`
`Anal. Chem. 1988, 60, 2781-2786
`
`2781
`
`transition time in the submillisecond region can be obtained
`by this technique. The transition time obtained with a 40-yA
`preset current is 885 us. Since the geometry of the gold
`microwire electrode is uncertain, the calculation of the the-
`oretical value for the transition time is not attempted. The
`effective scan rate for this chronopotentiogram is about 1100
`V/s. Cyclic voltammograms at similar scan rates were only
`obtained with an ultramicroelectrode after signal averaging
`(28). It also shows that the top and bottom portions of the
`transient are partially removed due to the decay of the current
`transient to the presetlevel in less than one cycle time (125
`us). The cycle time and thus the background overcorrection
`could be minimized by collecting data at shorter time intervals
`with emaller potential pulses, We also note that the shape
`of the transients is sharper than deacribed by eq 2, Similar
`behavior was predicted in potentiometric stripping analysis
`(26) and CCC with an ultramicroelectrode (27).
`In conclusion, the very low background achieved by sam-
`pling the current on the pulse and bythe use of a dynamic
`background correction is essential for the characterization of
`chronopotentiograms in the millisecond level. ‘The analytical
`utility of SPPC is clearly demonstrated by the fast anodic
`stripping analysis of ppb-level Cd(II) and Pb(II) in aqueous
`solution.
`It appears that an accurate measurement of a
`submillisecond transition time may require a faster data ac-
`quisition frequency.
`
`LITERATURE CITED
`(1) Lingane,J. J. J. Eleciroanal. Chem. 1960, 7, 379.
`(2) Reinmuth, W. H. Ana/. Chem. 1964, 99, 485-487.
`GC. Anal, Chem. 1961, 39, 1123-1124,
`(4) Laky, R. W.; Molntyre, J. D. Ev. Am. Chem. Soc, 1965, 87,
`38086-3812,
`
`for
`
`(5) Shuits, W. D.; Haga, F. E.; Muetier, T. A.; Jones, H.C. Anal. Chem.
`1085, 37, 1415-1416,
`(6) Sturrock, P. E.; Hughey, J. L,; Vaudreuil, B.; O'Brien, G. E. J. Eleciro-
`chem. Soc. 1078, 122, 1196-1200.
`7) Sturrock, P. E.; Vaudreuil, B. J. Electrochem. Soc. 1875, 122,
`13911-1315.
`(8) Sturrock, P. E.; Gibeon, R. H, J. Electrochem, Soc. 1978, 123,
`629-631.
`(8) Kato, Y.; Yamada, A. Yoshido, N.; Unoura, K.; Tanaka, N, Bul,
`Chem, Soc, Jor, 1081, 64, 175-180,
`(10) Schreiber, M. A; Last, T, A, Anal. Gham. 1981, 53, 2095-2100,
`11) Hussam, A.; Guneratna, G. Anal, Cham, 1968, 60, 503-507,
`12) Hance, G. W.; Kuwana, T. Ang/, Cham. 1987, 59, 191-134.
`13) Sawyer, D. T.; Roberts, J. L., Jr.
`f
`te;
`7
`(14) Aubanel, E, E.; Oidham, K. 8. Byte February 1985, 207-218.
`(16) Olmstead, M. L.; Nicholson, R. S. J. Phys. Cham. 1988, 72,
`(16) de Vries,W. T. J. Electroanal. Chem, InterfacialElectrochem. 1668,
`7
`+
`(17) Rodgers, A. S.; Meltes, L. J. Eleciroanal. Chem. Interfacial Electro-
`chem. 1968, 16, 1-11.
`(18) re Electroanal. Chem. Interfacial Electrochem, 1072, 34,
`(19) Perone, S. P.; Brumfiek, A. J. Electroanal. Chem. Intorfaole! Electro -
`chem, 1967, 13, 124-131.
`(20) Hullang, H.; Yagner, D.; Renman, L. Ana/. Chim. Acts 1067, 202,
`117-122,
`(21) Hulllang, H.; Yagner, D.: Renman, L. Anal. Chim. Acta 1887, 202,
`123-128.
`(22) Beranaski, A. 8. Ana/. Chem, 1087, 59, 662-666.
`{23) Kounaves, 8, P.; O'Dea,J. J.; Chandrashekar, P.; Osteryoung, J. Anal.
`Ghem. 1987, 59, 386-389.
`(24) Aoki, K.; Akimoto, K.; Tokuda, .; Mateuda, J.; Osteryoung, J, J. Elee-
`troanal., Chem. Interfacial Electrochem. 108%, 182, 281.
`(25) Howell, J. ©.; Wightman, R. M. Anal. Chem. 1984, 56, 524-529.
`(26) Hussam, A.; Coatzrea, J. F. Anal. Cham, 1988, 67, 581-583.
