**9. Biosensors using dehydrogenase enzymes for continuous monitoring**

## **9.1. FAD-GDH biosensor for continuous glucose monitoring**

It has been well established that many of the diabetic complications leading to both chronic and acute health problems, such as adult blindness, end-stage renal disease, lower-limb amputations, and heart disease or stroke, can be reduced or even prevented through intensive blood glucose control. [27-29] Effective glycaemic control requires frequent measurements of blood glucose in order to take necessary therapeutic interventions. Such an approach is exemplified by the use of so-called 'continuous glucose monitoring' (CGM) apparatus and methodologies that are used by a growing number of patients to monitor their diabetes condition. Such systems are composed of a probe that is inserted into the body such that it contacts glucose containing liquids in the body, such as interstitial fluid. These CGM systems are designed to operate over extended periods of time, typically over a number of days or longer. In reality CGM is a misnomer, inasmuch as the device samples in an episodic manner, but on a sufficiently high frequency to distinguish such devices from single measurement episodic systems. Nevertheless, compared with episodic selfmonitoring blood glucose (SMBG), CGM follows blood glucose dynamics and hence, provides patients and healthcare professionals with not only current blood glucose levels, but also real-time rate and direction of changes. Blood glucose thresholds can be set to alert for possible dangerous trends, for instance rapid blood glucose descents that may lead to hypoglycemia. Advances in CGM research and development are also critical to realize "artificial pancreas" of a closed-loop system in conjunction with an insulin pump.

336 Dehydrogenases

NAD--hydroxybutyrate

incorporate dehydrogenase based enzymes.

**Enzyme Reacts with O2? Reacts with maltose?**  PQQ-Glucose Dehydrogenase No Yes (Not Genetically Engineered

NAD-Glucose Dehydrogenase No No

FAD-Glucose Dehydrogenase No No

**Table 1.** Commonly used dehydrogenase enzymes for Commercial Self testing Biosensors

dehydrogenase No No

Shown below are examples of commercially available test strips for home use which

a. The Roche Accu-Check Aviva test strip. This test strip uses PQQ-GDH as the enzyme of choice.

b. The Abbott Optium Xceed glucose test strip. This strip uses NAD-Glucose dehydrogenase in the sensing

c. The Abbott Optium Xceed ketone test strip. This strip uses NAD+ linked β-hydroxybutyrate dehydrogenase. Because it uses the same mediator chemistry as B, and hence operates at the same potential, the two strips can be

d. The LifeScan OneTouch Verio test strip. This strip incorporates FAD-GDH in the strip chemistry to measure

used in the same meter. http://www.abbott-diabetescare.com/AU/ProductDetail.aspx?product=57

**Figure 8.** Examples of commercially available test strips for self-testing utilising dehydrogenases.

**9. Biosensors using dehydrogenase enzymes for continuous monitoring** 

It has been well established that many of the diabetic complications leading to both chronic and acute health problems, such as adult blindness, end-stage renal disease, lower-limb amputations, and heart disease or stroke, can be reduced or even prevented through intensive

glucose in whole blood. http://www.lifescan.co.uk/ourproducts/teststrips/onetouchveriopro

http://www.roche.com/products/product-details.htm?type=product&id=2

chemistry. http://www.abbott-diabetescare.com/AU/ProductDetail.aspx?product=57

(Photograph courtesy of Mr Christopher Leach, LifeScan Scotland Ltd., Inverness, UK).

**9.1. FAD-GDH biosensor for continuous glucose monitoring** 

variant [23] )

