B/och/m/caet ~ ' . s / c a ,4cta, I ! 15(1991)157-165 © 1991 Elsevier Science PublishersB.V. All lights reserved03044165/91/$03.50
157
BBAGEN 236~
A liposomal enzyme electrode for measuring glucose M a r k F. R o s e n b e r g
la, M a l c o l m N. J o n e s 2 a n d P a n k a j M. V a d g a m a
I Department of Medicine, Unit~,'rsityof Manchem'r, Hope Hospital, Salford (U.KJ and z Department of Biochemistryand Molecular Biology, Unit~rsilyof MatwJu~er, Medical School. Manchester ( U.I~)
(Received 14 May 1991) (Revised manusclipt l0 September 1991)
Key vmrds: ~
electrode; Glucoseoxidase; Liposome;Phase transition temperature
electrodes have been described for measuring glucose but have been limited by the saturation kinetics of the glucose oxidase not allowing clinically relevant glucose concentrations to be measured (0-25 mM). One way of alleviating this problem is to use diffusion-controlled memtwanes which result in the enzyme experiencing a smaller substrate concentration t h a n t h a t of the bulk solution. A s a n extension of this concept w e h a v e encapsulated glucose oxidase in liposumes whereby the lipid bilayer wall provides the diffusion-limiting membrane as well as providing a biocompotable layer which is of particular relevance when blood glucose is to be measured. Linear ranges were f o t ~ d to e ~ - a e e the required glucose concentrations and moreover by using liposomes prepared from different lipids, e.g., dimyristoyl (14:0) phosphatidyleholine (DMPC), dipalmitoyl (16:0) phosphatidylcholine (DPPC) a n d disteamyl (18:0) p h o s p h a ~ (DSPC), the electrode response was shown to depend on the bilayer permeabilitles in relation to the lipid phase transition temperatures a n d a s a consequence the linear ranges were d u ~ attered.
Introduction A n enzyme electrcde can be defined as a chemical W~n~lucer which functions by combining an electrochemical sensor with an immobilised enzyme where an analyte is enzymatically converted to a product which the detector is re~q~onsive to. One of the prime ad~:antages of such a sensor is that it allows direct reagentless measurement in optically opaque samples thus providing real-time monitoring. Particular interest has been shown in measuring glucose concentrations in biological solutions and in tissues, but despite this, few glucose enzyme electrodes (with noh~ble exceptions, Refs. 1 and 2) have achieved routine use in medicine. Glucose electrodes have taken a prominent role in biosensor research due to their anticipated use as glucose monitoring devices for diabetics [3]. The measurement of glucose by electrochemical techniques was originally described by Updike and Hicks in 1967 [4], who measured glucose by following the changes in oxygen partial pressure using a Clark oxygen electrode.
Corresl~ndence: M.N. Jones. Department of Biochcm~j and Molecular Biology. University. of Manchester, Medical School. Manchester. MI3 9PT. U.K.
They found that when oxygen was in non-rate-limiting excess and the glucose concentration was well below the apparent K m for the inunobilised glucose oxidase, there was a linear relationship between glucose concentration and decrease in oxygen tension. However, the effect of background pO 2 variation on measurement required either an O 2 equilibration step or simultaneous pO 2 measurement. This was later obviated by measuring hydrogen peroxide from the oxidase reaction at the polarising voltage used for H 2 0 2 detection ( + 0 . 6 V relative to a standard calomel electrode) [5]. However, at this potential other electrochemically active species such as ascorbate, urate and tyrosine found in blood were shown to generate a significant response which was then countered by using a low-molecularweight 'cut-off' cellulose acetate membrane (200 daltons) interposed between the electrode surface and the assay medium [6]. The linear ranges of enzymatic methods are normally limited by the Michaelis constant of the enzyme (Kin), and thus sample dilution is necessary for clinical assays where glucose conce~ttrations in excess of 10 mM may be measured for diabetic patients; measurement of undiluted samples however would simplify instrumentation and eliminate any volume errors during dilution caused by the presence of cellular ele-
158 ments in blood. The linear range may be extended through a reduction in the local glucose concentrmion, while maintaining sufficient oxygen (co-substrate) for the enzymic reaction; any increase in membrane permeability for oxygen relative to that. of glucose will increase the ratio of oxygen to glucose within the enzyme layer, so mitigating the effects of oxygen limitation on the e n z ~ a ~ reaction at high glucose levels in bulk solution. Therefore the use of barrier membranes in extending the linear range was proposed and exampies include the use of polyurethane [7] and organosilane reagents [8] both of which reduce the substrate accessability to the enzyme while maintaining sufficient oxygen availability for the enzymic reaction. Moreover oxygen has been shown to have a significant solubility. in lipid [9] while not allow/ng glucose partitioning; this concept was therefore used to construct a glucose enzyme electrode by supporting lipid in a microporous membrane over a glucose o:ddasc layer [10]. It therefore was apparent that an ideal membrane for enhancing the linear range of an enzyme electrode might be composed of lipid and in addition should be able to reduce glucose permeability to the enzyme; such a structure was thought to be a liposome encapsulating glucose oxidase. Liposomes are vesicles in which an aqueous volume is entirely enclosed by a membrane composed of lipid molecules (usually pbospbolipids) in a bilayer configumtiou. They form spontaneously when lipids are dispersed in aqueous media above their chain melting temperatures and can be constructed so that they entrap glucose oxidase within their aqueous compartmeuts. Polar solutes such as glucose have been shown to pass a c r ~ the lipusomal membrane, although only very slowly [11]. The encapsulation of glucose oxidase in phospholipid vesicles has been desen'bed previously [12,13] although there are no reports of such liposomes being used for measuring glucose with an enzyme electrode. Therefore in order to test out this possibility lipnsomes were immobilized on a Millipore membrane and interfaced with a Clark Oxygen electrode polarised for hydrogen peroxide detection. It was envisaged that the glucose would diffuse across the lipusomal bilayer resulting in the catalytic generation of hydrogen peroxide which diffuses out of the lipusome to be oxidized on the platinum surface of the electrode. It was hoped that this would in turn give rise to a current in proportion to the glucose concentration, according to the process; n-glucose + 0 2 + H 2 0 ~ v-gluconic acid + H 2 0 2 H 2 0 2 ~ 2H ÷ + 2 e - + O 2
Furthermore, in order to show that the electrode response was dependent on the surrounding lipid, the responses of liposomes composed of DMPC, DPPC
and DSPC respectively were monitored with regw.ct to temperature through the lipid phase transition temperatures, where a marked change in current due to an alteration in glucose permeability, arising from packing abnormalities in the liposumal bilayer should be observed [11]. In addition the extent of iinearity for liposomal enzyme electrodes composed of these different lipids was compared and correlated with their differing bilayer fluidities and permeab~lities. Materinls and Methods A basic Clark oxygen electrode system (Rank Brothers, Cambridge) was used. The electrode was lmlarised at +656 mV for hydrogen peroxide detection and the meter was linked to a chart recorder as descn'bed previously [8]. The electrode consisted of a central 2-nlm diameter platinum working electrode with an outer 12-ram diameter silver ring as the ~,eedoreference.
