Flow injection analysis as a diagnostic technique for development and testing of chemical sensors

  • Published on

  • View

  • Download


Analytica ChimicaActa, 204 (1988) 7-287Elsevier Science Publishers B .V., Amsterdam - Printed in The NetherlandsFLOW INJECTION ANALYSIS AS A DIAGNOSTIC TECHNIQUEFOR DEVELOPMENT AND TESTING OF CHEMICAL SENSORST.D. YERIAN, G .D. CHRISTIAN and J . RUZICKA*Center for Process Analytical Chemistry, Department of Chemistry, BG-10, University ofWashington, Seattle, WA 98195 (U.S.A )(Received 16 July 1987)SUMMARYAn immobilized urease sensor is developed for continuous, on-line analysis . The sensor consistsof the enzyme urease, cross-linked with bovine serum albumin into a cellulose pad, with an acid-base indicator dye covalently bound to the surface of the cellulose . The sensor is placed within aflow-injection optosensing system to monitor the changes in pH, and subjected to a thoroughevaluation, using the flow-injection technique; sensor stability (both dye and enzyme stability),speed of sensor response, sensor sensitivity, sensor-to-sensor reproducibility, response to a typicalinterferent, and sensor lifetime data are obtained . Sensor poisoning upon exposure to low levelsof mercury, and subsequent regeneration of the immobilized enzyme pad, is investigated for useas an on-line mercury sensor . The urea sensor is also evaluated for use as a continuous monitorfor urea in kidney dialysate . Enzyme Michaelis-Menten constants are determined for the immo-bilized urease, under given assay conditions, using a stopped-flow flow-injection technique .The recent literature on analytical chemistry abounds with reports on thedevelopment of electrochemical, fiber optic, piezoelectric and other sensors .Novel detection principles are frequently announced, and improved sensorquality in terms of selectivity, sensitivity, and lifetime are the goals of manyresearch efforts . The need for continuous monitoring of critical parameters inclinical chemistry, biotechnology, pharmaceutical, chemical and nuclear in-dustries, chemical warfare or environmental control is the driving force behindsensor research.While most such research has novel aspects, has sound theoretical basis, anduses high technology to fabricate the sensing devices, many of these chemicalsensors possess an Achilles' heel, a lack of time stability, reflected in deterio-ration or irreproducibility of their response . Surprisingly, although review au-thors point out the necessity for the use of common criteria for sensor evaluation[ 1 ], many research papers lack vital data on sensor performance, e .g., speedof response, reproducibility of response, sensitivity and selectivity of the sen-sor, as these parameters necessarily undergo changes during the lifetime ofeach device . Information on poisoning or recovery of sensors, as well as typicallifetimes under various operating conditions, is very scarce .0003-2670/88/$03 .50 1988 Elsevier Science Publishers B .V.8An automated, reproducible, and rigorous testing procedure would allowcritical testing of chemical sensors, and should become instrumental in theirfuture development . This work summarizes the characterization of an immo-bilized urease-based sensor, designed for continuous, on-line analysis of urea,in which such automated testing is applied .PRINCIPLESImpulse-response flow injection analysis (FIA) is based on the repetitiveaction of a well-defined zone of a selected chemical species (E) on a target(S), situated in an unsegmented carrier stream (Fig . 1) . The target may be asensor; the carrier stream is usually inert, i.e ., a solution to which the sensordoes not respond, or to which the sensor responds at the detection limit, thusestablishing a continuous baseline. The injected species (E) is the one to whichthe sensor is to respond, either directly, or indirectly through a series of chem-ical reactions . By continuously monitoring the sensor output, while repeatedlyinjecting a well-defined zone of the species at a fixed concentration, a series ofpeaks will be generated, which will reflect the time response of the sensor ; theresponse may be unchanged, or show deterioration within the selected timeperiod (Fig. 1) . The procedure should be automated, hence the response willnot be a function of subjective judgement or manual operational errors [ 21 .From the rate of change of the response under the well-defined impulse mea-surement, it should be possible to predict the usable lifetime of the sensor.time -0Fig. 1 . Impulse-response flow-injection manifold, consisting of two pumps (P), a timer (T), asensor (S), and a detector (D) . Carrier stream (C) is pumped through the system, carrying theinjected species (E) through the detector (D) and to waste (W) . The readout signal (I) is mon-itored continuously .Fig. 2 . Manifold for pH measurement, consisting of two peristaltic pumps (P1, P2), a timer (T),and the integrated microconduit shown in Fig . 4 (boxed area) . The flow cell communicates withthe light source and detector of the spectrophotometer by means of optical fibers .9When the sensor is subjected to a series of solutions, such as different con-centrations of the species (E), or injected zones of potential interferents, sen-sor poisons, or sensor regenerative solutions, more information is obtainedabout the sensor. In addition to examining the tendencies of a series of re-sponse peaks, changes in individual peak shapes may give information aboutphenomena occurring at the sensor surface . Because many chemical sensorsutilize a chain of chemical reactions in the sensing area, it should often bepossible to examine each set of sensing reactions separately, to differentiatethe components of the sensor response . An example of such systematic sensortesting and evaluation is the work on the fiber-optic urea sensor describedbelow.The sensor is based on the enzymatic degradation of urea, resulting in a pHchange which is sensed optically by means of a colored acid-base indicator . Anintegrated microconduit [3] is designed to contain a solid support with animmobilized acid-base indicator within the flow cell (Fig . 