Enzyme catalysed electrochemical biosensors have been the most successful technology for the commercialized SMBG products (as discussed in the preceding part of this chapter). For CGM, enzymes are also employed as the means of target analyte recognition, coupled to electrochemical transduction methods for determination of the analyte of interest. Correspondingly, such systems have been so far limited to the use of a redox enzyme, such as glucose oxidase (GOx) in which the prosthetic group is intimately associated with the enzyme, such that it cannot diffuse or leach away over the duration of sensor operation.[30] Recently commercialized flavin adenine dinucleotide-dependent glucose dehydrogenase (FAD-GDH, EC 1.1.99.10), where the FAD moiety forms an integral part of the enzyme, has attracted great attention for blood glucose monitoring because of its advantages over GOx in terms of insensitivity to molecular oxygen. However, like many redox enzymes, direct electron transfer between FAD-GDH and electrode surface cannot occur because the active centre FAD is insulated by the large proteins. Therefore, mediators are usually employed to shuttle electron between the FAD moiety and electrode surface. An example of such mediators is potassium ferricyanide used for LifeScan OneTouch Verio family of SMBG products. However, use of small molecular mediators in biosensors faces tremendous technical challenges for CGM which requires essentially all the reactive reagents immobilized without leaching out from the electrodes to achieve long-term measurement stability and to meet biocompatibility requirements, in particular for in-vivo applications. An attractive approach to tackle this challenge is to use a polymeric mediator which has mediator moieties chemically attached to polymer chains. Because of its large molecular size, the polymeric mediator can be co-immobilized with enzyme at electrode by various means, including surface grafting, [31-34] layer-by-layer surface adsorption, [35-37] retention behind semi-permeable dialysis membranes,[38-40] physical entrapment [41-43] or cross-link in hydrogels, [44-47] entrapment in electropolymerized [48,49] or chemically formed layers [50] or in inorganic layers,[51,52] and blend in carbon pastes. [53]

At LifeScan Scotland Limited, we have synthesized a ferrocene polymeric mediator which is a copolymer of vinylferrocene, acrylamide and 2-(diethylamino)ethyl methacrylate.[54] Like small molecular mediators, an effective polymeric mediator for biosensors should be able to exchange electrons with enzyme prosthetic group and then be re-oxidized/re-reduced at electrode in a reversible manor at a sufficient low potential to avoid or minimize interferences resulting from oxidation/reduction of other components in the sample fluid, such as uric acid, ascorbic acid etc. in a bodily fluid. Figure 9 shows cyclic voltammograms of the ferrocene polymeric mediator and FAD-GDH (from Toyobo Co. Ltd., Japan) in phosphate buffered saline (PBS). In the absence of glucose, cyclic voltammetry shows almost symmetric anodic and cathodic peaks which are attributed to oxidation and reduction of the ferrocene moieties of the polymeric mediator. After adding 2.5 mM glucose, the cyclic voltammogram changed dramatically, with a large increase in the anodic peak and a significant decrease in the cathodic peak. This was a typical phenomenon of enzymedependent catalytic reduction of ferrocenium generated during the oxidation half-cycle in the presence of glucose. The cathodic peak increased further as glucose concentration increased to 5.0 mM. These results clearly indicated that the ferrocene polymeric mediator exhibited preferential redox properties with respect to transfer of electrons from the reduced enzyme prosthetic group FADH2 to the glassy carbon electrode surface. The ferrocene moieties are reduced to ferrocenium moieties upon accepting the electrons from FADH2 and then are re-oxidized on transfer of electrons to the electrode.

Amperometric Glucose Sensors for Whole Blood Measurement Based on Dehydrogenase Enzymes 339

binder copolymer became water-insoluble and the ink layer evolved to form a 3-dimensional nano-porous structure which effectively entraps the large molecules of the enzyme and the ferrocene polymeric mediator whilst allows free diffusion of glucose and water molecules. [54]

**Figure 10.** Scanning electrode microscopy image of an electrode with immobilized FAD-GDH and the

The enzyme and ferrocene polymeric mediator are co-immobilized in vicinity of the

 The 3-dimensional nano-porous structure enables higher loading of the redox species and provides larger electrode surface area than a planar electrode with an enzyme layer. No chemical reactions are involved throughout the immobilization and hence, potential

The screen-printing technique suits mass production and has low manufacturing cost.

Because of the presence of ammonium hydroxide, the water-based carbon ink was basic in nature. FAD-GDH stability in the wet ink was studied by comparing glucose sensitivity of the modified electrodes fabricated from the same batch of formulated ink with varied standout time. The glucose sensitivity of the electrodes was assessed by testing current responses of a 3-electrode setup to glucose in a concentration range of 0 to 20 mM. The test conditions and results are presented in Figure 11. It is clearly seen that 2-day stand-out of the wet ink significantly reduced glucose calibration slope of the test. During the stand-out period, the wet ink containing FAD-GDH and the polymeric mediator was kept in a fridge at 4 C and negligible changes in ink rheology and solid content were detected. For comparison, the stand-out was also investigated for an ink using GOx as an enzyme and insignificant effect on glucose calibration slope was found. Therefore, the reduction in glucose sensitivity of the FAD-GDH working electrodes is attributed to decrease in the enzyme activity probably

Since Clark and Lyons reported the first enzymatic electrode for glucose measurement in 1962, [56] molecular oxygen has been involved in the enzymatic redox reactions as an electron acceptor for the first generation of biosensors. However, this type of biosensors is based on measuring generation of hydrogen peroxide or depletion of oxygen and hence, exhibits

electrode graphite particles, which is beneficial for fast electron transfer.

