Next Article in Journal
Comparing Proteolytic Fingerprints of Antigen-Presenting Cells during Allergen Processing
Previous Article in Journal
Interactive Roles of DNA Helicases and Translocases with the Single-Stranded DNA Binding Protein RPA in Nucleic Acid Metabolism
Previous Article in Special Issue
Bimolecular Rate Constants for FAD-Dependent Glucose Dehydrogenase from Aspergillus terreus and Organic Electron Acceptors
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Micro-Electromechanical Affinity Sensor for the Monitoring of Glucose in Bioprocess Media

1
Chair of Bioprocess Engineering, Department of Biotechnology, Technical University Berlin, ACK24, Ackerstr. 76, 13355 Berlin, Germany
2
IHP, Im Technologiepark 25, 15236 Frankfurt (Oder), Germany
*
Author to whom correspondence should be addressed.
Present address: RISE Acreo, Box 787, SE-60117 Norrköping, Sweden.
Int. J. Mol. Sci. 2017, 18(6), 1235; https://doi.org/10.3390/ijms18061235
Submission received: 8 February 2017 / Revised: 23 April 2017 / Accepted: 31 May 2017 / Published: 8 June 2017
(This article belongs to the Special Issue Biomolecular Engineering and Bioelectronics)

Abstract

:
An affinity-viscometry-based micro-sensor probe for continuous glucose monitoring was investigated with respect to its suitability for bioprocesses. The sensor operates with glucose and dextran competing as binding partner for concanavalin A, while the viscosity of the assay scales with glucose concentration. Changes in viscosity are determined with a micro-electromechanical system (MEMS) in the measurement cavity of the sensor probe. The study aimed to elucidate the interactions between the assay and a typical phosphate buffered bacterial cultivation medium. It turned out that contact with the medium resulted in a significant long-lasting drift of the assay’s viscosity at zero glucose concentration. Adding glucose to the medium lowers the drift by a factor of eight. The cglc values measured off-line with the glucose sensor for monitoring of a bacterial cultivation were similar to the measurements with an enzymatic assay with a difference of less than ±0.15 g·L−1. We propose that lectin agglomeration, the electro-viscous effect, and constitutional changes of concanavalin A due to exchanges of the incorporated metal ions may account for the observed viscosity increase. The study has demonstrated the potential of the MEMS sensor to determine sensitive viscosity changes within very small sample volumes, which could be of interest for various biotechnological applications.

1. Introduction

The continuous monitoring of specific metabolites and proteins is a key method towards establishing a thorough understanding and better control of bioprocesses in biotechnology. Continuously determined process data have the potential to improve the efficiency of methods and strategies like Design of Experiments (DoE), Quality by Design (QbD), and soft sensors. Continuous parallel and miniaturised sensors are especially of interest when miniaturised approaches are applied for bioprocess development, for example, when modern model-based DoE approaches with on-line refitting of model parameters are applied [1].
Regarding the substrates, glucose is of main interest, as it is the most commonly used carbon source in bioprocesses. Continuously operating enzyme-based glucose sensors have been reported since the 1980s [2] as has their use for the monitoring of the glucose concentrations in bioprocesses [3,4]. While the in-line use of biosensors in bioreactors is limited due to the need to resist the autoclavation process, other problems like enzyme stability, interfering and inhibiting substances in the medium [5], heat sensitivity of receptors [6], pH dependency of the enzyme activity [7], and sensor drifts have not been solved yet in ways that have led to sufficiently functional devices [8]. To date, only laboratory equipment and at-line systems (e.g., BioSenz Analyzer from Applikon, Delft, Netherlands) are used, and despite numerous publications and approaches, ready-to-use in-line biosensors for bioreactors are still unavailable.
Next to enzymatically operating biosensors, affinity-sensors for the continuous monitoring of biomolecules have gained attention in recent years. Similar to immuno-assays, an affinity-assay recognizes the analyte via specific binding to a receptor. However, as the interactions are non-covalent, non-catalytic, and reversible in nature, affinity-based sensors may be more robust and more easily applicable for continuous monitoring in bioprocesses compared to enzyme-based sensors. Binding constants of the analyte-receptor couple are normally in the range of 102 to 105 M−1. For comparison, binding constants of antibody-antigen couples are on the order of 108 M−1 and below.
One example for such an affinity system that has been studied thoroughly is a sensor for glucose based on the plant lectin concanavalin A (ConA) acting as the receptor. ConA, with a size of 237 amino acids and a molecular weight of 26.7 kDa, utilizes two divalent metal ions for carbohydrate binding, which are located close to the ligand binding pocket. Cation binding sites S1 and S2 are occupied by Mn2+ and Ca2+ in the protein’s active state [9], although other divalent transition metal ions like Zn2+, Co2+, Ni2+, Cd2+, etc. may equally reside on S1 [10,11]. The incorporation of Mn2+ and Ca2+ results in a binding constant of 320 M−1 for ConA’s target molecules, which are α-d-glucosyl and α-d-mannosyl residues [12].
For glucose detection, the competitive binding between glucose and dextran to ConA is utilized. Dextran molecules are of globular shape and consist of α-1,3-linked glucose chains, at which ConA can bind with a binding constant of 1.5 × 104 M−1 [12]. Different approaches were realized to measure the glucose concentration cglc in this competitive binding reaction [13]. Initial approaches made use of fluorescence detection [14,15] as did many of the following [16,17,18,19], but optical [20,21], osmotic pressure [22], and dielectric methods [23] were also investigated.
Reliable results were also obtained by transforming the cglc signal into a fluid-mechanical parameter, the dynamic viscosity η of the ConA/dextran assay [24,25]. This approach was first proposed and investigated by Ballerstädt and Ehwald [13,26]. Based on the tetrameric structure of ConA at physiological pH and the terminal glucosyl residues of a dextran molecule, mixtures of both become highly viscous. This viscosity depends on the mixing ratio and the number of bonds between ConA and dextran. Free glucose weakens the network, as glucose competes for carbohydrate binding sites, and, as a result, the viscosity decreases (see Table 1). To exploit the effect, the assay has to be separated from the analyte solution by a semipermeable membrane, which blocks the macromolecules [27,28,29]. In this way, the viscosity η of the assay becomes a function of the glucose concentration cglc outside the membrane.
Using modern microelectronic methods, a sensor chip for the measurement of viscosity was developed which utilizes a micro-electromechanical system (MEMS) [30]. The measurement of viscosity is carried out by using a quasi-electrostatic principle. For this purpose, a cantilever beam is moved by electric forces through the assay solution, with its velocity depending on the viscosity of the latter [31]. The sensor system was originally developed as a medical implant [32], i.e., for continuous glucose monitoring in human patients suffering from diabetes mellitus and has a very small size [33]. Various studies were performed on the affinity assay in contact with mammalian [34] and human tissue [35,36]. However, so far this approach has not been applied yet for glucose detection in bioprocess media.
In this work, we investigated if the sensor can be readily used in a typically applied defined mineral salt medium for the cultivation of Escherichia coli. Unexpectedly, and differing from the application in tissue environments, different medium components affect the sensor output and must be considered for application of the sensor.

