In-device enzyme immobilization: wafer-level fabrication of an integrated glucose sensor

https://doi.org/10.1016/S0925-4005(03)00552-5Get rights and content

Abstract

Wafer-level fabrication of integrated enzyme-based BioMEMS usually requires high temperature wafer-bonding techniques such as anodic bonding. Enzymes denature at comparatively low temperatures. Thus, enzymes need to be immobilized after wafer bonding. A convenient in-device immobilization method is presented allowing wafer-level patterning of enzymes inside micro-scale flow channels after wafer bonding. Enzymes are entrapped in a poly(vinyl alcohol)-styrylpyridinium (PVA-SbQ) membrane crosslinked by UV exposure through a transparent top wafer. The reaction kinetics of immobilized glucose oxidase is investigated in more detail. A low apparent Michaelis constant of 3.0 mM is determined indicating a rapid diffusion of glucose into the PVA-SbQ membrane as well as an oxygen-limited maximum catalytic rate. The entrapped glucose oxidase preserves its native properties since it is not chemically modified. Furthermore, the active PVA-SbQ membrane can be dehydrated in a vacuum and later rehydrated in buffer solution without significant loss of enzyme activity. An integrated enzyme-based glucose sensor fabricated on a wafer-level using in-device immobilization is described to demonstrate the potential of this novel technique. The sensor is part of a disposable microneedle-based continuous glucose monitor. The stability of glucose oxidase entrapped in PVA-SbQ is sufficient to continuously operate the sensor at 25 °C for 24 h.

Introduction

Integrated enzyme-based BioMEMS often require the enzymes being immobilized inside a micro-scale flow channel; one example is a disposable microneedle-based continuous glucose monitor. The glucose monitor consists of a microneedle array to sample interstitial fluid [1], an electrochemical enzyme-based glucose sensor and fluidic components such as valves and pumps (Fig. 1). An integrated porous polysilicon dialysis membrane [2] separates the interstitial fluid from dialysis fluid, which is pumped through the microneedles. Glucose can diffuse through this membrane into the system while larger proteins are retained improving the sensor long-term stability. The integrated glucose sensor measures the glucose concentration in the dialysis fluid, which is related to the glucose level in the interstitial fluid. The glucose monitor also includes a separate calibration fluid reservoir and pumping system to enable periodic sensor recalibration, which is required due to loss of enzyme activity during continuous operation.

Wafer-level fabrication of such complex BioMEMS usually requires high temperature wafer-bonding techniques such as anodic bonding (T≥350 °C), which allows bonding of two structured wafers. Enzymes denature at comparatively low temperatures; glucose oxidase for instance denatures above 65 °C [3]. Thus, enzymes need to be immobilized after wafer bonding. A novel in-device immobilization technique has been developed allowing wafer-level pattering of enzymes inside micro-scale flow channels after wafer bonding [4]. The technique is based upon poly(vinyl alcohol)-styrylpyridinium (PVA-SbQ) [5], a water-soluble photosensitive polymer. PVA-SbQ is already successfully employed in many biosensors, which have the enzymes immobilized on uncovered electrodes [6], [7], [8], [9], [10]. A mixture of enzymes and PVA-SbQ is either spin-coated or directly deposited onto the electrodes and then UV crosslinked. However, none of these sensors have the enzymes immobilized inside a micro-scale flow channel, as required for many BioMEMS. In-device immobilization takes advantage of the real benefit of photochemical crosslinking which is the integration of biochemical components inside micro-scale flow channels. The enzyme–polymer mixture is filled into the flow channels after wafer bonding and then selectively exposed to UV light through a transparent top wafer using a shadow mask. Thus, enzymes are entrapped in the locally formed membrane inside the flow channels. The design of auxiliary channels, which connect all sensor devices and allow easy filling and flushing of the entire wafer through a single inlet is critical for wafer-level enzyme immobilization. This simple technique is generally applicable for any biochemical components that can withstand low energy UV exposure and allows wafer-level fabrication of integrated biosensors.