`(27) Galue, 2.; Schenk, J. 0.; Adama, R. N. J. Electroanal. Chem. Interfa-
`cial Electrochem. 1982, 795, 1.
`
`RECEIVED for review April 11, 1988, Accepted September 23,
`1988,
`
`Covalent Enzyme Coupling on Cellulose Acetate Membranes
`for Glucose Sensor Development
`
`Robert Sternberg, Dilbir S. Bindra,? George S. Wilson,** and Daniel R. Thévenot*#
`Laboratoire de Bioélectrochimie et d'Analyse du Milieu, U.F.R. de Sciences et Technologie, Université Paris—Val de
`Marne, Avenue du Général de Gaulle, 94010 Créteil Cedex, France, and Department of Chemistry, University of
`Kansas, Lawrence, Kansas 66045
`
`Methods for immobilizing glucose oxidase (GOx) on cellulose
`acetate (CA) membranes are compared. The optimal method
`Involves covalent coupling of bovine eerum albumin (BSA) io
`CA membrane and a subsequent reaction of the membrane
`with @Ox, which has previously been activated with an ex-
`cess of p-benzoquinone. This coupling procedure Is fairly
`reproducible and allows the preparation of thin membranes
`(5-20 xm) showing high eurface activities (1-3 U/em*) which
`are stable over a period of 1-3 months. Electrochemical and
`radiolabeling experiments show that enzyme Inactivation as
`a result of immobilization Is negligible. A good correlation
`between surface activity of membranes and thelr GOx load
`Is observed.
`
`INTRODUCTION
`The performance of an enzyme electrodeis ultimately de-
`pendenton the ability of its enzymatic membrane to sustain
`
`* Author to whom correspondance should be directed.
`1Université Paris—Val
`de Marne.
`* University of
`
`and protect the enzyme. Among the available methods of
`enzyme immobilization, four methods are currently used:
`adsorption followed by reticulation with bifunctional reagents
`such as glutaraldehyde (1), covalent coupling between enzyme
`and activated support (2-5) or activated enzyme and support
`(6-8), and reversible immunological coupling (9). Those in-
`volving covalent coupling to solid supports are of great interest
`since they generally yield the best activity stabilities (10).
`Nevertheless two difficulties may be encountered:
`low levels
`of activatable or activated surface groupe on the support and
`denaturation of enzymeif covalent coupling is accomplished
`through functional groups of the enzyme which are essential
`to ita catalytic activity, Highly active and stable membranes
`may be prepared by acy) azide activation of reconstituted
`collagen films (2, 17). However such membranes have been
`found to be too thick and too fragile, especially at 37 °C, to
`be recommended for in vivo applications of enzyme electrodes.
`As cellulose acetate (CA) membranes ofdifferent thickness
`and permeability may easily be prepared byfilm casting or
`coating and because they exhibit significant permselectivity
`toward anions (12), we have studied their ability to support
`enzymeand be used for an in vivo implantable glucose sensor.
`
`0003-2700/88/0360-2761$01.60/0 © 1988 American Chemical Society
`
`Dexcom Exhibit 1009
`
`
`Dexcom Exhibit 1009
`
`

`

`2782 ¢ ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988
`
`(a)
`Figure 1. Schematic dlagram of rotating membrana electrode:
`rotating disk electrode shaft and elecirical contact, (b) Ke-F body, (c)
`threaded collar, (d) membrane support cap,(e) platinum electrode, and
`(f) membrane.
`
`The literature is abundant with reports of glucose sensors
`and many of these are used routinely for in vitro clinical
`determinations. Despite a history of more than 25 years (13),
`no completely reliable implantable sensor has yet been de-
`veloped (14). For such applications a miniaturized sensoris
`needed andthe fabrication of such devices through multilayer
`film deposition is a major problem. It is quite evident that
`miniaturized sensors will yield high sensitivity and stability
`only if the enzyme layer and the associated polymer layers
`can be deposited in a reproducible fashion. Commercially
`available membranes are not suitable for this purpose. The
`enzyme can be immobilized and protective polymer films
`deposited by dip coating (15). Because of the low success rate
`(<40%) associated with such a technique, there exists a need
`for optimization of procedures for both enzyme and protective
`layer preparation. Starting with an intact membraneprovides
`a defined system for the optimization process.
`This paper proposes and evaluates a method of glucose
`oxidase (GOx) coupling to CA membranes. These membranes
`have initially been activated and coupled to bovine serum
`albumin (BSA)in order to increase the numberof functional
`groups to which GOx maybe coupled and to impart a proteic
`environment to the enzyme, GOx is coupled to CA-BSA
`membranes by using p-benzoquinone (PBQ)as bifunctional
`reagent (9, 16, 17). Use of radiolabeled GOx,in order to keep
`a track of enzymatic activity and investigate the reasons for
`the loss of sensor response with time,is also discussed. This
`work allows for the correct choice of coupling procedure and
`support before the whole glucose sensor is assembled.