2. Results and Discussion

The functionality of the sensor probe was first tested by inserting it into standard electrolyte (SEL) at cglc = 0 mM, and the switching time tsw was recorded. The parameter tsw of the MEMS served as a measure for the viscosity of the solution under test. It stands for the time it takes to bend the metallic cantilever beam to a defined position, where the bending is caused by electrical forces between the cantilever and ground plate (for construction details, see the Materials and Methods section). Typical bending amplitudes amount to about 1 µm, since the distance between the cantilever and ground plate initially is 2.5 µm, while tsw values were found to range between 5 and 300 µm in previous investigations [29,31], depending on the viscosity of test solutions. Next to viscosity, the switching time is also affected by the temperature and the electrical conductivity of the test solution.
During the test measurement tsw was recorded every one minute for 16 days. It can be seen from Figure 1A that tsw stayed constant at a mean value of 28.9 ms with a standard deviation of 0.2 ms, i.e., with variations of less than 1%. The data demonstrate a remarkable stability of the assay and the sensor chip throughout more than 2.3 × 104 measurements. In particular, the stable sensor data indicates the semipermeable membrane between the sensoric assay and the test solution to be free of defects, which would cause leaking out of ConA and/or dextran and a drift of sensor signals.
Following this first functionality test, the sensor was placed into a typical bacterial phosphate buffered mineral salt medium [37] with cglc = 0 mM, abbreviated by MS or Noack in the following. Figure 1B shows the course of tsw for a time span of 3 days. The switching time tsw increases rapidly, exhibiting rates between 1 and 10 ms h−1 without reaching a stable end value within the duration of the experiment.
Returning the sensor to SEL with cglc = 0 mM (Figure 1C) results in a fast decrease of the switching time tsw which then stabilizes at a value higher than the one prior to exposure to the mineral salt medium. This shows that the response consists of a reversible and an irreversible part.
We safely exclude the effect caused by erroneous measurements, because it could be reproduced and because of the high stability of sensor operation as demonstrated in SEL. It has been shown by [31] that glucose concentrations cglc can be derived with 1% precision from the switching time tsw in SEL medium.
An important physical parameter is the electrical conductivity σ, which naturally depends on the concentrations of electrolytes and their mobility in the solution. The conductivity roughly doubles when going from SEL to Noack medium. An electrical field is operative between the MEMS cantilever and its bottom plate due to the quasi-electrostatic deflection scheme. The energy of the high-frequency (HF) field is also transformed into kinetic energy of the surrounding ions and the share of the latter increases with the ion concentration. It thus takes an increased amount of time for the cantilever to reach the switch-off position if the conductivity σ is higher (dtsw/dσ > 0).
In principle, the change of tsw is in accordance with the rise of conductivity. The conductivity difference, however, cannot explain the time constant of hours and days, on which tsw steadily increases. Diffusion constants D in the order of 1 to 2 × 10−9 m2 s−1 for electrolytes like Na+ and Cl would predict electrolyte concentrations cel to equilibrate within seconds in the cavity and the test solution after changing the medium.
Another difference between SEL and Noack medium is the pH value. It drops from 6.85 to 6.65. Although this could principally result in a certain disaggregation of ConA tetramers, this disaggregation should lead to a weakening of the ConA/dextran network and therefore to a drop of viscosity and tsw. In contrast, the viscosity remained nearly constant (data not shown).
In fact, the large time constants suggest biochemical rather than physical causes account for the observed tsw increase. A large part of the sensor response is reversible, and the speed in which the switching time drops after returning the sensor to SEL implies a mechanism that can be reversed simply by diffusion of Noack medium components out of the sensor cavity. The irreversible change of viscosity, on the other hand, seems to not be dependent on the concentration of the influencing medium components.
In order to investigate which component might be responsible for the change of viscosity, different components of the medium were separately added to SEL in the same concentration as they are present in Noack medium. The tests were performed at cglc = 0 for MgCl2, NH4Cl, FeCl3, and the phosphate buffer components. Each solution was tested by monitoring ∆tsw with the sensor probe, starting from the value initially measured in SEL. After monitoring for 24 h, the sensor was returned to unmodified SEL solution until no further change in tsw could be observed.
Figure 2A–D shows that all the components cause an increase in tsw. The strongest increase in tsw of around 9 ms in 5 h was caused by the addition of phosphate buffer (Figure 2D). For the addition of FeCl3, the effect of an increase in conductivity, as discussed above, is clearly obvious (Figure 2C). The short response time is caused by the fast diffusion of Fe3+ and Cl into the sensor cavity. Another fast increase in tsw occurs with MgCl2. However, a longer time constant of about 1 h can also be observed, but the increase stops after approximately 2 h. On the other hand, the presence of NH4Cl and FeCl3 leads to a constant increase of tsw during the observation time. Again, the initial increase can be understood by the higher conductivity of the solutions. The later increase has to have other reasons than the diffusion of media components into the sensor cavity, which should be at equilibrium at that point of time. In fact, the persistent tsw increase suggests an induced interaction of assay components among themselves.
Fit functions f of the measured transient curves fMg, fFe, fNH4, and fpho are also displayed exhibiting either linear or exponential behaviour. A summation of all four fit functions and a comparison of the ∆tsw response to the one in Noack medium are shown in Figure 3. The comparison shows that the component effects do not follow a linear superposition, but are reinforcing each other. As for possible causes, we suggest the following: (i) An exchange of the two divalent metal ions for carbohydrate binding within the lectin could lead to a decreased binding constant which would strengthen the ConA/dextran network. One or more chemical intramolecular binding reactions between (ii) dextran or (iii) ConA molecules could also strengthen the network of the assay components; (iv) The electro-viscous effect caused by a bigger protein ion-cloud. At the actual point of investigation, other possible explanations cannot be excluded, but the mechanisms mentioned appear the most likely ones based on our experimental findings.
(i)
When the two cation binding sites S1 and S2 are occupied by ions, the protein undergoes a conformational change that involves the isomerization of a non-proline peptide bond from trans- to its cis-conformation. The isomerization presents a certain energy barrier that needs to be overcome. This change has been called “locking” and it largely increases ConA’s affinity to the bound ions. It has been shown that only locked ConA can bind to carbohydrates, but also metal-free ConA can be in its locked/native state. Incubating with excess carbohydrates results in a shift of unlocked/locked equilibrium of metal free ConA from around 12% to over 60% of the locked conformation in an experimental set up. Varying the ion species in S1 and S2 has shown little change in binding strength to the ligand. However, the most important divalent metal cation Ca2+ is missing in the growth medium, and even though no direct explanations can be found in the literature, it is evident that ConA is a very cation sensitive protein with three cation binding sites, and therefore variations in ligand binding strength due to changes in specific cation concentrations cannot be ruled out.
(ii)
We also considered the phosphorylation of glucosyl residues that could lead to covalent bonds between dextran molecules, which would lead to a permanent increase in viscosity by strengthening the molecular network of the assays. In the lab, phosphorylation of dextran was achieved under exclusion of water at high temperatures (90 and 120 °C, respectively for 6 h, only approximately 10%) and low pH [38,39]. Therefore, phosphorylation appears rather unlikely to happen spontaneously under the conditions present in the sensor cavity.
(iii)
Some of the salts used in Noack medium are applied in protein crystallization protocols, namely NH4+, SO4+, and phosphate. Although their concentration in the medium is much smaller, the ions mentioned belong to the most effective salting-out agents of the Hofmeister series [40]. Thus, it might be expected that aggregation of ConA to higher mers than tetramers strengthens the intermolecular network between assay components and causes a higher viscosity. This assumption is supported by the fact that the applied concentration of ConA is close to its solubility limit in solutions of electrolytes used in the medium [41].
(iv)
Agglomeration induced by electrolytes and the associated increase in viscosity were already described by Kim and Myerson for lysozyme [42]. Concentrations of the involved ions and the apparent time constants were of the same order of magnitude as observed in this work. Their study also pointed to the electro-viscous effect that accounts for the property of proteins to surround themselves with a cloud of dissolved ions. The cloud composition, which depends on ionic species and concentrations, determines the forces necessary to move the protein molecule within the solvent, and thus the viscosity of the solution.
It is evident from the considerations given above that the intention to use the affinity-viscometric sensor probe for continuous glucose monitoring in bioprocesses faces several challenges. It has to be emphasized, however, that these objections do not apply to the usage of the sensor probe in human body implants, where variations in δT and δσ, for example, can be neglected and where pH is effectively buffered by bicarbonate.
In a next step, a new probe was used to monitor changes of cgls throughout a cultivation experiment. As a preparation, the sensor was calibrated in MS medium with glucose concentrations ranging from 1 to 50 mM, the results of which are shown in Figure 4. During the course of the calibration measurements, cglc = 1 mM was measured twice after 0.5 and 3 h, showing a mean tsw value of only 3.53 ms with a standard deviation increasing from 0.15 to 0.3 ms. Compared to the sensor response in Figure 1B, this results in only around 12% of the drift observed in Noack medium without glucose. Having glucose in the sensor cavity therefore decreased the sensor response to the MS medium by a factor of eight.
In order to monitor a real cultivation, nine samples were taken from the cultivation medium for off-line measurements. The increase in turbidity measured as absorbance at a wavelength of 600 nm (OD600) confirms bacterial growth throughout the cultivation. Meanwhile a steady decrease of glucose concentration in cultivation supernatant could be shown both by the sensor and reference measurement with an enzymatic assay (see Figure 5). It is important to note that during the monitoring experiment the sensor was submerged in the cultivation samples for intervals between 20 and 30 min.
The cglc values measured with the glucose sensor were similar to the measurements with the enzymatic assay, with a difference of less than ±0.15 g·L−1 for every measurement, with the exception of an outlier at 5 h. This shows that in a suitable application, the sensor produces both a reliable calibration curve and measuring results that are similar to an established method. The relative constancy of the glucose concentration, and thus the larger error in the glucose concentration, between 2 and 5 h of the cultivation is possibly related to the continuous pH decrease, which results in a lower viscosity and thus smaller tsw values, most probably due to a loss of the intramolecular interactions in the ConA tetramer. This pH dependency is critical for the application of the sensor, which thus would have its application potential mainly for pH controlled cultivations which are standard in industrial bioprocesses.
The difference in viscosity response to medium containing either glucose or no glucose points towards (a) agglomeration of ConA and (b) the change of the binding constant by an exchange of the two divalent metal ions for carbohydrate binding within the lectin. Both mechanisms would be affected by the presence of glucose. Statistically, glucose binds about 40 times stronger to ConA than glucosyl residues, so any strengthening of the ConA binding would not result in as much of a rise in viscosity.
While the viscosity-based MEMS sensor for glucose presented here seems to be promising, this study shows that, for applications which apply continuously changing media, special optimization of the viscous polymer network is necessary. However the presented sensor is directly applicable under conditions where a defined assay environment is created, for example, by the fast at-line analysis of glucose in samples collected from parallel cultivations in a liquid handling station in experimental approaches [1].

