We have developed an individual-cell-based on-chip cell-cultivation system consisting of an agarose microchamber (AMCA) cell cultivation array maintained at constant temperature, controlled atmosphere and humidity (Figure 1). We used the AMCA cell cultivation chip to create a neural cell network from individually grown cells. The AMCA was fabricated on the surface of 0.5 mm thick glass slide coated with 50 nm thick indium-tin oxide (ITO). Prior to agarose coating, we have coated the ITO layer with collagen I and Poly-D-Lysine to improve cells' adhesion to the surface (Figures 2(a),2(b),2(c)). In order to spatially separate cells, define their positions, and the network pattern, a 10-μm-thick agarose layer on the AMCA chip surface was etched by spot heating using the infrared laser to form microchambers and microchannels (Figure 1). Wells and tunnels were formed within the agarose layer according to the trace of the heating spot of laser. To create a precise microchamber design, we checked the position of the heating spot with an optical microscope prior to irradiating the chip with 1064 nm laser beam. We also confirmed whether the irradiation time and intensity was sufficient for desired size and shape by direct microscopic observation, as described previously [6,7]. Size and shape of the melted areas was controlled by manipulatingthe intensity of laser power and the magnification of objective lenses. First, as shown in Figures 2(d),2(e) and 2(h),2(i), we irradiated 35-mW focused laser using ×10 objective lens to form a 30-μm-diameter round microchambers for cultivating cells. The size of the wells was monitored by optical microscope and the heating continued until the wells reached a desired size. Next, we etched the 5-μm-wide tunnels (microchannels) between two adjacent holes with 17-mW laser using X20 objective lens (Figures 2(f) and 2(j)). It should be noted that at this stage we have connected only one chamber with the tunnel. The tunnel width was limited to 5 μm in order to prevent cell bodies passing through the tunnels. Finally, we melted a portion of agarose layer at the end of the tunnel to connect two microchambers (Figures 2(g) and 2(k)). For this experiment, we used low-melting-point (LMP) agarose that melted at about 65°C. The use of LMP It is required in order to minimize the damage to cells within the microchambers when we change the shape of microstructures during cultivation. Another important requirement for the agarose is an absence of contaminations, since impurities such as dust particles absorb 1064-nm-wavelength laser light and may cause unpredicted and uncontrolled melting of the agarose.

Figure 3 shows the schematic drawing of hippocampal cell cultivation in AMCA chip. Prior to cultivation microchannels were connected only to one side of two adjacent microchambers. Then, individual cells were introduced into each AMC and cultivation was started as described in Figs 2(f) and 2(j). To control the number of cells and cell types in each chamber, we used a micropipette to introduce individual single cells into each chamber. After neurites were sufficiently elongated into each microchannel, the other sides of the channel were connected by photo-thermal etching (see Figs 2(g) and 2(k)) to allow elongated neurite to connect to cells in the neighbouring chamber. Figure 4 illustrates an example of such a procedure. The micrographs show phase-contrast images of the growth of single hippocampal cells in three microchambers. When the cultivation started, single cells were transfered to AMCAs (Fig. 4(a)). Six hours later, single neurites elongated from cells into microchannels (Fig. 4(b)). At that time additional photo-thermal etching was used to connect two adjacent AMCAs (Fig. 4(c)). Two hours later (8 h total cultivation time), all of the cells kept their shape and continued their elongation (Fig. 4(d)). Eventually these cells were connected by their neurites counter-clockwise. (Figs 4(e) and 4(f), 14 and 28 h cultivation, respectively). The use of the stepwise photo-thermal etching has ensured the unidirectional elongation of cells. If no stepwise photo-thermal etching is use (i.e. if each microchambers had two channels), cells in such microchambers elongated in both directions (clockwise and counter-clockwise) simultaneously as shown in Figure 5(a) (5 h after cultivation started).