`
`EXPERIMENTAL SECTION
`
`Instrumentation. GOx activity of membranes was determined
`by continuous monitoring of enzymatically generated hydrogen
`peroxide on a combined platinum disk electrode(anode diameter
`2mm) (YSI Model 2150) covered with a general purpose cellulose
`acetate membrane. This electrode was connected to a YSI Model
`25 Oxidase Meter (Yellow Springs Instruments, Yellow Springg,
`OH)and to a potentiometric recorder (Curken). Determinations
`of solution and membrane radioactivity were performed on a LKB
`Wallac 80000 automated gamma counter, GOx membranes were
`mounted on a rotated platinum disk electrode (anode diameter
`3 mm)using a specifically designed membraneholder tip (Figure
`1). The rotation speed of the membrane and anode was controlled
`by a Solea-Tacussel (Lyon, France) EDI motor and Controvit
`power unit. A Ag/AgC]reference and a platinum wire auxiliary
`electrode were connected,together with the platinum disk anode,
`toa BAS LC4A amperometric detector (Bioanalytical Systems,
`West Lafayette, IN) and potentiometric recorder (Curken). All
`experiments were performed in thermostated cells at 37 °C.
`
`Chemicals. Reagents. Pure p-benzoquinone (PBQ) was ob-
`tained from Merck and was recrys
`from petroleum ether.
`Sodium borohydride from Sigma (98%) was used and Cerium(IV)
`sulfate-2-(sulfuric acid) was obtained from Alfa Division (Ven-
`tron). Acetone and ethanol (both AR 100%) were obtained from
`Mallinckrodt. Cellulose acetate (CA) with 39.8% acetyl content,
`was purchased from Aldrich Jansen Chimica. Glucose, hydrogen
`peroxide, and phosphate salts were reagent grade and all solutions
`were prepared in doublydistilled water.
`Proteins. Glucose oxidase (GOx) isolated from Aspergillus
`niger, (grade VII, 136000 U/g) and bovine serum albumin (BSA)
`fraction V were obtained from Sigma. Radiochemicals. Na!
`(23 mCi/mL in 0.01 N NaOH) was obtained from ICN Radio-
`chemicals (Irvine, CA).
`Glucose Oxidase Membrane Preparation. a. Ceilulose
`Acetate Membrane Preparation. With stirring, 1.8 g of CA was
`dissolved in 15.8 g of acetone, and 2.8 mL. ofdistilled water was
`added. After homogenization, this solution was cast on a glass
`plate and evaporated for 60 s at. 22 °C td form a thin membrane
`(0.015 mm in the wet state and 0.010 mm in the dry state). The
`CA membrane was removed from the glass plate byimmersing
`the plate in distilled water. The membrane was cut into smaller
`pieces and stored in distilled water.
`b. Cellulose Acetate Membrane Activation. Procedure A.
`Four CA membranes (2.5 cm X 2.5 cm) were suspended for 3.-h
`in 100 mL of 1 mM Ce({IV) solution in 0.1 M nitric acid prepared
`just prior to use. Then the membranes were washed in distilled
`water for 5 min and immersed in 10 mL of 10 mg/mL BSA in
`0.1 M borate buffer, pH 9.0. The volume of BSA solution was
`reduced to about 1 mL and three additions of 2 mg of sodium
`borohydride were made every 20 min at room temperature. After
`asecond washing of 5 min in distilled water, the membranes were
`stored in 0,1 M phosphate buffered saline solution (PBS), pH 7.4,
`at room temperature until the GOx coupling reaction was per-
`formed.
`Procedure B. This method is the same as procedure A except
`that a higher Ce(IV) concentration (100 mM)and a shorter ox-
`idation time (20 min) were employed.
`Procedure C, In this case 100 mL of 100 mM sodium periodate
`(20 min) was employed as oxidizing agent. The membranes were
`washed and BSA added as in procedure A. The Schiff base formed
`was reduced for 2 h by adding 4 mg of sodium cyanoborohydride
`to the 1 mL of BSA solution containing the membrane.
`c. Glucose Oxidase Activation, Freshly prepared PBQ solution
`0.1 mL (15 mg in 1 mL ethanol) was introduced in an alumi-
`num-foil-covered tube containing 0.5 mL of GOx solution (20 mg
`in 0.1 M PBS pH 7.4). After 30 min of incubation at 37 °C, the
`mixture was filtered through a G-25 Sephadex column (1 X 10
`cm) coupled with a peristaltic pump (20 ml/h) and equilibrated
`with 0.15 M sodium chloride. Thefirst fraction, a pink-brown
`band of 2~3 mL, was collected and used as the enzyme coupling
`solution.