3. Materials and Methods

The preparation of sensor chips on CZ-Si 200 mm wafers was conducted by semiconductor technology using the IHP-proprietary 0.25 µm SGB25V technology [43]. The MEMS cantilever made from bio-stable TiN was initially embedded with interlayer dielectric layers [44] from SiO2, but was released in a chemical etching step after bringing the wafer out of the clean room. For this purpose, the liquid from the last rinsing procedure was superseded with CO2 by critical point drying in order to prevent unwanted static friction or stiction of the cantilever to the bottom plate [45]. Afterwards, the bond pads for the electrical connection were provided with Au stud bumps and the sensor chips were separated using stealth dicing [46]. Every chip was then flip-chip bonded to a flexible circuit board for electrical connection.
Bonded chips were glued into a cooling body in order to dissipate the thermal energy generated during measurements. Integrating a semipermeable membrane with a cut-off at 2.2 nm [29] into the cooling body forms a cavity with a volume of about 1 µL (see Figure 6A,B). The cavity was filled with the affinity assay through a channel at the back, which was sealed afterwards. The affinity assay consisted of ConA and dextran dissolved in SEL solution added with functional cations [47] and Tris buffer.
The measured quantity of the sensor MEMS is the switching time tsw, which is the time needed to deflect the cantilever to a defined position. The sensor chips are subjected to a direct current (DC) voltage Vdd of 3.3 V, which is transformed into a high-frequency alternating current (AC) voltage of 3.2 GHz by a ring oscillator circuit that is applied to the cantilever [31]. To compensate for production variations, a control voltage Vctrl is added that enables chip-specific adjustments of switching time tsw. Variations of environmental conditions are compensated by using two MEMS cantilevers: one for the measurement and a second one with a rigid, unbendable beam acting as reference.
In an extensive study of the sensor chips, X-formed cantilevers were investigated, but also Y- and I-shaped geometries can be prepared [33]. In this work, I-shaped cantilevers with a width of 6 µm and a length of 150 µm were used (Figure 6C). Switching time tsw and glucose concentration cglc are related by the characteristic function
tsw = k1 exp(−cglc/k2) + k3
with coefficients ki to be determined for each sensor probe in a calibration step prior to usage. The viscosity strongly depends on temperature T of the measured liquid. Therefore, a band gap circuit for T measurement was integrated into the sensor chip [48] in order to allow for a T-dependent estimation of ki.
Measurements were made by placing the sensor probe in a 1.5 mL tube containing the sample. The tube was placed inside a thermostat (Labnet Accublock Mini Digital Dry Bath, Edison, NJ, USA) to keep the temperature at 37 °C. The flex cable was connected to the system board that houses a microcontroller (Texas Instruments MSP430, Dallas, TX, USA) and a radio module. The system board received its power from an Agilent E360A DC power supply, Santa Clara, CA, USA.
Conductivity measurements were made at 37 °C with a Mettler-Toledo FiveEasy FE30, Gießen, Germany. Measurements of pH were performed at room temperature with a Schott pH meter CG824, Mitterteich, Germany. Standard electrolyte (SEL) solution, the affinity assay, and mineral salt medium [37] were prepared according to Table 1. Deionized water (ρ ≥ 18 MΩ·cm) was used throughout all experiments.
The cultivation for a monitoring experiment was performed in a 250 mL magnetically stirred double wall glass bioreactor (Ochs Laborfachhandel, Bovenden, Germany) with a BL 21 E. coli strain. A starting OD600 of 0.1 and an initial glucose concentration of 1 g·L−1 in mineral salt medium (in accordance with [37]) were used. Medium temperature was adjusted to 30 °C. Samples were taken hourly. OD600 was measured photometrically with an Ultraspec 3300 Pro from Amersham Biosciences, Freiburg, Germany. After centrifugation of the samples at 105 g for 5 min, the supernatant was filtered and conductivity, pH, and cglc were measured. The latter was determined with the MEMS sensor and referenced to the Glucose Hexokinase FS kit (DiaSys Diagnostic Systems, Holzheim, Germany).