Figure 10 shows scanning electron microscopy image of the electrode surface.

ferrocene polymeric mediator

The main advantages of these electrodes are

enzyme denaturing is avoided.

resulting from the enzyme denaturing in the basic wet ink.

**Figure 9.** Cyclic voltammograms: 1.5mg/mL FAD-GDH, 9 mg/mL ferrocene polymeric mediator in 0.01M pH7.4 PBS, scan rate 5 mV/s, without (solid line), with 2.5 mM glucose (broken line) & 5.0 mM glucose (dotted line)

We have also developed a technique for co-immobilization of FAD-GDH and the ferrocene polymeric mediator in modified electrodes which were fabricated by screen-printing a waterbased carbon ink containing both the enzyme and the ferrocene polymeric mediator. The ink contained graphite particles as conductive pigments and a pH sensitive copolymer as a binder. [55] The copolymer binder was water-soluble in the presence of ammonium hydroxide and hence, the ink was miscible with dissolved FAD-GDH and the ferrocene polymeric mediator during the ink formulation. As ammonia evaporated upon drying a screen-printed ink pad, the binder copolymer became water-insoluble and the ink layer evolved to form a 3-dimensional nano-porous structure which effectively entraps the large molecules of the enzyme and the ferrocene polymeric mediator whilst allows free diffusion of glucose and water molecules. [54] Figure 10 shows scanning electron microscopy image of the electrode surface.

**Figure 10.** Scanning electrode microscopy image of an electrode with immobilized FAD-GDH and the ferrocene polymeric mediator

The main advantages of these electrodes are

338 Dehydrogenases

glucose (dotted line)

At LifeScan Scotland Limited, we have synthesized a ferrocene polymeric mediator which is a copolymer of vinylferrocene, acrylamide and 2-(diethylamino)ethyl methacrylate.[54] Like small molecular mediators, an effective polymeric mediator for biosensors should be able to exchange electrons with enzyme prosthetic group and then be re-oxidized/re-reduced at electrode in a reversible manor at a sufficient low potential to avoid or minimize interferences resulting from oxidation/reduction of other components in the sample fluid, such as uric acid, ascorbic acid etc. in a bodily fluid. Figure 9 shows cyclic voltammograms of the ferrocene polymeric mediator and FAD-GDH (from Toyobo Co. Ltd., Japan) in phosphate buffered saline (PBS). In the absence of glucose, cyclic voltammetry shows almost symmetric anodic and cathodic peaks which are attributed to oxidation and reduction of the ferrocene moieties of the polymeric mediator. After adding 2.5 mM glucose, the cyclic voltammogram changed dramatically, with a large increase in the anodic peak and a significant decrease in the cathodic peak. This was a typical phenomenon of enzymedependent catalytic reduction of ferrocenium generated during the oxidation half-cycle in the presence of glucose. The cathodic peak increased further as glucose concentration increased to 5.0 mM. These results clearly indicated that the ferrocene polymeric mediator exhibited preferential redox properties with respect to transfer of electrons from the reduced enzyme prosthetic group FADH2 to the glassy carbon electrode surface. The ferrocene moieties are reduced to ferrocenium moieties upon accepting the electrons from FADH2 and

then are re-oxidized on transfer of electrons to the electrode.