2) . A variation inthe solution pH results in a color change on the sensor surface, which is mon-itored via a fiber-optic bundle . The theory of this approach is well known, anddescribed elsewhere [ 4, 5 ] .For this work, a number of factors must be considered . It is desirable toimmobilize the enzyme directly to the cellulose pad, both to keep from com-plicating the system design with a separate enzyme reactor, and to allow directobservation of the enzyme via the fiber-optic "window" positioned in the flowcell. Also, the "sensing" part of the system is easily accessible for replacement,dispersion is minimal so that very small sample volumes can be introduced,and the sensor has fast response and fast washout characteristics . The prob-lems are: not to destroy the acid-base indicator during the cellulose activation ;to attach sufficient enzyme to the relatively small surface area of the sensingpad; to produce uniform pads ; and finally, to produce pads with an appropriatelifetime during use and during storage .Thus, the lifetime and response characteristics of the urea sensor are depen-dent on the following features : (a) the stability of the optical components(light source, detector) and light losses in the optical fibers, as well as thequality of the fibers, and any physical changes occurring on the surface of thereflecting material in the flow cell; (b) indicator stability in terms of its cova-lent attachment to the reflecting material, and as affected by photobleaching ;(c) enzyme stability in terms of its physical attachment to the reflecting ma-terial, and activity loss caused by thermal and chemical degradation duringuse; (d) the ability of the indicator/enzyme couple to respond to analyte overa wide range of concentrations and to be regenerated/washed if a maximumlevel of analyte is repeatedly exceeded; (e) the speed of response and reversi-bility of the enzyme-indicator couple as they might change during sensorlifetime .10EXPERIMENTALApparatusAll experiments were done with a Bifok-Tecator FIAstar 5020 flow-injectionanalyzer. The manifold used (Fig . 2) incorporated integrated microconduitsinto the FlAstar system, replacing the original injection valve . For all work butthe dialysate monitoring, the integrated microconduit was composed of a flow-through detector and miniaturized injection valve, with variable sample vol-ume (Fig. 4) . The dialysate manifold (Fig . 3) incorporated an integrated mi-croconduit comprised of impressed channels for flow path, and a flow-throughcell (Fig . 5) which interfaces with bundles of plastic optical fibers, for trans-mission of source illumination to the flow cell and reflected signal from theflow cell to the detector .Reflected absorbances were measured at 605 nm for Merck indicators 9583and 9582 [ 7 ] . The absorbance maximum was found by scanning through thevisible spectra of the two colors of the indicator, and subtracting the blue spec-trum from the yellow (baseline) spectrum, a feature of the Tecator FlAstar5023 detector. The reflectance measurements (Ar ) were registered automati-cally on the FlAstar 5023 spectrophotometer, and digitally displayed andprinted on the FIAstar 5020 . The results were concurrently fed to a recorder(Radiometer, Servograph REC-61, furnished with an REA-112 high-sensitiv-ity interface) .ReagentsAll chemicals used were of analytical-reagent grade; all water was deionized.The carrier solution for the buffer pH determinations was 5 X 10 -4 M hydro-chloric acid. The carrier solution for urea determinations was a dilute TrisSampleCarrierFig. 3 . Manifold for urea measurement in kidney dialysate, consisting of three pumps(P1, P2,P3), a timer (T), and the integrated microconduit shown in Fig. 5 (boxed area) .Fig. 4. Integrated microconduit for measurement of pH, consisting of a teflon injection valve anda flow cell . Connecting channels are impressed into the underside of the PVC block. The sensorpad is placed at the endface of the fiber-optic bundles (inset)- the "flow cell".1 1WFig. 5. Integrated microconduit for urea measurement in kidney dialysate . The sensor (inset) hasurease cross-linked within the pad, as well as the acid-base indicating dye .buffer solution, containing 1 X 10 -3 M Tris [ tris- (hydroxymethyl) - amino-methane] in 0.140 M sodium chloride and 1.0 mM EDTA, adjusted to pH 6.5with 0.1 M HCl, unless otherwise stated.The urease (Sigma U-2125) contained 71 000 micromolar units per gram,unless otherwise stated . Immobilization procedures are outlined below . The12urea standards were prepared by dilution of a urea stock solution with carrier .Stock urea solution (100 mM) was prepared by dissolving 0 .600 g of urea(Sigma) in 100 ml of deionized water .Buffer solutions used were prepared as tabulated in Perrin and Dempsey [ 6 ](citric acid/potassium phosphate, Table 10 .47 ; Tris/HCI, Table 10 .32) .Mercury solutions were prepared by dilution of a stock solution of mer-cury (II) chloride (Alfa) with carrier . The stock mercury (II) (10 mM) solu-tion contained 0 .271 g of mercury (II) chloride dissolved in 100 ml of deionizedwater.The enzyme-regenerating solution was 0 .1 M Tris/10 mM thioacetamide/10mM EDTA adjusted to pH 7.0 with 0.1 M hydrochloric acid .Dialysate solution was prepared according to COBE Centry instructions,which involve the premixing of an acid concentrate solution (263 g 1-1 NaCl,9 .9 g 1 -1 CaC12i 6 .7 g 1 -1 KCI, 5 .4 g 1 -1 Ch3COOH, 3.4 g 1 -1 MgCl, and 90 g 1 -1dextrose) and solid sodium hydrogencarbonate ; the final solution was 25-35mM in hydrogencarbonate, pH 7 .4 .Covalent immobilization procedure . The procedure applied was adapted froma method used to immobilize enzymes to porous glass, described by Adams andCarr [8] . Silanization of the cellulose surface is a modification of the proce-dure described by Weibel et al. [ 9 ] : the cellulose pads were stirred, at reducedpressure, in 20 ml of 10% (v/v) 3-aminopropyltriethoxysilane (Sigma) in tol-uene, at 75 C for 3 h . The cellulose pads were removed from the mixture, washedwith 96% ethanol, then placed in 20 ml of 2 .5% (v/v) glutaraldehyde (Sigma)in pH 7.0 phosphate buffer, and kept at room temperature and reduced pres-sure for 1 .5 h. Finally, the activated pads were soaked in a strong enzyme so-lution (20 mg of urease in 1 .0 ml of pH 7.0 phosphate buffer) for 3 .5 h, atreduced pressure to remove