636
Anal.
chem.1993. 65, 638-639
Continuous-Flow Sensor Strategy Comprising a Rotating Bioreactor and a Stationary Ring Amperometric Detector Kiyoshi Matsumoto,' J. J. Baeza Baeza,! and Horacio A. Mottola' Department of Chemistry, Oklahoma State Uniuersity, Stillwater, Oklahoma 74078-0447
A rotatlng kmnobillr.6.nzyme reactor and a static PI-rlng electrode heve been d d g n e d In a waled cell for uw under contlnu~ow/stopped-flowopwatlon. The operatlng characterlslksolthelntegratedbloreaclor/detectorunH havebeen studled for determlnatlon of glucose usingImmoMllzedglucose oxidase [EC 1.1.3.41 and amperornetrlc rnonkorlng of the H,02 produced. Determinations are based on the rate of response under stopped-flow conditions. The trend In values of the apparent Mlchaells-Menten constant conllrms the valldlty o( the approach to sensitive determinations utlllzlng very small amounts 01 blocatalysts. v
Thecombinationofanimmobilized biocatalyst (bioreaetor) and an electrode collector system bas proved useful in sensor design.'-3 These contributions to biosensing do not incorporate, however, a moving reactor with continuous-flow sample processing. The approach presented here shows analogies with the rotating disk electrode and the rotating ring-disk electrode. The hydrodynamic conditions at these electrodes are clearly defined and rigorously described by the dependence of the current on angular velocity.' In the assembly described here, a product of an enzymecatalyzedreactionatabioreact~r(rotatedataconstantspeed) is hydrodynamically transported toa stationaryringelectrode where it is electrochemically monitored. The sample is transported to detection by an unsegmented continuous-flow stream. The design circumvents problems associated with systemsemploying shaftrotation, which are difficulttoadapt to continuous-flowsample/reagent(s)processing, but permits taking advantage of the good hydrodynamic environment in the vicinity of the ring electrode. The conditions there result in enhanced sensitivity with only a very small amount of immobilized biocatalyst. The method is illustrated by measuring,ataplatinumelectrode,thelevelofH2O2produced in the glucose oxidase-catalyzed oxidation of glucose hy dissolved oxygen. The enzyme is immobilized on controlledpore glass which is easy to uniformly deposit on one side of adouble-coated stickingtape (Scotchdoublecoatedtape665, 3M, St. Paul, MN)(Figure 1). The tape is then affixed to a disk that can be rotated. Rotation of the "enzyme reactor" minimizesiremoves the problem aeaociated with the slowing down of reactions in low-dimensional spaces. This slowing is characteristic of immobilized reagents confined to onedimensional spaces such as those in pores or other low dimensional fractal surfaces. This is reflected by the fact thattheMichaelis-Mentenconstant for immobilized enzymes is generally somewhat higher (lower initial reaction rates)
'
Permanent address: Department of Food Science and Technology. Faculty of Agriculture, Kyuahu University, 46-09. Fukuoka 812. Japan. I Permanent address: Department of Analytical Chemistry. Faculty of Chemistry, University of Velencia, Burjaesot (Valencial, Spain. (1)Ksmin, R. A,; Wilson, G. Anal. Chem. 1980,52, 1198-1205. (2) Wang. J.; Lin. M. S. A n d . Chim. Acta 1989,218,281-290. (31 Bonskdar. M.; Vilehez, J. L.; Mottola, H.A. J. Electroonal. Chem. Interfoeid Electroehem. 1989,266.47-55. (4) Opeksr. F.: Beran. P. J. Electroanof.Chem. Interfaciol Electro-
ehem. 1976,69, 1-105.
0003-2700/93/0365-0S36$04.0010
1mm Flg~ne1.
photograph of Um rota-
dlsk enzyme reactor Sean fmm
the top and showing the cornpact packing 01 lhe controlleaporeglass chips covering one slde of lhe doublecoated tape.
thanfor tbesameenzymeinsolution,asaresultofdiffusional limitations. Minimization of such limitations can be accomplished by decreasing the particle size of theinert supportor, as done here, by efficient stirring. It should be noted that in packed reactors the decrease in particle size may result in impedance to free flow.
