Pregnant ICR mice (E18.5) were purchased from DBL (Eumseong, Korea) and sacrificed for primary culture of neurons and glia, following a previously described protocol with minor modifications [5]. Briefly, embryos were decapitated and the entire hippocampi were dissected out and were treated with papain (Worthington) and serially triturated. Dissociated cells were counted and seeded at a density of 1×10⁵ cells ml^-1 for 2D and 4×10⁶ cells ml^-1 in a collagen mixture for 3D. Cells were cultured in plating medium consisting of neurobasal media supplemented with 5% fetal bovine serum (GIBCO), 2% B27-supplement (Invitrogen), 2 mM Glutamax-I (GIBCO) and 1% penicillin-streptomycin (GIBCO). After 1 day, plating medium was replaced by serum-free medium and maintained at 37℃ in a 5% CO₂ humidified incubator. One-half of the medium was replaced with fresh culture medium every 2~4 days. Cultured cells growing in 500 µl media were treated with 1 µl adeno-CMV-mCherry virus (viral stock titer: 4.5×10¹² genome copies (GC) ml^-1) for 4 days.

Commercial type I collagen (8~11 mg ml^-1 in acetic acid (0.1% [w/v] for custom-extracted collagen and 0.02 N for commercial collagen), rat tail; Corning or custom-extracted) was used and neutralized (pH 7.5) collagen solutions were prepared as reported previously [5]. Total of 500 µl of a 2.5 mg ml^-1 collagen solution was prepared by adding 110~160 µl collagen stock (depending on the concentration of collagen stock solution), 50 µl of 10x Dulbecco Modified Eagle Medium (DMEM; Sigma-Aldrich). 10~20 µl of 0.5 N NaOH for neutralization, and 50 µl of cell-suspension in DMEM. Final volume of 500 µl was matched by adding 1x DMEM (Lonza). Cells were seeded in collagen at a density of 4×10⁶ cell ml^-1. All solutions were added on ice to minimize undesired gelation. Collagen seeded with cells was then immediately processed for gelation (37℃, 30 min), and media was added.

To stain neurons and astrocytes, 3D constructs were fixed in 4% [w/v] paraformaldehyde and blocking was performed with 2% [w/v] BSA in PBS containing 0.1% [w/v] Triton X-100 for 2 hr. 3D constructs were then incubated sequentially with primary and secondary antibodies diluted in the blocking solution at 4℃ overnight. Following primary antibodies were used: mouse anti-βIII-tubulin (TuJ1) (1:1,000; Abcam), chicken anti-GFAP (1:500; Millipore). Alexa Fluor conjugates (Alexa Fluor 488 and 594) (1:1,000; Molecular Probes) were used for secondary antibodies. Nuclei were stained with Hoechst 33342 (1: 5,000; Molecular Probes). All samples were rinsed with PBS between the incubation steps.

Fluorescence images were acquired using an inverted confocal laser scanning microscope (LSM 700; Carl Zeiss) equipped with solid-state lasers (405, 488 and 555 nm). Post-image processing such as maximum intensity projection and 3D reconstruction was performed using ZEN 2012 software (Carl Zeiss). Z-stacked images (stack size, 17~176 µm; step size, 0.94~3.4 µm) were acquired with 10x and 20x objectives.

Electrophysiological recordings were performed with co-culture of hippocampal neurons and astrocytes at DIV (days in vitro) 7 and 14. Whole-cell patch-clamp recordings were achieved by applying collagenase as previously described [5]. For tonic GABA measurement, whole-cell recordings were made from neurons. The holding potential was -60 mV. Pipette resistance was typically 5~8 MΩ and the pipette was filled with an internal solution (in mM): 135 CsCl, 4 NaCl, 0.5 CaCl₂, 10 HEPES, 5 EGTA, 2 Mg-ATP, 0.5 Na₂-GTP, 10 QX-314, pH adjusted to 7.2 with CsOH. For passive conductance measurement, whole-cell recordings were made from astrocytes. The holding potential was -70 mV. Pipette resistance was typically 5~8 MΩ for granule cells and the pipette was filled with an internal solution (in mM): 140 K-gluconate, 10 HEPES, 7 NaCl, and 2 MgATP adjusted to pH 7.4 with CsOH. Electrical signals were digitized and sampled at 50 µS intervals with Digidata 1440A and Multiclamp 700B amplifier (Molecular Devices) using pCLAMP 10.2 software. Data were filtered at 2 kHz.

Mice (8~10 weeks old, C57BL/6 background) were anesthetized by intraperitoneal injection of 2% avertin (20 µl g^-1) and placed into stereotaxic frames. Adenovirus containing mCherry was loaded into a micro dispenser (VWR, USA) and injected bilaterally into the hippocampal dentate gyrus (DG) region at a rate of 0.3 µl min^-1 (total 2 µl) with a 25 µl syringe using a syringe pump KD Scientific, USA). The stereotaxic coordinates of the injection site were 1.7 mm away from the bregma and the depth was 2 mm beneath the skull. All experimental procedures described below were conducted according to the animal welfare guidelines approved by Institutional Animal Care and Use Committee of the Korea Institute of Science and Technology.

The significance of data for comparison was assessed by Student's two-tailed unpaired t-test between two groups and one-way ANOVA test between three groups. Analyses were performed with Prism (GraphPad Software, Inc.) and Clampfit software. The data distribution was assumed to be normal. Data are presented as mean±SEM (standard error of the mean). Levels of statistical significance are indicated as follows: ^*(p<0.05), ^**(p<0.01), ^***(p<0.001).

