INTRODUCTION Glucose sensing is widely used as one of the evaluation methods for cellular activities and functions, since glucose plays an important role in principal source of biological energy. A cellular glucose uptake rate is frequently determined from culture media collected at specific times by conventional analytical instruments. However, due to the low time resolution, dynamic cellular responses are not evaluated in detail. Meanwhile, most of glucose sensors in the conventional analytical instruments often utilize glucose oxidase (GOx) as a biological recognition element for the analysis of glucose. From the amount of H2O2 generated via the GOx-catalyzed oxidation of glucose by oxygen, glucose concentration is indirectly determined. However, even if such GOx-based glucose sensor was used for cell culture systems, the accurate concentration of glucose would not be determined due to the cellular oxygen consumption based on respiratory metabolism. Additionally, H2O2 is well-known to induce apoptotic or necrotic cell death. In the present work, we prepared an oxygen-independent glucose dehydrogenase (GDH)-modified single-walled carbon nanotube (SWCNT) electrode, at the surface of which a cellulose dialysis membrane was modified to improve the stability of the electrochemical signal, and demonstrated its performance of direct glucose monitoring in the cell culture system. EXPERIMENTAL An electrode for monitoring of glucose levels based on GDH-adsorbed SWCNTs (GDH/SWCNT electrode) was prepared as follows. SWCNT dispersion was cast onto the surface of a glassy carbon electrode. After physically adsorption of 1,4-naphthoquinone (NQ), which promotes direct electron transfer between GDH and SWCNT, at the surface of SWCNTs, an aliquot of a mixture of aqueous solution containing GDH and hexamethylenediamine was cast onto the NQ-adsorbed SWCNT electrode surface, subsequently followed by casting of a glutaraldehyde solution to conjugate GDH with NQ-adsorbed SWCNTs rigidly. The obtained electrode surface was covered with Nafion, further followed by covering cellulose dialysis membrane (CDM). Monitoring changes in glucose levels during hepatic glucose metabolism was performed by means of amperometry at +0.4 V vs. Ag|AgCl in a Dulbecco’s Modified Eagle Medium (DMEM)-based culture medium containing 5 mM glucose and 10% fetal bovine serum (FBS). The working electrode was located several handled micrometers from human hepatoma cell line Hep G2. RESULTS AND DISCUSSION We first examined the glucose concentration dependence of the steady-state anodic current densities for GDH/SWCNT electrodes with and without CDM in the phosphate buffered saline (pH 7.4). The responses at the both electrodes linearly increased with increasing glucose concentration up to 30 mM and then leveled off. However, the slope for the GDH/SWCNT electrode with CDM was slightly smaller than that without CDM. This indicates that CDM might restrictsupplying glucose to the GDH/SWCNT surface. We next confirmed the feasibility of the GDH/SWCNT electrode to amperometric monitoring of glucose level for real-time cell-based assays. We examined long-term stability of the GDH/SWCNT electrode in the DMEM-based culture medium containing 5 mM glucose and 10% FBS. In the absence of CDM, the anodic current rapidly decreased less than 40% after 23 h, but in the presence of CDM, the anodic current was observed 70% or larger even after 24 h. We therefore employed the GDH/SWCNT electrode with CDM for a real sample. Figure 1 shows anodic current responses observed at the CDM-covered GDH/SWCNT electrode located near the bottom surface of a culture dish in the presence and absence of Hep G2 cells in the DMEM-based culture medium with and without cisplatin, which is well-known as one of the anti-cancer drugs. In the absence of Hep G2, an anodic current in the presence of cisplatin showed the similar behavior for that in the absence of cisplatin, so that cisplatin unaffected to the current responses. Meanwhile, in the presence of Hep G2 cells without cisplatin, anodic current rapidly decreased because of their glucose metabolism. In contrast, a current response first rapidly, and then, slightly decreased in the presence of both Hep G2 and cisplatin. The former current change rate was the similar case to the presence of Hep G2 cells without cisplatin, indicating cellular glucose metabolism because cisplatin hardly worked soon after its exposure to Hep G2 cells. The latter current change rate was almost the same as that in the absence of Hep G2 cells with cisplatin, indicating the inhibition of their glucose metabolism due to damage of Hep G2 cells by cisplatin. Thus, the CDM-covered GDH/SWCNT electrode would be applicable to non-invasive cell-based safety assays as an index of changes in cellular glucose metabolism. Figure 1