Furthermore, medium evaporation effect, due to the unsaturated moisture-content of the environmental air, may also induce hyperosmolarity. These issues get worse when the duration of the experiments increases from hours to days, which is the time extent required in order to study network functional plasticity (Chiappalone et al., 2008; Shahaf and Marom, 2001), long-term neuronal

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

2553

network phenomena (Hofmann and Bading, 2006; van Pelt et al., 2004a,b) and chronic treatment effects (Howell et al., 2011; Morefield et al., 2000; Xiang et al., 2008). The shortcomings of the abovementioned setup cause a gradual decline in the health of these cell cultures and a decrease of data reproducibility, for which the environmental stability is critical (Gross and Schwalm, 1994). Environmental chambers can be classified as closed or open systems depending on their air tightness. Open systems are not suitable for long-term experiments because of the unstable environmental control and the potential risk of contamination. Therefore, closed systems should be preferred for long-term experiments (Ho et al., 2005). Prototypes of environmental chambers appeared in the 1950s (Christiansen et al., 1953; Rose, 1967; Sykes and Moore, 1959) and during the 1980s several systems improving cell survival were developed. Temperature controlled bath chambers were designed to provide an accurate and uniform heating during live cell imaging (Heidemann et al., 2003; Ho et al., 2005) and patch clamp experiments (Forsythe and Coates, 1988; Toyotomi and Momose, 1989), which is required because of the temperature dependence of neuronal properties. During the same years, devices for the control of the gaseous environment and for the improvement of the bicarbonate–carbon dioxide buffer system, which aimed at pH maintenance, were also proposed (Bavister, 1988; Fantini et al., 1987; Vukasinovic et al., 2009). Many efforts were also spent to improve the reliability of MEA recordings and flow chambers with controlled medium environment were coupled to MEA chips (Blau and Ziegler, 2001; Gross and Schwalm, 1994; Pancrazio et al., 2003). Moreover, sealing membranes, able to prevent infections, hyperosmolarity and to preserve a physiological pH, were proposed (Blau et al., 2009; Potter and DeMarse, 2001). Particularly, the lid proposed by Potter and colleague allowed to use a dry incubator for cell culturing, through the decrease of the evaporation effect. Consequently, it was possible to introduce the recording setup inside an incubator. In contrast, the electronics of the recording systems would be quickly damaged by a standard humid incubator with a moisture content higher than 65% (Hales et al., 2010), which prevents from long-term recordings. A similar system, which coupled an incubator with an electrical recording setup and a perfusion pump, was described by Mukai et al. (2003). However, so far there are no portable environmental chambers able to grow neuronal cultures since the beginning of their maturation, as done by incubators, and to provide long-term multichannel recordings of neuronal activity. In this work we present the technological development of a novel environmental chamber which measures the electrical activity of neuronal networks grown on MEAs. The behavior of the system was characterized in terms of heating control and CO2 flow rate. Moreover, the ability of the custom chamber to preserve cell vitality was quantitatively evaluated. Finally, dissociated hippocampal cells were cultured on MEAs and their activity was recorded both

2554

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

inside the custom chamber and with the standard equipment. In the following sections we describe the chamber design and a lumped-parameter thermal model, used as design tool. Then, we demonstrate that the custom chamber does not alter the network development nor the neuronal viability and we give the proof of concept of the feasibility of the recording inside the chamber for long periods. To conclude, we discuss about the advantages of our environmental device and we provide future improvements which should be introduced to perform long-term continuous recordings of spontaneous neuronal activity.

Design System Design The system was designed with the following general requirements: (1) temperature control, (2) humidity maintenance, (3) pH stabilization, (4) sterility preservation, (5) optical transparency, (6) external accesses for medium change and pharmacological stimulation, (7) integrated acquisition card to acquire neuronal signals, (8) suitability for cell monitoring on an inverted microscope. A schematic representation of the chamber architecture and the whole system layout are shown in Figure 1. The chamber consists of an outer and an inner box, which are separated by a watertight cavity filled with water (Fig. 1A) and covered by a top plate. The size of the outer box is 180 mm
180 mm, with a height of 45 mm. The inner box, leant on the outer one by means of four small backings, is formed by a 150 mm
150 mm base with a height equal to 30 mm; a 50 mm
50 mm MEA housing is provided inside. More details about geometrical features of the chamber are reported in Figure S1. A silicone membrane is located between the top plate and the boxes beneath, to guarantee the sealing of the closure, which is provided by means of eight small Rolez clamps. Proper openings for medium and pharmacological agent addition are provided into the top, assuring sterility with pierce silicone membranes for needle insertion. Moreover, connectors for air inlet and outlet are provided as shown in Figure 1A. The chamber heating is obtained by means of a circulating bath (E306, Ecoline, Lauda GmbH, Lauda-Ko¨nigshofen, Germany) which pumps water in the watertight cavity by means of two metal connectors (Fig. 1B and C). A PT100 thermo-resistance is inserted in a reference well, filled with water, through a sealed opening in the top plate. The probe provides the feedback loop for the proportional-integral (PI) temperature controller, which is integrated into the commercial circulating bath. Furthermore, this well is placed symmetrically to the MEA housing to guarantee its reliability as reference (Fig. 1A). To maintain the pH of the medium in a physiological range (7.4), a gas tank containing 20% O2, 70% N2, and 10% CO2 is used. A percentage of humidity higher than 95%

Figure 1. Schematic illustrations of two section of the chamber, whose points of observation are reported in the simple scheme on the right (top view). A: Cross-sectional view of the chamber. The picture illustrates the symmetrical allocation of the MEA chip and the reference well inside the inner box, surrounded by heating water (light blue), which flows in the cavity between the inner and external box. Through the top plate, air inlet and outlet and the opening for PT100 probe insertion are visible. B: Longitudinal section of the chamber. It depicts the position of the gold pins contacting MEA pads, the connectors and the external board which connects the MEA to the external recording equipment. Inlet and outlet of heating water are also illustrated. C: Configuration of the whole system. It shows the connections between the chamber, the circulating bath and the bubbling module, which enhances the moisture-content of 10% CO2 air. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/bit]

is maintained using a commercial bubbling module (Okolab s.r.l., Pozzuoli (Na), Italy), which enhances the moisture-content of the mixture before delivering it into the chamber through the top plate. In order to prevent condensation, a tubing structure connected to the heating bath is placed over the top plate (Fig. 1C). Gold pins for electrical signal acquisition are located inside the chamber, in contact with MEA pads. Signals are carried outside to an external custom board (76 mm
115 mm) placed on the top plate (Fig. 1B), which arranges them in a single 68-pin socket that matches the standard Multi Channel Systems cable (MCS GmbH, Reutlingen, Germany). The electrical connection between gold pins and the external circuitry is allowed by means of three

connectors, which pass through the top of the chamber. For this purpose, three openings were designed with ProEngineer Wildfire, manufactured using subtractive rapid prototyping (Roland Modela MDX-40) and sealed by means of a silicone glue (Elastosil E43, Wacker Chemie AG, San Donato Milanese (Mi), Italy). The same milling machine was used to realize openings for the insertion of inverted microscope objectives beneath both the MEA housing and the reference well. The chamber is made of polymethylmethacrylate (PMMA) plates, glued by means of an acrylic adhesive. This material was selected due to its optical transparency. The whole system can be sterilized with Ultra Violet rays or with Ethylene Oxide (EtO).