**Figure 9.** Cyclic voltammograms: 1.5mg/mL FAD-GDH, 9 mg/mL ferrocene polymeric mediator in 0.01M pH7.4 PBS, scan rate 5 mV/s, without (solid line), with 2.5 mM glucose (broken line) & 5.0 mM

We have also developed a technique for co-immobilization of FAD-GDH and the ferrocene polymeric mediator in modified electrodes which were fabricated by screen-printing a waterbased carbon ink containing both the enzyme and the ferrocene polymeric mediator. The ink contained graphite particles as conductive pigments and a pH sensitive copolymer as a binder. [55] The copolymer binder was water-soluble in the presence of ammonium hydroxide and hence, the ink was miscible with dissolved FAD-GDH and the ferrocene polymeric mediator during the ink formulation. As ammonia evaporated upon drying a screen-printed ink pad, the


Because of the presence of ammonium hydroxide, the water-based carbon ink was basic in nature. FAD-GDH stability in the wet ink was studied by comparing glucose sensitivity of the modified electrodes fabricated from the same batch of formulated ink with varied standout time. The glucose sensitivity of the electrodes was assessed by testing current responses of a 3-electrode setup to glucose in a concentration range of 0 to 20 mM. The test conditions and results are presented in Figure 11. It is clearly seen that 2-day stand-out of the wet ink significantly reduced glucose calibration slope of the test. During the stand-out period, the wet ink containing FAD-GDH and the polymeric mediator was kept in a fridge at 4 C and negligible changes in ink rheology and solid content were detected. For comparison, the stand-out was also investigated for an ink using GOx as an enzyme and insignificant effect on glucose calibration slope was found. Therefore, the reduction in glucose sensitivity of the FAD-GDH working electrodes is attributed to decrease in the enzyme activity probably resulting from the enzyme denaturing in the basic wet ink.

Since Clark and Lyons reported the first enzymatic electrode for glucose measurement in 1962, [56] molecular oxygen has been involved in the enzymatic redox reactions as an electron acceptor for the first generation of biosensors. However, this type of biosensors is based on measuring generation of hydrogen peroxide or depletion of oxygen and hence, exhibits

Amperometric Glucose Sensors for Whole Blood Measurement Based on Dehydrogenase Enzymes 341

**Figure 12.** Effect of oxygen content on current response to 5 mM glucose in 0.01M pH7.4 PBS for screen-printed electrodes containing FAD-GHD (☐) and GOx (■), tested by using an Ag/AgCl reference

**Figure 13.** Calibration plots for screen-printed electrodes before (☐) and after (■) 25 KGy e-beam sterilization, tested in 0.01M pH7.4 PBS by using an Ag/AgCl reference electrode and a platinum

0 5 10 15 20

glucose concentration (mM)

counter electrode, the error bars are first standard deviation of 7 electrodes

0

3

6

current (A)

9

12

Like many medical devices, CGM biosensors normally need sterilization, in particular for *in vivo* or *ex-vivo* applications to eliminate any harmful contaminants such as fungi, bacteria, viruses, and spore forms etc. There are different established methods for sterilizing medical devices. Some of them involve the use of high energy means which can cause damages to materials in certain circumstances. It is essential that the key redox species, including FAD-GDH and the ferrocene polymeric mediator in the biosensor are not subject to any significant damages during a chosen sterilization process. Figure 13 shows comparison of glucose calibration plots for the same batch electrodes fabricated by screen-printing the water-based carbon ink containing FAD-GDH and the ferrocene polymeric mediator before and after 25 KGy e-beam sterilization. The results indicate 10-16% reduction in steady state current in the tested range of glucose concentrations. A sample of the ferrocene polymeric mediator also went through the sterilization process alongside the electrodes. 1H nuclear magnetic resonance and size exclusion chromatography indicated no change to the composition and molecular weight of the ferrocene polymeric mediator after the e-beam sterilization. The reduction in the current responses to glucose is probably attributed to e-beam induced decrease in the enzyme activity.

5 10 15 20 25

oxygen content (Kpa)

electrode and a platinum counter electrode

0.0 0.2 0.4 0.6 0.8 1.0

sensor current change

**Figure 11.** Calibration plots for screen-printed electrodes prepared by using the ink printed after ink formulation without delay (☐) and the ink printed 2 days (■) after ink formulation, tested in 0.01M pH7.4 PBS by using an Ag/AgCl reference electrode and a platinum counter electrode, 0.3 V potential, room temperature, the error bars are first standard deviation of 7 electrodes