Michaelis and Menten proposed a simple model to account for the kinetic properties of free enzymes in solution [11]. Critical feature of their theory is that an enzyme E combines with a substrate S, with a rate constant k1, to form an enzyme–substrate complex ES, which is a necessary intermediate product in catalysis (1). The enzyme–substrate complex can dissociate, with a rate constant k2, or the product P is formed, with a rate constant k3.E+Sk2k1ESk3E+PThe catalytic rate V of the proposed reaction varies with the substrate concentration [S] following the Michaelis–Menten Eq. (2), which describes the kinetic characteristic of many enzymes including glucose oxidase under specific conditions. The Michaelis constant Km is equal to the substrate concentration at which the catalytic rate is half of its maximum value. It also measures the affinity of the enzyme–substrate complex in case of k2k3, which applies for most enzymes. The Michaelis constant of free enzymes in solution is independent of the enzyme concentration but depends on the particular substrate and also on ambient conditions such as temperature, pH value and ionic strength. The Km of glucose oxidase from Aspergillus niger with respect to glucose is reported to be 33 mM in sodium phosphate buffer (25 °C, pH 5.6, in air) [12].V=Vmax[S][S]+Km,Km=k2+k3k1The maximum catalytic rate Vmax and the Michaelis constant Km can be determined by plotting the reaction rate versus various substrate concentrations. Non-linear curve fitting with the Michaelis–Menten equation yields both parameters. However, there are two concerns interpreting the measuring results of enzyme-based sensors using the Michaelis–Menten model.

The kinetic characteristic of immobilized enzymes differs from the Michaelis–Menten model since the reaction rate is limited by the diffusion rate of the substrate into the membrane for low substrate concentrations. In this diffusion-controlled range, a linear correlation between the substrate concentration and reaction rate results (Fig. 2a). The Michaelis–Menten model still applies for high substrate concentrations, where the diffusion rate is much faster than the catalytic rate of the immobilized enzymes. Thus, the effect of immobilization on the enzyme properties can be quantified by fitting the enzyme-controlled part of the curve. Fitting of the entire curve yields an apparent Michaelis constant Km,app higher than Km for free enzymes in solution under identical ambient conditions (Fig. 2a). However, this apparent Michaelis constant measures both the effect of substrate diffusion into the membrane as well as the effect of immobilization on the catalytic performance of the enzymes. A large increase in Km,app combined with inaccurate curve fitting indicates a large diffusion-controlled range rather than a loss in enzyme activity. Correspondingly, the apparent Michaelis constant decreases with an increasing substrate diffusion rate. Furthermore, Km,app depends on the enzyme concentration since Vmax scales with the total enzyme activity while the diffusion-controlled part of the curve remains unchanged. Thus, the apparent Michaelis constant decreases with a decreasing enzyme concentration. It equals Km of soluble enzymes for very low enzyme concentrations.

In the case of a multi-substrate reaction such as the oxidation of glucose (3) another effect on the kinetic characteristic needs to be addressed. The catalytic rate can also be limited by stoichiometric restrictions. Thus, the oxidation rate of glucose depends on the oxygen and glucose concentration as well as on the catalytic efficiency of glucose oxidase. The Km of glucose oxidase from A. niger with respect to oxygen is reported to be 0.51 mM in potassium phosphate buffer (0.01 M, 25 °C, pH 6.6, 0.2 M glucose) [13].glucose+O2GODgluconicacid+H2O2

Low oxygen concentrations or high enzyme concentrations can lead to an oxygen-limited maximum catalytic rate Vmax,Ox. The apparent Michaelis constant Km,app drops below Km under such conditions (Fig. 2b). Furthermore, the apparent Michaelis constant decreases with an increasing enzyme concentration or decreasing oxygen concentration. However, the maximum catalytic rate is not oxygen-limited at high oxygen concentrations and low enzyme concentrations. The kinetic characteristic is described by the Michaelis–Menten equation under these conditions and the apparent Michaelis constant equals Km of soluble enzymes.

Rather than fitting the non-linear kinetic characteristic, a convenient way to determine the Michaelis constant Km and maximum catalytic rate Vmax is to plot the experimental data in a linear correlation such as the Lineweaver–Burke (4), Eadie–Hofstee (5) or Hanes form (6) of the Michaelis–Menten equation [14], [15]. Linear regression of the data then easily yields both parameters.1V=1Vmax+KmVmax1[S]V[S]=VmaxKmVKm[S]V=[S]Vmax+KmVmaxHowever, since the kinetic characteristic is not following the Michaelis–Menten model in case of a diffusion-limited catalytic rate V or an oxygen-limited Vmax, the effect of linearization itself on the apparent Michaelis constant needs to be considered under these conditions. Linear regression of the experimental data plotted in the Lineweaver–Burke form for instance yields a higher apparent Michaelis constant than non-linear curve fitting since low substrate concentrations are more weighted. This emphasizes the diffusion-controlled part of the curve. Linear regression of the same data plotted in the Hanes form on the other hand yields a lower Km,app since high substrate concentrations are more weighted [14]. Non-linear curve fitting yields an apparent Michaelis constant with a more balanced measure of the effect of substrate diffusion and oxygen limitation.