Biffi et al.: Neuronal Environmental and Recording Chamber Biotechnology and Bioengineering

2555

Thermal Model Design A lumped parameter thermal model was realized to test the chamber design and to choose (1) the wall thickness of the inner and the outer boxes and (2) the controller characteristics. The electrical circuit that represents the analogue model of the system is shown in Figure 2A. In this

scheme, voltage represents temperature, and current represents heat transfer. Thermal features of each block are condensed in one resistor and one capacitor, whose values are determined by its size and material (Fig. 2A, dashed box). Accordingly, the MEA and the reference well were modeled with the same components due to their symmetry. Then, the transfer function between the

Figure 2. A: Circuitry representing the lumped parameter thermal model of the chamber. Q: Heat generator. R1, C1: inner walls resistance and capacitance; R2, C2: outer walls resistance and capacitance; Ca: capacitance of the air inside the chamber; Rt, Ct: top plate resistance and capacitance; Rm, Cm: resistance and capacitance of both cell culture medium (MEA well) and water (reference well); Rp, Cp: resistance and capacitance of the walls beneath the medium and the reference water. Dashed box: R and C are determined by the size and material of each block. d is the wall thickness [m]; k is the thermal conductivity [Wm1 K1]; S is the wall surface [m2]; c is the specific heat capacity [Jg1 K1]; r is the density [gm3] and V is the wall volume [m3]. B: Blocks of the Simulink model. C: Result of a thermal simulation with outer wall thickness ¼ 10 mm and inner walls thickness ¼ 5 mm (simulating time of 200 min, P ¼ 10 and I ¼ 0.1). The blue line is Tm; the red one is the controlled driving temperature. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/bit]

2556

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

temperature of the water ðTH2 O Þ, forced by the heat generator Q, and the temperature of the medium ðTm Þ was computed. This block was introduced into the model of the whole system, which was implemented in Matlab (Simulink). The input is composed by a ramp block, whose slope models the experimental heating curve of the commercial bath, and a saturation block limited to 37, which is the temperature set point. The chamber is described by the transfer function block, which is connected to the PI controller. A saturation block, whose lower limit is equal to 20 (i.e., the average environmental temperature), is interposed between them. This simulates the circulating thermostat controller which does not have the cooling function (Fig. 2B).

Materials and Methods Temperature, pH, and Osmolarity Tests The system was tested in terms of temperature profile, gas flow rate, and medium osmolarity. Temperature values measured by the PT100 probe were acquired and visualized with Wintherm1 Plus software (Lauda GmbH). Then, the chamber temperature profile was compared to the profile of the standard recording system (TC02, MCS GmbH). Concerning flow control, a flow meter (60–600 mL/min, Platon NG series, CT Platon SaS) was interposed between the air cylinder and the bubbling module. Ten Petri dishes, filled with medium (2 mL), were kept half in the chamber and half inside a standard incubator. The airflow in the chamber was varied from 60 to 600 mL/min and each flow value was maintained for 24 h. After this period, the pH of the medium was measured by a Beckman 350 pH meter (Beckman Coulter S.p.A, Cassina De’ Pecchi (Mi), Italy). A non-parametric statistical analysis (Mann–Whitney test) was performed with Statistica (StatSoft, Inc., Vigonza (Pd), Italy) to compare the two data sets. The significance level was established at P < 0.05. Median values and coefficients of variation (ratio between the interquartile range and the median value) are reported in the Results Section. Once the flow was selected, pH values of culture medium kept (i) inside the chamber, (ii) on a lab bench, and (iii) on a commercial equipment for neuronal signal recording (MEA-1060-Inv-BC-Standard MCS GmbH, 378C, without CO2 and humidity control) were measured at different time points (30 min, 1, 2, 4, 6, 15, and 24 h; n ¼ 3 in each group). During all the experiments, changes in pH values were evaluated qualitatively by monitoring phenol red, a wide used pH indicator, contained in the cell culture medium. Finally, osmolarity values of culture medium (n ¼ 4 in each group) kept (i) inside the chamber, (ii) in a standard incubator, and (iii) on a commercial equipment for neuronal signal recording (MEA-1060-Inv-BC-Standard MCS GmbH, 378C, without CO2 and humidity control) were measured over 48 h (OM-6050 Osmo Station, Arkray, Europe, B.V., Amstelveen, The Netherlands).