L-a-dipalmhoylphosphatidylcholine (DPPC, product no. P-0763), L-a-distearoylphosphatidylcholine (DSPC, product no. P-1138), L-a"-dm~Tisteeylplmshatidylcholine (DMPC, product no. P-0888), Sepharose 4B and o-dian/sidine tablets were all suppl/ed by Sigma (Poule, U.K.). Phosphatidylinositol (Pl) from wheat germ (as the sodium salt) was Grade I from Lipid Products (South Nuttield, U.IC). Chloroform and methanol were purchased from BDH (BDH, Poole, U.K.). PBS buffer was supplied by Oxoid (Oxoid limited, London, UK.). Latex beads were supplied by Polysciences (Warrington, PA, U.S.A.). [3H]DPPC (specific radioactiv/ty 80 C i / n m ~ , code T R K 673) was purchased from Amersham International (Amersham International PLC, Amersham, U.IC). Glucose oxidase (EC 1.1.3.4 from ~ n/get; 128 IU m g - ' protein) and horseradish per~qidase (EC 1.11.1.7 from horseradish; 200 IU mg -1 protein)were supplied by Sigma (Poole, U.K.). The scintillation fluid (Ecosinct A) was supplied by National Diagnostics (Manville, N J, U.S.A.). All other chemicals used were of analytical reagent grade where available. Solutions were prepared with double-distilled water.
Preparation of reuerse-phase ecaporation r'esicles (REV) These were prepared by a modification of the method described by Szoka and Pamlhaffmpouh3s []4] and were used primarily because of their large encapsulated volume for entrapping glucose oxidase. DMPC, DPPC or DSPC (9 mg), depending upon the fiposome suspension being prepared, was added to PI (1 rag) and I00 /zl of [3H]DPPC (2 p.Ci/ml) in 3 ml of distilled chloroform: methanol mixture (4:1). The purity of the lipids was checked by thin-layer chromatography on
159 silica gel plates and the solvents were distilled in order to remove any traces of peroxides. The mixture was rotary evaporated at 6 0 " C to yield a thin lipid film. This was then re-dispersed in 6 ml of 4 : 1 chloroform:methanol and 3 mi of nitrogen saturated 1/10 phosphate-buffered saline (PBS, t =0.15 tool/l) containing 10 mg of glucose oxidas¢. The mixture was vortexed, followed by 4 rain sonication in a bath-tEpe sonicator (Decon l ~ 100, Decon Ultrasonics, Sussex, U.K.) under nitrogen. The temperature of the bath was maintained at 30, 50 and 6 0 " C for liposomes composed of D M I ~ , DPPC and DSPC, respectively, (i.e. above the phase transition temperature of the respective lipid). The resulting homogeneous emulsion was rotary evaporated at a temperature above the phase transition temperature of the lipids. After a viscous gel-phase, gel inversion occurred with the formation of an aqueous suspension. This was purged with nitrogen fox a further 15 rain at a temperature above the phase transition temperature to remove traces of organic solvent and to anneal defects in the liposomes. chroma~ To separate the encapsulated glucose o~dase from the non-encapsulated enzyme the preparation was passed through a Sepharose 4B column (30 × 1.5 can) and oluted using PBS (pH 7.3) at a flow rate of 0.2 m l / m i n and fractions of 2 cm 3 were collected using an LKB Redirac F r a ~ o n Collector (Bronnna, Sweden). Prior to the application of the liposome suspension, the column was pre-treated with a lipnsome dispersion to prevent adsorption of liposomes onto the gel matrix [151. ges/c/e ~ e d/str/bution ana/ys/s by photon condat/on spectroscopy and ves/c/e/ame//ar~y The size ~f the vesicles was determined by a dynamic light scattering technique using a Malvern Autosizer, model number (RR146) Malvern Instruments (Malvern, U.K-). This determined the hydrodynamic diameter on the basis of fluctuations in scattered light intensity with time [16]. The autosizer gives the equivalent normal weight dism'bution of particle diameters from which the weight-average diameter (dw) of the lipusmnes is obtained. Checks of the accuracy of the d,~ values were made by using monedisperse latex beads with nominal diameters of 50, 150 and 244 nm. Measurements were made on liposomes dispetsecl in PBS (pH 7.3). Vesicles prepared by reverse-phase evaporation have been shown to be largely unilamellar [14]. In the case of the DPPC-PI REV the vesicles were investigated by electron microscopy using negative staining with phosphotungstic acid (2% w / v in water) and ammonium molybdate (2% w / v in water) and also by freeze fracture techniques. The micrographs gave no evidence to
suggest that the vesicles were other than largely unilamellar.