EXPERIMENTAL SECTION Reagents and Solutions. Allchemicaleused.excepta8 noted, were of analytical reagent grade. The water used for solution preparation was deiunized and further purified by distillation in an all-borosilicate-glassstill with aqumz immersion heater. The O-l)-glucosewas from Sigma Chemical Co. (St. Louis, MOL Glutaraldehyde (25'0 w w aqueous solution) waq obtained from Aldrich Chemical Co. (Milwaukee. WII. The carrier elertrolyte solution was a 0.10 M phosphate buffer of pH 7.00. Reactor Electrode Flow-Throughcell. Figure2 illustrates the design of the flowthrough chamlrer containing the rotating enzyme reactor and the ring electrode. The reactor is a disk of Teflon in whichaminiaturemagneticstirring bar (Teflon-coated Micro Stir har from Markson Science, Inc., Phoenix, AZ, has beenembedded. Aneedle i*attached tothecenterofthe bottom part oft he Teflon disk. The point of this needle is made to rest on aspecially machined inverted conical indentation t hat permits smooth rotation of the disk when it is driven by a common laboratory magnetic stirrer (a Mag-Mix unit from Precision Scientific, Chicago. IL,in our case,. Typically. a reactor disk carried 1.4 mg of controlled-pore glass on its surface. The main hody of the cell was made of Plexiglass, and the platinum ring from the wire rim reinforcement of a discarded platinum gauze electrode. As shown in Figure 2. the Pt-ring electrdewasplaced around the rotsting reactor disk in acentrally symmetrical position. The volume of the reactor electrode cell could be varied by insertion of a Teflon spacer between the top and bottom parts of the cell. The minimum cell volume (no spacer used) was 450rrL and the maximum (with a 3-mm-thick sparer) was 1.8 mL. The volume comprising the area just ahove the reactor was calculated tu be 113 PI,. Estimation of the Velocity of Rotation of the Reactor. The rotation of the reactor. as explained above. was done with the aid ofa laboratory magneticstirrer and the rotation velocity controlled by means of a variable transformer with an output 0 1993 A n w k a n Chemical Society
ANALYTICAL CHEMISTRY, VOL. 65, NO. 5, MARCH 1, 1993
Flow in
=v
t-54
T
Flow out
I
,
kRd
I
A
WE
8
t 1
4
3
U
C
4 P t - r i n e electrode
L
Figure 2. Schematlc representation of the components of the flowthrough cell containing the rotating reactor and the platlnum rlng electrode. A: Upper cell body defining the flow-through reactlonl detectlon chamber. B: Rotating reactor showlng posltlonlng needle polnt. C: Lower cell body showlng the well for the rotatlng reactor and the concentric Pt-ring electrode. D: Top view of the lower cell body. All measurements In millimeters. The lower cell body Is screwed to the upper cell body for cell assembling.
Table I. Rslationship between Rotation Velocity of the Reactor and the Voltage Setting in the Variable TransformerP voltane settine rDm % relative standard deviation 60 258 & 12 4.7 70 452 f 25 5.5 80 664 f 21 3.2 90 874 f 26 3.0 a
637
Figure 3. Block dlagram of continuous-flow system and detec#on arrangement. A and B represent the two Intakesof buffer solutlonand C the Intake of glucose-contalnlngsample. P&V representsthe pump and valve unit (FIAtron SHS-200). W = waste outlets. WE = platlnum worklng electrode and reactor cell. D = potentlostatlampwometrk detector (Model LC4B from Bloanalytlcal Systems, West Lafayette, IN): filtering settlngs wlth frequency cutoffs of 0.1 and 1.0 Hz dld not affect the signal response. RE = reference electrode (AglAgCI, 3 M NaCI, from Bioanalytical Systems). AE = stainless steel tube used as auxiliary electrode. RD = strlp-chart recorder, SuperScrlbe Model 4910 from Houston Instruments, Austin, TX.
Table 11. Explanation of Typical Time Intervals and Owratina Conditions of the System time interval,s 120 5 60 120 15
operating conditions carrier passed through the system until sample introduction sample loading carrier transports sample to optimum position in reactor/detectorcell stopped flow carrier pushes sample out of cell cycle repeated for introduction of next sample
Pump on on on off
on
Measurements based on five replicates at each voltage setting.