`d, Coupling Reaction. The CA-BSA membranes were sus-
`pended in 2-3 mL of activated GOx solution, whose pH was
`adjusted to 8.0-9.0 with 0.25 mL of 1,0 M sodium carbonate
`solution. After 38 h of incubation at room temperature, the
`membranes were removed, washedin stirred 0.15 M potassium
`chloride solution for 24 h andfinally atored in 0.1 M PBS, pH
`7.4, containing 1.6 mM sodium azide. The azide was added to
`all storage and experimental buffers to limit microbial contam-
`ination and catalase activity.
`Radiolabeled Glucose Oxidase Preparation. The iodine
`monochloride method described by Helmkamp,Contreras, and
`Bale (18) was chosen to incorporateI into GOx. A 0.5 mM IC]
`reagent was prepared by adding 4 uL of pure ICI to 100 mL of
`2M NeCl. Ina 1-mLconical vial was placed 80 nL of 0.5 mM
`ICI and the desired amount of Na“I solution. Typically, 1,0-2.5
`uCiof !] was used to iodinate 1.0 mg of protein, depending upon
`the specific activity needed, The desired amount of glucose
`oxidase in a volumeof 0.20 mL was added to the vessel with the
`room lights turned off. The contents of the vessel were mixed
`by tapping and allowed to react for 2 min in the dark. The
`reaction mixture was applied to the top of a G-26 Sephadex
`column (25 X 1 em) that had been equilibrated with PBS, pH
`7.4. Elution was continued with PBS at a flow rate of 0.6 mL/min
`and 1-mLfractions were collected. The effluent was monitored
`
`Dexcom Exhibit 1009
`
`
`Dexcom Exhibit 1009
`
`

`

`ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988 « 2783
`
`OR ee CH; NH-
`
`NH:
`
`NH,
`
`NH,
`NH;
`NH
`
`with a gamma counter and the first fraction, containing labeled
`protein, was collected. The pooled protein was concentrated to
`about 1.5 mL by using pressure dialysis. The ratio of protein-
`boundiodideto total iodide was determined bythe trichloroacetic
`acid (TCA) precipitation method (19). The value, expressed as
`efficiency of labeling, was usually 95-98%. The same data were
`used to calculate the specific activity of the iodinated glucose
`oxidase in microcuries of I attached per milligram of glucose
`oxidase. The spectrophotometric determination of enzymatic
`activity of radiolabeled glucose oxidase solution showed retention
`of 80-85% of tho original activity.
`Glucose Oxidase Membrane Characterization. a. Enzy-
`matie Surface Activity Determination. The hydrogen peroxide
`anodic sensor, covered with a nonenzymatic CA membrane, was
`immersed into a thermostated cell (87 °C) containing 10 mL of
`PBS. Whenthe background current was stabilized,i.e. after 10
`min of polarization, 0.1 mL of 1 M glucose standard solution waa
`added to the buffer solution. The incremental production of
`hydrogen peroxide was measured by placing the CA-BSA-
`PBQ-GOx membranein the stirred buffer solution and then
`removing it. This process was repeated several times. There-
`sulting increase in current, after suitable calibration with hydrogen
`peroxide standards, could be used to estimate the membrane
`surface enzymatic activity.
`b, Sensor Response. The rotating membraneelectrode was
`dipped into a thermostated call at $7 °C containing 25 mL of PBS.
`Whenthe background current was stabilized,i.e, after about 20
`min of polarization, several 0,026-0.125 mL additions of 1 M
`glucose standard colution were performed until a 15-20 mM final
`concentration was obtained. Then one or two 26-uL aliquots of
`0.1 M hydrogen peroxide were added to the buffer solution.
`Response was calculated by comparing steady-state current either
`to the background current(/,) prior to any glucose addition or
`to the ateady-state current corresponding to the previous addition.
`Thus either J - J, vs C or AI/AC va C or log C curves were plotted.
`¢. Immobilized Enzyme Determination. The mass of immo-
`bilized GOx was estimated from the y activity of the membranes
`prepared by using radiolabeled GOx and the specific activity of
`radiolabeled enzymesolution. With the initial enzymeactivity
`of lyophilized powder (U/g) and a 15-20% decrease of this activity
`due to the radiolabeling procedure taken into account, a further
`estimation of membrane activity expreseed in U/cm? was ac-
`complished.
`
`RESULTS AND DISCUSSION
`
`The purpose of this work was to find a suitable method that
`allows the efficient and reproducible coupling of GOx to a
`support suitable for an in vivo glucose sensor. We were
`particularly concerned with the methods and support mate-
`rials that could easily be adaptable to the preparation of
`needle-type microsensors. Thus all commercially available
`solid membranes were avoided and we restricted ourselves to
`cast CA films as support material.