Assay procedures Fractions (2 ml) from gel chromatography were assayed for protein according to the procedure described by Lowry et al. [17] using a glucose oxidase standard ( 0 - 2 0 0 / t g / 1 0 0 p.I). In order to disperse the liposomal glucose oxidase, 100/tl SDS 1% (w/v) was added to each liposomal fraction and protein standards. SDS at this concentration has been shown not to denature glucose oxidase in neutral media [18]. The fractions were assayed for lipid from the [3H]DPPC counts obtained by scintillation counting (Beckman LS 9800 U.S.A.). These parameters were then used to express the encapsulated glucose oxidase concentration in/~g glucose oxidase//t tool lipid. Immobilization of the h'posomes and membrane characterisation In order for the liposomes to be used in conjunction with a hydrogen peroxide sensing electrode, the liposomes were immobilised by filtering a fixed volume (1 ml) of the appropriate liposome solution under vacuum onto a Millipore nitrocellulose membrane (1 cm2), 0.45 /zm pore size (Millipore, Watford, U.K.) followed by washing of the membrane with PBS buffer (pH 7.3). The radioactivity of the Millipore membranes was determined by scintillation counting by adding 4 ml of scintillation fluid to the membranes and the counts were then used to assess the liposome density on each membrane. By relating the DPM for the membrane to the D P M / m o l of lipid, the number of tool of lipid on the membrane could be evaluated and since the numbor of tool of protein per tool of lipid was known, the number of tool of bound protein could also be ascertained. This was undertaken after the glucose responses of the appropriate membrane had been tested. Procedure for glucose determination at various temperatares An isotonic PBS buffer was used for aqueous sampies. The surface of the hydrogen peroxide sensor was moistened with buffer to allow electrical contact between working and pseudoreference electrodes. The appropriate Millipore immobilised liposomal membrane was then positioned over the electrode and was held in position by the screw-fit top of the electrod~ body. Aliquot,, of a stock solution of 1 M o-glucc~se were added to an appropriate volume of buffer in the sampie chamber and the medium stirred to homogeneity. The steady state currents were then recorded. This was repeated for a range of liposome membranes; D M P C / P I , D P P C / P I and D S P C / P I (all of weight ratio 9:1). In addition the glucose responses were
160 tested over a range of temperatures using a thermostatted reaction cell. No hysteresis effects of the electrode response were observed on heating and cooling. The stability of the electrode responses were tested for D P P C / P ! electrodes. The electrode currents were found to be stable over a 2 h period with respect to a glucose standard. This long-term stability imp~es firstly, that the continuous production of hydrogen peroxide does not have any detrimental effects on the iiposomal electrodes, such as the poss~le oxidation and degradation of unsaturated acyl chains which might be present in the PI and secondly, that the integrity of the liposomes in the electrode is maintained, since loss of integrity with release of encapsulated glucose oxidase would have resulted in a significant increase in signal (if the electrode cell was not washed out) as occurred on addition of detergent (see below). Furthermore the electrodes showed no hysteresis effects when the glucose concentration was reduced which was also indicative of stability.
to the conditions prevailing when measuring the temperature variation of the electrode responses. (c) The effect of the presence of the organic solvents (chloroform and methanol), son~ation and heating on the activity of glucose oxidase during the reverse-phase evaporation procedure was assessed by subjecting the enzyme to the preparation procedure, but ha the absence of lipid. This d e m o ~ r a t e d that the free enzyme lost approx. 50% of its activity during the process. The encapsulation in liposomes would however offer a degree of protection to the enzyme so that the experiments do not necessarily imply that the encaps~lat:':.i' enzyme would have lost 50% of its activity. (d) The integrity of the i i ~ m e s o n imm~ailizatinn on the filters was assessed using a histocbe~rdcal stain (pbenazine methosulphate plus nitroblue tetmzolinm). No colonr was found to develop on addition of filters with adsorbed iiposomes encapsulating glucose oxidase to the dye, but addition of Triton X-100 to disrupt the liposomes released protein turning the dye blue.
Electrode fouling Control systems The effect of temperature on the intrinsic e ~ kinetics as well as electrochemical kinetics was evaluated: (a) by immobllising glucose oxidase (0.35 m g / m l ) on a Millipore filter by vacuum fdtration. The membrane was applied over the electrode and the glucose responses were evaluated over a range of temperatures in a similar manner as described for the liposomal membranes using comparable enzyme concentrations to that of the lilxr~omal systems. In these experiments the electrodes were maintained at the required temperature for 10-20 rain during which the electrode responses were measured. (b) The glucose oxidase activity in solution was assayed by a coupled enz3qmatic assay employing the chromogen o-dianisidine [19]. o-Dianisidine (1% w / v ) was diluted by dissolving 100 ttl in 12 ml of phusl~hate-buffer (pH 6.0) and 2.5 m! of this solution was added to 0.3 ml of glucose solution (18% w / v ) previously equilibrated overnight. Aliquots (3 nil) of the glucose .~iution were placed in the spectrophotometer cuvette and equih'brated at the required temperature for 20-30 rain. Finally 10 /zl of 0.4 m g / m l glucose oxidase/horseradish peroxidase solution was added, the solution stirred, and the absorbance vs. time profile was monitored over a range of constant temperatures at 460 nm using a CaD" 210 recording spectrophotometer in the double-beam mode. The control consisted of dye solution with glucose and horseradish peroxidase but with the glucose oxidase omitted. The glucose oxidase was only maintained at the measured temperature for a short period (5-10 rain) after its addition to the chromogen so that measurements relate
Should the liposomal electrode be used for the measurement of glucose concentration in clin~al samples the effects of serum components on the response becomes important. There are three sites for potential fouling; the platinum electrode, the immobilized lipoand the membrane containing the iiposomes. A cellulose acetate filter can be used to protect the platinum electrode although it reduces the electrode response, alternatively both the filter holding the immobilized liposomes and the platinum electrode could be protected from serum components by use of a small pore s ~ e polycarbonate filter [3]. Addition of serum to electrodes protected only with a cellulose filter between the platinum and immobilized l i ~ results in a signal reduction at high glucose concentration (25 raM) by approx. 40% suggesting that the immobilized liposomes are being fouled by serum components. Results The formation of RF.V liposomes in the presence of glucose oxidase resulted in the encapsulation of the enzyme in the liposomes. This was shown by the separation of :the liposomally encapsulated gluco~ oxidase from the excess free enzyme by gel filtration through a Se~haruse 4B column (Fig. 1). The liposomes elnted in the void volume of the column; a small amount of lipid eluted at (Vo + V~) which may have formed as a result of liposome disruption. The amount of encapsulated enz~ne depends on the lioosome concentration and hence on the lipid concentration. In these preparations the percentage encapsulation was a few percent of the total glucose oxidase used which is sufficient for the preparation of a functional electrode.