To assess the morphology of cells in the 3D neuron-astrocyte co-culture, we performed immunocytochemistry by using antibodies against Tuj1, a neuron-specific class III β tubulin, and GFAP, an astrocytic filament protein. In striking contrast to neurons that show similar morphology in 2D and 3D cultures, astrocytes in the two cultures exhibited a remarkable difference. In the standard 2D culture, astrocytes had a flat, polygonal morphology, which is very different from their in vivo counterparts. Collagen-based 3D culture supported more complex cellular morphology (Fig. 1A). Astrocytes possessed a small soma and several, clearly distinct processes, resembling a stellate morphology of resting-state astrocytes observed in vivo. With a side view of 3D-cultured astrocytes, we could readily identify processes branching out in several directions from soma (Fig. 1B). These morphological properties of 2D- and 3D-cultured astrocytes are similar with previous findings [11,21] and show that astrocytes can maintain key morphological features of the in vivo or ex vivo astrocytes when cultured in 3D.

High-titer adenoviral transduction of astrocytes has been shown to cause local and selective virus-induced reactive astrocytosis in vivo [22]. Although it has been speculated that interactions between viral DNA or capsid with receptors expressed in astrocytes might lead to preferential astrogliosis, the mechanism underlying the induction of astrogliosis is not understood. To examine if we could mimic the induction of reactive astrocytes in our 3D culture system, we infected the culture with adenovirus expressing mCherry as a reporter gene under the control of a CMV promoter. Upon treatment of cells with high-titer virus (1 µl of 4.6×10¹² GC ml^-1 viral stock solution into a total of 500 µl of culture media), astrocytes displayed a hypertrophic morphology typical of reactive astrocytes (Fig. 2A, C, D and E). The cell population in neurons and astrocytes was not affected by adenoviral infection as compared to control condition (Fig. 2B). Increase in GABA contents has also been reported in hippocampal reactive astrocytes from Alzheimer's disease model mice [16,19]. In our 3D culture, we observed a significant increase of GABA signal in adenovirus-infected reactive astrocytes as compared to control astrocytes (Fig. 3A and B).

We next compared morphological and biochemical characteristics of reactive astrocytes in adenovirus-infected hippocampal brain slice with those in our 3D cell culture system. Similar to our 3D culture (see Fig. 1, 2, 3A and 3B), we found cellular hypertrophy and significant increase of GABA signal in reactive astrocytes from the DG region of hippocampal brain slice infected with adenovirus (Fig. 3C and 3D). These ex vivo results were reminiscent of a previous report from in vivo studies [22]. However, adenoviral induction of astrogliosis was rather less selective in hippocampal slice. In addition to astrocytes that were directly infected with adenovirus, as evidenced by the expression of mCherry, astrocytes in the vicinity of the infected cells showed hypertrophy and up-regulation of GFAP, which might be due to inflammatory responses resulting from physical injury caused by injection of adenovirus into the hippocampus.

The increased intracellular GABA content in reactive astrocytes can be released tonically from astrocytes and detected in neighboring neurons in the form of tonic inhibition current [16,19]. To test this in 3D culture at DIV 7, we measured the tonic GABA current using the GABA_A antagonist bicuculline from a neuron in the vicinity of viral infection (Fig. 4A). We found a significant increase in tonic GABA current in adenovirus treated condition compared to control condition (Fig. 4A, B), indicating an increase in tonic GABA release from reactive astrocytes. It has been suggested that astrocytes synthesize GABA through monoamine oxidase B (MAO-B) pathway with putrescine as an initial substrate [16,23,24]. To test if putrescine pathway is intact in our 3D culture system, we measured the tonic GABA current in normal 3D-cultured astrocytes treated with putrescine. We found a significant enhancement of tonic GABA current after putrescine treatment (Fig. 4C). We found a very similar trend in the DIV 14 culture, compared to DIV 7 condition, except that the magnitude of putrescine- and adenovirus-induced tonic GABA current was much higher at DIV 14 (Fig. 5A~D). These results suggest that tonic GABA release from reactive astrocytes is detected in adenovirus-infected reactive astrocytes.

Fully differentiated, mature astrocytes have been shown to display a leaky membrane property with extremely low membrane resistance (ranging from 1~10 MΩ), namely a passive conductance with a linear current-voltage relationship [25]. It has been recently demonstrated that the most of passive conductance is mediated by the heterodimer of two-pore potassium (K2P) channel subunits, TREK-1 and TWIK-1 [26]. Therefore, the presence of passive conductance can serve as a useful electrophysiological marker of astrocytes. To test whether our 3D-cultured astrocytes display this electrophysiological property, we measured the passive conductance from individual astrocytes by whole-cell patch-clamping. We found that the passive conductance was readily observed in 3D-cultured normal and reactive astrocytes (Fig. 4E-G, 5E-G). Interestingly, passive conductance was significantly increased in reactive astrocytes at DIV 14 (Fig. 5G), but not at DIV 7 (Fig. 4G). This result is consistent with previous reports that showed an enhancement of passive conductance in ischemic condition accompanying reactive astrocytes due to an increase in K2P channel expression [27,28]. Taken together, these results suggest that 3D-cultured reactive astrocytes display the electrophysiological features as evidenced by the increased tonic GABA release and passive conductance, resembling pathophysiological conditions in vivo.