microbubbles from the pad and facilitate diffusionof the enzyme to the reactive aldehyde group of the glutaraldehyde . The padswere kept in 0 .1 M pH 7 .0 phosphate buffer, and stored at 0-5 C .Cross-linked enzyme procedure . The same glutaraldehyde buffer solution wasused to which 30 mg bovine serum albumin (BSA) was added .RESULTS AND DISCUSSIONVarious components and applications of this system were subjected to sys-tematic evaluation by repetitive measurements of injected species . They in-cluded the response and stability of the pH indicator and the immobilizedenzyme, evaluation of different methods of enzyme preparation, the effect ofenzyme inhibitors, and the use of the system in a continuous monitoring mode .IndicatorType of indicator . The solid support, or "active surface" implemented in thesystem is the ColorpHast (non-bleeding) indicator strips . These are cellulosefibers on which the acid-base indicators have been covalently immobilized .0.70A r0.350.00Iclw time ,4 min .1 3Fig. 6. Seventy-five representative injections of pH 6 .8 phosphate buffer into 1 X 10 - `' M HClcarrier. Sensor is Merck indicator 9582 ; color change (yellow to blue) is monitored at 605 nm .Injections: (a) 1-25, (b) 675-700, and (c) 1175-1200 .The acid-base indicators are azo dyes, AR-N=N-AR, where AR is an arylgroup with one or more reactive groups capable of forming a bond with cellu-lose ; examples are -SO 2CH2CH2OSO2OH and - (CH3 ) NSO2CH2CH2OSO2OH.The dyes are made non-bleeding by incorporating additional sulfonic acidand/or carboxylic acid groups into the aryl groups [ 7 ] .The indicator strips are commercially available Merck products ; the differ-ent strips available collectively span the pH range 0 .0-13 .0, and each individ-ual cellulose pad will change color over a range of 2 .0-3 .5 pH units .To test the response and stability of the immobilized Merck indicators, 1200repetitive injections of buffer solution into dilute carrier were performed . Merckindicator 9853, which changes pH over the range 6 .5-10 .0, was used in con-junction with urease. The reflectance spectrum, pH response characteristics,and calculated pKa of the bound dye were previously reported, as was the dyeresponse to repetitive injections of buffer solutions [ 4 ] . The reproducibility ofthe response to 1200 injections of buffer using Merck indicator 9582 (Fig . 6),which changes pH over the range 4 .0-7.0, is virtually identical to previous re-sults for indicator 9583 . When 100,ul of pH 6.8 buffer was injected into 5 X 10 -4M HC1 carrier, a very stable dye response was observed to 1200, or indeed12 000, repetitive injections of the analyte. However, there was a system drift,evident over a long series of injections of the same sample . The relative stan-140.140 timeFig. 7. Response of Merck indicator 9583 to Tris buffer, pH 8.5, at 0.1 M, 0.01 M, and 0.001 M.Carrier is dilute hydrochloric acid, pH 3.0 .Fig. 8. Calibration curves for injections of 50 MI of urea standards (0.312, 0.625, 1.25, 2.50, 5 .0,10.0 and 25.0 mM) at two different enzyme concentrations in the cross-linking solution : ( ) 20mg of enzyme; (0) 10 mg of enzyme .dard deviation (r.s.d.) of the response to ten injections of pH 6 .8 buffer was0.66%; the r.s .d. of the response to 25 injections of pH 6 .8 buffer was 0.86%, aresult of the slow decrease in sensor response . Comparing the means of foursets of 99 injections performed over the course of a day, the r .s .d. increased to6.4% . This was observed every day, and at the beginning of each set of mea-surements, the dye response was recorded at the same high A r, indicating thatthe slow loss in signal was instrumental, and not loss in dye concentration onthe cellulose .Response to buffer concentration . The recorded optical signal is a response tochange in hydrogen ion concentration . It is not strictly "pH", which is an ac-tivity measurement (pH = -logaH+ ) ; aH+ ismeasured on a numerical scaleof potential, called the hydrogen scale, where :EH=EH 0+RT In aHwhere EHO is the standard potential of the reversible hydrogen electrode [ 10 ] .The dye/surface/solution measurement can be quite complicated, becausethe dye response is dependent on the pKa and charge of the indicator, and onthe ionic strength of the solution . The buffer or sample injected will have acharacteristic response peak, dependent on its molecular charge (which affectsits retention time within the pad) and on the ionic strength of the carrierstream. The response of the dye to a sample plug is also a strong function ofthe buffer concentration (buffer capacity) (Fig. 7), which will be to some01 5extent a function of sample size, i.e ., dispersion. To avoid sample dilution forsimple pH measurements, dispersion should be minimized; this will also min-imize interactions of the sample with the carrier stream .Enzymatic monitor for ureaUrease has been successfully immobilized to many surfaces, taking advan-tage of many of the existing immobilization procedures . In most situations, theimmobilized urease has been shown to be extremely stable [ 11-15 ] . Here, sev-eral methods were tested: inorganic bridge formation [ 16 ] , cyanuric chlorideactivation [ 17 ] , adsorption [ 4 ], cross-linking (with and without bovine serumalbumin) the enzyme into the pad with glutaraldehyde [ 12 ], and covalentattachment via silanization of the cellulose, followed by glutaraldehyde cou-pling of the enzyme to the activated surface [ 7 ] .Two immobilization methods proved successful : glutaraldehyde cross-link-ing of the urease and bovine serum albumin within the fibrous pad, and cova-lent attachment via silanization and glutaraldehyde coupling to the cellulose .The reagents involved in inorganic bridge formation were found to destroy thedye attachment to the cellulose . While the conditions for cyanuric chlorideactivation could be made mild enough to preserve most of the immobilized dye,very little activity was observed, thus the sensor had very low sensitivity . Whenurease was cross-linked without any bovine serum albumin (BSA), initial re-sponse was similar to urease cross-linked with equal amounts of bovine serumalbumin and enzyme, but the resulting sensors were completely inactive in justa few days .The two successful