161
DMPC/PI
DMPC/PI 14
~eoo¢"Lipidl~ "1 L'
OD
s
0.4
n
i
0,3
1
O.= '
16
20
fraction number
25 (2nd)
30
O
' 5
~
r
r
lo 15 glucose concentration mM --G--19 c
--~---24 c
, 2o
--~- 28 c
DPPC/PI
DPPC/PI 2OOO[
O.5
(b)
off o
~
12 lO
l
(b)
0.2
o
o
6
~0 1~ 20 25 fraction number (2mlj
lo 16 20 215 g l u c o s e c o n c e n t r a t i o n mM
6
"-6-" 29 C
3o
"e- 35 C
ao
"~" 38 C
DSPC/PI DSPC/PI 40001
0.5
ast (c)
~
~
0.2 0
10
0.1 --D- 21C
ol fraction number (2roll Fig. |. Separation o i iipusom¢~ co,tRaining e,JIrappcd glucose oxidant from free glucose oxidase on a Sepharose 4B column: (a) D M P C / P ! IJposomes; (b) D P P C / P i iip~o~es; and (c) D S P C / P | liposomes. [], lipid and O, prmein. The fract~o~Lswere collected at a flow rate of 0.2 ml /min.
20 30 4o glucose concentration mM - ~ - 34 C
"49"-40 C
"m- 55 C
"-+-"64 C
Fig. 2. Temperature effect on the liposomal electrode response as a function of glucose concentration for liposomal enzyme electrodes incorporating glucose oxidas¢: (a) DMPC/PI liposomes; (b) DPPC/P| liposomes; and (c) DSPC/PI liposomcs (9: ! weight ratio). ]For clarity some data has been omitted from (a) and (b). The liposomes (1 ml)were immobilised by adsorption on a Millipore filter (1 cm 2) and then interfaced with a Clark Ox','gen electrode polarhed for hydrogen peroxide detection ( + 650 mV).
162 TABLE |
glucose oxidase to a Millipore filter was tested over a range of temperatures (Fig. 4). The glucose oxidase concentration used to produce the electrode was coln-
Liposome and membrane cfiaracterisad.m
Liposome mg glucose composition ox/dase/ p:mol lipid
Weightaverage diameterand standard deviation of the weight d/stn-out/ond~ (am_+standard d~iation )
gg glucose ox/dase/m2
DMPC/P! DPPC/P! DSPC/PI
150± 60 134± 79 199± 114
60 42 30
0.33 0.19 0.03
,~
5.o
~ 4~ Characterisation data for these iiposomal preparations and the resulting membranes are shown in Table I. Fig. 2 shows the electrode responses as a function of glucose concentration and temperature for each iiposoreal electrode ~'stem. from which it is seen that the range o f iinearity of response verses glucose concentration decreases in the series DSPC > DPPC > DMPC. Fig. 3 shows the electrode response curves at a constant glucose concentration (5 mM) as a function of temperature. These e,,~eriments demonstrate that the mesomorphic state of me liposomal bilayer has a significant effect on the dectrode response. The electrode currents showed an iacrease as the gel to liquid-crystalline phase transition temperature (T~) of the phospholipid is approached (the values of T~ for pure DMPC, DPPC and DSPC bilayers are 24, 41 and 55 o C, respectively, [20]). Addidon of Pl is known to decrease 7~ for DPPC bilayers by several degrees [21] and would be expected to have an analogous effect on DMPC and DSPC bilayers. For DPPC the experiments were carfled out through the phase transRion and showed that the electrode response passed through a maximum. The permeabilRies of DPPC liposomal dispersions to D-glucose are known to pass through maxima in the region of T~ [22,23]. In this regard it might have been expected that the electrode response for the DMPC electrodes would decrease at higher temperatures ( > 25-30 ° C) however this appeared not to be so, possibly because of the intrinsic greater fluidity of the shorter chain lipid bilayer. In the case of DSPC it was not practicable to extend measurements to higher temperatures. As a control, the response to 5 mM glucose of an electrode made by adsorption of free (unencapsulated)
Fig. 3. The response ~o 5 mM glucose over .z range of temperatures for liposomal electrons of various compositions incorporating gincose oxidase: (a) DMPC/PI GIx~omes; (b) DPPC/Pi fiposomes; and
(c) DSPC/P] lil~somes. The liI~somes were immobi|isedas descn~ed in Fig. 2. The results are means_+S.E.(n =3). T¢ indicates the gel to liquid-cD~tallinetransition temperatureof the pure pho,~ pholipid.
~ 4.0 3..5 3
2o
iS
25
30
temperatureeC
3.z
(b)
3.0
< 2.8 : ~= 2-s o ~ 2.4
2 2°
25
30
35
40
45
temperatureOC 5
4.5
157
BBAGEN 236~
A liposomal enzyme electrode for measuring glucose M a r k F. R o s e n b e r g
la, M a l c o l m N. J o n e s 2 a n d P a n k a j M. V a d g a m a
I Department of Medicine, Unit~,'rsityof Manchem'r, Hope Hospital, Salford (U.KJ and z Department of Biochemistryand Molecular Biology, Unit~rsilyof MatwJu~er, Medical School. Manchester ( U.I~)
(Received 14 May 1991) (Revised manusclipt l0 September 1991)
Key vmrds: ~
electrode; Glucoseoxidase; Liposome;Phase transition temperature
electrodes have been described for measuring glucose but have been limited by the saturation kinetics of the glucose oxidase not allowing clinically relevant glucose concentrations to be measured (0-25 mM). One way of alleviating this problem is to use diffusion-controlled memtwanes which result in the enzyme experiencing a smaller substrate concentration t h a n t h a t of the bulk solution. A s a n extension of this concept w e h a v e encapsulated glucose oxidase in liposumes whereby the lipid bilayer wall provides the diffusion-limiting membrane as well as providing a biocompotable layer which is of particular relevance when blood glucose is to be measured. Linear ranges were f o t ~ d to e ~ - a e e the required glucose concentrations and moreover by using liposomes prepared from different lipids, e.g., dimyristoyl (14:0) phosphatidyleholine (DMPC), dipalmitoyl (16:0) phosphatidylcholine (DPPC) a n d disteamyl (18:0) p h o s p h a ~ (DSPC), the electrode response was shown to depend on the bilayer permeabilitles in relation to the lipid phase transition temperatures a n d a s a consequence the linear ranges were d u ~ attered.
Introduction A n enzyme electrcde can be defined as a chemical W~n~lucer which functions by combining an electrochemical sensor with an immobilised enzyme where an analyte is enzymatically converted to a product which the detector is re~q~onsive to. One of the prime ad~:antages of such a sensor is that it allows direct reagentless measurement in optically opaque samples thus providing real-time monitoring. Particular interest has been shown in measuring glucose concentrations in biological solutions and in tissues, but despite this, few glucose enzyme electrodes (with noh~ble exceptions, Refs. 1 and 2) have achieved routine use in medicine. Glucose electrodes have taken a prominent role in biosensor research due to their anticipated use as glucose monitoring devices for diabetics [3]. The measurement of glucose by electrochemical techniques was originally described by Updike and Hicks in 1967 [4], who measured glucose by following the changes in oxygen partial pressure using a Clark oxygen electrode.
Corresl~ndence: M.N. Jones. Department of Biochcm~j and Molecular Biology. University. of Manchester, Medical School. Manchester. MI3 9PT. U.K.
They found that when oxygen was in non-rate-limiting excess and the glucose concentration was well below the apparent K m for the inunobilised glucose oxidase, there was a linear relationship between glucose concentration and decrease in oxygen tension. However, the effect of background pO 2 variation on measurement required either an O 2 equilibration step or simultaneous pO 2 measurement. This was later obviated by measuring hydrogen peroxide from the oxidase reaction at the polarising voltage used for H 2 0 2 detection ( + 0 . 6 V relative to a standard calomel electrode) [5]. However, at this potential other electrochemically active species such as ascorbate, urate and tyrosine found in blood were shown to generate a significant response which was then countered by using a low-molecularweight 'cut-off' cellulose acetate membrane (200 daltons) interposed between the electrode surface and the assay medium [6]. The linear ranges of enzymatic methods are normally limited by the Michaelis constant of the enzyme (Kin), and thus sample dilution is necessary for clinical assays where glucose conce~ttrations in excess of 10 mM may be measured for diabetic patients; measurement of undiluted samples however would simplify instrumentation and eliminate any volume errors during dilution caused by the presence of cellular ele-
158 ments in blood. The linear range may be extended through a reduction in the local glucose concentrmion, while maintaining sufficient oxygen (co-substrate) for the enzymic reaction; any increase in membrane permeability for oxygen relative to that. of glucose will increase the ratio of oxygen to glucose within the enzyme layer, so mitigating the effects of oxygen limitation on the e n z ~ a ~ reaction at high glucose levels in bulk solution. Therefore the use of barrier membranes in extending the linear range was proposed and exampies include the use of polyurethane [7] and organosilane reagents [8] both of which reduce the substrate accessability to the enzyme while maintaining sufficient oxygen availability for the enzymic reaction. Moreover oxygen has been shown to have a significant solubility. in lipid [9] while not allow/ng glucose partitioning; this concept was therefore used to construct a glucose enzyme electrode by supporting lipid in a microporous membrane over a glucose o:ddasc layer [10]. It therefore was apparent that an ideal membrane for enhancing the linear range of an enzyme electrode might be composed of lipid and in addition should be able to reduce glucose permeability to the enzyme; such a structure was thought to be a liposome encapsulating glucose oxidase. Liposomes are vesicles in which an aqueous volume is entirely enclosed by a membrane composed of lipid molecules (usually pbospbolipids) in a bilayer configumtiou. They form spontaneously when lipids are dispersed in aqueous media above their chain melting temperatures and can be constructed so that they entrap glucose oxidase within their aqueous compartmeuts. Polar solutes such as glucose have been shown to pass a c r ~ the lipusomal membrane, although only very slowly [11]. The encapsulation of glucose oxidase in phospholipid vesicles has been desen'bed previously [12,13] although there are no reports of such liposomes being used for measuring glucose with an enzyme electrode. Therefore in order to test out this possibility lipnsomes were immobilized on a Millipore membrane and interfaced with a Clark Oxygen electrode polarised for hydrogen peroxide detection. It was envisaged that the glucose would diffuse across the lipusomal bilayer resulting in the catalytic generation of hydrogen peroxide which diffuses out of the lipusome to be oxidized on the platinum surface of the electrode. It was hoped that this would in turn give rise to a current in proportion to the glucose concentration, according to the process; n-glucose + 0 2 + H 2 0 ~ v-gluconic acid + H 2 0 2 H 2 0 2 ~ 2H ÷ + 2 e - + O 2
Furthermore, in order to show that the electrode response was dependent on the surrounding lipid, the responses of liposomes composed of DMPC, DPPC
and DSPC respectively were monitored with regw.ct to temperature through the lipid phase transition temperatures, where a marked change in current due to an alteration in glucose permeability, arising from packing abnormalities in the liposumal bilayer should be observed [11]. In addition the extent of iinearity for liposomal enzyme electrodes composed of these different lipids was compared and correlated with their differing bilayer fluidities and permeab~lities. Materinls and Methods A basic Clark oxygen electrode system (Rank Brothers, Cambridge) was used. The electrode was lmlarised at +656 mV for hydrogen peroxide detection and the meter was linked to a chart recorder as descn'bed previously [8]. The electrode consisted of a central 2-nlm diameter platinum working electrode with an outer 12-ram diameter silver ring as the ~,eedoreference.