fundamental limitations. Measurement of hydrogen peroxide requires relatively high operational potential (normally >0.45V) and suffers from significant interference resulting from oxidation of other substances in the bodily fluid. [57, 58] For oxygen measurement, the test result is sensitive to the variations in oxygen supply and test conditions. [59] The second generation biosensors use mediators to shuttle electrons between the enzyme prosthetic groups and electrodes. Typically, mediators have the attractive property of being selected for a particular desirable redox potential at which the mediators readily undergo redox reactions at the electrode whilst the redox reactions of the interferents are insignificant. However, for some enzymes, such as GOx, oxygen can compete with the mediators to accept electrons from the reduced prosthetic group FADH2 to form hydrogen peroxide which cannot be oxidized at the electrode under the applied potential for the mediator re-oxidation. As a result, the biosensor response to the analyte is dampened in the presence of oxygen. As shown in Figure 12, tested in PBS with a 3-electrode setup, the current response to 5 mM glucose for the screen-printed electrode modified with GOx and the ferrocene polymeric mediator decreased almost 60% as blood oxygen content increased from 8 Kpa to 23.14 Kpa. In contrast, the screen-printed electrode modified with FAD-GDH and the ferrocene polymeric mediator had little change in the current response in a range of the blood oxygen content from 9.8 Kpa to 20.1 Kpa.

The oxygen insensitivity of FAD-GDH makes the enzyme very attractive for the development of CGM biosensors. This is because the biosensors using an oxygen-sensitive enzyme in general has more profound response dampening effect at low glucose concentrations due to the generation of hydrogen peroxide, which can contribute to significant accuracy error for the CGM and impose critical challenges for CGM to provide patients and healthcare professionals with a true picture of ambient glycemia profile which is critical for reliable detection and/or prediction of hypoglycemia, an important step to good diabetes management.

**Figure 11.** Calibration plots for screen-printed electrodes prepared by using the ink printed after ink formulation without delay (☐) and the ink printed 2 days (■) after ink formulation, tested in 0.01M pH7.4 PBS by using an Ag/AgCl reference electrode and a platinum counter electrode, 0.3 V potential,

0 5 10 15 20

glucose concentration (mM)

fundamental limitations. Measurement of hydrogen peroxide requires relatively high operational potential (normally >0.45V) and suffers from significant interference resulting from oxidation of other substances in the bodily fluid. [57, 58] For oxygen measurement, the test result is sensitive to the variations in oxygen supply and test conditions. [59] The second generation biosensors use mediators to shuttle electrons between the enzyme prosthetic groups and electrodes. Typically, mediators have the attractive property of being selected for a particular desirable redox potential at which the mediators readily undergo redox reactions at the electrode whilst the redox reactions of the interferents are insignificant. However, for some enzymes, such as GOx, oxygen can compete with the mediators to accept electrons from the reduced prosthetic group FADH2 to form hydrogen peroxide which cannot be oxidized at the electrode under the applied potential for the mediator re-oxidation. As a result, the biosensor response to the analyte is dampened in the presence of oxygen. As shown in Figure 12, tested in PBS with a 3-electrode setup, the current response to 5 mM glucose for the screen-printed electrode modified with GOx and the ferrocene polymeric mediator decreased almost 60% as blood oxygen content increased from 8 Kpa to 23.14 Kpa. In contrast, the screen-printed electrode modified with FAD-GDH and the ferrocene polymeric mediator had little change in

the current response in a range of the blood oxygen content from 9.8 Kpa to 20.1 Kpa.

prediction of hypoglycemia, an important step to good diabetes management.

The oxygen insensitivity of FAD-GDH makes the enzyme very attractive for the development of CGM biosensors. This is because the biosensors using an oxygen-sensitive enzyme in general has more profound response dampening effect at low glucose concentrations due to the generation of hydrogen peroxide, which can contribute to significant accuracy error for the CGM and impose critical challenges for CGM to provide patients and healthcare professionals with a true picture of ambient glycemia profile which is critical for reliable detection and/or

room temperature, the error bars are first standard deviation of 7 electrodes

0

3

6

current (A)

9

12

**Figure 12.** Effect of oxygen content on current response to 5 mM glucose in 0.01M pH7.4 PBS for screen-printed electrodes containing FAD-GHD (☐) and GOx (■), tested by using an Ag/AgCl reference electrode and a platinum counter electrode