between 0 and 140 V and 7.5 maximum amperage (The Superior Electric Co., Bristol, CT). To determine the relationship between the variable transformer setting and the revolutions per minute (rpm)of the reactor itself, the rotation at different voltage setting was video-filmed with a Memorex 127 camera (Tandy Corp., Fort Worth, TX) and the time concurrently recorded with a laboratorystopwatch/countdown timer (“The Omniwatch”from Cole-Parmer, Chicago, IL). The filmed video record was then played back at slow motion, and the number of revolutions in a fixed time interval deduced from the appearance of a black ink dot positioned on one side of the top of the reactor. Table I summarizes the results obtained. Each voltage setting was fixed after allowing the reactor to rotate freely for a few seconds at an applied voltage of 110 V. This operation was repeated before each replicate was obtained, and the uncertainty given by the standard deviation is considered to represent mainly the reproducibilityin setting the voltage by this procedure. Measurements performed with and without solution flow through the reactor gave similar rpm values. Continuous-Flow System. A FIAtron SHS-200 microprocessor-controlled solution-handling system (FIAtron Systems, Milwaukee,WI) was used for pumping,sample introduction, and stopping of the flow. Figure 3 illustratesschematically the overall configurationof the continuous-flowsetup utilized. Tubing inside the FIAtron unit, Teflon, 0.50-mm i.d.; pump tubing, Fisher AccuRated 1.0-mmi.d. (Fisher Scientific Co., Pittsburgh, PA);tubing in the rest of the setup, 1.0-mm i.d., Teflon. Table I1 gives the time intervals typically used in the continuous-flow/stoppedflow operation programmed via the FIAtron SHS-200. Procedure for Enzyme Immobilization. The enzyme reactor was prepared by immobilizing glucose oxidase [EC
1.1.3.41, from Aspergillus niger (TypeVII, fromSigmaChemical) containing 1.25 x 105 I.U. per gram of solid, on aminopropylmodified controlled-pore glass (APCPG) with a mean pore diameter of 1273 A, a surface area of 24 m2/g,and 48.2 rmol/g of amino groups (Electro-Nucleonics, Fairfield, NJ). The APCPG, smoothly spread on one side of the double-coated tape, was allowed to react with an aqueous solution of 5 % w/w glutaraldehyde at pH 10.00 (0.20 M carbonate) for 2 h at room temperature. After being washed with purified water and 0.10 M phosphate buffer of pH 7.00, the enzyme (5 mg of enzyme preparation in 0.50 mL of pH 7.00 phosphate buffer) was coupled to the residual aldehyde groups in phosphate buffer (0.10M, pH 7.00) overnight at 5 “C. Unreacted aldehydegroups were capped by reaction with an aqueous solution of glycine (15 mg/mL, 2 h at room temperature) after washing with phosphate buffer. The immobilized enzyme preparation was finally washed with phosphate buffer (pH 7.00) and stored at 5 “C between uses. This procedure was the most effective of four different approaches to immobilization as illustrated in Figure 4. The preparations were perfectly stable for at least 1 month of daily use. All pH measurements were made with an Orion Research (Cambridge, MA) Model 601A digital pH meter equipped with an epoxy-body combination electrode (Sensorex, Westminster, CA). The potential applied to the Pt-ring electrode for HzOz measurement was +0.60 V vs a Ag/AgCl, 3 M NaCl reference.
RESULTS AND DISCUSSION A perusal of the literature, selectively covered in some review papers596 and in a monograph? reveals that the ( 5 ) Mottola, H. A. Quim. Anal. (Salananca) 1989,8, 119-128.
698
ANALYTICAL CHEMISTRY, VOL. 85, NO. 5, MARCH 1, 1993
Normalized Activity
08
08
/ 02 0 0
4
7 Ini
B * 1
2
3
4
Ini
StOQ
5
Flguro 4. Comparison of Immobilizetion strategies tested. “Activity” of each approach was normaked to the “activity” of the same welght of lmmoblllzed enzyme preparation suspended In an equal volume of subatrate solution used for testlng each of the preparations. (1) Reference: suspended lmmoblllzed enzyme (enzyme preparation Is not flxed to one si de^ of the doublecoeted tape), (2) lmmobllization on the APCW before attaching to the tape, (3) filling a well In the center of the dlsk with lmmoblllzed enzyme-CW powder and then covering with a cellulose nitrate membrane (pore size 0.45 p), (4) lmmobllizetlon on APCW after attaching to the tape (procedure described In the Expewlmental Section), and (5) fUng the APCW on the tape and then the enzyme and bovine serum albumin (1 % ) colmmoblllzedas to make a protelnlc membrane.
hydrodynamic characteristics of an integrated rotating bioreador/detwtor system have apparently received little attention in the development of biosensors in continuous-flow situations. This paper is aimed a t such development and the establishing of the characteristics of an assembly as to its performance in a continuous-stopped-flowprogrammed operation. The design was inspired by the rotating disk-ring electrode, but in order to avoid the use of a rotating drive shaft, which is not compatiblewith operation in a sealed cell configuration, two radical modifications were introduced. These modifications comprised: (1)the ring electrode is stationary and only the reactor part rotates and (2) the reactor/electrode is located a t the bottom of the cell. This configuration retains the basic hydrodynamic conditions, during detection, that prevail in the vicinity of the ring electrode in a rotating ring/ diskelectrode. Mass transport at the surface is controlled by convection of the active species caused by rotation. If &rb is defined as a diffusion coefficient characterizing mass transfer in a turbulent moving fluid and differing from D (the molecular diffusion coefficient assumed to be independent of concentration), three characteristic regions can be distinguishedin the vicinity of the solid phase: (1)a turbulent diffusion layer in which transfer by turbulent pulsations predominates (&b >> D), (2) a viscous diffusion sublayer in whichDt,b varies with the fourth power of the distances from the surface up to a region in which &b is approximately equal to D,and (3)a laminar diffusion sublayer in which Dtwb is unimportant and mass transfer is determined by molecular diffusion. Consider that the following generalized chemical transformation occurs on the surface of the reactor: E .
R-P in which R is the analyte, E the immobilized enzyme, and P the electroactiveproduct to be detected a t the ring electrode. (6) Gorton, L.; Csoregi, E.; Emneus, J.; Jonsson-Petersson, G.; MarkoVarga, G.; Persson, B. Anal. Chim. Acta 1991,250,203-248. (7)Stulik, K.;Pacakova, V. Electroanulytical Measurements in Flowing Liquids; Ellis Horwood: Chichester, UK, 1987;Chapter 3.
Ini
D-T---Flow
Ini
t
Ini
Flgure 5. Effect of reactor rotation under contlnuous- and stoppedflow conditions. A: Stopped flow with rotation. B: Stopped flow without rotatlon. C: Continuous flow wlth rotatlon. D Contlnuous flow without rotation. Qlucose concentration 0.50 mM; flow rate, 0.83 mL/mln; sample size, 69.2 pL; velocity of rotatlon, 874 rpm.
Product P is carried from the reactor into the bulk of the solution under convective transport and a certain number of P particles reach the ring, where
P f ne-+A and the current developed a t the detector should be directly proportional to the concentration of analyte in the bulk of the solution and should increase with increasing rotation velocity. If the flow is stopped when the sample plug transported by continuous flow reaches the center of the reactor,detectiontakes place under conditionss i m i i to those of batch detection.192 Effect of Reactor Rotation. Figure 5 shows the effect of rotation under continuous- and stopped-flow conditions. Responses under continuous flow are relatively small but comparativelylarger if the reactor is rotated (comparetraces C and D in Figure 5). Under stopped-flow conditions there is practically no significant response without rotation of the reactor, but a significant signal that increases almost linearly with time develops when the reador is rotated. This behavior confirms the hydrodynamic convective characteristics discussed above. In view of the responses illustrated in Figure 5, the rest of the results here will be based on the initial rate of response of the signal under stopped-flow conditions. Effect of Sample Size. The rate of response increased almost linearly with sample size up to 250 pL in a cell with a volume of 450 pL. For convenience a sample size of 69.2 pL was used to evaluate the other parameters. Sensitivity is almost tripled, however, with samples of about 200 pL. Effect of Geometric Cell Volume. The effect of cell volume on the rate of response was investigated by changing the spacer thickness from 1up to 3 111111. The rate of response decreased with cell volume, as should be expected because of the dilution effectfavored by rotation and the current response to bulk concentration. Consequently,the smallest cell volume of 460 p L (no spacer in use) was selected for further measurements. Effect of Reactor Rotation Velocity on the Rate of Response. According to the hydrodynamics of a rotating disk immersed in a solution,the amount of substrate reaching the surface of the disk per unit of area and unit of time should be proportional to the square root of the rotation velocity.* In the range 170 to 900 rpm, the initial rate of response did (8) Levich, V. G. Physicochemical Hydrodynamics; Prentice-Halk Englewood Cliffs, NJ, 1962.
ANALYTICAL CHEMISTRY, VOL. 65, NO. 5, MARCH 1, 1993
Table 111. Values of K,’ (Apparent Michaelis-Menten Constant) Determined As Discussed in the Text (TemDerature 20 1 “C) rotation velocity, rpm
K,’,”mM
420 840 1010 free enzyme in solutionb
9.79 6.59 6.83 1.97 1.77 1.38 15.0
120 240
linear regression standard deviation f2.54 f1.29 f1.54 f0.19 f0.15 *0.11 *0.4
0 Eachvalue of K,’based on triplicates of eight different substrate concentrations. Estimated using spectrophotometric monitoring (436 nm) of o-dianisidine oxidation.