To evaluate the ability and limit of the stepwise photo-thermal etching method used during cultivation, we need to measure the extent of damage incurred on the collagen layer by localised heating. The collagen layer is essential for the stable attachment of neural cells onto the chip, but it is heated by the focused infrared laser during photo-thermal etching. We have therefore examined the effect of heating on the collagen layer using a collagen coated chip (Figures 6(a),6(b),6(c),6(d)). Figure 6(a) is the micrograph of the 20-hour-cultivation of hippocampal cells on the collagen layer. Figure 6(b) shows the cultivation of the same cells on the collagen layer, which was incubated for 60 min at 65°C before cultivation started. The incubation temperature, 65°C, was determined by the melting of LMP agarose (at 65°C). Figure 6(c) illustrates the cultivation of the same cells on the collagen layer, which was incubated for 60 min at 90°C before cultivation started. In this case the incubation temperature, 90°C, was determined by the possible highest temperature by spot heating of infrared beam, ca. 90°C. As a reference, we also cultivated the same cells on the chip without a collagen coating (Figure 6(d)). As shown in Figures 6(a) and 6(b), the difference in the cell growth on the pre-heated and not heated chips. Even when the chip was pre-heated 90°C for one hour, the growth of neuritis was nearly identical on all chips (pre-heated and non pre-heated). In this experiment chips were irradiated for one hour, in contrast to a regular experiment when no more than few seconds of irradiation is required to produce a channel. Thus the effect of photo-thermal etching on collagen layer was smaller than the result shown in Figures 6(b) and 6(c). If the collagen molecules were diffused, the result must be same as the result shown in Figure 6(d), where cells could not attach.

We have also checked the suitability of a collagen layer for cell cultivation using an AMCA cultivation chip. In the AMCA chip, the collagen layer is between ITO layer and top agarose layer. To manufacture the cultivation wells, we used 1064-nm 35-mW focused laser for 2 s with ×10 objective lens to form 50-μm wide round microchambers. Figure 6(e) is the micrograph taken after 20-hour of cultivating hippocampal cells in these microchambers. Figure 6(f) is the micrograph of the hippocampal cells grown under the same conditions as Figure 6(e), except for the AMCA chip re-coated with collagen after photo-thermal etching. As seen on Figures 6(e) and 6(f) little or no difference was observed between cells grown in different chambers. However, the neuritis on the collagen re-coated chip climbed over the microchambers (Figure 6(f). These results indicate that the collagen layer between ITO layer and agarose layer kept their performance even after photo-thermal etching.

AMCA chips were kept at in a constant temperature and under controlled atmosphere and humidity (37°C, 5%, respectively) Aphase-contrast/fluorescent optical microscope (IX-70; with a phase-contrast objective lens, ×20, Olympus, Tokyo, Japan) with a focused 1064-nm infrared laser irradiation unit (max. 1 W; PYL-1-1064-M, IPG Photonics, Oxford, MA, USA) was used to melt the agar layer on the chip. The objective lens in the microscope was used to simultaneously observe the chip surface and to focus the 1064-nm laser. A series of phase-contrast images of cell growth and network formation was acquired by using a charge-coupled device (CCD) camera (CS230, Olympus) and recorded in the computer system with a video capture board.

To attach the collagen onto the ITO surface, the chips were washed twice with 80% ethanol and with PBS, air dried and treated with 2 ml of 150-μg/ml collagen solution (pH 3.0) (Collagen type I-C (from pig skin): Nitta Gelatin, Tokyo, Japan). Following 24 h incubation at room temperature, the chip was washed with PBS once and incubated with 2 ml of a 100-μg/ml Poly-D-Lysine solution (Poly-D-Lysine: SIGMA) for 24 h at room temperature. Following the incubation, the chip was washed with PBS. Collagen and Poly-D-Lysine treated surfaces were coated with 2% (w/v) agarose (ISC BioExpress, GenePure LowMelt: melting temp. 65°C) using a spin-coater (500 rpm for 5 s followe by 4000 rpm for 20 s). The agar-coated chips were placed in a refrigerator at 4°C. The microstructures within the layer were designed using a photo-thermal etching procedure.

Rat hippocampal cells were obtained from 18-day-old fetuses (E18) following a dissection protocol as described previously [7]. The isolated tissue was incubated in 0.25% trypsin (Sigma) in Ca²⁺- and Mg²⁺-free Hank's balanced salt solution (HBSS, Gibco) for 8 min at 37°C. After trypsination the tissue was rinsed in a 2-ml plating medium (Neurobasal medium with B27 supplement, Gibco) five times for 5 min and mechanically dissociated with a fire-polished pipette into single cells. The cells were placed one by one into each agar microchamber with a micropipette and incubated at 37°C with 5% CO₂ at saturated humidity (Figure 3). We used a conditioned serum-free medium (Neurobasal medium with 0.074 mg/ml L-glutamine, 50 μg/ml gentamycin, 2% (v/v) B-27 supplement 25 μg/ml, Gibco and 50 μg/ml L-glutamic acid hydrochloride, Sigma). Hippocampal glial cells were cultivated for 2 weeks.