Substrate Preparation Round glass coverslips (diameter 24 mm) and standard 60-electrode MEA biochips (electrode spacing 200 mm, electrode diameter 30 mm; Multi Channel Systems, MCS GmbH) were used as substrate for cell plating. All the reagents and materials used were provided by Invitrogen (Monza (MB)), Italy if it is not differently specified. Coverslips were bathed 12 h in HNO3, thoroughly cleaned in MilliQ water and autoclaved. After the sterilization procedure, each coverslip was located in a small Petri dish, treated with poly-L-lysine (1 mg/mL, Sigma–Aldrich, srl, Milan, Italy) in 100 mM borate buffer pH 8.5 and placed overnight in a humidified incubator at 378C. Then, coverslips were carefully rinsed with sterile phosphatebuffered saline (PBS) to remove residues and finally incubated overnight with the plating medium (neurobasal medium (NBM) supplemented with 10% fetal bovine serum (FBS; Lonza, Group Ltd, Basel, Switzerland) and 1% penicillin/streptomycin (P/S)) the day before the dissection. MEA biochips were sterilized by means of an overnight treatment in oven at 1108C and filled with plating medium 4 h to increase their hydrophilicity. Afterward, they were treated overnight with poly-L-lysine (2 mg/mL) in a humidified incubator. After careful washings in PBS, MEAs were filled and incubated with the plating medium (NBM, 10% FBS, 1% P/S). Neuronal Culture Preparation Primary neuronal cultures were obtained from CD1 mice at E17.5. Hippocampi were extracted, rinsed with Hank’s Balanced Salt Solution (HBSS) and treated with Trypsin (0.25%; Sigma-Aldrich; 15 min, 378C). Then, cells were re-suspended in the plating medium (NBM, 10% FBS, 1% P/S) and mechanically dissociated using glass pipettes. Dissociated cells were plated at 200 cells/mm2 on coverslips and 800 cells/mm2 on MEAs and incubated for 4 h at 378C. Then, the plating medium was replaced with the cell culture medium (NBM, B-27 1
, 1% P/S, Glutamax 1 mM and 10 mM HEPES pH 7.4 (Lonza)). Neuronal cultures were split in two equal groups. One group was grown and maintained in a humidified incubator, while the other in the custom chamber, previously sterilized, up to 21 days in vitro (DIV). The 30% of the total amount of medium was changed every 2 days until the end of the experiments. Immunostainings Fourteen DIV neuronal cultures were fixed for 5 min in 4% paraformaldehyde, 4% sucrose, phosphate buffer 120 mM at pH 7.4 and 378C. Cells were rinsed in PBS 1
three times after fixing. Non-specific sites were blocked and cells were permeabilized for 1 h in 16.6% normal goat serum, 0.3% Triton X-100 in PBS. Cells were incubated in the same buffer with primary antibodies overnight at 48C as follows: rabbit-anti-(glial fibrillary acid protein) (GFAP, 1:2,000,

Biffi et al.: Neuronal Environmental and Recording Chamber Biotechnology and Bioengineering

2557

Dako, Aachen, Germany); mouse-anti-tubulin b-III (Tuj-1 clone, 1:1,000, Millipore Corporation, Billerica, MA). After rinsing five times with phosphate buffer at room temperature, cells were incubated for 2 h at room temperature with either Alexa-fluor 546 goat anti-mouse IgG (1:1,000) together with conjugated Alexa-fluor 488 goat anti-rabbit IgG (1:1,000). After rinsing four times in phosphate buffer with decreasing NaCl molarity, the nuclei of cells were stained for 5 min at room temperature with Hoechst-33342 diluted in phosphate buffer to a final concentration of 1:10,000. After one final rinse, coverslips were mounted with fluorescent mounting medium (Dako) and imaged through an epifluorescence microscope (Axiovert 135 TV, Zeiss s.p.a, Milan, Italy). A standard filter set was used to image Hoechst-33342, Alexa-fluor 546, and Alexa-fluor 488. Cell Viability Assays Cell viability was verified by propidium iodide/fluorescein diacetate staining at 7 DIV, 14 DIV, and 21 DIV. Cells grown on glass coverslips were incubated for 4 min in ringer solution (in mM: NaCl 130; KCl 5; KH2PO4 1.2; MgSO4 1.2; CaCl2 2; HEPES 25, Glucose 6) containing 15 mg mL1 of fluorescein diacetate (FDA, Sigma-Aldrich), 5 mg mL1 of propidium iodide (PI), and 3 mg mL1 of Hoechst-33342. Then, cells were washed with ringer and multiple images were taken by a CCD Camera (Exi-Blue Fluorescence microscopy camera, QImaging) mounted on an epifluorescence microscope (Axiovert 135 TV, Zeiss). A standard filter set was used to image Hoechst-33342, FDA, and PI, respectively. Images were then acquired and visualized with Volocity software (Perkin Elmer, Waltham, MA). At each time point, 20
images of glass coverslips cultured in the custom chamber (n ¼ 6) and in the control incubator (n ¼ 6) were acquired. Four fields per coverslip were taken and the percentage of living cells per field was counted and averaged. In order not to bias cell counting results, four fixed positions were previously selected in the microscope reference system. These positions matched the four corners of a 9 mm
9 mm, whose center was aligned with the coverslip housing and the microscope objective beneath. The percentage of healthy cells was calculated as the percentage of cells expressing FDA on the total nuclei expressing Hoechst-33342. Percentage of living cells was reported at every time point as mean  SD of six independent plates and a one-way ANOVA test was performed with Statistica (StatSoft, Inc.). The significance level was established at P < 0.05. Electrophysiological Recordings The standard equipment used for electrophysiological recordings consisted of a commercial preamplifier stage (MEA-1060-Inv-BC-Standard, gain: 55, bandwidth: 0.02 Hz–8.5 kHz, MCS GmbH), a Programmable Gain Amplifier (PGA64, gain: 50, bandwidth: 0.1 Hz–5 kHz, MCS GmbH) and a data acquisition system (USB-ME64,

2558

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

MCS GmbH). Recordings with the custom system were performed with the same equipment but the commercial preamplifier was replaced with the custom chamber. This was connected to the PGA (gain: 1,000, bandwidth: 0.1 Hz– 5 kHz) by means of the 68-pin socket placed on the external board. First the two recording systems were characterized and compared in terms of intrinsic electronic noise. Specifically, the root mean square (RMS) noise was measured (i) with all inputs grounded, (ii) with inputs connected to a MEA, which contained cell culture medium but not cells on the electrodes, (iii) with inputs connected to the same MEA chip and both the heating system and the gas cylinder switched on (measured only with the custom equipment), and finally (iv) with the input signals connected to a MEA chip with a neuronal network grown on electrodes. All RMS noises were estimated by computing the integral of the signal power spectral density (mV Hz1/2) in the band of the action potentials (between 300 Hz and 3 kHz), as previously done (Rolston et al., 2009). A nonparametric statistical analysis (Wilcoxon matched pair test) was performed with Statistica (StatSoft, Inc.) to compare the two data sets (significance level at P < 0.05). Median valued and coefficients of variation are reported in the Results Section. Then, to assess the recording feasibility inside the chamber, the spontaneous electrical activity of 3 hippocampal neuronal networks (13 DIV) grown on MEAs was recorded repeatedly for 5 min by both the standard and the custom setup at 20 kHz. The signal to noise ratio (SNR) was computed for each setup. Then, spikes were detected from raw data with MC_Rack Software (MCS GmbH), using for each channel a fixed threshold equal to five times the standard deviation of average noise amplitude in the first 500 ms of signal. Furthermore, signals acquired with the standard system were defined as the gold standard and the detection sensitivity (SE) of the custom setup was computed as follows (Equation 1): SB ¼