`Cellulose Acetate Membrane Activation. Bovine serum
`albumin (BSA)(1, 3, 6, 7), gelatin (20), or collagen (2, 8) are
`frequently used in enzyme immobilization techniques (21).
`They bring to the enzymea proteic environment which seems
`favorable for its stability (11). As CA membranes usually
`possess low levels of acceasible hydroxyl groups for covalent
`enzyme coupling, we have decided to increase the number of
`reacting groups by first covalently coupling BSA to CA (Figure
`2). BSA contains more than 57 active amino groups per
`molecule and may be covalently grafted onto the free aldehyde
`groups obtained by CA hydroxyl group oxidation (Figure 3).
`The imine functions are subsequently reduced with boro-
`hydride. We have compared three different procedures for
`CA membraneactivation and CA-BSA membranes from each
`procedure which were reacted with identical activated GOx
`solutions (vide supra). Comparisons of the properties of GOx
`membranes prepared by these three procedures (Table I) show
`that the best activities are obtained with procedure C,i.e.
`periodate oxidation followed by cyanoborobydride reduction.
`The use of cyanoborohydride, which is a milder reducing agent
`
`H-PBO-NH~)NH,
`
`NH
`
`.
`
`Pa
`
`NH
`
`CH;NH (ssa ym 2
`
`“
`
`H,
`
`H-PBQ-N
`
`Figure 2. General scheme of GOx coupling on CA membranes: en-
`hancement of CA surface groups by covalent BSA coupling.
`
`Table I, Activation Procedures of CA Membranes and
`Properties of CA-BSA-PBQ-GOx Membranes Obtained*
`
`activation
`
`CA membraneoxidation
`reactant
`concentration
`duration (mii)
`reaction with BSA
`concentration (g/L)
`reduction
`Teactant
`concentration (g/L)
`duration (min)
`radiolabeled GOx attached
`(ug-em"*)
`
`GOrxsurface activity
`(U-em™)
`
`CA-BSA-PBQ-GOx membrane
`preparation
`procedure procedure
`b
`c
`
`procedure
`a
`
`Ce(IV)
`1
`180
`
`10
`
`BH,
`8x2
`3x 20
`
`Ce(IV)
`100
`20
`
`10
`
`BH,
`8x2
`3x 20
`
`10,
`100
`20
`
`10
`
`BH,CN-
`4
`120
`
`19218
`80228
`3.8 + 1.5
`(n = 5)
`(nm = 6)
`(n = 3)
`0.502010 102040 2820.40
`(n = 3)
`(n = 5)
`(n = 8)
`
`*Amounts of GOx and activities are expressed as mean +
`standard deviation for a set of membranes prepared identically.
`All GOx masses and activities are referred to membrane surface,
`ie, 0.8 cm.
`
`than borohydride (procedure A and B), probably prevents
`breaking of disulfide bonds of the BSA protein.
`Glucose Oxidase Activation and Coupling. Following
`a procedure described by Avrameas et al. (16), we used p-
`benzoquinone (PBQ)for activating GOx solutions. GOx was
`treated with an excess of PBQ, reagent known to react with
`proteins (17), and the excess PBQ was separated bysize ex-
`clusion chromatography (Figure 3). Most of the above-men-
`tioned coupling reactions take place at low temperature, low
`ionic strength, and within the physiological pH range. The
`reason for treating GOx and not the amino groups of the
`CA-BSAsupport with PBQ is thatit allows GOx-PBQ to react
`
`Dexcom Exhibit 1009
`
`Dexcom Exhibit 1009
`
`

`

`2764 « ANALYTICAL CHEMISTRY, VOL. @0, NO. 24, DECEMBER 15, 1988
`
`membrane activation
`
`jou
`
`Co(lv) or 105
`oxidation
`
`4-cHo
`
`BSA-Uu,— {-cH=n-Bsa + H,0
`
`BHzorBH,CN-
`
`{cHen-BsA eereduction
`
`{-CHpNH-BSA-NH,
`
`anzyme activatien
`
`p~benzequinens
`———————
`
`GOx-NH,
`
`OH
`
`OH
`
`GOXx-NH
`Intermediate
`form
`
`0
`oxcess of pBQ GOx-NH
`EN
`activated ( )
`
`form
`
`0
`
`Coupling reaction
`
`separation ef excens sf pQQ (sephadax 025)
`
`{-CHyNH-BSA-NH,
`
`activated GOx—
`
`deneNH~-BSA-NH
`
`OH
`
`OH
`
`NH- GOx
`
`Figure 3. Chemical reactions involved In CA-BSA-PBQ-GOx membrane preparation.
`
`a
`
`6
`
`o
`
`12
`
`iP
`
`is
`
`4
`
`2:
`
`4iax B
`
`o 0
`
`Immobilized radiolabeled GOx ( ug/cm2 )
`Figure 4, Correlation of GOx surface activities with Immobllized
`1987_GOx amount in different CA-BSA-PBQ-GOx membranes.