4. Conclusions

In this study, a sensor intended for continuous glucose monitoring and developed for medical applications was tested in a bioprocess environment. Its functionality relies on a MEMS chip and a biochemical affinity assay containing ConA and dextran. A mineral salt culture medium for E. coli bioprocesses was chosen as a representative model system. The study combines an investigation focused on the interactions of the affinity assay with the components of the medium and an off-line monitoring experiment of an E. coli cultivation. Whereas an unprecedented stability of sensor operation could be observed in standard electrolyte, the measurement signal tsw experienced a significant drift after bringing it in contact with bioprocess medium in the absence of glucose. Interestingly, the addition of glucose resulted in an eight times lower response. The effect may be caused by the aggregation of ConA, the cross-linking of dextran molecules, the change of binding constants due to exchange of metal ions in ConA, or the electro-viscous effect, which would modulate the viscosity independently of analyte concentration.
While the sensor may be directly applied off-line with diluted samples at higher glucose concentrations, the application of the sensor as an on-line tool for glucose monitoring in bioprocess media appears challenging at the current state of development. While the pH dependency of the signal is important to consider, it is not a problem, as most biotechnological processes are performed at constant pH, although the shift of tsw in a phosphate buffered medium has to be considered in the calculation of glucose concentrations. Also, changes in divalent ions over the run of the cultivation may eventually influence the signal. Thus, further developments may have to consider a second reference chamber with a membrane which cannot be penetrated by glucose.
The presented measurements of the new sensor MEMS have also demonstrated its ability to determine viscosity within very small sample volumes in the pL range. This should be of interest for, e.g., the investigation of aggregation in the early stages of protein crystallization and other applications in bio-rheology or biotechnology in general.

Acknowledgments

We thank the companies Acuros GmbH, Berlin, BioSensor Technology GmbH, Berlin, and Sitec Sensortechnik, Brandenburg, for the integration of sensor probes, the provision of semipermeable membranes, and transponder boards, respectively. We also thank Thomas Voss for chip-flex-bonding and the colleagues at Innovations for High Performance Microelectronics, who took part in the preparation of sensor chips. The investigations were performed in the frame of the Joint Lab Bioelectronics between Technical University Berlin and IHP.

Author Contributions

All authors conceived and designed the experiments; Lorenz Theuer and Micha Lehmann performed the experiments; all authors analysed the data; Lorenz Theuer contributed dedicated analysis tools; Mario Birkholz, Stefan Junne, and Peter Neubauer supervised the experiments and wrote the paper.

Conflicts of Interest

Mario Birkholz is co-inventor of patents related to the affinity-viscometric sensor.

Abbreviations

ConAConcanavalin A
MEMSMicroelectromechanical system
OD600Optical density at 600 nm