L-a-dipalmhoylphosphatidylcholine (DPPC, product no. P-0763), L-a-distearoylphosphatidylcholine (DSPC, product no. P-1138), L-a"-dm~Tisteeylplmshatidylcholine (DMPC, product no. P-0888), Sepharose 4B and o-dian/sidine tablets were all suppl/ed by Sigma (Poule, U.K.). Phosphatidylinositol (Pl) from wheat germ (as the sodium salt) was Grade I from Lipid Products (South Nuttield, U.IC). Chloroform and methanol were purchased from BDH (BDH, Poole, U.K.). PBS buffer was supplied by Oxoid (Oxoid limited, London, UK.). Latex beads were supplied by Polysciences (Warrington, PA, U.S.A.). [3H]DPPC (specific radioactiv/ty 80 C i / n m ~ , code T R K 673) was purchased from Amersham International (Amersham International PLC, Amersham, U.IC). Glucose oxidase (EC 1.1.3.4 from ~ n/get; 128 IU m g - ' protein) and horseradish per~qidase (EC 1.11.1.7 from horseradish; 200 IU mg -1 protein)were supplied by Sigma (Poole, U.K.). The scintillation fluid (Ecosinct A) was supplied by National Diagnostics (Manville, N J, U.S.A.). All other chemicals used were of analytical reagent grade where available. Solutions were prepared with double-distilled water.
Preparation of reuerse-phase ecaporation r'esicles (REV) These were prepared by a modification of the method described by Szoka and Pamlhaffmpouh3s []4] and were used primarily because of their large encapsulated volume for entrapping glucose oxidase. DMPC, DPPC or DSPC (9 mg), depending upon the fiposome suspension being prepared, was added to PI (1 rag) and I00 /zl of [3H]DPPC (2 p.Ci/ml) in 3 ml of distilled chloroform: methanol mixture (4:1). The purity of the lipids was checked by thin-layer chromatography on
159 silica gel plates and the solvents were distilled in order to remove any traces of peroxides. The mixture was rotary evaporated at 6 0 " C to yield a thin lipid film. This was then re-dispersed in 6 ml of 4 : 1 chloroform:methanol and 3 mi of nitrogen saturated 1/10 phosphate-buffered saline (PBS, t =0.15 tool/l) containing 10 mg of glucose oxidas¢. The mixture was vortexed, followed by 4 rain sonication in a bath-tEpe sonicator (Decon l ~ 100, Decon Ultrasonics, Sussex, U.K.) under nitrogen. The temperature of the bath was maintained at 30, 50 and 6 0 " C for liposomes composed of D M I ~ , DPPC and DSPC, respectively, (i.e. above the phase transition temperature of the respective lipid). The resulting homogeneous emulsion was rotary evaporated at a temperature above the phase transition temperature of the lipids. After a viscous gel-phase, gel inversion occurred with the formation of an aqueous suspension. This was purged with nitrogen fox a further 15 rain at a temperature above the phase transition temperature to remove traces of organic solvent and to anneal defects in the liposomes. chroma~ To separate the encapsulated glucose o~dase from the non-encapsulated enzyme the preparation was passed through a Sepharose 4B column (30 × 1.5 can) and oluted using PBS (pH 7.3) at a flow rate of 0.2 m l / m i n and fractions of 2 cm 3 were collected using an LKB Redirac F r a ~ o n Collector (Bronnna, Sweden). Prior to the application of the liposome suspension, the column was pre-treated with a lipnsome dispersion to prevent adsorption of liposomes onto the gel matrix [151. ges/c/e ~ e d/str/bution ana/ys/s by photon condat/on spectroscopy and ves/c/e/ame//ar~y The size ~f the vesicles was determined by a dynamic light scattering technique using a Malvern Autosizer, model number (RR146) Malvern Instruments (Malvern, U.K-). This determined the hydrodynamic diameter on the basis of fluctuations in scattered light intensity with time [16]. The autosizer gives the equivalent normal weight dism'bution of particle diameters from which the weight-average diameter (dw) of the lipusmnes is obtained. Checks of the accuracy of the d,~ values were made by using monedisperse latex beads with nominal diameters of 50, 150 and 244 nm. Measurements were made on liposomes dispetsecl in PBS (pH 7.3). Vesicles prepared by reverse-phase evaporation have been shown to be largely unilamellar [14]. In the case of the DPPC-PI REV the vesicles were investigated by electron microscopy using negative staining with phosphotungstic acid (2% w / v in water) and ammonium molybdate (2% w / v in water) and also by freeze fracture techniques. The micrographs gave no evidence to
suggest that the vesicles were other than largely unilamellar.