Section snippets

Enzyme immobilization

Enzyme solutions with glucose oxidase (GOD) concentrations of 0.6–45.6 μg/ml were obtained by mixing the initial glucose oxidase solution (5.7 mg/ml, glucose oxidase from A. niger, dissolved in sodium acetate buffer solution, 0.1 M, pH 4.0, Sigma–Aldrich, USA) with phosphate buffered saline (PBS) solution (0.01 M, pH 7.4, containing 2.7 mM potassium chloride and 0.137 M sodium chloride, Sigma–Aldrich, USA).

Poly(vinyl alcohol)-styrylpyridinium (SPP-13-H Bio, Toyo Gosei, Japan) was mixed 1:1 with the

Kinetic analysis of glucose oxidase immobilized in PVA-SbQ

As mentioned in Section 1, the effect of immobilization on the reaction kinetics can be quantified by comparing the apparent Michaelis constant with Km of soluble enzymes under identical ambient conditions. Since the exact conditions inside the PVA-SbQ membrane are unknown, especially during the glucose oxidation, a pH value of 7.4 and a temperature of 20 °C are assumed to be approximate ambient conditions. The pH value of 7.4 corresponds to that of intersitial fluid. The catalytic rate of

Conclusion

In-device enzyme immobilization is demonstrated as a convenient method for wafer-level patterning of enzymes inside micro-scale flow channels. A photosensitive polymer, poly(vinyl alcohol)-styrylpyridinium is mixed with buffer solution containing the enzyme, filled into the channels via capillary action after wafer bonding and then selectively crosslinked by UV exposure through a transparent top wafer using a shadow mask. Thus, enzymes get entrapped inside the locally formed gel. The unlinked

Acknowledgements

The Alexander von Humboldt Foundation and the DARPA BioFlips program have funded this research project.

Stefan Zimmermann received his diploma in electrical engineering in 1996 and his PhD in electrical engineering in 2001 from the Technical University Hamburg-Harburg (TUHH), Germany. He worked in the Department of Microsystems Technology at the TUHH from 1996 to 2001. His research focused on MEMS design and fabrication. In 2002, he received the research award “Promotionspreis der Metall- und Elektroindustrie” of the Nordmetall, Verband der Metall- und Elektroindustrie, Germany, for the

References (19)

There are more references available in the full text version of this article.

Cited by (48)

  • Microfluidic mechanoporation for cellular delivery and analysis

    2022, Materials Today Bio
    Citation Excerpt :

    The skin is also a highly beneficial location for vaccination because of its large surface area and presence of immune cells in abundance [85,91,106,117]. Apart from its ability to transdermally deliver a wide range of biomolecules, microneedle arrays have been studied for diverse applications, such as cosmetology, ocular and gastrointestinal delivery [117], glucose sensing [93,118] and transdermal fluid extraction [93,119]. There have been some notable developments in microfluidic integrated microneedle arrays [109,120,121].

  • Fabrication of a microdialysis-based nonenzymatic microfluidic sensor for regular glucose measurement

    2021, Sensors and Actuators, B: Chemical
    Citation Excerpt :

    They used the amperometry method in their attempt to detect the glucose in the fluid, which was being pumped into the channel. However, a major problem emerged as the passing fluid gradually washed away the immobilized enzyme and sent it out of the channel [12]. Zahn et al. utilized microdialysis method in their research for designing an integrated enzyme-based glucose-measuring system.

  • Microneedle-assisted technology for minimally invasive medical sensing

    2021, Microchemical Journal
    Citation Excerpt :

    Hydrogel or hollow MNs provide a more direct way to access ISF [92]. Zimmerman et al. reported the use of silicon hollow MNs for ISF sampling [93]. Hollow MNs can extract ISF with of help of capillary action [32,91].