References

  1. Cruz Bournazou, M.N.; Barz, T.; Nickel, D.; Glauche, F.; Knepper, A.; Neubauer, P.; Lopez Cárdenas, D.C. Online optimal experimental re-design in robotic parallel fed-batch cultivation facilities for validation of macro-kinetic growth models at the example of E. coli. Biotechnol. Bioeng. 2016, 114, 610–619. [Google Scholar] [CrossRef] [PubMed]
  2. Cleland, N.; Enfors, S.-O. Control of glucose-fed batch cultivations of E. Coli by means of an oxygen stabilized enzyme electrode. Eur. J. Appl. Microbiol. Biotechnol. 1983, 18, 141–147. [Google Scholar] [CrossRef]
  3. Brooks, S.L.; Ashby, R.E.; Turner, A.P.F.; Calder, M.R.; Clarke, D.J. Development of an on-line glucose sensor for fermentation monitoring. Biosensors 1987, 3, 45–56. [Google Scholar] [CrossRef]
  4. Bradley, J.; Schmid, R.D. Optimisation of the biosensor for in situ fermentation monitoring of glucose concentration. Biosens. Bioelectron. 1991, 6, 669–674. [Google Scholar] [CrossRef]
  5. Lüdi, H.; Garn, M.B.; Haemmerli, S.D.; Manz, A.; Widmer, H.M. High-tech analyses in modern biotechnology flow injection analysis and in-line biosensors for bioprocess control: A comparison. J. Biotechnol. 1992, 25, 75–80. [Google Scholar] [CrossRef]
  6. Schügerl, K. Progress in monitoring, modeling and control of bioprocesses during the last 20 years. J. Biotechnol. 2001, 85, 149–173. [Google Scholar] [CrossRef]
  7. Wang, J. Electrochemical glucose biosensors. Chem. Rev. 2008, 108, 814–825. [Google Scholar] [CrossRef] [PubMed]
  8. Biechele, P.; Busse, C.; Solle, D.; Scheper, T.; Reardon, K. Sensor systems for bioprocess monitoring. Eng. Life Sci. 2015, 15, 469–488. [Google Scholar] [CrossRef]
  9. Kalb, A.J.; Levitzki, A. Metal-binding sites of concanavalin a and their role in the binding of α-methyl d-glucopyranoside. Biochem. J. 1968, 109, 669–672. [Google Scholar] [CrossRef] [PubMed]
  10. Weinzierl, J.; Kalb, A.J. The transition metal-binding site of concanavalin A at 2.8 Å resolution. FEBS Lett. 1971, 18, 268–270. [Google Scholar] [CrossRef]
  11. Becker, J.W.; Reeke, G.N.; Wang, J.L.; Cunningham, B.A.; Edelman, G.M. The covalent and three-dimensional structure of concanavalin A. III. Structure of the monomer and its interactions with metals and saccharides. J. Biol. Chem. 1975, 250, 1513–1524. [Google Scholar] [PubMed]
  12. Mansouri, S.; Schultz, J.S. A miniature optical glucose sensor based on affinity binding. Nat. Biotechnol. 1984, 2, 885–890. [Google Scholar] [CrossRef]
  13. Ballerstädt, R.; Ehwald, R. Suitability of aqueous dispersions of dextran and concanavalin A for glucose sensing in different variants of the affinity sensor. Biosens. Bioelectron. 1994, 9, 557–567. [Google Scholar] [CrossRef]
  14. Schultz, J.S.; Sims, G. Affinity sensors for individual metabolites. Biotechnol. Bioeng. Symp. 1979, 9, 65–71. [Google Scholar]
  15. Schultz, J.S.; Mansouri, S.; Goldstein, I.J. Affinity sensor: A new technique for developing implantable sensors for glucose and other metabolites. Diabetes Care 1982, 5, 245–253. [Google Scholar] [CrossRef] [PubMed]
  16. Meadows, D.L.; Schultz, J.S. Design, manufacture and characterization of an optical fiber glucose affinity sensor based on an homogeneous fluorescence energy transfer assay system. Anal. Chim. Acta 1993, 280, 21–30. [Google Scholar] [CrossRef]
  17. Rolinski, O.J.; Birch, D.J.S.; McCartney, L.; Pickup, J.C. Molecular distribution sensing in a fluorescence resonance energy transfer based affinity assay for glucose. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 2001, 57, 2245–2254. [Google Scholar] [CrossRef]
  18. Ballerstadt, R.; Polak, A.; Beuhler, A.; Frye, J. In vitro long-term performance study of a near-infrared fluorescence affinity sensor for glucose monitoring. Biosens. Bioelectron. 2004, 19, 905–914. [Google Scholar] [CrossRef] [PubMed]
  19. Müller, A.J.; Knuth, M.; Nikolaus, K.S.; Krivánek, R.; Küster, F.; Hasslacher, C. First clinical evaluation of a new percutaneous optical fiber glucose sensor for continuous glucose monitoring in diabetes. J. Diabetes Sci. Technol. 2013, 7, 13–23. [Google Scholar] [CrossRef] [PubMed]
  20. Ballerstadt, R.; Kholodnykh, A.; Evans, C.; Boretsky, A.; Motamedi, M.; Gowda, A.; McNichols, R. Affinity-based turbidity sensor for glucose monitoring by optical coherence tomography: Toward the development of an implantable sensor. Anal. Chem. 2007, 79, 6965–6974. [Google Scholar] [CrossRef] [PubMed]
  21. Nielsen, J.K.; Christiansen, J.S.; Kristensen, J.S.; Toft, H.O.; Hansen, L.L.; Aasmul, S.; Gregorius, K. Clinical evaluation of a transcutaneous interrogated fluorescence lifetime-based microsensor for continuous glucose reading. J. Diabetes Sci. Technol. 2009, 3, 98–109. [Google Scholar] [CrossRef] [PubMed]
  22. Boss, C.; Meurville, E.; Sallese, J.-M.; Ryser, P. A viscosity-dependent affinity sensor for continuous monitoring of glucose in biological fluids. Biosens. Bioelectron. 2011, 30, 223–228. [Google Scholar] [CrossRef] [PubMed]
  23. Huang, X.; Li, S.; Schultz, J.S.; Wang, Q.; Lin, Q. A dielectric affinity microbiosensor. Appl. Phys. Lett. 2010, 96, 033701. [Google Scholar] [CrossRef]