`
`estimated from the amount of GOx immobilized, and the
`specific activity of the initial lyophilized GOx powder, were
`compared to experimental activity values, an excellent cor-
`relation was obtained. In fact the measured/calculated ac-
`tivity ratio was equal to 0.84 + 0.29 for six membranes and
`22 experiments indicating that the radiolabeling procedure
`allows a good evaluation not only of total GOx mass butalso
`of its enzymatic activity.
`Several experiments involving the use of radiolabeled GOx
`were carried out in order to understand the reasonsfor the
`loss of enzymatic activity with time, a phenomenon observed
`for most membranes prepared. Figure 5 compares the
`time-dependent loss of enzymatic activity with the loss of
`enzyme mass from the membrane surface, For thefirst 7 days,
`the removal of enzyme from the membraneis directly reflected
`in the loss of biological activity. After this time, however, the
`biological activity decays much more rapidly. It was surmised
`that this latter phenomenon was due to time dependent de-
`naturation of the immobilized enzyme resulting from multiple
`covalent attachment. It is possible for the activated groups
`on the membranesurface to react with protein amine functions
`even after the coupling reaction is supposedly terminated.
`Such further BSA-GOx or GOx-GOx coupling may modify
`the enzymeconfiguration and produce some denaturation. To
`prevent this from occurring, 1 M lysine was reacted with
`membranesfor 6 h immediately after coupling. The purpose
`of this procedure is to deactivate remaining groups on the
`
`with CA-BSAas well as with other GOx molecules allowing
`multilayer enzymatic fixation. In order to check the possible
`effect of PBQ on the enzymeactivity, GOx activities of all
`solutions were measured during different steps of this acti-
`vation procedure, Taking into accountvariation of solution
`volumes, spectrophotometric and electrochemical determi-
`nation of GOx activities in 0.1 M glucose solutions showed
`that GOx-PBQ solutions yield 92-97% initial activity. This
`result indicates that no significant reduction of GOx activity
`ia occurring during this activation procedure. As reaction
`between.GOx and PBQis performed iin an excess of PBQ,the
`GOx-PBQ product exhibits quinone properties and may
`continue to react with proteins such as CA-BSAto yield
`CA-BSA-PBQ-GOx membranes. Control experiments with
`nonactivated CA membranes showed nosignificant attach-
`ment of the enzyme.
`Characteristics of Glucose Oxidase Membranes Used
`for Glucose Sensors. The characterization of the enzymatic
`membranes is important for the evaluation of various enzyme
`attachment procedures. We used three different approaches
`for such a characterization:
`(a) membrane surface activity,
`determination using continuous electrochemical monitoring
`of hydrogen peroxide in a stirred 0.1 M glucose solution; (b)
`mounting the membrane on a rotating disk electrode tip
`(Figure 1), for well-defined hydrodynamics, and determination
`of sensor calibration curve to glucose; (c) incorporation of a
`radiolabel inGOx and determination of the maas of GOx in
`solution or on the membrane.
`Thefirst method is simple and rapid but not very accurate
`since hydrodynamic conditions are not well defined when
`membranes are freely stirred in the measuring cell. The
`second metnod is more accurate but takes into account
`membrane permeability to substrates and reaction products;
`furthermore a good contact between these thin membranes
`and the platinum disk is not trivial. Finally, the third method
`is definitely the most complicated because it requires “I-GOx
`preparation and characterization, but it is also the most ac-
`curate and sensitive. Figure 4 shows the correlation between
`enzymaticactivity, as monitored by surface activity, and the
`mass of immobilized enzyme. The mass of enzyme on the
`membraneof different membranes was controlled such that
`the amountof enzyme immobilized could be varied. All these
`data obtained just after coupling show an excellent correlation
`between surface activity and the mass of GOx immobilized.
`Furthermore when enzymeactivities of these membranes,
`
`Dexcom Exhibit 1009
`
`
`Dexcom Exhibit 1009
`
`

`

`
`
`Table II. Pertinent Parameters for GOx Membranes
`
`ANALYTICAL CHEMISTRY, VOL. 60, NO, 24, DECEMBER 16, 1988 © 2765
`
`collagen—GOx
`reconstituted
`collagen
`acy] azide
`activation
`
`Col-GOx + CA
`collagen +
`cel acetate
`acyl azide
`activation
`
`CA + GOx
`cellulose
`acetate
`entrapment
`wi
`CA
`
`CA-BSA-PBQ-GOx
`cellulose
`acetate
`covalent coup-
`ling with PBQ
`
`I
`|
`
`parameters
`membrane
`material
`attachment
`procedure
`membrane thickness (4m)
`
`100
`200-400
`
`&10
`
`100 + 15
`300 + 15
`
`0.01-0.1
`
`0.1
`40
`50-90
`3-6
`
`5-25
`5-25
`
`0.01-0.1
`
`5-25
`5-25
`
`4-10
`
`swollen
`glucose responses:
`sensitivity? (mA-M"-cm™)
`linear range (mM)
`0.001
`0.1
`0.001
`lower limit
`2-3
`5-30
`2-3
`upper limit
`10-60
`10-60
`20-50
`response time transient (s)
`0.5-8
`2-4
`2-4
`steady state (min)
`10-150
`4-8
`120-2000
`stability’ (days)
`* All sensor sensitivities are referred to anode ares,i.e. 7.06 mm®, >Stabilities are evaluated for a 50% decrease of sensitivity.