techniques, i .e., covalent attachment via silanization,and cross-linking with BSA and glutaraldehyde, possessed virtually identicalresponse characteristics, and sensor lifetimes . The main difference betweenthe two techniques is ease of preparation ; the cross-linking procedure takesless than an hour, with a 100% success rate, whereas the covalent immobili-zation takes all day, with a 25% success rate (successful in one of four attempts) .Enzyme concentration. To examine the effect of increasing the enzyme"loading" on an individual pad, two different cross-linking solutions were pre-pared. In two 1 .0-ml aliquots of pH 7 .0 phosphate buffer, 30 mg of bovine serumalbumin and either 20 mg or 10 mg of urease were dissolved ; ca. 50,U1 of 2 .5%glutaraldehyde (in pH 7 .0 phosphate buffer) was added to each aliquot, andthe resulting solutions were dispensed by dropper onto the cellulose pads . Thisis the "pad equivalent" of the cross-linking procedure for the preparation ofimmobilized enzyme electrode sensors [ 11 ] .The response of the system with the higher enzyme loading (20 mg ml -1 )was 45% greater than that with the pad containing 10 mg ml-1 at the level ofsaturation of the response . At a low concentration of substrate (up to 0 .6 mM1 6urea), in the linear region of the response curve, the enzymatic response wasincreased over 250% (Fig. 8) . Both curves are very similar in shape, i.e., theupper limit of the linear region is the same, and the enzyme demonstratessaturation to the substrate at very nearly the same concentration of urea . Theadvantage of the higher enzyme concentration, then, is that the magnitude ofthe enzymatic response is significantly increased at low levels of urea . Thiseffectively increases the linear range of the system, by allowing low concentra-tions to be determined above the detection limit .For all subsequent work, 20 mg ml -1 urease was used in the immobilizationprocedure for all the urea sensors . The molecular weight of urease is 60 000 . Ifa drop of solution is assumed to be 20 ,ul, and all of the enzyme in the drop wassuccessfully cross-linked into the cellulose, this corresponds to 5 .4 X 10-9molof enzyme on each individual sensing pad .Stopped-flow response . The stopped-flow approach was selected for enzy-matic measurements, for a very simple reason : the immobilized enzyme reac-tion with urea required 30-60 s for a measurement with good sensitivity, theactual stop time depending on enzyme activity (a function of the age of thesensor) and the room temperature (the system is not thermostated) . At thecarrier flow rate of 1 .2 ml min -1 , the "stop" is initiated 3-4 s after sampleinjection. The urea sample volume for all experiments is 50 l. Usually thepeak height is recorded, but if there is any variation in background (sample)pH, a kinetic measurement is preferred [ 4 ] . The shape of the response to dif-ferent concentrations of urea is shown in Fig . 9 .Precision. Figure 10 demonstrates the precision of response of successfullyimmobilized urease . These 25 replicate injections of 50 ul of 10 mM urea, witha stop-flow time of 60 s, exhibit a relative standard deviation of 1 .10% .In contrast, the result of an unsuccessful immobilization attempt using animmunoaffinity membrane as support is shown in Fig . 11 . This response re-sults from the enzyme adsorbed (instead of covalently attached) to the com-mercially available Biodyne immunoaffinity membrane . After the enzymeimmobilization attempt (as per Biodyne instructions), the material was sim-ply cut to the same shape as the Merck indicator pad and sandwiched togetherwith the indicator pad in the flow cell . With this method, immobilization trialson any type of surface could be rapidly evaluated ; even strongly adsorbed en-zyme is washed out of this sensor, demonstrated by a decay in response torepetitive impulses of urea standard .Increasing sensor volume . When two pads were placed in a widened versionof the described flow cell, an increase in A r was observed, especially at low ureaconcentrations (11% at 10 mM urea, 26% at 1 .0 mM urea) . However, the pre-cision of the response was greatly decreased . Increasing the depth of the flowcell results in significant "noise", or variation in A r, while the flow is stopped .This does not appear to be a practical way to increase sensitivity .More active urease . To investigate the effect of using a higher activity of time0 400 200 00ArAr0500 2500IIa time11I4min1'11111411111111111111111I1 7Fig. 9. Stop-flow response (60-s stop) of cross-linked urease (20 mg ml - ') to different urea stan-dards: (a) 10.0 mM, (b) 5.0 mM, (c) 2.5 mM, and (d) 1 .25 mM. The abrupt signal drop occurswhen flow is started to wash the sample from the cell .Fig. 10. Response of cross-linked urease sensor to 25 injections of 10 .0 mM urea (50-,ul volume) .Stop time is 60 s .0 .6030 .0TimeFig. 11 . Response of Biodyne immunoaffinity membrane sensor with Merck indicator 9583 toreplicated injections of 10 mM urea standards (50,u1) . Stop time is 60 s .enzyme, two types of urease sensors were prepared on the same day, underidentical immobilization conditions . In one set of sensors, the urease previ-ously described was used; the immobilization solution contained 20 mg of U-2125 urease (1420 micromolar units) . The other set of sensors was preparediiPy1 8Ca-4m timeFig. 12 . Immobilized enzyme response to changing concentrations of sodium chloride in carrierand sample solutions . Two urea concentrations were investigated . The first series of peaks (onthe time axis) is for 10 mM urea in (a) 0 .0, (b) 70.0, (c) 560.0, (d) 140 .0 mM NaCl. The secondseries of peaks is for 1 mM urea in (a) 0 .0, (b) 35.0, (c) 70.0, (d) 560.0, (e) 140.0 mM NaCl .from more active urease (Sigma U-0376 with 1 .1 molar units per gram) . Aquantity of 5 .5 mg of U-0376 was used (6000 micromolar units) . Nine daysafter immobilization, the pads were tested and the pads with the more activeenzyme showed a 28% greater response to 10 mM urea and a 256% greaterresponse to 1 .0 mM urea .Interference from sodium chloride . Both dye and enzymatic responses areaffected by the ionic strength of the medium . The dye has been shown to besomewhat sensitive to carrier ionic strength [ 4 ] , but the concentration of so-dium chloride even more seriously affects the sensor response when the en-zyme response is being monitored (Fig . 