In vitro long-term performance study of a near-infrared fluorescence affinity sensor for glucose monitoring

Abstract

The long-term in vitro performance of a fluorescence affinity sensor for transdermal blood glucose monitoring was investigated. Affinity binding of fluorescently labeled concanavalin A (ConA) was used in this application, as previously described by Ballerstadt and Schultz [Anal. Chem. 17 (2000) 4185–4192). In this paper, the fluorescence emission of the sensor was extended to the near infrared (670 nm) using Alexa647™ as the fluorochrome conjugated to concanavalin A. Sensors were alternately exposed to glucose solutions having concentrations of 2.5 and 20 mM with a dwell time of 3 h. The optical output of the sensors was monitored over a 4-month period. The sensors showed an initial increase in fluorescence over the first 3–4 weeks before gradually decreasing, with an approximately linear drop of 25% per month. In order to understand the reasons for the decrease in fluorescence output, further experiments were conducted, including time-dependent membrane leakage tests, solubility tests of ConA, temperature-dependent activity tests of ConA, and fluorescence photo-bleaching tests.

From these results, it became evident that the decrease in fluorescence was not due to denaturation of the ConA. The most likely cause was leakage of the fluorescently labeled ConA through the interface between the outer sealant and the membrane. This problem is considered to be solvable and future publications will address this issue. Extrapolation of the experimental data suggests that a leak-proof sensor would be remarkably stable with a fluorescence decrease of only 15% over a 1-year period.

Introduction

A long-term implantable glucose sensor is a desirable alternative to the ex vivo blood glucose tests strips, and the 3-day sensor implants currently available for monitoring blood glucose. Various sensor concepts for long-term glucose monitoring have been proposed, including enzyme-based glucose sensors (Abel et al., 1984, Gerritsen et al., 1998, Heller, 1999, Pickup and Thevenot, 1993, Wilkins and Atanasov, 1995) and numerous noninvasive spectroscopic techniques (Arnold, 1996, Khalil, 1999, Klonoff, 1997, Koo et al., 1999). Schultz and co-workers demonstrated the use of the glucose-specific binding protein concanavalin A (ConA), a lectin from the plant canavalia ensiformis (Sumner and Howell, 1936), in various fluorescence based affinity glucose sensors (Mansouri and Schultz, 1984, Meadows and Schultz, 1993, Schultz et al., 1982). Other groups have employed ConA in viscometric affinity assays (Ballerstadt and Ehwald, 1994, Beyer et al., 2001), glucose-sensitive hydrogels (Lee and Park, 1996, Obaidat and Park, 1997), piezoelectric crystal device (Barnes et al., 1991), and in fluorescence energy transfer assays (Ballerstadt and Schultz, 2000, Rolinski et al., 1999, Russell et al., 1999, Tolosa et al., 1997, Ballerstadt and Schultz, 1997). The principal advantages of employing ConA as a glucose sensitive sensor element are its oxygen independence and reversible binding mechanism (no glucose consumption).

In a paper by Ballerstadt and Schultz (2000), a novel fluorescence affinity sensor with dramatically improved fluorescence yield was described. The concept was to implant a fluorescently labeled ConA containing sensor capsule just under the skin and to irradiate it with light of an appropriate wavelength, through the skin (Brumfield et al., 1998). A portion of the fluorescence generated by the implanted sensor penetrates back through the skin and is measured with a photodetector (the intensity of the fluorescent light being proportional to the local glucose concentration). In this scheme, a semipermeable sensor capsule is fabricated, which is filled with a high surface area substrate containing glucose-residues (e.g., Sephadex beads). This is then dyed with a dye that strongly absorbs light at the excitation and emission wavelengths of the fluorescent dye bound to ConA (in the case of Ballerstadt and Schultz, 2000, the fluorescent dye was fluorescein that emits at 520 nm). In the absence of glucose, a majority of ConA binds to the large fraction of glucose residues inside the dyed porous beads.

Since the incoming excitation light is absorbed by the dye on the Sephadex beads, the non-bound fluorochrome–ConA generates only minimal fluorescence. However, in the presence of sufficient glucose the bound fluorochrome–ConA is displaced and is free to diffuse out of the beads, leading to a large increase in fluorescence. The authors reported that three different sensors were still functional after a period of 3 months when stored at room temperature, at constant glucose concentration, and not exposed to light. In this particular design, the optimal excitation wavelength was 480 nm, while the peak in the fluorescence emission was in the 520 nm range. While this design showed that a sensor based on affinity binding could detect glucose in vitro, under laboratory conditions, it would have difficulty measuring glucose in vivo due to the strong absorption of 480–600 nm light by skin tissue components, such as hemoglobin. In addition, it would be inappropriate to draw any conclusions as to the sensors long-term in vivo performance at body temperatures of 37 °C under continuous glucose exposure and long-term exposure to laser light.

In this work, we monitored the fluorescence response of the affinity sensor while simulating the in vivo conditions of the human body (continuously cycling glucose concentrations at body temperature) over a period of months. Taking into account the strong light absorption of skin in the 480–600 nm range, we chose to conjugate ConA with the fluorophore Alexa647™, having a peak absorption at 647 nm, and an emission peak at 670 nm wavelengths at which skin is nearly transparent.

Our main objective was to study the long-term stability of the sensor. This included studying the photostability of the fluorochrome, the inherent stability and binding activity of ConA as a function of temperature, light exposure, and glucose concentration. We also investigated the stability of the semipermeable membrane, and package design.