Assay procedures Fractions (2 ml) from gel chromatography were assayed for protein according to the procedure described by Lowry et al. [17] using a glucose oxidase standard ( 0 - 2 0 0 / t g / 1 0 0 p.I). In order to disperse the liposomal glucose oxidase, 100/tl SDS 1% (w/v) was added to each liposomal fraction and protein standards. SDS at this concentration has been shown not to denature glucose oxidase in neutral media [18]. The fractions were assayed for lipid from the [3H]DPPC counts obtained by scintillation counting (Beckman LS 9800 U.S.A.). These parameters were then used to express the encapsulated glucose oxidase concentration in/~g glucose oxidase//t tool lipid. Immobilization of the h'posomes and membrane characterisation In order for the liposomes to be used in conjunction with a hydrogen peroxide sensing electrode, the liposomes were immobilised by filtering a fixed volume (1 ml) of the appropriate liposome solution under vacuum onto a Millipore nitrocellulose membrane (1 cm2), 0.45 /zm pore size (Millipore, Watford, U.K.) followed by washing of the membrane with PBS buffer (pH 7.3). The radioactivity of the Millipore membranes was determined by scintillation counting by adding 4 ml of scintillation fluid to the membranes and the counts were then used to assess the liposome density on each membrane. By relating the DPM for the membrane to the D P M / m o l of lipid, the number of tool of lipid on the membrane could be evaluated and since the numbor of tool of protein per tool of lipid was known, the number of tool of bound protein could also be ascertained. This was undertaken after the glucose responses of the appropriate membrane had been tested. Procedure for glucose determination at various temperatares An isotonic PBS buffer was used for aqueous sampies. The surface of the hydrogen peroxide sensor was moistened with buffer to allow electrical contact between working and pseudoreference electrodes. The appropriate Millipore immobilised liposomal membrane was then positioned over the electrode and was held in position by the screw-fit top of the electrod~ body. Aliquot,, of a stock solution of 1 M o-glucc~se were added to an appropriate volume of buffer in the sampie chamber and the medium stirred to homogeneity. The steady state currents were then recorded. This was repeated for a range of liposome membranes; D M P C / P I , D P P C / P I and D S P C / P I (all of weight ratio 9:1). In addition the glucose responses were
160 tested over a range of temperatures using a thermostatted reaction cell. No hysteresis effects of the electrode response were observed on heating and cooling. The stability of the electrode responses were tested for D P P C / P ! electrodes. The electrode currents were found to be stable over a 2 h period with respect to a glucose standard. This long-term stability imp~es firstly, that the continuous production of hydrogen peroxide does not have any detrimental effects on the iiposomal electrodes, such as the poss~le oxidation and degradation of unsaturated acyl chains which might be present in the PI and secondly, that the integrity of the liposomes in the electrode is maintained, since loss of integrity with release of encapsulated glucose oxidase would have resulted in a significant increase in signal (if the electrode cell was not washed out) as occurred on addition of detergent (see below). Furthermore the electrodes showed no hysteresis effects when the glucose concentration was reduced which was also indicative of stability.
to the conditions prevailing when measuring the temperature variation of the electrode responses. (c) The effect of the presence of the organic solvents (chloroform and methanol), son~ation and heating on the activity of glucose oxidase during the reverse-phase evaporation procedure was assessed by subjecting the enzyme to the preparation procedure, but ha the absence of lipid. This d e m o ~ r a t e d that the free enzyme lost approx. 50% of its activity during the process. The encapsulation in liposomes would however offer a degree of protection to the enzyme so that the experiments do not necessarily imply that the encaps~lat:':.i' enzyme would have lost 50% of its activity. (d) The integrity of the i i ~ m e s o n imm~ailizatinn on the filters was assessed using a histocbe~rdcal stain (pbenazine methosulphate plus nitroblue tetmzolinm). No colonr was found to develop on addition of filters with adsorbed iiposomes encapsulating glucose oxidase to the dye, but addition of Triton X-100 to disrupt the liposomes released protein turning the dye blue.
Electrode fouling Control systems The effect of temperature on the intrinsic e ~ kinetics as well as electrochemical kinetics was evaluated: (a) by immobllising glucose oxidase (0.35 m g / m l ) on a Millipore filter by vacuum fdtration. The membrane was applied over the electrode and the glucose responses were evaluated over a range of temperatures in a similar manner as described for the liposomal membranes using comparable enzyme concentrations to that of the lilxr~omal systems. In these experiments the electrodes were maintained at the required temperature for 10-20 rain during which the electrode responses were measured. (b) The glucose oxidase activity in solution was assayed by a coupled enz3qmatic assay employing the chromogen o-dianisidine [19]. o-Dianisidine (1% w / v ) was diluted by dissolving 100 ttl in 12 ml of phusl~hate-buffer (pH 6.0) and 2.5 m! of this solution was added to 0.3 ml of glucose solution (18% w / v ) previously equilibrated overnight. Aliquots (3 nil) of the glucose .~iution were placed in the spectrophotometer cuvette and equih'brated at the required temperature for 20-30 rain. Finally 10 /zl of 0.4 m g / m l glucose oxidase/horseradish peroxidase solution was added, the solution stirred, and the absorbance vs. time profile was monitored over a range of constant temperatures at 460 nm using a CaD" 210 recording spectrophotometer in the double-beam mode. The control consisted of dye solution with glucose and horseradish peroxidase but with the glucose oxidase omitted. The glucose oxidase was only maintained at the measured temperature for a short period (5-10 rain) after its addition to the chromogen so that measurements relate
Should the liposomal electrode be used for the measurement of glucose concentration in clin~al samples the effects of serum components on the response becomes important. There are three sites for potential fouling; the platinum electrode, the immobilized lipoand the membrane containing the iiposomes. A cellulose acetate filter can be used to protect the platinum electrode although it reduces the electrode response, alternatively both the filter holding the immobilized liposomes and the platinum electrode could be protected from serum components by use of a small pore s ~ e polycarbonate filter [3]. Addition of serum to electrodes protected only with a cellulose filter between the platinum and immobilized l i ~ results in a signal reduction at high glucose concentration (25 raM) by approx. 40% suggesting that the immobilized liposomes are being fouled by serum components. Results The formation of RF.V liposomes in the presence of glucose oxidase resulted in the encapsulation of the enzyme in the liposomes. This was shown by the separation of :the liposomally encapsulated gluco~ oxidase from the excess free enzyme by gel filtration through a Se~haruse 4B column (Fig. 1). The liposomes elnted in the void volume of the column; a small amount of lipid eluted at (Vo + V~) which may have formed as a result of liposome disruption. The amount of encapsulated enz~ne depends on the lioosome concentration and hence on the lipid concentration. In these preparations the percentage encapsulation was a few percent of the total glucose oxidase used which is sufficient for the preparation of a functional electrode.