12) . The effects are not as serious asthe soluble urease dependence on sodium chloride concentration, but it is clearthat there is an optimum salt concentration at low concentrations of substrate,and large variations in sample ionic strength will affect the sensor response .Sensor-to-sensor reproducibility. To prepare the urease sensors, the indicatorpad is first "peeled" off of the plastic backing to which it has been glued forone-time use. This results in a piece of cellulose of non-uniform thickness . Forsimple pH measurements, this has not been a problem, presumably becausethere is such a high dye concentration on the fibers that the actual volume ofpad (hence the amount of dye) in the flow cell does not influence the magni-tude of the response . However, the amount of support present when the en-zyme is cross-linked into the pad will affect the amount of enzyme that isactually immobilized within the flow cell. When four sensors (pads) were pre-pared at the same time, with the same enzyme/albumin/glutaraldehyde solu-tion, the response did vary from sensor to sensor (Table 1) .Effect of glutaraldehyde concentration . The response of the sensor `in terms0.60Ar0.300.0of sensitivity as well as sampling frequency' is very dependent of the concen-tration of glutaraldehyde used in the cross-linking procedure. With an increasein the amount of glutaraldehyde used, sampling frequency decreases and sen-sitivity decreases (Fig. 13) . This can be attributed to a decrease in mass-trans-port capabilities within the pad moiety, because of the increase of thickness ofthe membrane of enzyme/BSA coating the cellulose fibers .Sensor lifetime . If the enzyme is immobilized to the cellulose pads in phos-phate buffer, tested with a dilute Tris carrier, and stored at room temperaturein buffer, the sensors have a lifetime of 1-2 weeks . If EDTA is used duringimmobilization, and if 1 .0 mM EDTA is present in the carrier, and if the padsare stored at 0-5 C in buffer and EDTA, the lifetime is extended to 1-2 months11leilp.4min4n time0/IblLI0 60Ar0.30000 102030Injection number (time .I19Fig. 13. The effect of glutaraldehyde concentration used in cross-linking the enzyme with BSA onthe sensor response. Glutaraldehyde (2.5% solution) added to the 1 .0 ml of enzyme/BSA solution :(a) ca. 40l; (b) ca. 80l .Fig. 14. Response of the cross-linked urease sensor to injections of 10 mM urea, as a function of[He' ] in the carrier stream; (0) 0.04, (/) 0.20, () 0.40, (p) 2 .0MHe, .TABLE 1Comparison of sensor-to-sensor responsesSensor 1 2 3 410 mM ureaAvg. A, 0.635 0.513 0.556 0.536R.s.d. (%) 1.10 0.63 0.96 1 .51 mM ureaAvg. A, 0.404 0.287 0.313 0.225R.s.d . (%) 1.00 1 .40 2.20 3.70aAverage and r .s.d. for n = 5 .20(at least 50% activity retained) . Attempts to improve the sensor lifetime with2-mercaptoethanol (4-8 pM concentrations in the immobilization solution andstorage buffer) were unsuccessful ; in fact, addition of the mercaptoethanolcaused lower initial enzyme activity as well .Other supports . At this point, some justification for using cellulose as thesupport matrix may be in order . While it is convenient to use the commerciallyavailable Merck products, it was found to be a fairly simple matter to immo-bilize soluble Merck indicator dyes with covalenty binding arms to Nylon-6membranes, gelatin films, and various ultrafiltration memebranes, as well ascellulose . However, immobilization of sufficient quantities of enzyme is an-other matter. The available surface area for these other membranes is muchlower than for the fibrous cellulose pad . Nylon has very few free amino groups(necessary for linking through a bifunctional reagent like glutaraldehyde) andnylon membranes cannot be partially hydrolyzed without destroying its struc-tural integrity, so that immobilization relies on entrapment or cross-linkingwithin the membrane itself. The resulting system has been shown to have lowurea sensitivity [ 18 ] , and would probably not give sufficient enzyme activityon the small sensor used here .Michaelis-Menten constants. Flow injection analysis can be readily imple-mented to evaluate the kinetic parameters of an immobilized enzyme surface .Because in this particular instrumental design the immobilized enzyme is withinthe detection window, the kinetic data can be obtained almost immediatelyupon enzyme interaction with the substrate, without delays caused by the timerequired to mix enzyme and substrate solutions, or by transport of the productfrom an immobilized enzyme reactor to the detector . The reaction is monitoredcontinuously, while the flow is stopped when the injected substrate volumereaches the immobilized enzyme . Each injection of substrate generates a typ-ical pseudo- Michaelis-Menten curve, i .e., initially a linear absorbance vs . timeresponse, and then curving of the response to approach a steady-state value .The kinetic response of an immobilized enzyme is more complicated thanthat of homogeneous systems, mainly because of mass transport considera-tions at low concentrations of substrate . When an enzyme is immobilized, onegenerally observes an increase in Km. This increase is usually related to thecharge on the substrate and/or support, diffusion effects, or even tertiarychanges in enzyme configuration [ 19 ] . It has been suggested that "apparent"Km, Km (app) and "apparent" Vmaxf Vm.. (app), be reported for insoluble en-zymes [ 20 ] . To determine the Michaelis-Menten constants under the follow-ing experimental conditions, the linear portions of the reflected absorbance(A r ) vs . time curves were evaluated to determine the initial velocity (A r s -1 ),which is then plotted vs. substrate concentration. The Km and Vmax values canbe obtained more accurately from the reciprocal Lineweaver-Burke plot, wherethe y and x intercepts correspond to 1/ 1tmax and -1/Km, respectively, and theslope of the line is Km/ Vmax. With five data points (the high concentrations ofTABLE 2Kinetic data for cross-linked urease1/velocity (s/A,) 8.26, 8.33, 9 .80, 11.89, 16.03, 31 .55, 80.0, 149 .0, 322 .01/[substrate](mM-1 ) 0 .01, 0 .04, 0.10, 0.20, 0.40, 0.80, 1 .60, 3.20, 6.402 1substrate), a plot of the reciprocal velocity (s/A r ) versus the reciprocal sub-strate concentration (1/mM) generated a straight line with a slope of 20 .6 andwith a y intercept of 7 .78 (s/A r ), and a correlation coefficient of 0 .996 .At room temperature, with a carrier 10 -3 M in Tris buffer and 140 mM insodium chloride at pH 6.5, theVmaxobtained was 0 .128 A r s -1 , and the Kmobtained was 2.64 mM urea. The Km of type III urease (soluble), determinedby a pH-stat method, in diluted sodium perchlorate at pH 7 .0, is reported as2.25 mM [ 211 . The Km of soluble urease in unbuffered solution, pH 6 .5, 25 C(determined by pH measurements), is 2 .56 mM [ 22 ] .While the data points at high urea concentration show good linearity, andthe resulting Km value is in good agreement with soluble urease constants, thedata obtained at lower concentrations of substrate show a very different slope(Table 2) . This slope reflects the effect of external mass transfer limitations[23], and is equal to 1/Kmam+Km/ Vmax,rather than Km/Vmax,where Km isthe mass transfer coefficient (dimensionless units of length/time) and am isthe surface area per unit volume . These Kmam values can be determined fromthe Lineweaver-Burke data [ 24 ], and used to evaluate the relative contribu-tions of different surfaces to mass transfer .Mercury monitorEnzymes on inert pads have been reported in the literature for use in themonitoring of enzyme inhibitors present in air or water . Cholinesterase, en-trapped in starch gel onto a polyurethane foam pad, was used by Goodson andJacobs [25] for collection, concentration and detection of cholinesteraseinhibitors .Enzymes are very sensitive to traces of metals, and inhibition of the enzy-matic reaction has been used analytically for determination of the metal in-hibitor [26.27] . Silver (1), Hg (II), Cu (II), Cd (II), Co (II), Ni (II), Mn (II),and Pb (II) have all been quantified by using data from the inhibition of theurea/urease reaction by the metal [ 22 ] . Metal ions generally are assumed toinactivate urease by reaction with a sulfhydryl group, and the relative toxicityof each of these ions has been correlated to the solubility product of the cor-responding metal sulfide [ 26 ] . The urease inhibition has been shown to berelatively selective for mercury (II) [ 27 ], with a selectivity of 1000 (or greater)over Ni (II), Cd(II), Pb (II), and Zn (II) . Selectivity for Hg (II) over Cu (II)22varied from 50 to 1000 with experimental conditions, and the enzyme showedvery little selectivity for Hg (II) over Ag (I) ; Hg (II) was determined in therange of 0-0.7 nmol [ 21,27 ] .Two types of inhibition exist : reversible and irreversible . In irreversible in-hibition, there is permanent modification of a functional group on the enzymerequired for catalysis, making the enzyme inactive . The inhibition cannot betreated by Michaelis-Menten kinetics. The three major types of reversible en-zyme inhibition are usually called competitive, uncompetitive, and noncom-petitive. Noncompetitive inhibition can be experimentally distinguished bythe effects of inhibition on the reaction kinetics of the enzyme. For valid ki-netic analysis, the inhibitor must combine rapidly and reversibly with the en-zyme or the enzyme/substrate complex [ 28 ] .Inhibition of urease with metals has been described as non-competitive atpH 8.9; the observed rate of the enzymatic reaction is inversely proportionalto the inhibitor concentration . The relative inhibition of the enzyme has beenshown to be a linear function of the mercury (II) concentration for a givensample size, or a linear function of amount of Hg (II) if the sample size isvariable . The immobilized enzyme has also been shown to give definite andreproducible levels of activity after the enzyme/metal complex has been treatedwith a regenerating solution of thioacetamide, EDTA, and Tris buffer [ 27 ] .With the system recommended here, full enzyme activity can be restored,which means that a reversible sensor is available for metal determination . Todemonstrate this, the immobilized urease was exposed to various levels of mer-cury (II) contamination in the carrier stream (without the EDTA), and theresponse to a 10 mM urea standard was monitored . At low levels of mer-cury (II) (0.2-2.0 pM) , injections of 10 mM urea showed an exponential de-crease in peak height with increase in time of exposure . When the log of thepeak height is plotted against the injection number (time), the plot is linear,and the slope of these plots can be used to determine the concentration ofHg (II) in the carrier stream (Fig . 14) . With 0.04,uM mercury in the carrierstream, there was no loss in urease activity after 30 injections of sample (ca .30 min) . Sensitivity could possibly be increased by using as carrier somethingother than Tris; there is probably some degree of complexing of Hg (II) by theTris molecule, as Hg (II) does tend to complex with amines [ 29, 30 ] .The other possible mode of sample monitoring would be to alternate injec-tions of urea standard with a metal sample, similarly to a reported approachwith an immobilized enzyme thermistor [ 31 ] . At low levels of metal, resultssimilar to those in Fig . 14 would be obtained . At higher concentrations of metal(20MM),a single injection of metal will completely inactivate the sensor, serv-ing to indicate "metal present" (above some level) . The concentration of metalat which complete inactivation occurs could be adjusted by varying the enzymeactivity on the sensor .23Dialysate monitorIt is usual to characterize a sensor by generating a calibration curve for whicha series of standards is injected [ 32 ] . In testing the practical applicability ofsensors, however, it is often more appropriate to simulate a true monitoringsituation. For that purpose, the following situation is considered . Over 250 000people in the world are treated on a kidney dialysis system, either at home orat organized kidney centers [ 33 ] . During dialysis, patients have their blood-stream cleared of small, toxic molecules that build up because of their impairedkidney function. One of these small molecules is urea . The process usuallytakes 3-5 h, three times a week . This dialysis process is not