`
`zsi
`vg
`4
`oa
`5
`S
`"
`§
`i
`a
`3
`
`if
`3,
`»
`$
`3
`x4
`g
`a
`is
`Z
`
`Be
`
`3
`
`8
`
`10
`Time ( days )
`Figure 6. Relative evolution of surface GOx activity (O} and °1-GOx
`arount (X) In a CA-BSA-PBQ-GOx membrane treated with lysine
`atter coupling.
`2.3
`
`18
`
`no
`
`1
`(U/cm2) 0
`
`Surfaceaclivity
`
`ne
`"
`i
`5
`8
`Time ( days )
`Figure 5, Relative evolution of surface GOx activity (O) and 11-Gox
`(%) in a CA-BSA-PBQ-GOx membrane not treated with lysine after
`coupling.
`
`membrane, thus preventing further coupling. The effect of
`this treatment is shown in Figure 6 where it will be noted that
`the enzymatic activity tracks the mass of immobilized protein.
`Protein is still lost from the membraneat an approximately
`comparable rate. However, in the absence of radiochemical
`measurements, membranes showed significant improvement
`on these stability values, probably because of decreased
`handling. A relative activity value as high as 80-85% was
`frequently observed after passage of 2-3 weeks (Figure 7). As
`the membranes were very fragile and unsupported, handling
`them on a routine basis must have caused their partial dis-
`integration. These results demonstrate that enzymeinacti-
`vation is negligible after coupling and that any slight surface
`activity decrease, when observed, is probably related to im-
`mobilized enzyme loss from the membrane.
`Microbial degradation as a possible explanation for the
`observed behavior was ruled out on the basis that azide was
`present as an inhibitor in all storage buffers. The fact that
`nosignificant release (<1%) of iodide was observed, as con-
`firmed by the repetitive TCA assays (16) on stored iodinated
`enzymesolution also supports this belief.
`Table II presenta pertinent analytical parameters for GOx
`membranes when mounted on a platinum anode. The sensor
`sensitivities, as referred to anode area, of the CA~BSA-
`PBQ-GOx membranes are very close to those for highly active
`acyl azide activated collagen membranes (2). As CA mem-
`branes are significantly thinner than reconstituted collagen,
`especially when the latter are swollen, both transient and
`steady-state response times for them are correspondingly much
`shorter. The stability at 37 °C and anion permselectivity of
`CA membranes are also preferable to those of collagen for
`
`2
`
`a
`
`e
`e
`Time (days)
`Figure 7. Evolution of surface GOx activity In a CA~-BSA-PBQ-GOx
`membrane treated with lysine after coupling and not routinely handied
`for radiochemical measurements.
`
`10
`
`Ww
`
`a
`
`development, of in vivo glucose sensors.
`Figure 8 presents calibration curves of sensors prepared with
`such membranes, Thelinear ranges usually reach 2-3 mM
`glucose for such active collagen—-GOx or CA-BSA-PBQ-GOx
`membranes indicating that, for higher glucose concentrations,
`the enzymatic reaction is the rate-limiting step. High linear
`ranges have been obtained either with less active CA mem-
`branes using a GOx entrapment procedure or with collagen~-
`GOx membranes covered with nonenzymatic CA membranes
`allowing external diffusion restriction (see Table II). Indeed
`linear ranges as high as 15 mM are needed for potentially
`implantable glucose seneors (14).
`
`Dexcom Exhibit 1009
`
`
`Dexcom Exhibit 1009
`
`

`

`2786 @ ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988
`
`]7|
`
`
`
`&=sSeneorresponse{(uA/cm2)
`
`(4) Moody, G. J.; Sanghera, G. 8.; Thomaa, J. D. R. Analyst (London)
`1906, Neeaceey
`(5) Moody, G.
`J; Banghera, G. 8,.; Thomas, J. D. A. Analyst (Lorwon)
`1086, 111,“Tosst2s8,
`(6) Kulya, J. J.; Gureviciane, V. V.; Laurinaviclus, V. A.; Jonuska, A, V.
`Biosensors "1088, 2, 35-44,
`(7) Mullen, W. H.; Keedy, F. H.; Churchouse, S. J.; Vadgama, P. M, Anal.
`Chim. Acta 1088, 183, 69-66.
`(8) Maseinl, M.; Mateescu,M. A.: Pilloton, R. Bloelectrocham. Bloenerg.
`1986, 16, 149-157,
`(8) DeAiwa, W. U: Hill, B, S.; Melklajohn, 6. 1; Wilson, G. S$. Anal,
`Caets59, 2688-2691.
`(10) Thévenot, D. R. Diabetes aei 5(3), eee:
`(11) Thévenot, D. R.; Sternberg, R.; Coulet, P
`Diabetes Care 1982,
`&(3), 203-208.
`(12) Sitarnpalam, G.; Wilson, oS ae Chem. 1883, a 1608-1610.
`(13) Scheller, F, W.; Pielffer, D. 5 cee Renneberg, Fi + Kirstein, D,
`In Biosensors: Fundamentals and Applications; Turner,A. P. F., Ka-
`rubs, J., Wilson, G. S., Eds.; Oxford University Press: New York,
`
`1987;pp315-346.
`18
`s
`®
`10
`.
`es
`a0
`(14) Velho,
`G; Reach, G.; Thévenot, D. A. In Biosensors: Fundamentals
`Glucose concentration ( mM )
`and
`tions; Turner, A. P. F., Karube, I., Wilson, G. &., Eds.;
`University Presa: New York, 1987; pp 390-408,
`(18) polio Kawamorl, R.; Yamasakl, Y. In Biosensors: Fundemen-
`andApplications; Turner, A, P, F., Karube, outing aye Eds,;
`Ontord University Press: New York, 1987; pp409-424
`(16) Ternynck, T.; Avrameas, S. Ann.
`Immunol, (Paris) 1976, 127C,
`197-208,
`(17) Webb,J. L. In Quinones in Enzyme.and Metaboke Inhibitors 117; Aca-
`demic: New York, London, 1966; pp 421-594,
`(18) Helnkamp, R. W.; Conteras, M. A.: Bale, W. F. Int. J. Appi. Radiat.
`fsot, 1067.
`(18) Der-Ballan, G. P. Anal, Blohom, 1980, 106, 411-418.
`(20) Romette, J. L.; Yang, J. S.; Kusakabe, H.; Thomas, D. Biotechnol.
`Bloong. 1883, 25, 2857-2668,
`(21) Barker, S. A, In Bosensors, Fundamentals and Applications; Tumer,
`A. P, F., Karube, I., Wilson, @ S., Eds.; Oxford University Press; New
`York, 1987; pp 86-09.
`(22) Stemberg, R.; Barrau, M.-B.; Ganglott, L.; Thévenot, D. R.; Bindra, D.
`S.; Wilson, G. 8.; Vetho, G; Froguel, P; Reach, @ Biosensors.
`In
`press.
`
`Figure. 8. Glucose calibration ourves of sensore prepared with (O)
`collagen—GOx, (x) CA with entrapped GOx, and (A) CA~BSA-PBQ-
`GOx membranes.
`
`This study underlines the importance of careful control of
`coupling conditions in order to achieve reproducible immo-
`bilization of enzyme with high activity and stability. The
`coupling methods described here are successfully adapted to
`needle-type microsensors (22). An extended linear range,
`necessary for subcutaneous implantation, is achieved by
`adding an external membrane (polyurethane) with concom-
`itant sacrifice of sensitivity and to some extent response time.
`A complete evaluation of these microsensors is currently in
`progress.
`
`ACKNOWLEDGMENT
`We thank Uditha de Alwis for numerous discussions and
`suggestions concerning the immobilization chemistry.
`LITERATURE CITED
`(1) Amok, M. A.; Rechnitz, @ A. Anal, Gham. 1982, 54, 2315-2317,
`(2) Thévenot, D, A. Coulet, P. Ru: Sternberg, F.; Gautheron, D.C. Ana.
`Chem. 1678, Bt,96-100.
`(3) Wingard, L. B.,
`Jr; Casiner, J. F.; Yao, 8. J; Wolfson, S.
`K., uri
`Drash, A. L.; tuCeG, Appl. Biochem. Biotechnol. 1084, 9,‘ooida
`
`RECEIVED for review December9, 1987. Accepted September
`26, 1988. This work has been supported by “Caisse Nationale
`de’Assurance Maladie des Travailleurs Salariés” (Grant C,
`N.A.M.T.S.-LN.S.E.R.M. No. 85,3,.54.8.E), French Ministry
`of Foreign Affairs, National Institutes of Health (U.S.) (Grant
`DK 30718) and “Aide aux Jeunes Diabétiques”. Their fi-
`nancial help is gratefully acknowledged.
`
`Dexcom Exhibit 1009
`
`
`Dexcom Exhibit 10