161
DMPC/PI
DMPC/PI 14
~eoo¢"Lipidl~ "1 L'
OD
s
0.4
n
i
0,3
1
O.= '
16
20
fraction number
25 (2nd)
30
O
' 5
~
r
r
lo 15 glucose concentration mM --G--19 c
--~---24 c
, 2o
--~- 28 c
DPPC/PI
DPPC/PI 2OOO[
O.5
(b)
off o
~
12 lO
l
(b)
0.2
o
o
6
~0 1~ 20 25 fraction number (2mlj
lo 16 20 215 g l u c o s e c o n c e n t r a t i o n mM
6
"-6-" 29 C
3o
"e- 35 C
ao
"~" 38 C
DSPC/PI DSPC/PI 40001
0.5
ast (c)
~
~
0.2 0
10
0.1 --D- 21C
ol fraction number (2roll Fig. |. Separation o i iipusom¢~ co,tRaining e,JIrappcd glucose oxidant from free glucose oxidase on a Sepharose 4B column: (a) D M P C / P ! IJposomes; (b) D P P C / P i iip~o~es; and (c) D S P C / P | liposomes. [], lipid and O, prmein. The fract~o~Lswere collected at a flow rate of 0.2 ml /min.
20 30 4o glucose concentration mM - ~ - 34 C
"49"-40 C
"m- 55 C
"-+-"64 C
Fig. 2. Temperature effect on the liposomal electrode response as a function of glucose concentration for liposomal enzyme electrodes incorporating glucose oxidas¢: (a) DMPC/PI liposomes; (b) DPPC/P| liposomes; and (c) DSPC/PI liposomcs (9: ! weight ratio). ]For clarity some data has been omitted from (a) and (b). The liposomes (1 ml)were immobilised by adsorption on a Millipore filter (1 cm 2) and then interfaced with a Clark Ox','gen electrode polarhed for hydrogen peroxide detection ( + 650 mV).
162 TABLE |
glucose oxidase to a Millipore filter was tested over a range of temperatures (Fig. 4). The glucose oxidase concentration used to produce the electrode was coln-
Liposome and membrane cfiaracterisad.m
Liposome mg glucose composition ox/dase/ p:mol lipid
Weightaverage diameterand standard deviation of the weight d/stn-out/ond~ (am_+standard d~iation )
gg glucose ox/dase/m2
DMPC/P! DPPC/P! DSPC/PI
150± 60 134± 79 199± 114
60 42 30
0.33 0.19 0.03
,~
5.o
~ 4~ Characterisation data for these iiposomal preparations and the resulting membranes are shown in Table I. Fig. 2 shows the electrode responses as a function of glucose concentration and temperature for each iiposoreal electrode ~'stem. from which it is seen that the range o f iinearity of response verses glucose concentration decreases in the series DSPC > DPPC > DMPC. Fig. 3 shows the electrode response curves at a constant glucose concentration (5 mM) as a function of temperature. These e,,~eriments demonstrate that the mesomorphic state of me liposomal bilayer has a significant effect on the dectrode response. The electrode currents showed an iacrease as the gel to liquid-crystalline phase transition temperature (T~) of the phospholipid is approached (the values of T~ for pure DMPC, DPPC and DSPC bilayers are 24, 41 and 55 o C, respectively, [20]). Addidon of Pl is known to decrease 7~ for DPPC bilayers by several degrees [21] and would be expected to have an analogous effect on DMPC and DSPC bilayers. For DPPC the experiments were carfled out through the phase transRion and showed that the electrode response passed through a maximum. The permeabilRies of DPPC liposomal dispersions to D-glucose are known to pass through maxima in the region of T~ [22,23]. In this regard it might have been expected that the electrode response for the DMPC electrodes would decrease at higher temperatures ( > 25-30 ° C) however this appeared not to be so, possibly because of the intrinsic greater fluidity of the shorter chain lipid bilayer. In the case of DSPC it was not practicable to extend measurements to higher temperatures. As a control, the response to 5 mM glucose of an electrode made by adsorption of free (unencapsulated)
Fig. 3. The response ~o 5 mM glucose over .z range of temperatures for liposomal electrons of various compositions incorporating gincose oxidase: (a) DMPC/PI GIx~omes; (b) DPPC/Pi fiposomes; and
(c) DSPC/P] lil~somes. The liI~somes were immobi|isedas descn~ed in Fig. 2. The results are means_+S.E.(n =3). T¢ indicates the gel to liquid-cD~tallinetransition temperatureof the pure pho,~ pholipid.
~ 4.0 3..5 3
2o
iS
25
30
temperatureeC
3.z
(b)
3.0
< 2.8 : ~= 2-s o ~ 2.4
2 2°
25
30
35
40
45
temperatureOC 5
4.5