monitored, exceptfor occasional blood sample analysis . Models for quantification of dialysis havebeen developed to measure the adequacy of dialysis; the model developed tomeasure urea-clearance kinetics is used as a basis for study to understand ure-mia, and improve dialysis therapy [ 34 ] . These models do not allow for time-dependent changes that may accompany the dialysis process, nor has such databeen available . Here, a method is proposed that may allow time-dependentdata to be collected and included in new models . Continuous monitoring of theurea clearance effected by a dialysis machine would provide information onthis process that is not currently available, information that would allow eachpatient to be dialyzed for only that length of time required for her/his individ-ual situation . This should prove cost-effective for centers, as the machinescould then service more patients, and would be therapeutic for patients, whomay be able to spend less time on the dialysis machines .To apply the immobilized urease sensor to kidney dialysate fluid, a new in-tegrated microconduit was designed (Figs . 3 and 5) . For continuously oper-ating situations like kidney dialysis, it is felt that a simple design, in which theinjection valve is eliminated from the microconduit, will result in a systemrequiring less regular maintenance . When the sample volume is containedwithin a valve bore or the external loop of a valve, moving parts are requiredto introduce the sample, and these parts eventually wear and leak [ 35 ] . Valve-less systems have been developed, where the sample is introduced by differ-ential pumping techniques [ 36 ] or controlled aspiration techniques [ 37 ] . Themethod used here for the introduction of dialysate is valveless, and combinesaspects of the hydrodynamic and controlled-aspiration techniques . The carrieris pumped by P1 (Fig . 3) across the sensing pad, to establish the baselineresponse, as well as wash out the previous sample . The carrier stream is thenstopped, and the second pump (P2) draws (aspirates) sample from the con-tinuously flowing dialysate stream across the sensor, in the opposite directionto the carrier flow . Once the sample reaches the flow cell, the second pump isalso stopped, and the urease reaction takes place . When enough reaction timehas been allowed (usually 60 s), the first pump is switched on, and carrierwashes through the flow cell . When the sample is not being pumped across theflow cell, all of it is pumped directly to waste. When buffer solutions are used240.90A r0.450.00lal4Iblrrrl rr-'a timeFig. 15 . Response of the urea monitor system to buffer samples : (a) pH 6.8, (b) pH 8 .0, (c) pH9 .0, (d) pH 10.0 .as the "sample" instead of urea standards or dialysate, and without stoppingP2, the signals have the shape observed in Fig . 15. The width of the "peak"(which is actually a steady-state signal) produced by aspiration of the buffers,depends only on the "on" time interval for P2 (10 s in this case) .To simulate a patient being monitored on dialysis, a flask (reactor) con-taining a urea solution was continuously sampled and diluted at the same rateby an external pump (P3), to maintain constant reactor volume while contin-uously removing urea from the reactor (note that the flow rate of P2 must beless than the dialysate flow rate, or waste will be drawn across the sensor aswell, diluting and possibly contaminating the sample) . Under these condi-tions, the urea concentration will decrease exponentially : C = C exp ( - t/tree)where C is the urea concentration at time t, C is the initial urea concentration,and tLes is the mean residence time ; tree can be replaced by Vr/Q, where Vr is thereactor volume, and Q is the pumping rate [ 38 ] . Taking the logarithm of bothsides giveslogC= 2.303 log C - (2 .303 Q/Vr ) tThus, a plot of log C vs. time will be linear, with a slope dependent on thereactor volume and pumping rate, and a y intercept dependent on the initialurea concentration (CO ) .ICI1d.Idlrrrr rI*timeFig. 16 . Response of the urea monitoring system to a reactor with an initial urea concentration of25 mM and reactor volume of 50 ml .Fig. 17 . Response of the urea monitoring system to 10 .0 mM urea in 25 ml of (0) pH 6 .5 Triscarrier, (A) pH 7.4 Tris carrier, and (0) pH 7 .4 dialysate solution .To verify that these equations do indeed apply to the monitoring system usedhere, six systems were tested with the same urease sensor: four systems at 50-ml reactor volume with urea concentrations of 2 .5, 5 .0, 10.0, and 25.0 mM; andtwo systems with 25 mM urea at reactor volumes of 25 ml and 100 ml . Thesensor response to 50 ml of 25 mM urea is shown in Fig . 16 . For reference, thenormal adult serum urea range is 1 .78-6 .08 mM [ 39 ], and the correspondingconcentration in the dialysate solution will depend on the membranes used inthe filtration system . At low concentrations of urea (below ca . 1 .0 mM urea),the exponential decrease in enzymatic response is observed. At high concen-trations of urea, however, the enzyme is saturated, and no change in signal canbe observed. When the logarithm of the peak is plotted against time, the systemresponse is clearly composed of two parts : the linear portion predicted by thedilution equations, and a flat region which is the result of enzyme saturationbehavior . This flat region is not without merit ; it verifies, without the use ofcalibrating standards, that the immobilized enzyme is sufficiently active (i .e .,that the individual sensor is responding as it should) .When the slopes of the linear portions of the plots are calculated and com-pared (Table 3), it is clear that the system responds as expected . When aconstant volume is maintained, and urea concentration is varied, the slope isrelatively constant . When both volume and urea concentration are varied, theslope changes .When the same experiment was repeated at pH 7 .4, the pH of the dialysatefluid, a smaller signal was observed (Fig . 