Like many medical devices, CGM biosensors normally need sterilization, in particular for *in vivo* or *ex-vivo* applications to eliminate any harmful contaminants such as fungi, bacteria, viruses, and spore forms etc. There are different established methods for sterilizing medical devices. Some of them involve the use of high energy means which can cause damages to materials in certain circumstances. It is essential that the key redox species, including FAD-GDH and the ferrocene polymeric mediator in the biosensor are not subject to any significant damages during a chosen sterilization process. Figure 13 shows comparison of glucose calibration plots for the same batch electrodes fabricated by screen-printing the water-based carbon ink containing FAD-GDH and the ferrocene polymeric mediator before and after 25 KGy e-beam sterilization. The results indicate 10-16% reduction in steady state current in the tested range of glucose concentrations. A sample of the ferrocene polymeric mediator also went through the sterilization process alongside the electrodes. 1H nuclear magnetic resonance and size exclusion chromatography indicated no change to the composition and molecular weight of the ferrocene polymeric mediator after the e-beam sterilization. The reduction in the current responses to glucose is probably attributed to e-beam induced decrease in the enzyme activity.

**Figure 13.** Calibration plots for screen-printed electrodes before (☐) and after (■) 25 KGy e-beam sterilization, tested in 0.01M pH7.4 PBS by using an Ag/AgCl reference electrode and a platinum counter electrode, the error bars are first standard deviation of 7 electrodes

CGM biosensors were fabricated by sequentially screen-printing and drying a carbon ink, an Ag/AgCl ink and the water-based carbon ink containing FAD-GDH and the ferrocene polymeric mediator on the same plane of a plastic foil to form two carbon tracks, a pseudoreference electrode and a working electrode, respectively. Then the plastic foil was heatlaminated onto a plastic disc which had pre-formed micro-channels (typical channel dimension: width x depth = 0.3 x 0.3 mm) in a way that the micro-channel runs over the two electrodes with the working electrode upstream of the pseudo-reference electrode. Surface areas exposed to the micro-channel are 0.3 x 2.0 mm and 0.3 x 5.5 mm for the working electrode and the pseudo-reference electrode, respectively. [54] For comparison, biosensors with GOx were also fabricated in the same way except for replacing FAD-GDH with GOx in the water-based carbon ink. Loadings of the two types of enzymes in the water-based carbon inks were identical in weight.

Amperometric Glucose Sensors for Whole Blood Measurement Based on Dehydrogenase Enzymes 343

and 30 hour test time. This is probably due to the oxygen effect on the GOx biosensor as

The steady-state current responses of the two biosensors over the 54-hour test are plotted against the YSI measured blood glucose concentration in Figure 15. For both of the biosensors, the test results demonstrated good linear correlations between the current responses and blood glucose concentration over the tested glucose concentration range. However, the FAD-GDH biosensor had significantly higher current responses than the GOx biosensor, leading to higher sensitivity to glucose concentration variations and hence, potentially better measurement accuracy. This suggests that FAD-GDH has higher activity than GOx given that the enzyme loadings were the same for the two types of biosensors.

**Figure 14.** Current response variations of sensors with screen-printed working electrodes containing FAD-GDH (solid line) and GOx (dotted line) and Ag/AgCl reference electrode, with varied blood glucose concentrations (), tested with continuous flow of human blood at a flow rate of 15 μL/min

23 24 25 26 27 28 29 30

test time (hours)

y = 0.0468x + 0.0525 R² = 0.985

> y = 0.0263x - 0.0346 R² = 0.9605

0

6

12

YSI blood glucose concentration (mM)

18

24

30

**Figure 15.** Calibration plots of sensors with screen-printed working electrodes containing FAD-GDH (□) and GOx (■) respectively and Ag/AgCl reference electrode, tested with continuous flow of human

0 5 10 15 20 25 30

YSI blood glucose concentration (mM)

Because the blood cells continuously consumed glucose and result in gradual decrease in blood glucose concentration at room temperature, direct assessment on measurement

blood at a flow rate of 15 μL/min

0.0

0.0

0.3

0.6

sensor current (A)

0.9

1.2

0.3

0.5

sensor current (A)

0.8

1.0

1.3

oxygen content of the blood sample gradually increased under continuous agitation.