TP TP þ FN

(1)

where TP is the number of true positive and FN is the number of false negative. Afterward, off-line analyses were implemented in Matlab (The Mathworks, Natick, USA) as previously described (Biffi et al., 2011). Parameters extracted as descriptors of the spontaneous activity were the mean firing rate (MFR), the bursting rate (number of bursts per minute), the burst duration (s), the number of electrodes involved in network bursts and the network burst duration (s). These parameters were analyzed for each recording and median values of the three repeated data sets were computed. A non-parametric statistical analysis (Wilcoxon matched pair test) was performed with Statistica (StatSoft, Inc.) to identify differences between the two recording setups. The significance level was established at P < 0.05. Subsequently, recordings of spontaneous electrical activity of two 18 DIV neuronal networks, identical for cell type,

density and treatments, were performed. Specifically, one MEA was acquired for 3 h by the standard setup while the other one was recorded directly inside the custom chamber. Spikes were detected with MC_Rack. Then, the recording time courses were evaluated in terms of MFR. Frequency values were computed for each 1 min bin in Matlab and normalized with respect to the value of the first minute.

Results The Thermal Model The outcome of an acceptable thermal simulation (simulating time: 200 min) is depicted in Figure 2C. The picture shows that Tm takes about 30 min to grow from 208C (external environmental average temperature of the laboratory) to 378C and that, after a small overshoot at 398C, it starts to oscillate around the set point. This result was obtained defining the wall thicknesses equal to 10 mm for the outer wall and 5 mm for the inner one. Furthermore, the controller gains P and I were set to 10 and 0.007, respectively. Then, these values were chosen for the chamber production and the PI programming. Values of all parameters used in these simulations are listed in Table I.

The System Assembly The system was assembled as shown in Figure 3A and then connected to a 50 L/150 bar gas cylinder (10% CO2) by means of a rigid tube (white arrow). Figure 3B depicts the chamber from the top. On the left it shows the PT100 probe and syringes for medium refreshment. Next, there are the air inlet (yellow tube) and the air outlet, connected to a 0.2 mm filter to maintain sterility inside the chamber. On the right, the acquisition board that gets the 60 signals from MEA electrodes through the top slide is shown. Furthermore, Figure 3C is a bottom view of the chamber and displays the opening for an inverted microscope objective. Finally, Figure 3D shows the inner part of the chamber, which accommodates the MEA and the reference well. Gold pins for signal detection and the temperature probe are illustrated. The temperature tests revealed that the model nearly approximated the real system. Specifically, a ramp-like increase of the bath temperature caused the medium temperature to rise and to reach the set point after about 45 min. This value slightly depended on the initial temperature of the water in the reference well (data not shown). Furthermore, after a typical overshoot peak at Table I. Parameter Value

Figure 3. A: System assembly: the custom chamber is connected to the heating bath and to a 50 L/150 bar gas cylinder (not depicted) by means of an air tube (white arrow). B: The chamber view from the top: on the left the PT100 probe, syringes for medium and drug addition, the air inlet (yellow tube) and the air outlet; on the right the acquisition board. C: Bottom view of the chamber: opening for the inverted microscope objectives. D: Inside the chamber: the MEA housing and the reference well. [Color figure can be seen in the online version of this article, available at http:// wileyonlinelibrary.com/bit]

about 388C, the temperature measured by PT100 made few oscillations between 36.5 and 37.58C and finally reached a stable value around 378C (Fig. 4A, blue line). In order to reduce even more these temperature ripples, after the transient phase P was lowered from 10 to 2.5, a value found empirically. Figure 4A also shows the typical temperature profile of the standard recording system (MCS GmbH) with the temperature control switched on (red line). It oscillates between 36.9 and 378C, with an oscillation period of 1 min (resolution: 0.18C). No relevant differences between the

Values of components of the electric analogue model used during temperature simulations. R1 2.63

C1 354.23

R2 4.41

C2 839.66

Ca 16.18

Rt 2.69

Ct 342.86

Rm 20.57

Cm 4.15

Rp 116.96

Cp 1.30

R values are in K W1, C values are in J K1.

Biffi et al.: Neuronal Environmental and Recording Chamber Biotechnology and Bioengineering

2559

chamber (dots), on a lab bench (diamonds) and on a commercial equipment for neuronal signal recording (stars, MEA-1060-Inv-BC-Standard MCS GmbH, 378C, without CO2 and humidity control) for 24 h (median values, n ¼ 3 inside each group). These results highlight that the pH of medium maintained in environmental air (0.05% CO2) rises above 8 in less than 2 h and reaches values around 8.8, as previously reported (Potter and DeMarse, 2001). In contrast, the pH of medium kept in the chamber proposed here is stable for many hours, as in a standard incubator. Furthermore, the 70 mL/min flow induced a sufficient bubbling to humidify the air in order to avoid hyperosmolarity. Figure 4C depicts median osmolarity values of medium maintained inside the environmental chamber (dots), in a standard incubator (diamonds) and on a commercial equipment for neuronal signal recording (stars, MEA-1060-Inv-BC-Standard MCS GmbH, 378C, without CO2 and humidity control) over 48 h. The time points at 0, 24 and 48 h are shown in the graph (median values of 4 measurements). Osmolarity increases of about 13 mOsm day1 inside the custom recording chamber and 10 mOsm day1 in a standard incubator, as previously described (Potter and DeMarse, 2001). No significant differences were identified between the two data sets. In contrast, osmolarity increases of 55 mOsm day1 on the commercial recording equipment. Flow increase due to water evaporation inside the bubbler was prevented by periodically refilling it. With 70 mL/min flow, the 50 L gas cylinder, used to provide the 10% CO2 air, lasted more than 10 weeks. Finally, the optical monitoring requirement was verified. Figure 5 shows a neuronal network on microelectrodes of a MEA chip. The image was taken by using an inverted microscope (Axiovert 135 TV, Zeiss). The 5
differential interference contrast objective was inserted beneath the Figure 4.