17) . This reflects the lowered ureaseactivity, the increased buffer capacity of the carrier at this pH, and the smallerpH range available for the enzyme reaction to occur (the reaction products areself-buffering at pH 9 .0) [ 22 ] . When the system monitors dialysate solutionIr P1IItI+++ (+% I .0.55Ar0.320025Ar26TABLE 3Dialysate monitor system: the effect on the linear portion of the log peak height vs . time plot fromchanging the reactor volume and changing the initial urea concentration in a constant reactorvolumewith the same concentration of urea as these solutions, the signal is muchfurther reduced, because of the much higher buffer capacity of the hydrogen-carbonate buffer in the dialysate. To be truly applicable to this solution, everyeffort to increase sensitivity would have to be employed : higher activity ofurease, possibly higher enzyme loading, and possibly dilution of the dialysateprior to enzymatic reaction .CONCLUSIONResponse of a sensor is a complex function of sensor qualities, of a monitoredenvironment, and of the mode in which sensor and monitored species arebrought together. Manual testing in a batch mode is inadequate and inefficientbecause it lacks the exact timing and reproducibility of events. Automated FIAprovides well-reproduced conditions by allowing the sensor to be tested in aflowing stream of a carrier, into which well-defined zones of testing materialare periodically injected . This allows response properties of a sensor to be ex-actly characterized.Such use of FIA is not limited to optical sensors . It has been used previouslyto test response of ion-selective electrodes [ 40, 41 ], microelectrodes [42 ] andISFETs [ 43 ] . Recently, the surface treatment of glass carbon electrodes wasevaluated by means of the voltametric response observed in a flow-injectionmode [ 44 ] . The present work is, however, the first to introduce several aspectsof the techniques of FIA to a systematic evaluation and development of a chem-ical sensor .To conclude, one may observe that physical modulation of a transmitted andobserved signal is a basis of all modern techniques . Beams are chopped, laserspulsed, acoustic and electric signals are modulated at wide frequencies . It isappropriate that chemical modulation in a flow stream, which is the basis ofReactor vol .(ml)Urea conc .(mm)No. ofdata pointsRegression line r 2100 25 18 y= -0.15x+2.48 0.997925 25 20 y=-0.42x+2.09 0.995050 25 23 y=-0.28x+2.45 0.995950 10 13 y=-0.34x+2.70 0.998050 5 18 y=-0.29x+1 .80 0.997650 2 .5 14 y=-0.30x+1.52 0.9945all flow-injection techniques, should become a tool for research and evaluationof chemical sensors .The authors express their gratitude to Merck, Darmstadt, Germany, and theNorthwest Kidney Center, Seattle, WA, for donation of research chemicals, toTecator A/B Sweden for the loan of equipment, and to the Danish Council forScientific and Industrial Research for travel support of J . R. and G.D . C .REFERENCES1 G.B . Brown, Enzyme Engineering, Vol . 3, Plenum, New York, 1975 .2 J . Ruzicka and E.H. Hansen, Anal . Chim. Acta, 179 (1986) 1 .3 J. Ruzicka, Anal. Chem., 55 (1983) 1040A .4 T.D . Yerian, G.D . Christian and J . Ruzicka, Analyst, 111 (1986) 865 .5 J. Ruzicka and E .H. Hansen, Anal . Chim. Acta, 173 (1985) 3 .6 D .D. Perrin and B . Dempsey, Buffers for pH and Metal Ion Control, Chpaman and Hall,London, 1974 .7 K. Neisius and W. Baumer, U .S. Patent No. 4 029 598, June, 1977 .8 R.E. Adams and P .W. Carr, Anal. Chem., 50 (1978) 944 .9 M.K. Weibel, W. Dntschilo, H .J. Bright and A .G. Humphrey, Anal. Biochem., 52 (1973)402 .10 R.G . Bates, Determination of pH : Theory and Practice, Wiley, New York, 1964 .11 C. Tran-Minh and G . Brown, Anal . Chem., 47 (1975) 1359 .12 K . Mitsugi and N. Mimura, Enzyme Engineering, Vol . 4, Plenum, New York, 1978 .13 G.G . Guilbault and G . Nagy, Anal. Chem., 43 (1973) 417 .14 M. Mascini and G . Palleschi, Anal . Chim. Acta, 145 (1983) 213 .15 J .P. Joseph, Mikrochim . Acta, Part II, (1984) 473 .16 R.A. Messing, Biotechnology and Bioengineering, Vol . XVI, Wiley, New York, 1974, pp .1419-1423 .17 L.A. Saari and W .R. Seitz, Anal. Chem., 56 (1984) 810 .18 D.J. Inman and W.E. Hornby, Biochem. J., 129 (1972) 255 .19 H.H. Weetal, Anal. Chem., 46 (1974) 602A .20 K . Mosbach, Applications of Biochemical Systems in Organic Chemistry, Wiley, New York,1976.21 E.C. Toren, Jr. and F.J . Burger, Mikrochim . Acta, (1968) 1049 .22 G.B. Kistaiakowsky and W .H .R. Shaw, J. Am. Chem. Soc., 75 (1953) 2751 .23 K.J. Laider and P .S . Bunting, Methods in Enzymology, Vol . 64, Academic, New York, 1980 .24 R.A. Messing, Immobilized Enzymes for Industrial Reactors, Academic, New York, 1975 .25 L.H . Goodson and W .B. Jacobs, in E.K.Pye and L.B. Wingard, Jr . (Eds.), Enzyme Engi-neering, Vol . 2, Plenum, New York, 1973 .26 W.H.R. Shaw and D .N. Raval, J . Am . Chem. Soc ., 83 (1961) 3184 .27 L. Ogren and G . Johansson, Anal. Chim. Acta, 96 (1978) 1 .28 A .L. Lehninger, Biochemistry, Worth Publishers, New York, 1975 .29 A . Cabrera-Martin, J .L. Peral-Fernandez, S. Vinente-Perez and F . Burriel-Marti, Talanta,16 (1969) 1023 .30 A. Ringbom, Complexation in Analytical Chemistry, Interscience, New York, 1963 .31 B. Mattiasson, B . Danielsson, C . Hermansson and K. Mosbach, FEBS Lett ., 85 (1978) 203 .272832 J. Ruzicka and E.H. Hansen, Anal . Chim. Acta, 173 (1985) 3 .33 C. Blagg, personal communication .34 P.S . Malchesky, P . Ellis, C . Nosse, M. Magnusson, B . Lankhorst and S. Nakamoto, DialysisTransplant., 11 (1982) 42 .35 J .J. Harrow and J . Janata, Anal. Chem ., 55 (1983) 2461 .36 J . Ruzicka and E .H. Hansen, Anal. Chim. Acta, 145 (1983) 1 .37 C . Riley, L.H. Asslett, B .F. Rocks, R.A. Sherwood, J .D . Watson and J . Morgan, Clin . Chem .,29 (1983) 332 .38 J. Ruzicka and E.H. Hansen, Flow Injection Analysis, 2nd edn ., Wiley-Interscience, NewYork,1988.39 L .A. Kaplan and A.J . Pesce, Clinical Chemistry: Theory, Analysis, and Correlation, C .V.Mosby Company, St. Louis, 1984 .40 E.H. Hansen, F .J. Krug, A.K. Ghose and J. Ruzicka, Analyst, 102 (1977) 705 .41 R.Y. Xie, V.P.Y . Gadzepko, A.M. Kadry, Y .A. Ibrahim, J. Ruzicka and G.D. Christian, Anal .Chim. Acta, 184 (1986) 259 .42 A. Haemmeric and J. Janata, Anal. Chim. Acta, 144 (1982) 115 .43 R. Smith, R.J. Huber and J. Janata, Sensors & Actuators, 5 (1984) 127 .44 J. Wang and P. Tuzhi, Anal. Chem., 58 (1986) 1787 .page 1page 2page 3page 4page 5page 6page 7page 8page 9page 10page 11page 12page 13page 14page 15page 16page 17page 18page 19page 20page 21page 22


View more >