The two types of biosensor discs were tested side-by-side at room temperature. Human blood samples were collected on the same day for the test. Upon testing, a continuous flow of the human blood samples were pumped from blood sample reservoirs through the channels of each biosensor at 15 μL per minute by using a peristaltic pump (Ismatec). Sensor current responses to blood glucose were recorded by using a potentiostat (Uniscan Instruments) operated at 0.3 V potential. Step changes of blood glucose concentration were realized by switching the blood sample reservoirs which were open to the atmosphere and under gentle magnetic agitation. Low glucose blood samples were prepared by standing the blood sample reservoirs in a 37 C water-bath to accelerate glucose consumption by blood cells, whilst high glucose blood samples were prepared by spiking with a 20% wt/wt glucose stock (from Sigma-Aldrich). YSI measurements (YSI 2300 STAT Glucose Analyzer) of glucose concentrations were performed for the blood samples as references. In order to minimize error caused by blood mixing in the disc channel upon changing glucose concentration, the YSI measurements were performed by taking a sample from the reservoir when the previous blood sample in the biosensor micro-channel was completely depleted indicated by steady sensor current. Given the slow decrease in blood glucose concentration with time at room temperature, two YSI measurements were normally performed for each blood sample during the period it flowed through the biosensor disc channel.

In a typical test, the two types of biosensors were tested continuously over 54 hours with daily changes of blood samples from different donors. The disc channels were flushed with PBS (0.01 M, pH7.4) prior to the blood changes. Figure 14 shows typical recorded current responses of one FAD-GDH and one GOx biosensors to the same blood samples with varied glucose concentrations in a range of 2.29 to 25.64 mM between 23 and 30 hour test time. YSI measured blood glucose concentrations are also shown. For both biosensors, the currents clearly followed the step-changes in blood glucose concentration and reached "steady-state" in 3 to 5 minutes. It is expected that the real response time of the biosensors to the blood glucose concentration changes is shorter than 3 to 5 minutes because the recorded steady-state currents can only be obtained after complete depletion of the "old" blood samples from the disc channels upon the reservoir switching. By a close look, the current response of the GOx biosensor clearly drifted whilst the current response of FAD-GDH sensor did not between 28.5 and 30 hour test time. This is probably due to the oxygen effect on the GOx biosensor as oxygen content of the blood sample gradually increased under continuous agitation.

342 Dehydrogenases

carbon inks were identical in weight.

CGM biosensors were fabricated by sequentially screen-printing and drying a carbon ink, an Ag/AgCl ink and the water-based carbon ink containing FAD-GDH and the ferrocene polymeric mediator on the same plane of a plastic foil to form two carbon tracks, a pseudoreference electrode and a working electrode, respectively. Then the plastic foil was heatlaminated onto a plastic disc which had pre-formed micro-channels (typical channel dimension: width x depth = 0.3 x 0.3 mm) in a way that the micro-channel runs over the two electrodes with the working electrode upstream of the pseudo-reference electrode. Surface areas exposed to the micro-channel are 0.3 x 2.0 mm and 0.3 x 5.5 mm for the working electrode and the pseudo-reference electrode, respectively. [54] For comparison, biosensors with GOx were also fabricated in the same way except for replacing FAD-GDH with GOx in the water-based carbon ink. Loadings of the two types of enzymes in the water-based

The two types of biosensor discs were tested side-by-side at room temperature. Human blood samples were collected on the same day for the test. Upon testing, a continuous flow of the human blood samples were pumped from blood sample reservoirs through the channels of each biosensor at 15 μL per minute by using a peristaltic pump (Ismatec). Sensor current responses to blood glucose were recorded by using a potentiostat (Uniscan Instruments) operated at 0.3 V potential. Step changes of blood glucose concentration were realized by switching the blood sample reservoirs which were open to the atmosphere and under gentle magnetic agitation. Low glucose blood samples were prepared by standing the blood sample reservoirs in a 37 C water-bath to accelerate glucose consumption by blood cells, whilst high glucose blood samples were prepared by spiking with a 20% wt/wt glucose stock (from Sigma-Aldrich). YSI measurements (YSI 2300 STAT Glucose Analyzer) of glucose concentrations were performed for the blood samples as references. In order to minimize error caused by blood mixing in the disc channel upon changing glucose concentration, the YSI measurements were performed by taking a sample from the reservoir when the previous blood sample in the biosensor micro-channel was completely depleted indicated by steady sensor current. Given the slow decrease in blood glucose concentration with time at room temperature, two YSI measurements were normally performed for each

blood sample during the period it flowed through the biosensor disc channel.