  24. Beyer, U.; Schäfer, D.; Thomas, A.; Aulich, H.; Haueter, U.; Reihl, B.; Ehwald, R. Recording of subcutaneous glucose dynamics by a viscosimetric affinity sensor. Diabetologica 2001, 44, 416–423. [Google Scholar] [CrossRef] [PubMed]
  25. Zhao, Y.; Li, S.; Davidson, A.; Yang, B.; Wang, Q.; Lin, Q. A mems viscosimetric sensor for continuous glucose monitoring. J. Micromech. Microeng. 2007, 17, 2528–2537. [Google Scholar] [CrossRef]
  26. Ehwald, R.; Ballerstadt, R.; Dautzenberg, H. Viscosimetric affinity assay. Anal. Biochem. 1996, 234, 1–8. [Google Scholar] [CrossRef] [PubMed]
  27. Krushinitskaya, O.; Tønnessen, T.I.; Jakobsen, H.; Johannessen, E.A. Characterization of nanoporous membranes for implementation in an osmotic glucose sensor based on the concanavalin a–dextran affinity assay. J. Membr. Sci. 2011, 376, 153–161. [Google Scholar] [CrossRef]
  28. Boss, C.; Meurville, E.; Sallese, J.-M.; Ryser, P. Size-selective diffusion in nanoporous alumina membranes for a glucose affinity sensor. J. Membr. Sci. 2012, 401–402, 217–221. [Google Scholar] [CrossRef]
  29. Birkholz, M.; Ehwald, K.-E.; Fröhlich, M.; Kulse, P.; Basmer, T.; Ehwald, R.; Guschauski, T.; Stoll, U.; Siegel, H.; Schmaderer, S.; et al. Minimal-invasiver Blutzuckersensor. In Proceedings of the GMA/ITG-Fachtagung Sensoren und Messsysteme 2012, Nürnberg, Germany, 22–23 May 2012; pp. 177–187. [Google Scholar]
  30. Birkholz, M.; Ehwald, K.-E.; Ehwald, R.; Kaynak, M.; Borngräber, J.; Drews, J.; Haak, U.; Klatt, J.; Matthus, E.; Schoof, G.; et al. Mikroviskosimeter zur kontinuierlichen Glucosemessung bei Diabetes mellitus. In Proceedings of the Mikrosystemtechnik Kongress 2009, Berlin, Germany, 12–14 October 2009; pp. 124–125. [Google Scholar]
  31. Birkholz, M.; Ehwald, K.-E.; Basmer, T.; Reich, C.; Kulse, P.; Drews, J.; Genschow, D.; Haak, U.; Marschmeyer, S.; Matthus, E.; et al. Sensing glucose concentrations at GHz frequencies with a fully embedded BioMEMS. J. Appl. Phys. 2013, 113, 244904. [Google Scholar] [CrossRef] [PubMed]
  32. Birkholz, M.; Glogener, P.; Basmer, T.; Glös, F.; Genschow, D.; Welsch, C.; Ruff, R.; Hoffmann, K.P. System integration of a silicone-encapsulated glucose monitor implant. Biomed. Eng. Biomed. Tech. 2014, 59, S1089–S1092. [Google Scholar]
  33. Birkholz, M.; Ehwald, K.-E.; Kulse, P.; Drews, J.; Fröhlich, M.; Haak, U.; Kaynak, M.; Matthus, E.; Schulz, K.; Wolansky, D. Ultrathin TiN membranes as a technology platform for CMOS-integrated MEMS and BioMEMS devices. Adv. Funct. Mater. 2011, 21, 1652–1656. [Google Scholar] [CrossRef]
  34. Ballerstadt, R.; Evans, C.; Gowda, A.; McNichols, R. In vivo performance evaluation of a transdermal near-infrared fluorescence energy transfer affinity sensor for continuous glucose monitoring. Diabetes Technol. Ther. 2006, 8, 296–311. [Google Scholar] [CrossRef] [PubMed]
  35. Diem, P.; Kalt, L.; Haueter, U.; Krinelke, L.; Fajfr, R.; Reihl, B.; Beyer, U. Clinical performance of a continuous viscosimetric affinity sensor for glucose. Diabetes Technol. Ther. 2004, 6, 790–799. [Google Scholar] [CrossRef] [PubMed]
  36. Ballerstadt, R.; Evans, C.; McNichols, R.; Gowda, A. Concanavalin A for in vivo glucose sensing: A biotoxicity review. Biosens. Bioelectron. 2006, 22, 275–284. [Google Scholar] [CrossRef] [PubMed]
  37. Noack, S.; Klöden, W.; Bley, T. Modeling synchronous growth of bacterial populations in phased cultivation. Bioprocess Biosyst. Eng. 2008, 31, 435–443. [Google Scholar] [CrossRef] [PubMed]
  38. Sacco, D.; Klett-Zygmunt, D.; Dellacherie, E. A re-investigation of the phosphorylation of dextran with polyphosphoric acid: Evidence for the formation of different types of phosphate moieties. Carbohydr. Res. 1988, 184, 193–202. [Google Scholar] [CrossRef]
  39. Suflet, D.M.; Chitanu, G.C.; Desbrières, J. Phosphorylated polysaccharides. 2. Synthesis and properties of phosphorylated dextran. Carbohydr. Polym. 2010, 82, 1271–1277. [Google Scholar] [CrossRef]
  40. Drenth, J. Principles of Protein X-ray Crystallography, 2nd ed.; Springer: Berlin, Germany, 1999; p. 341. [Google Scholar]
  41. Agena, S.M.; Pusey, M.L.; Bogle, I.D.L. Protein solubility modeling. Biotechnol. Bioeng. 1999, 64, 144–150. [Google Scholar] [CrossRef]
  42. Kim, Y.-C.; Myerson, A. Diffusivity, viscosity, and cluster formation in protein solutions. Biotechnol. Bioprocess Eng. 1997, 2, 64–67. [Google Scholar] [CrossRef]
  43. Birkholz, M.; Mai, A.; Wenger, C.; Meliani, C.; Scholz, R. Technology modules from micro- and nanoelectronics for the life sciences. WIREs Nanomed. Nanobiotechnol. 2016, 8, 355–377. [Google Scholar] [CrossRef] [PubMed]
  44. Birkholz, M.; Ehwald, K.-E.; Wolansky, D.; Costina, I.; Baristiran-Kaynak, C.; Fröhlich, M.; Beyer, H.; Kapp, A.; Lisdat, F. Corrosion-resistant metal layers from a CMOS process for bioelectronic applications. Surf. Coat. Technol. 2010, 204, 2055–2059. [Google Scholar] [CrossRef]
  45. Kulse, P.; Birkholz, M.; Ehwald, K.-E.; Bauer, J.; Drews, J.; Haak, U.; Höppner, W.; Katzer, J.; Schulz, K.; Wolansky, D. Fabrication of MEMS actuators from the BEoL of a 0.25 µm BiCMOS technology platform. Microelectron. Eng. 2012, 97, 276–279. [Google Scholar] [CrossRef]