A: Commercial recording system (MCS GmbH, red line) and custom chamber reference well (blue line) temperature profiles over a 20-h period. After the initial transient phase, no differences between the two systems were measured (inset). B: pH values of cell culture medium kept inside the chamber (dots), on a lab bench (diamonds) and on a commercial equipment for neuronal signal recording (stars, MCS, GmbH, 378C, without CO2 and humidity control) over 24 h. Data are reported as median values (n ¼ 3). C: Osmolarity values of cell culture medium kept inside the custom chamber (dots), inside a standard humidified incubator (diamonds) and on a commercial equipment for neuronal signal recording (stars, MCS GmbH, 378C, without CO2 and humidity control) over 48 h. Data are reported as median values (n ¼ 4). [Color figure can be seen in the online version of this article, available at http:// wileyonlinelibrary.com/bit]

temperature control of the chamber and of the standard recording system were determined at steady state (Fig. 4A, inset). With regard to the flow control, the gas flow value of 70 mL/min maintained medium pH in the physiological range. Accordingly, no significant differences concerning pH values were found between the custom chamber (7.49  0.07) and a standard incubator (7.6  0.12; P > 0.05, n ¼ 10 inside each group) after 24 h. Figure 4B shows pH values of cell culture medium kept inside the

2560

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

Figure 5. 5
differential interference contrast image (DIC) of a neuronal network grown on MEA. Scale bar 100 mm.

MEA housing through the opening in the bottom wall. The quality of the picture attests the possibility of duly monitoring cells inside the chamber during network maturation.

Chamber Effect on Cell Viability No biological contaminations of both coverslips and MEAs were observed during the culturing period. Moreover, the moisture content of the air was adequate to avoid considerable evaporation in between successive medium changes. A first qualitative assessment about the morphology of neuronal networks was performed comparing the immunostaining of two 14 DIV neuronal cultures. Figure 6A and B shows the fluorescence of a 14 DIV neuronal culture grown inside a standard incubator and the fluorescence of a 14 DIV neuronal culture grown inside the custom chamber, respectively. The green staining identifies the astrocytes, the red marks the neurons, and the blue labels the nuclei of the cells. The two 20
pictures highlight the same neuronal cell density, network organization and astrocyte morphology and proliferation. Then, the influence of the custom chamber on the cell vitality was quantitatively evaluated by propidium iodide/fluorescein diacetate stainings. Figure 7A and B shows the staining of a 7 DIV neuronal culture grown inside the standard incubator and a 7 DIV neuronal culture grown inside the custom chamber. No significant differences were found in cell viability (Fig. 7C, P ¼ 0.31, one-way ANOVA, n ¼ 6 in each group) under both culture conditions and at different time points. This indicates that the custom chamber neither alters the network development nor the neuronal viability over the network maturation time.

Chamber Effects and Advantages on Electrophysiological Experiments The RMS noise of the custom system was equal to 3.59  0.02 mV (n ¼ 5) with all inputs grounded (a measure of the electronic noise) and to 4.07  0.11 mV (n ¼ 5) with inputs connected to an empty MEA chip (with cell culture medium but not cells on the electrodes) and both the heating system and the gas cylinder switched off (a measure of both electronic and electrode thermal noises). No differences were observed with the input signals connected to an empty MEA chip and all the peripherals turned on (RMS noise equal to 4.04  0.21 mV, n ¼ 5). Finally, the RMS noise was equal to 4.15  0.17 mV (n ¼ 5) with the input signals connected to a MEA chip with a neuronal network grown on the electrodes. Afterwards, RMS noise values of the standard system used in our lab (MCS GmbH) were measured. RMS noise was equal to 1.73  0.02 mV (n ¼ 5) with all inputs grounded and to 2.93  0.13 mV (n ¼ 5) with the inputs connected to an empty MEA chip. These results evidence that both the electronic noise and the thermal noise of the electrodes are higher in the recording system proposed here than in the commercial equipment. This was ascribed to the absence of a metal shield around the custom board and to the presence of free wires, which acted as antennas collecting environmental noise. In contrast, the RMS noise evaluated with the input signals connected to a MEA chip with a neuronal network grown on the electrodes was equal to 3.15  0.63 mV (n ¼ 5), as measured with the custom chamber (Wilcoxon test, P > 0.05). Furthermore, these measures highlight that the intrinsic electronic noise of the system proposed here is smaller than electrode characteristic noises (5–20 mV peak to peak of thermal and biological

Figure 6. Immunofluorescence of two neuronal cultures at 14 DIV grown (A) inside a standard incubator and (B) in the custom chamber. Scale bars 100 mm. Red label: tubulin b-III; green label: GFAP; blue label: Hoechst-33342. Standard 40 -6-diamidino-2-phenylindole (ex: 365/10, em: LP397, dic: 395dclp), green fluorescent protein (ex: 450/40, em: 535/50, dic: 505dclp) and tetramethylrhodamine isothiocyanate (ex: 546/10, em: 575–640, dic: 560dclp) filter set was used to image Hoechst-33342, GFAP, and tubulin b-III (Tuj-1 clone), respectively. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/bit]

Biffi et al.: Neuronal Environmental and Recording Chamber Biotechnology and Bioengineering

2561

Figure 8. A: Equipment used during electrophysiological experiments. (1) Standard setup: MEA-1060 preamplifier (G: 55, bandwidth: 0.02 Hz–8.5 kHz), PGA64 programmable gain amplifier (G: 50, bandwidth: 0.1 Hz–5 kHz), USB-ME64 A/D card, computer. (2) Custom setup: custom incubating and recording chamber, PGA64 (G: 1,000, bandwidth: 0.1 Hz–5 kHz), USB-ME64 A/D card, computer. B: Circuit diagram of equivalent impedances referred to microelectrodes, MEA-1060 input and output (top), custom recording board (bottom), MCS cable and PGA64 input. R and C values were derived from datasheets. [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/bit]

Figure 7.

Propidium iodide/fluorescein diacetate staining of two neuronal cultures at 7 DIV grown (A) inside a standard incubator and (B) in the custom chamber. Green, blue, and violet arrowheads identify viable cells, necrotic/apoptotic cells, and necrotic cells, respectively. Scale bars 100 mm. Red label: propidium iodide; green label: fluorescein diacetate; blue label: Hoechst-33342. Standard 40 -6-diamidino-2phenylindole (ex: 365/10, em: LP397, dic: 395dclp), fluorescein isothiocyanate (ex: 450– 490, em: 515–565, dic: 510dclp), and rhodamine (ex: 546/10, em: 575–640, dic: 560dclp) filter set was used to image Hoechst-33342, fluorescein diacetate and propidium iodide, respectively. C: Statistical analysis of cell viability. Cell cultures grown inside a standard incubator (black) or in the custom chamber (gray) at three time points: 7 DIV, 14 DIV, and 21 DIV (P ¼ 0.31, one-way ANOVA, n ¼ 6 in each group). [Color figure can be seen in the online version of this article, available at http://wileyonlinelibrary.com/ bit]