In a typical test, the two types of biosensors were tested continuously over 54 hours with daily changes of blood samples from different donors. The disc channels were flushed with PBS (0.01 M, pH7.4) prior to the blood changes. Figure 14 shows typical recorded current responses of one FAD-GDH and one GOx biosensors to the same blood samples with varied glucose concentrations in a range of 2.29 to 25.64 mM between 23 and 30 hour test time. YSI measured blood glucose concentrations are also shown. For both biosensors, the currents clearly followed the step-changes in blood glucose concentration and reached "steady-state" in 3 to 5 minutes. It is expected that the real response time of the biosensors to the blood glucose concentration changes is shorter than 3 to 5 minutes because the recorded steady-state currents can only be obtained after complete depletion of the "old" blood samples from the disc channels upon the reservoir switching. By a close look, the current response of the GOx biosensor clearly drifted whilst the current response of FAD-GDH sensor did not between 28.5 The steady-state current responses of the two biosensors over the 54-hour test are plotted against the YSI measured blood glucose concentration in Figure 15. For both of the biosensors, the test results demonstrated good linear correlations between the current responses and blood glucose concentration over the tested glucose concentration range. However, the FAD-GDH biosensor had significantly higher current responses than the GOx biosensor, leading to higher sensitivity to glucose concentration variations and hence, potentially better measurement accuracy. This suggests that FAD-GDH has higher activity than GOx given that the enzyme loadings were the same for the two types of biosensors.

**Figure 14.** Current response variations of sensors with screen-printed working electrodes containing FAD-GDH (solid line) and GOx (dotted line) and Ag/AgCl reference electrode, with varied blood glucose concentrations (), tested with continuous flow of human blood at a flow rate of 15 μL/min

**Figure 15.** Calibration plots of sensors with screen-printed working electrodes containing FAD-GDH (□) and GOx (■) respectively and Ag/AgCl reference electrode, tested with continuous flow of human blood at a flow rate of 15 μL/min

Because the blood cells continuously consumed glucose and result in gradual decrease in blood glucose concentration at room temperature, direct assessment on measurement stability of the biosensors by recording sensor response change with time at a fixed glucose concentration proved to be unreliable. Nevertheless, closely aligned data points to the linear regression line in Figure 15 illustrate stable biosensor response during the 54-hour test, suggesting good FAD-GDH stability during the continuous measurement.

Amperometric Glucose Sensors for Whole Blood Measurement Based on Dehydrogenase Enzymes 345

**Figure 16.** Variation of UV absorbance at 340 nm of 0.3 mM NADH in 0.01 M pH7.4 PBS by carbon electrode with immobilized ferrocene polymeric mediator at 0.3 V potential, by using an Ag/AgCl

0 300 600 900 1200 1500

oxidation time (min)

**Figure 17.** UV spectroscopy monitoring reduction of oxidized NADH in the presence of ADH and EtOH, 0.65ml NADH (0.2 mM) + 0.25ml EtOH (99.5%) + 20ul ADH (control) (solid line), 0.65ml oxidized NADH + 0.15mL EtOH (before reduction) (broken line), 0.65ml oxidized NADH + 0.25ml EtOH + 20 ul

300 350 400 450 500

wave length (nm)

For continuous monitoring, use of native NAD as a coenzyme apparently is not ideal due to its low molecular weight and high water-solubility. Direct immobilization of free NAD at electrode has been a long-standing challenge. [64, 65] There are a large number of publications in the literature concerned with NAD retention for various purposes. In the field of biosensors, one promising approach is covalently attaching NAD moieties to polymer chains which can be immobilized at electrode by different means. This can be achieved by directly coupling NAD to an electrode modified with a polymer bearing functional groups,[66] entrapping polymeric NAD in semipermeable membranes [67, 68] or synthesizing a charged polymeric NAD which was then physically adsorbed at electrode surface in conjunction with counter-charged polymer(s) in a manner of layer-by-layer. [69] At LifeScan Scotland Limited, we intend to develop continuous monitoring biosensors using NAD-dependent enzymes by immobilizing all the redox species, i.e.

reference electrode and a platinum counter electrode, at room temperature

0.0

0.0

0.5

1.0

absorbance (A)

1.5

2.0

0.5

1.0

absorbance at 340 nm

1.5

2.0

ADH (4mg/ml) (dotted line)