  • Micro and nano-needles as innovative approach in nanomedicine

    2019, Nanostructured Biomaterials for Regenerative Medicine
  • Biosensors and Related Bioanalytical Tools

    2017, Comprehensive Analytical Chemistry
    Citation Excerpt :

    They have been demonstrated to be pain free and potentially low-cost and easy-to-use [106,107]. Over the last decade, they have been widely used for the transdermal delivery of a number of drugs and some studies have been reported on their use for glucose measurements by either extraction of biological fluid and successive analysis or in situ analysis [106–111]. In a recent work [112], hollow silicon-dioxide microneedles that are up to one order of magnitude both smaller (needle borehole with diameter of 4 mm) and more densely packed (1 × 106 needles/cm2) than those commonly used for biosensing have been proposed.

  • A flexible electrochemical glucose sensor with composite nanostructured surface of the working electrode

    2016, Sensors and Actuators, B: Chemical
    Citation Excerpt :

    New technologies are being pursued to develop novel glucose sensors. Zimmermann et al. [13] proposed an enzyme-based electrochemical flow-through glucose sensor that could be integrated into a flow channel and reported a detection limit of 20 mg/dL. Huang et al. [14] proposed a capacitive MEMS affinity sensor for continuous glucose monitoring at physiologically relevant concentrations ranging from 30 to 360 mg/dL.

View all citing articles on Scopus

Stefan Zimmermann received his diploma in electrical engineering in 1996 and his PhD in electrical engineering in 2001 from the Technical University Hamburg-Harburg (TUHH), Germany. He worked in the Department of Microsystems Technology at the TUHH from 1996 to 2001. His research focused on MEMS design and fabrication. In 2002, he received the research award “Promotionspreis der Metall- und Elektroindustrie” of the Nordmetall, Verband der Metall- und Elektroindustrie, Germany, for the development of a micro-flame ionization detector and a micro-flame spectrometer. In 2001, Dr. Zimmermann joined the Berkeley Sensor and Actuator Center at the University of California, Berkeley, USA, as a post-doctoral research engineer with support of a Feodor–Lynen Fellowship of the Alexander von Humboldt Foundation. His research is focused on BioMEMS. Dr. Zimmermann is currently working on a disposable continuous glucose monitor.

Doerte Fienbork received her qualification as a pharmaceutical technical assistant in 1994 from the Dr. von Morgenstern School, Lueneburg, Germany, and her practicing license as a pharmacist in 2000 from the Christian-Albrechts University of Kiel, Germany. In 1997 during her studies, she worked at the School of Pharmacy at the University of London, England, focusing on advanced biochemical analysis methods. While working in different pharmacies from 1999 to 2001 in Hamburg, Germany, she specialized in the treatment of diabetes. In 2001, Doerte Fienbork joined the Berkeley Sensor and Actuator Center at the University of California, Berkeley, USA, as a junior specialist. She is currently working on the application of BioMEMS to medical problems.

Albert William (Bill) Flounders received his BS degree in chemical engineering from The Johns Hopkins University, and his PhD in chemical engineering from the University of California, Berkeley. Dr. Flounders completed post-doctoral research in immunology at the US Department of Agriculture, Western Regional Research Center, Albany, CA. His research involved immobilization, stabilization and patterning of proteins for biosensor development. From 1996 to 2001, Dr. Flounders was a senior member of technical staff at Sandia National Laboratories in Livermore, CA. His research focused upon integration of multiple biological receptors and adaptation of MEMS fabrication tools to biochemical processing. Dr. Flounders is presently the technology manager of the Microlab at the University of California, Berkeley, and is responsible for design and development of Berkeley’s new fabrication laboratory planned for opening in 2006.

Dorian Liepmann is the vice chair and Lester John and Lynne Dewar Lloyd distinguished professor of bioengineering at UC, Berkeley. He received his BS and MS in chemical engineering from Caltech, and his PhD from the University of California, San Diego. Before coming to Berkeley in 1993, Dr. Liepmann worked in the Advanced Technology Group at SAIC in La Jolla, on projects for the Navy. Previously Dr. Liepmann was a member of the Arroyo Center providing technology and policy analysis for the Army. He joined the Mechanical Engineering Department at UC, Berkeley in 1993 and the Bioengineering Department in 1999. Dr. Liepmann is a director of the Berkeley Sensor and Actuator Center (BSAC). He is currently working on the application of MEMS to biological and medical problems.

View full text