noise), which is an important requisite of all custom systems (Bai and Wise, 2001). Finally, the measured noises were comparable to the literature, both with grounded electrodes (Charvet et al., 2010; Pancrazio et al., 2003; Rolston et al., 2009) and with electrodes connected to an empty MEA chip (Charvet et al., 2010). Figure 8A depicts the two setups used during electrophysiological experiments. The evaluation of SNR showed higher values in the standard setup (about 20 dB) than in the custom chamber (18 dB). As aforementioned, peak to peak noise values were comparable in the two configurations, whereas the signal amplitude was reduced in the custom setup (Fig. 9A and B). This can be ascribed to the absence of a preamplification board, located close to the signal sources, in the custom chamber. Indeed, the stage before the PGA acts as a low-pass filter (Fig. 8B) whose cutoff frequency is given by Equation (2): fcutoff ¼

1 2pRC

(2)

where R and C are the equivalent resistance and capacitance presented to the PGA input. C is mainly due to the parasitic shunt capacitance of the MCS cable (128 pF) while R-value is different in the two configurations. Considering the

2562

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

Figure 9.

Single channel activity (1s) acquired with the standard (A) and the custom (B) equipment (y-axes ranges from 100 to þ100 mV). Raster plot of 5 min activity recorded with the standard (C) and the custom (D) equipment.

commercial setup, R is the MEA1060 output resistance (300 V) while in the custom system it is mainly given by the source impedance, which results from electrode impedance and cell coupling/adhesion and exceeds 1 MV (Merril and Tresco, 2005). Therefore, fcutoff is equal to 4 MHz for the commercial setup (no signal components are cut) and to 1.2 kHz for the custom recording equipment, which is within neuronal spike frequency content. The signal decrease was further observed in the sensitivity values of the recordings performed by the custom system with respect to the gold standard. Indeed, the sensitivity, which measures the percentage of action potentials correctly identified as spikes, was about 85%. Moreover, a significant difference in spiking frequencies was detected (Wilcoxon test P < 0.05, Table II), which was certainly due to the lower SNR value of the recording chamber. In contrast, neuronal electrical activity recorded by the custom equipment showed a spiking organization qualitatively similar to standard

Table II.

recordings. Figure 9C and D shows the spiking structures (raster plot) of one network, whose activity was acquired by the standard and the custom equipment, respectively. The number of active electrodes and patterns of bursts and network bursts are clearly comparable. Furthermore, the quantitative data analysis confirmed these qualitative evaluations. Accordingly, no significant differences in burst and network burst features were identified (Table II). To verify the improvements in cell viability during neuronal activity recording by the custom chamber, 3 h acquisitions of two 18 DIV cultures grown on MEAs were performed. Figure 10 shows the kinetics of these recordings in terms of normalized mean firing rate. The time course of the recording performed by the standard setup (Fig. 10, stars) highlighted a slow decrease of spiking frequency after 30 min. Moreover, after less than 2 h, this network was completely silent, due to pH and osmolarity changes in the cell culture environment (Blau et al., 2009; Potter and

Results of single channel, burst and network burst (NB) analysis.

Standard system Custom chamber

Mean firing rate (Hz)

Burst length (s)

Burst rate (Hz)

NB electrodes

NB length (s)

1.7 W 0.6 0.7 W 1.1

0.19  0.1 0.16  0.0

2.4  1.4 1.5  1.6

0.6  0.1 0.6  0.1

0.6  0.2 0.4  0.1

Median  coefficient of variation (ratio between the interquartile range and the median value). No significant differences were found between the two recording systems, except for the mean firing rate (bold data Wilcoxon test P < 0.05).

Biffi et al.: Neuronal Environmental and Recording Chamber Biotechnology and Bioengineering

2563

Figure 10.

Time course of the normalized MFR for two MEAs recorded by the standard (stars) or the custom (dots) equipment for 3 h. Each marker is the MFR in 1 min of

activity.

DeMarse, 2001). On the contrary, the activity of the MEA placed inside the custom chamber (Fig. 10, dots) was stable for more than 3 h.

Discussion In the literature, several closed chambers designed to improve the reliability of long-term data are described, which attests to the necessity of similar devices in the biological field (Forsythe and Coates, 1988; Toyotomi and Momose, 1989). However, these chambers rarely provide a stable control on all the environmental parameters (Blau and Ziegler, 2001) and they are generally used only during the experimental sessions (Gross and Schwalm, 1994; Pancrazio et al., 2003; Potter and DeMarse, 2001) or they are not independent from bulky standard incubators (Mukai et al., 2003). In this work a closed chamber for neuronal growth and electrical activity recording is described. The chamber was realized in PMMA plates and it was assembled with a heating bath and a gas cylinder to provide a controlled environment for cell maturation. Temperature, pH, and osmolarity tests were performed to verify the fulfillment of the requirements. Results underlined that the time extent necessary to reach the temperature set point was as good as the one estimated by the thermal model, in agreement with literature data (Vukasinovic et al., 2009) and with data described by few commercial environmental chambers (e.g., ‘‘Chambered Slide Micro-Incubator,’’ Harvard Apparatus, Holliston, MA). Moreover, after few oscillations, which did not affect cell survival, as verified by vitality tests, temperature values

2564

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

remained stable for many days. Regarding pH measurements, they confirmed that the pH of medium kept inside the chamber proposed here was stabilized around 7.4 (as occurs in a standard incubator) with a constant 10% CO2 humidified gas flow and it was kept stable for days even at slow flow rate (less than 70 mL/min), which assured a small gas cylinder consumption. In contrast, pH values of medium kept on the lab bench and on the standard recording system drifted quite fast and rose above 8. This induces changes in the electrophysiological activity of neurons and provokes cell viability decline, which limits the experimental session duration (Potter and DeMarse, 2001), as then verified by long-term recording experiments. The moisture content of the air introduced inside the chamber, was adequate to avoid evaporation and the consequent hyperosmolarity in between successive medium changes, which was verified by measuring osmolarity values. These underlined the ability of the chamber proposed here to maintain the osmolarity increase in a range comparable to standard incubators and under the lethal value (50 mOsm day1, Potter and DeMarse, 2001). Therefore, the system showed an adequate control of environmental parameters that remained stable for many days, which is a prerequisite for long-term electrophysiological recordings. Finally, the system provides inlets for medium replacement and drug stimulation, which is an important requisite. Indeed, if a perfusion system is not available, the medium change is generally performed after the opening of the chamber under a sterile hood (Hales et al., 2010). This provokes mechanical stress and changes in cell homeostasis, due to the fast drop in CO2 air content. The system proposed here does not require those operations and thus assures the environmental stability for the whole cell