  46. Birkholz, M.; Ehwald, K.-E.; Kaynak, M.; Semperowitsch, T.; Holz, B.; Nordhoff, S. Separation of extremely miniaturized medical sensors by IR laser dicing. J. Optoelectron. Adv. Mater. 2010, 12, 479–483. [Google Scholar]
  47. Ehwald, R. Affinitätsassy in der mikrodialysekammer—Ein generelles Prinzip für kontinuierlich messende Biosensoren. Tech. Mess. 2004, 71, 24–28. [Google Scholar] [CrossRef]
  48. Brokaw, A.P. A simple three-terminal IC bandgap reference. IEEE J. Solid State Circuits 1974, 9, 388–393. [Google Scholar] [CrossRef]
Figure 1. (A) Switching time tsw of the sensor probe in standard electrolyte (SEL) over 16 days. (B) Measurement in Noack medium exhibiting with drifting tsw values (C). Subsequent measurement in SEL brought tsw values almost back to its starting point, the red line marks the mean value of the 16-day measurement shown in (A).
Figure 1. (A) Switching time tsw of the sensor probe in standard electrolyte (SEL) over 16 days. (B) Measurement in Noack medium exhibiting with drifting tsw values (C). Subsequent measurement in SEL brought tsw values almost back to its starting point, the red line marks the mean value of the 16-day measurement shown in (A).
Ijms 18 01235 g001
Figure 2. Difference in switching time ∆tsw(t) during the observation period measured in SEL including the individual components of mineral salt (MS) medium: (A) MgCl2; (B) NH4Cl; (C) FeCl3; (D) phosphate buffer components.
Figure 2. Difference in switching time ∆tsw(t) during the observation period measured in SEL including the individual components of mineral salt (MS) medium: (A) MgCl2; (B) NH4Cl; (C) FeCl3; (D) phosphate buffer components.
Ijms 18 01235 g002
Figure 3. Comparison of ∆tsw transients after transferring the sensor probe from SEL to MS medium (top) and the sum of individual components of medium as derived from their fitting functions (bottom).
Figure 3. Comparison of ∆tsw transients after transferring the sensor probe from SEL to MS medium (top) and the sum of individual components of medium as derived from their fitting functions (bottom).
Ijms 18 01235 g003
Figure 4. Calibration curve measured before the cultivation monitoring experiment in Noack medium with defined glucose concentrations between 1 and 50 mM during 6.5 h.
Figure 4. Calibration curve measured before the cultivation monitoring experiment in Noack medium with defined glucose concentrations between 1 and 50 mM during 6.5 h.
Ijms 18 01235 g004
Figure 5. Growth curve of a BL 21 Escherichia coli cultivation in Noack medium. The cell growth was monitored by optical density at 600 nm (OD600). For glucose measurements, samples were removed from the cultivation and analysed with the glucose sensor and a commercial enzymatic glucose hexokinase assay.
Figure 5. Growth curve of a BL 21 Escherichia coli cultivation in Noack medium. The cell growth was monitored by optical density at 600 nm (OD600). For glucose measurements, samples were removed from the cultivation and analysed with the glucose sensor and a commercial enzymatic glucose hexokinase assay.
Ijms 18 01235 g005
Figure 6. (A) Sensor probe (white circle); (B) Magnification of the sensor probe showing the cooling body, cover plate and semipermeable membrane; (C) Optical microscope picture of the micro-electromechanical system (MEMS) part of the sensor chip prior to integrating the semipermeable membrane and filling with the affinity assay. The measurement MEMS is given on the left and the reference MEMS is provided on the right-hand side of the figure. Golden-coloured cantilevers can be seen to stretch over the ground plates (orange-brown) in both cases. Due to the architecture of the semiconductor layer stack, the cantilevers have a height of 2.5 µm above the ground plate, such that the configuration of both forms a kind of parallel plate capacitor.
Figure 6. (A) Sensor probe (white circle); (B) Magnification of the sensor probe showing the cooling body, cover plate and semipermeable membrane; (C) Optical microscope picture of the micro-electromechanical system (MEMS) part of the sensor chip prior to integrating the semipermeable membrane and filling with the affinity assay. The measurement MEMS is given on the left and the reference MEMS is provided on the right-hand side of the figure. Golden-coloured cantilevers can be seen to stretch over the ground plates (orange-brown) in both cases. Due to the architecture of the semiconductor layer stack, the cantilevers have a height of 2.5 µm above the ground plate, such that the configuration of both forms a kind of parallel plate capacitor.
Ijms 18 01235 g006
Table 1. Composition of media used. SEL, standard electrolyte.
Table 1. Composition of media used. SEL, standard electrolyte.
Standard Electrolyte SELMineral Salt Medium [37]
Componentg·L−1mMComponentg·L−1mM
Tris/HCl1.217.7KH2PO42.7220
NaCl7.76133Na2HPO4 × 2H2O3.5620
MnSO4 × H2O0.1691.00NaCl5.0085.6
CaCl20.1471.32Na2SO41.107.74
NaN30.57.7NH4Cl0.509.35
MgCl2 × 6H2O0.040.197
Affinity assay = SEL + ConA + dextranFeCl3 × 6H2O0.0080.030
Dextran T2000700.035MnSO4 × H2O0.0040.024
ConA70.264Thiamine0.050.148