*
indeed vary linearly with the square root of the rotation velocity. The Pearson’s correlation coefficient of the straight line obtained in plots of the rate of response vs the square root of the rotation velocity was 0.993, and slightly higher than the same coefficient (0.986) for plots of the rate of response vs the rotation velocity itself. The Apparent Michaelis-Menten Constant. As noted in the introduction to this paper, rotation is expected to decreasethe value of the apparent Michaelis-Menten constant with a concomitant increase in initial reaction rates. This makes possible sensitive determinations with relatively very small amounts of catalyst. To verify this, the apparent Michaelis-Menten constant, K,’, was determined by stopping the flow for 2 min, in the center of the reactor, and thus holding samples of 69.2 pL of glucose solution there and monitoring the increase in signal amperometrically with the stationary ring electrode. The final concentration of glucose in the reactor was estimated by taking into consideration that the cell volume equaled 450 pL. The values of K,’ were obtained a t five different velocities of rotation, and the results are illustrated in Table 111. The calculation of K,’ was performed by fitting data to the linear relationship: (US) = (m/[glucosel) + n
(1)
690
the strategy presented here for utilization of rotating enzyme reactors in continuous-flow systems. Effect of pH and Glucose Concentration. The pH profile of activity of immobilized glucose oxidase is usually broad, with a maximum occurring around pH 6. The results of the experiments performed during this work reflect a different pH profile. The rate of response dramatically increased (almost doubled) from pH 6.5 to pH 7.00 and continued a moderate increase up to pH 8.00, the highest value tested. This behavior may be the result of a combination of factors. First, immobilization may shift the optimum pH depending upon the nature of the carrier, chemical modification of the enzyme, the type of enzymatic reaction, and the buffer capacity in the vicinity of the enzyme active site.’O The oxidation potential of HzOzis also pH dependent and shifts toward more anodic potentials with a decrease in pH. The rapid decrease in enzyme activity at pH 6.5 may be due in part to moving away from the region of optimum potential for HzOz detection. A linear relation was observed between the rate of response and the glucose concentration in the range of 0.0010 to 1.00 mM a t 847 rpm. The equation of the regression line obtained a t 847 rpm was rate of response (nA/min) = 0.679 + 660.0(C,,,,,,,
mM)
The correlation coefficient for this type of plot was typically equalto 0.998, the detection limit (estimated from 3 standard deviations of the blank) was 2.3 pM, and the coefficient of variation for 10 successive measurements of 0.50 mM glucose was 1.59%. Linearity, however, depends on the rotation speed; the lower the rotation speed, the wider the dynamic range. If we consider the minimum measurement time (about 30 s), the transport time, and the wash-out time (about 30-60 s), the minimum time between determinations can be estimated roughly as 2 min. This corresponds to a satisfactory sample rate of about 30 samples per hour.
ACKNOWLEDGMENT
All signal values were corrected by subtracting background readings and the plot of 1/S vs l/[glucosel represents a graphical approach similar to the Lineweaver-Burk plot. As expected, the values in Table I11 confirm the trend to larger K,’values as rotation rate decreases and justify further
The authors acknowledge the loan of the SHS-200 samplehandling unit from FIAtron Systems (Milwaukee, WI) and partial support from the Oklahoma State University Water Research Center. They thank Mike Lucas (Oklahoma State University machine shop) for the construction of the reactor/ electrode cell. One of the authors (J.J.B.B.) gratefully acknowledgesthe Conselleriade Cultura, Educacion y Ciencia de la Generalitat Valenciana, for a FPI grant. The research described in this paper was presented at the 4th Symposium on Kinetics in Analytical Chemistry, September 28, 1992, Erlangen, Germany, and one of the authors (H.A.M.) acknowledges travel support from the Halliburton Foundation Inc. (Dallas, TX).
(9)de Levie, R. J. Chem. Educ. 1986,63, 10-15. (10)(a) Goldstein, L.; Levin, Y.; Katchalski, E. Biochemistry 1964,3, 1913-1919. (b) Goldstein, L.;Katchalski, E. Fresenius’ 2.A n d . Chem. 1968,243, 375-396.
RECEIVED for review June 12, 1992. Accepted November 17, 1992.
where S = rate of response; K,’ = m/n. Because relationship 1is not a direct one between the concentration of substrate and the reciprocal of response signal, it was necessary to use weighting to correct for this.9 In this case the weight used Was
wi = (SJ4