culture period. Concerning the environmental control, some improvements are still needed. Indeed, it would be useful to integrate a CO2 sensor and to maintain the acid-base homeostasis with pulses of CO2. This would consequently reduce the air consumption and it would allow the use of smaller cylinders. Furthermore, the integration with a closed perfusion system would maintain a constant medium level, which would assure an improved control of osmolarity values. Further, we quantitatively evaluated the influence of the custom chamber on the cell vitality by immunostainings and dead/alive labeling of neurons grown on standard coverslips. Our results highlight that in the custom chamber neuronal networks grow with morphology and viability comparable to cell cultures grown in standard incubators. Moreover, it alters neither the network development nor the neuronal viability over the culturing period. We maintained cells in culture up to 21 DIV since this is generally the period of in vitro development where modulation and shaping in the synaptic connectivity occur (Chiappalone et al., 2006). Anyway, we believe that our system, providing a controlled and stable environment in terms of temperature, pH and humidity, could guarantee cell viability even with longer experiments. Measures concerning the intrinsic noise of the electronics did not evidence relevant differences between the chamber, the commercial system used in our lab (MCS GmbH) and other custom systems proposed in the literature (Charvet et al., 2010; Pancrazio et al., 2003; Rolston et al., 2009). Furthermore, no influences due to the presence of the peripherals for the environment control were observed. The chamber allowed to repeatedly record neuronal electrical activity reducing mechanical disturbances and cellular stress. This activity showed physiological features, indicating that the custom chamber does not alter the network electrophysiology. Moreover, we compared a 3-h recording that was performed with the custom chamber, to an acquisition obtained by means of a commercial system (MCS, GmbH). This system integrates the temperature control but suffers from disadvantages concerning pH stability and osmotic balance maintenance. Therefore, it induces changes in neuronal activity in about 30 min and cell culture decline in less than 2 h (Blau et al., 2009; Potter and DeMarse, 2001). Our results confirmed the literature and showed that the custom chamber, that maintains temperature, pH, and osmolarity values stable for days, does not alter spontaneous electrophysiological activity and neuronal viability. Therefore, the system proposed hereby can be a solution for long-term recordings. The main problem observed during electrophysiological measurements was the value of the SNR, which influenced the sensitivity of the detection. Measures concerning the noise that overlaps with action potentials did not evidence significant differences between the chamber and thee commercial recording system (MCS GmbH). Therefore, the reduced SNR was attributed to the lack of a decoupling and preamplification stage immediately close to the

electrodes. Indeed, signal attenuation and noise coupling can be significant when external cables are used before any off-chip buffering (Bai and Wise, 2001). A schematic representation of the custom circuitry highlighted that the stage before the PGA acts as a low-pass filter whose cutoff frequency is within neuronal spike frequency content. Therefore, this causes signal attenuation and prevents from performing signal acquisitions comparable to the commercial system. Hence, a custom decoupling stage, placed next to signal sources, is necessary. A future improvement of this system will be the development of a dedicated preprocessing board, housed directly inside the chamber to reduce the effect of parasitic shunt capacitance and electrode impedance levels. The design of a single channel prototype is already ongoing. Besides, the development of an electronic board for data compression, as the one which was previously described by our group (Biffi et al., 2010), would give several advantages concerning data storage. Another strength of the system is that the electronic board did not show injuries caused by temperature and humidity, after a period longer than o1 month. Generally, the humid environment found in standard incubators (more than 60%) may cause electrical shorts, changes in component properties, and degradation of materials commonly used in electronic devices (Potter and DeMarse, 2001). For these reasons, few systems described in the literature proposed to introduce recording equipment inside dry incubators (Potter and DeMarse, 2001). In contrast, thanks to the electronic board sealing obtained with a silicone elastomer, we could increase the air content moisture of the chamber without the use of a cumbersome incubator. As a result, our system is also compact and portable. Finally, the ability of the system to observe cell cultures grown inside the chamber was demonstrated. The quality of the images guaranteed the possibility of monitoring the network maturation despite, the fact that the wall thickness beneath the MEA limited the use of short working distance objectives. Particularly, the 5 mm minimum working distance leads to the choice of phase contrast or DIC objectives with a maximum magnification of 20
. Although the system cannot work at higher magnification, 20
objectives allow studying the morphological and structural features of a neuronal network by bright field or fluorescence, coupling optical monitoring to electrophysiological studies. To conclude, the proposed system is a technological platform able to grow a neuronal network on a MEA chip since the very beginning of its maturation and to acquire at the same time its electrical activity during long-term experiments. The device assures environmental parameter stability and eliminates mechanical shock problems. Hence it should be considered an innovative platform for an electrophysiological laboratory, being independent from cumbersome incubators. Finally, since the chamber is not a medium-filled closed system, it is possible to adapt it for multiple MEA housings. In the literature, both the importance of this requirement and the limits imposed

Biffi et al.: Neuronal Environmental and Recording Chamber Biotechnology and Bioengineering

2565

by other configurations have already been discussed (Mukai et al., 2003; Pancrazio et al., 2003). Our platform would allow performing parallel long-term experiments with neuronal networks grown on MEA chips maintaining both the control and the treated cell cultures in the same environment. This configuration assures that variations observed among cell cultures are strictly dependent on the type of treatment and it offers more reliable and reproducible experiments within a compact and portable device. Authors would like to thank Dr. Luca Muzio for helpful discussions; they are also grateful to Dr. Marco Rasponi and Sergio Caspani for their help in the fabrication steps. Finally, authors would like to thank people from the Alembic facility for their support. The work was developed within the research line—biosensors and artificial bio-systems—of the convention between the Italian Institute of Technology and Politecnico di Milano. Emilia Biffi PhD is partly funded by Fondazione Confalonieri.

References Bai Q, Wise K. 2001. Single-unit neural recording with active microelectrode arrays. IEEE Trans Biomed Eng 48:911–920. Bavister B. 1988. A minichamber device for maintaining a constant carbon dioxide in air atmosphere during prolonged culture of cells on the stage of an inverted microscope. In Vitro Cell Dev Biol 24:759–763. Biffi E, Ghezzi D, Pedrocchi A, Ferrigno G. 2010. Development and validation of a spike detection and classification algorithm aimed at implementation on hardware devices. Comput Intell Neurosci 2010:659050. Biffi E, Menegon A, Piraino F, Pedrocchi A, Fiore GB, Rasponi M. 2011. Validation of long-term primary neuronal cultures and network activity through the integration of reversibly bonded microbioreactors and MEA substrates. Biotechnol Bioeng 109:166–175. Blau A, Ziegler C. 2001. Prototype of a novel autonomous perfusion chamber for long-term culturing and in situ investigation of various cell types. J Biochem Biophys Methods 50:15–27. Blau A, Neumann T, Ziegler C, Benfenati F. 2009. Replica-moulded polydimethylsiloxane culture vessel lids attenuate osmotic drift in long-term cell cultures. J Biosci 34:59–69. Charvet G, Rousseau L, Billoint O, Gharbi S, Rostaing JP, Joucla S, Trevisiol M, Bourgerette A, Chauvet P, Moulin C, Goy F, Mercier B, Colin M, Spirkovitch S, Fanet H, Meyrand P, Guillemaud R, Yvert B. 2010. BioMEA: A versatile high-density 3D microelectrode array system using integrated electronics. Biosens Bioelectron 25:1889–1896. Chiappalone M, Bove M, Vato A, Tedesco M, Martinoia S. 2006. Dissociated cortical networks show spontaneously correlated activity patterns during in vitro development. Brain Res 1093:41–53. Chiappalone M, Massobrio P, Martinoia S. 2008. Network plasticity in cortical assemblies. Eur J Neurosci 28:221–237. Christiansen G, Danes B, Allen L, Leinfelder P. 1953. A culture chamber for the continuous biochemical and morphological study of living cells in tissue culture. Exp Cell Res 5:10–15. Fantini E, Athias P, Courtois M, Grynberg A. 1987. A simple gas-flow chamber for cultured cell electrophysiology in a controlled atmosphere. Pflugers Arch 409:632–634. Forsythe I, Coates R. 1988. A chamber for electrophysiological recording from cultured neurones allowing perfusion and temperature control. J Neurosci Methods 25:19–27. Gross G, Schwalm F. 1994. A closed flow chamber for long-term multichannel recording and optical monitoring. J Neurosci Methods 52: 73–85.

2566

Biotechnology and Bioengineering, Vol. 109, No. 10, October, 2012

Hales C, Rolston J, Potter S. 2010. How to culture, record and stimulate neuronal networks on micro-electrode arrays (MEAs). J Vis Exp 39:2056. Heidemann S, Lamoureux P, Ngo K, Reynolds M, Buxbaum R. 2003. Opendish incubator for live cell imaging with an inverted microscope. Biotechniques 35:708–714, 716. Ho C, Mou T, Chiang P, Weng C, Chow N. 2005. Mini chamber system for long-term maintenance and observation of cultured cells. Biotechniques 38:267–273. Hofmann F, Bading H. 2006. Long term recordings with microelectrode arrays: Studies of transcription-dependent neuronal plasticity and axonal regeneration. J Physiol Paris 99:125–132. Howell KR, Kutiyanawalla A, Pillai A. 2011. Long-term continuous corticosterone treatment decreases VEGF receptor-2 expression in frontal cortex. PLoS ONE 6(5):e20198. Johnstone A, Gross G, Weiss D, Schroeder O, Gramowski A, Shafer T. 2010. Microelectrode arrays: A physiologically based neurotoxicity testing platform for the 21st century. Neurotoxicology 31:331–350. Merril DR, Tresco PA. 2005. Impedance characterization of microarray recording electrodes in vitro. IEEE Trans Biomed Eng 52:1960–1965. Morefield S, Keefer E, Chapman KD, Gross GW. 2000. Drug evaluations using neuronal networks cultured on microelectrode arrays. Biosens Bioelectron 15:383–396. Morin F, Takamura Y, Tamiya E. 2005. Investigating neuronal activity with planar microelectrode arrays: Achievements and new perspectives. J Biosci Bioeng 100:131–143. Mukai J, Shiina T, Jimbo Y. 2003. Continuous monitoring of developmental activity changes in cultured cortical networks. Electr Eng Jpn 145:28– 37. Pancrazio J, Gray S, Shubin Y, Kulagina N, Cuttino D, Shaffer K, Eisemann K, Curran A, Zim B, Gross GW, O’Shaughnessy TJ. 2003. A portable microelectrode array recording system incorporating cultured neuronal networks for neurotoxin detection. Biosens Bioelectron 18:1339– 1347. Potter S, DeMarse T. 2001. A new approach to neural cell culture for longterm studies. J Neurosci Methods 110:17–24. Rolston JD, Gross RE, Potter SM. 2009. A low-cost multielectrode system for data acquisition enabling real-time closed-loop processing with rapid recovery from stimulation artifacts. Front Neuroeng 2:1–17. Rose G. 1967. The circumfusion system for multipurpose culture chambers. I. Introduction to the mechanics, techniques, and basic results of a 12-chamber (in vitro) closed circulatory system. J Cell Biol 32:89–112. Shahaf G, Marom S. 2001. Learning in networks of cortical neurons. J Neurosci 21:8782–8788. Sykes J, Moore E. 1959. A new chamber for tissue culture. Proc Soc Exp Biol Med 100:125–127. Toyotomi S, Momose Y. 1989. Temperature-controlled perfusion apparatus for microscope using transparent conducting film heater. Am J Physiol 256:C214–C217. van Pelt J, Corner M, Wolters P, Rutten W, Ramakers G. 2004a. Longterm stability and developmental changes in spontaneous network burst firing patterns in dissociated rat cerebral cortex cell cultures on multielectrode arrays. Neurosci Lett 361:86–89. van Pelt J, Wolters P, Corner M, Rutten W, Ramakers G. 2004b. Long-term characterization of firing dynamics of spontaneous bursts in cultured neural networks. IEEE Trans Biomed Eng 51:2051– 2062. Vukasinovic J, Cullen D, LaPlaca M, Glezer A. 2009. A microperfused incubator for tissue mimetic 3D cultures. Biomed Microdevices 11:1155–1165. Xiang K, Earl DE, Davis KM, Giovannucci DR, Greenfield LJ, Jr., Tietz EI. 2008. Chronic benzodiazepine administration potentiates high voltageactivated calcium currents in hippocampal CA1 neurons. J Pharmacol Exp Ther 327:872–883.