Share and Cite

MDPI and ACS Style

Theuer, L.; Lehmann, M.; Junne, S.; Neubauer, P.; Birkholz, M. Micro-Electromechanical Affinity Sensor for the Monitoring of Glucose in Bioprocess Media. Int. J. Mol. Sci. 2017, 18, 1235. https://doi.org/10.3390/ijms18061235

AMA Style

Theuer L, Lehmann M, Junne S, Neubauer P, Birkholz M. Micro-Electromechanical Affinity Sensor for the Monitoring of Glucose in Bioprocess Media. International Journal of Molecular Sciences. 2017; 18(6):1235. https://doi.org/10.3390/ijms18061235

Chicago/Turabian Style

Theuer, Lorenz, Micha Lehmann, Stefan Junne, Peter Neubauer, and Mario Birkholz. 2017. "Micro-Electromechanical Affinity Sensor for the Monitoring of Glucose in Bioprocess Media" International Journal of Molecular Sciences 18, no. 6: 1235. https://doi.org/10.3390/ijms18061235

APA Style

Theuer, L., Lehmann, M., Junne, S., Neubauer, P., & Birkholz, M. (2017). Micro-Electromechanical Affinity Sensor for the Monitoring of Glucose in Bioprocess Media. International Journal of Molecular Sciences, 18(6), 1235. https://doi.org/10.3390/ijms18061235

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop