u ... m Au...” . nu .V. ..~..,i-tir1u~.v- . .w.‘ mun-"M watchman?" .m .::~ . 72;. ‘5‘ - £31; 3% 67;. q 5; “V‘” 2 n.» ‘ ‘ "\ ‘ , 4., “4' Fair w‘i": air a ' (“wt-u v, ‘3‘ egg“ 2&3.“ Afi‘dh“. ". ‘3 t ‘ {C5143- . _.I v‘. r- n at 1"”?(13‘ (7% .4”. ‘ .. Q . . ”13>" fié‘fi'mum ' “(fa * ~‘ “‘m‘ . r 3-. ‘ g ’70 n X» ”Nylwf 1M flu Lg." I . V“. V '3}!- F3“: ’{V' ,. L.)- . , . x ' ‘ :“F{“"",. .V a ’y‘. .r 1 ,' u , . ' g: ‘.;.-'V with; .1“ ‘ m».-;‘. 3.13.: 1 $3"! . .r _ , a... ’g‘é‘a; ' “173$?“ z‘éagk‘x'lv‘n tr:"§:§:§ 2... Erfic 1" u...- my 'L“::‘“:'“1§" u; ' '4‘ u- ""x\..- ' "o :3 V .3.‘ I ‘. 1 «M. V t "I , . - V_‘ V n 5“".- 7 -" ' ' a Afln‘x‘. s -. “VA m, 3 I“ ‘ gm} 39% ~ n§1 '1 Eh:- fi‘ “’5‘" “2:517:21 1 ... ‘3‘. v .1... i x ,_ "-3."? 1 J . ‘ . L, rarrfi ' ‘ “W “A” E‘ “La mit- 35% \u’s "fl&‘%fi “a .44u~.. H; -u‘ 5%? u .3 23%;?“ ‘1 I 'I ' ’1‘: ‘éfitfi new? 3' ”.31 33‘" . ," m’u‘vfl ‘.:'1~.‘ .u ‘ x ”A" r‘ ‘45: n-« « ,sk-‘fih.Kf V at “ V..,n 532:3 w. ..;_ -‘. E" “I?“ 1 .-.1 r n. :gira. nu . . ...._ mg'.‘ 3,“? N ’. ' “nu llllHillllllllllIllIUWIHIIWIHIllllllllllllllllllllll L 301026 3394 LIBRARY Michigan Stale University This is to certify that the dissertation entitled Design and Optimization of a Flow Injection System for Enzymatic Determination of Sugars presented by Kristine S. Kurtz has been accepted towards fulfillment of the requirements for Ph . D . degree in Chemistry ’ Km J41 ajor professor Date 2 ’)$.4; MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE ll RETURN BOX to move this checkout tron your rooord. TO AVOID FINES rotum on or baton dot. duo. DATE DUE DATE DUE DATE DUE MSU lo An Affirmative Action/Equal Opportunity Institution mafia-9.1 DESIGN AND OPTIMIZATION OF A FLOW INJECTION SYSTEM FOR ENZYMATIC DETERMINATION OF SUGARS By Kristine S. Kurtz A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1992 6 7:: - W 7/ ABSTRACT DESIGN AND OPTIMIZATION OF A FLOW lNJECTION SYSTEM FOR ENZYMATIC DETERMINATION OF SUGARS By Kristine S. Kurtz The principles of flow injection analysis and immobilized enzymes were used to develop methodologies for the determination of six nutritionally important sugars (glucose, galactose, sucrose, lactose, maltose, and fructose). Indirect detection of the sugars was accomplished by means of enzymatic reactions that yielded hydrogen peroxide. The hydrogen peroxide formed was subsequently reacted with a leuco-dye/peroxidase reagent and the dye product was monitored spectrophotometrically. In developing the methodologies, performance aspects such as enzyme immobilization procedures, FIA manifold designs, and system optimizations were considered. Initially, investigations were conducted to improve enzyme immobilization efficiencies. Optimal conditions for immobilization were evaluated based on support pretreatment, support silylation solvent, and silylation atmosphere. In addition, various types of glass supports were investigated for their potential enzyme loading capacities. Four immobilized enzyme reactor configurations were investigated for their kinetic and flow characteristics. Reactor characteristics were studied for systems with and without an enzymatic reaction to gain insight into the dispersion contributions from both chemical and physical processes. Reactors were evaluated based on their resulting activity and the dispersion introduced. Reaction-rate dependent parameters were investigated and optimized for each sugar determination manifold. A Composite Modified Simplex (CMS) routine was employed as a means of multivariate optimization for system parameters. The parameters optimized include carrier pH, carrier flow rate, detection reagent flow rate, activator concentrations, and modifier pH. System factors considered in the response function of the Simplex optimization were sensitivity, precision, and sample throughput. Optimized system performance was obtained for each sugar determination by combining the results from the support evaluations, the enzyme reactor configuration comparisons, and the Simplex optimizations. Analytical figures of merit such as working range, detection limit, precision, and sample throughput were evaluated for each sugar under the optimized conditions. TABLE OF CONTENTS CHAPTER PAGE LIST OF TABLES .......................................................................................... vii LIST OF FIGURES ....................................................................................... ix I INTRODUCTION ......................................................................................... 1 A) Parallel Multichannel Analyzer ............................................. 2 B) Project Goals ................................................................................. 4 II HISTORICAL BACKGROUND ................................................................ 7 A) Carbohydrate Determinations in Food Analysis ............... 7 1. Non-specific Methods ...................................................... 8 2. Specific Methods ............................................................... 9 a. Separation Methods ............................................. 9 b. Selective Chemical Methods ............................. 1 1 B) Immobilized Enzymes in Analytical Chemistry ................ 14 1. Enzyme Immobilization Methods ................................ 15 2. Immobilized Enzyme Applications .............................. 1 7 a. Enzyme Probes ...................................................... 1 7 b. Enzyme Reactors .................................................. 18 c. Enzyme Membranes ............................................ 19 C) Flow Iniection Analysis ............................................................. 20 D) Previous Work in Our Laboratory ......................................... 26 1. Glucose Oxidase Immobilization ................................. 26 2. Glucose Determination .................................................. 2 7 3. Optimizations ................................................................... 29 III INSTRUMENTATION AND ANALYZER MANIFOLD DEVELOPMENT ......................................................................................... 30 A) Introduction and Instrumentation ....................................... 30 1. Glucose and Galactose Determinations ................... 35 2. Sucrose,Lactose,Maltose, and Fructose Determinations ................................................................ 35 B) Manifold Development .............................................................. 38 1. Enzyme Immobilization ................................................ 38 2. Reactor Configurations ................................................. 39 3. Enzymatic Reaction-Rate Parameters ..................... 39 4. Interferences .................................................................... 40 iv IV OPTIMIZATION OF ENZYME IMMOBILIZATION AND SUPPORT CONSIDERATIONS .............................................................. 4 1 A) Enzyme Immobilization ............................................................ 42 1. Glucose Oxidase Immobilization ................................. 45 2. Galactose Oxidase Immobilization ............................. 48 B) Immobilization Optimization .................................................. 5 1 Etching ............................................................................... 51 2. Silylation Solvent Comparison .................................... 52 3. Pretreatment, Silylation, and Glutaraldehyde Activation ........................................................................... 55 4. Optimized Immobilized Procedure ............................. 64 C) Trinder Reaction versus Malachite Green Reaction ....... 66 D) Preliminary Support Investigations .................................... 69 1. Immobilization Reproducibility .................................. 7 1 2. Investigations using Controlled-pore Glass ........... 73 a. Glucose Oxidase .................................................. 75 b. B-Galactosidase .................................................... 79 3. Reactive Amino Group Determinations ................... 83 a. Determination Method ...................................... 83 b. Support Silylation Investigations .................. 84 c. Results and Discussion ...................................... 88 4. Reactor Dimension Comparisons: Glucose Oxidase 91 V MANIFOLD OPTIMIZATIONS ............................................................... 96 A) Introduction ................................................................................. 96 1. Reactor Configurations ................................................. 97 a. Designs and Dimensions .................................. 97 b. Flow Characteristics .......................................... 99 2. Enzymatic Reaction-Rate Parameters ..................... 102 a. Introduction to Simplex Optimization ......... 102 b. General Response Function ............................. 103 3. Selectivity .......................................................................... 106 B) Manifolds ...................................................................................... 1 10 1. Glucose (Glucose Oxidase) ........................................... 110 a. Reactor Configuration Evaluations .............. 112 b. Simplex Optimization ........................................ 1 14 0. Characterization ................................................. 12 1 d. Selectivity ............................................................. 123 2. Galactose (Galactose Oxidase) .................................... 125 a. Simplex Optimization ........................................ 126 b. Reactor Configuration Evaluations .............. 132 c. Characterization ................................................. 136 d. Selectivity ............................................................. 136 3. Sucrose (Invertase/Mutarotase) ................................. 138 a. Reactor Configuration Evaluations ............... 141 b. Simplex Optimization ......................................... 144 c. Characterization .................................................. 150 d. Selectivity .............................................................. 153 4. Lactose (B—Galactosidase) .............................................. 155 a. Reactor Configuration Evaluations ............... 157 b. Simplex Optimization ......................................... 159 c. Characterization .................................................. 165 d. Selectivity ............................................................... 167 5. Maltose (a-Glucosidase) ................................................. 169 a. Reactor Configuration Evaluations ............... 169 b. Simplex Optimization ......................................... 172 c. Characterization .................................................. 17 8 d. Selectivity ............................................................... 178 6. Fructose (Glucose Isomerase) ....................................... 180 a. Simplex Optimization ......................................... 184 b. Reactor Configuration Evaluations ............... 191 c. Characterization .................................................. 193 d. Selectivity ............................................................... 194 VI CONCLUSIONS AND FUTURE PROSPECTS ................................... 198 A) Automated Multichannel Parallel Sugar Determinations ........................................................................... 198 B) Sugar Determination Applications and Interference Minimization ............................................................................... 203 C) Enzyme Immobilization ........................................................... 207 D) Alternative Manifold Designs ............................................... 209 LIST OF REFERENCES ............................................................................ 2 1 1 vi LIST OF TABLES TABLE PAGE 2- 1 Applications and versatility of FIA ..................................................... 22 2-2 Flow injection/immobilized enzyme-based carbohydrate determinations - ........................................................ 24 & 25 4- 1 Initial enzyme immobilization procedure .......................................... 44 4-2 Optimized enzyme immobilization procedure .................................. 65 4-3 Supports investigated for enzyme immobilization ......................... 87 4-4 Amine functional group concentrations on glass support surfaces ........................................................................................ 89 5-1 Initial and boundary experimental conditions for glucose optimization _ . ....... _ ........ . ............... .116 5-2 Initial and optimal experimental conditions for glucose determination ........................................................................... 117 5-3 Glucose manifold selectivity ................................................................ 124 5-4 Initial and boundary experimental conditions for galactose optimization ........................................................................... 128 5-5 Initial and optimal experimental conditions for galactose determination ....................................................................... 130 5-6 Galactose manifold selectivity ............................................................ 139 5-7 Initial and boundary experimental conditions for sucrose optimization .............................................................................. 146 5-8 Initial and optimal experimental conditions for sucrose determination ........................................................................... 147 5-9 Sucrose manifold selectivity ................................................................ 154 5- 10 Initial and boundary experimental conditions for lactose optimization ................................................................................ 160 5-11 Initial and optimal experimental conditions for lactose determination ............................................................................ 162 vii TABLE PAGE 5-12 Lactose manifold selectivity ................................................................ 168 5-13 Initial and boundary experimental conditions for maltose optimization ............................................................................. 173 5-14 Initial and optimal experimental conditions for maltose determination .......................................................................... 175 5- 15 Maltose manifold selectivity ................................................................ 181 5-16 Initial and boundary experimental conditions for fructose optimization ............................................................................. 186 5-17 Initial and optimal experimental conditions for fructose determination .......................................................................... 187 5- 18 Fructose manifold selectivity .............................................................. 197 6- 1 Simplex optimization summary ........................................................ 200 6-2 Analytical figures of merit for the determination manifolds 204 viii LIST OF FIGURES FIGURE PAGE 1-1 Molecular structures for the six nutritionally important sugars of interest ............................................................... 5 2-1 Enz tic pathways for glucose determination, a) g ucose oxidase,Iperoxidase pathway; b) hexokinase, G-6- dehydrogenase pathway ............................ 13 2-2 Glucose oxidase / Trinder reaction ................................................. 28 3- 1 Enzymatic reaction schemes ............................................................ 3 1 3-2 Detection reactions ............................................................................. 33 3-3 General instrumentation for sugar determinations ................. 34 3-4 Dispersion comparisons for a single-bead string reactor (SBSR), a coiled open tubular reactor (OTC), and a straight open tubular reactor (OTR) .................................. 36 3-5 Instrumentation for glucose and galactose determination (a); and for sucrose, lactose maltose, and fructose determination (b) ........................................ 37 4-1 General procedure for 3-APTS / glutaraldehyde immobilization of enzymes onto glass supports .......................... 43 4-2 Single-bead string reactor ................................................................. 46 4-3 Glucose calibration curve with Trinder reaction detection ....47 4-4 Galactose calibration curve with Trinder reaction detection ................................................................................ 50 4-5 Silylation solvent comparison for non-porous beads ................. 54 4-6 Glass surface hydroxylation/dehydroxylation equilibrium ............................................................................................ 56 4-7 Common bonding models for 3-APTS ............................................. 58 4-8 Immobilization scheme for simultaneous investigation of support pretreatment, silylation atmosphere, support rinse, and glutaraldehyde solvent .................................. 60 ix FIGURE PAGE 4-9 4-10 4-11 4-12 4-13 4-14 4-15 4-16 4-17 5-1 5-2 5-5 5-6 5-7 5-8 5-9 Single-bead string reactor comparisons for support pretreatment, sil lation atmosphere, support rinse, and glutaraldehy e solvent .............................................................. 62 Malachite green reaction ................................................................... 68 Comparison of Trinder and malachite green (MG) detection for hydrogen peroxide (a) and glucose (b) ................... 70 Enzyme immobilization reproducibility for glucose oxidase ...72 Packed bed reactor design .................................................................. 76 Reactor sensitivity (a) and disp rsion ) comparisons for a SBSR and two PBRs (327 544 ) containing giucose oxidase ...................................................................................... 78 B-Galactosidase activity comparison for three difi'erent carrier conditions ............................................................... 80 Reactor sensiti 'ty (a) and dispersion (b) comparisons for a PBR (327 ) and SBSR containing B-galactosidase ......... 82 Packed bed reactor dimension comparison for glucose oxidase ..................................................................................... 95 Immobilized enzyme reactor configurations ................................ 98 Reactor configuration comparison without enzymatic reaction ............................................................................ 10 1 General response function used in CMS optimizations ......... 105 Response trends as a function of various order of magnitude values for constants a and b ................................. 107 Molecular structures for raffinose, melibiose, and melezitose ..................................................................................... 109 Flow injection manifold for glucose determination ................. 111 Reactor configuration comparisons for glucose oxidase, sensitivity (a), dispersion (b) .......................................... 1 13 Response function progress of the Simplex optimization for glucose ................................................................... 1 19 Scatter diagrams for individual lucose Optimization variables, carrier pH (a), carrier ow rate (b), reagent flow rate (c) ........................................................................... 120 FIGURE PAGE 5-10 Glucose calibration curves for initial and optimized manifold conditions ........................................................ 122 5-11 Flow injection manifold for galactose determination .............. 127 5-12 Response function progress of the Simplex optimization for galactose ................................................................ 131 5-13 Scatter diagrams for individual galactose optimization variables, carrier pH (a), carrier flow rate (b), reagent flow rate(c), ratio Fe III/II (d), concentration Fe(CN)6 ............ 133 5-14 Reactor configuration comparisons for galactose oxidase, sensitivity (a), dispersion (b) .......................................... 135 5-15 Galactose calibration curves for initial and optimized manifold conditions ....................................................... 137 5-16 Flow injection manifold for sucrose determination ................. 140 5-17 Reactor configuration comparisons for invertase, sensitivity (a), dispersion (b) .......................................................... 143 5-18 Response function progress of the Simplex optimization for sucrose ................................................................... 148 5-19 Scatter diagrams for individual sucrose optimization variables, carrier pH (a), carrier flow rate (b) ........................... 149 5-20 Sucrose calibration curves for initial and optimized manifold conditions ....................................................... 152 5-21 Flow injection manifold for lactose determination .................. 156 5-22 Reactor configuration comparisons for B-galactosidase, sensitivity (a), dispersion (b) .......................................................... 158 5-23 Response function progress of the Simplex optimization for lactose .................................................................... 163 5-24 Scatter diagrams for individual lactose optimization variables, carrier pH (a), carrier flow rate (b), Mg+2 concentration (c) ...................................................................... 164 5-25 Lactose calibration curves for initial and optimized manifold conditions ....................................................... 166 5-26 Flow injection manifold for maltose determination ................. 170 5-27 Reactor configuration com arisons for a-glucosidase, sensitivity (a , dispersion (b) .............................. 17 1 xi FIGURE PAGE 5-28 Response function progress of the Simplex optimization for maltose .................................................................. 176 5-29 Scatter diagrams for individual maltose optimization variables, carrier pH (a), carrier flow rate (b) ........................... 177 5-30 Maltose calibration curves for initial and optimized manifold conditions ....................................................... 179 5-31 Flow injection manifold for fructose determination ................ 182 5-32 Response function progress of the Simplex optimization for fi'uctose .................................................................. 188 5-33 Scatter diagrams for individual fructose optimization variables, carrier pH (a), carrier flow rate (b), modifier pH (c) ..................................................................................... 190 5-34 Reactor configuration comparisons for glucose isomerase, sensitivity (a), dispersion (b) ..................................... 192 5-35 Fructose calibration curves for initial and optimized manifold conditions ....................................................... 195 6-1 Parallel, multichannel flow injection analyzer ........................ 202 xii CHAPTER I INTRODUCTION Automated multicomponent determinations are becoming increasingly important in analytical chemistry as demand continually rises for more analytical results in shorter time frames. Typically, multiple components are identified and determined in complex matrices by three general approaches. The traditional and most popular methodologies incorporate a separation step at some point in the analysis. Gas chromatography (GC), high performance liquid chromatography (HPLC), supercritical fluid chromatography (SFC), and more recently, capillary zone electrophoresis (CZE) are very powerful techniques utilized in multicomponent determinations for most sample types. In addition to separation techniques, which generally employ a universal method of detection, a second approach to multicomponent determinations involves selective detection of sample components. Techniques which provide selective detection include atomic absorption spectroscopy (AAS), atomic emission spectroscopy (AES), and mass spectrometry (MS). Combined separation / selective detection methods such as GC/MS are very common in modern analytical laboratories. Several selective detection methods (e.g. ICP emission spectroscopy and tandem mass spectrometry) are capable of simultaneous multicomponent determinations without a prior separation. The third general approach to multicomponent analyses employs automated, multichannel, parallel systems which rely on selective chemistries to determine components of interest within a 2 complex sample matrix. Due to the selectivity, separations are not necessary in this approach, and these analyses can result in faster and more specific determinations. Food analysis is an extremely diverse area and presents a unique challenge to analytical chemists due to the complex chemical composition and dynamic nature inherent to various foodstuffs. An important area within food analysis is that of carbohydrate determinations. Identifying and determining various carbohydrates is essential to many areas of food science including nutrition, agriculture, formulation, preparation, and industrial processing. There is a continuing demand for rapid, accurate, and precise analytical methods for carbohydrate determinations. As a result of society‘s growing awareness in the importance of health and nutrition, food industries and government officials are concerned with the nutrient content of foods and their subsequent labeling. Carbohydrate determinations are very important in the various stages of food processing, from the raw materials used to the final processed products. Every stage of the process must be evaluated with respect to Food and Drug Administration (FDA) requirements and product shelf life. A) Parallel Multichannel Analyzer In an effort to improve upon existing methods and explore new methods of analyses, an automated, multichannel, parallel approach to carbohydrate determinations in foodstuffs is the focus of this work. The system is designed for the simultaneous determination of six nutritionally important sugars: glucose, galactose, sucrose, lactose, maltose, and fructose. 3 In the analyzer under development, immobilized enzyme reactors are used in a flow injection analysis (FIA) system to provide selective, automated determinations of the six sugars in complex mixtures, such as food samples, without prior separation. Separate FIA manifolds containing the appropriate enzyme reactors allow for selective and simultaneous determinations of the individual sugars. Within each manifold, the sugar of interest is reacted enzymatically to produce hydrogen peroxide which subsequently yields a colored product via a leuco-dye, peroxidase reaction. The resulting dye is monitored colorimetrically. Development of a highly parallel sugars analyzer, such as the six- channel system described, can lead to improved technology over current methods for sugars and also to advances in immobilized enzyme methodologies. The specificity of the enzymatic reactions allows for selective sugar determinations with minimal sample preparation in most cases. The FIA system is readily automated, providing a significant sample throughput advantage over conventional methods. Sample pretreatment processes, such as extractions, dilutions, and reagent additions, may also be automated and incorporated into the FIA system. The instrumentation and methodologies developed should be well suited to meet analytical requirements in agriculture, food processing, and nutritional research. The sugars analyzer is designed to be highly modular in nature, resulting in great versatility. This versatility makes it possible to determine a single sugar, all six in parallel, or any combination in between. By substituting or adding further manifolds, the analyzer is readily adapted to provide additional sugar determinations. The 4 instrumentation and selective methodologies developed may also be employed in other multicomponent determination applications such as clinical assays and general food analyses. B) Project Goals The primary goals of this research were to develop methodologies for selective determination of the six sugars of interest (Figure 1-1). This includes aspects such as enzyme immobilization, FIA manifold designs, and system optimizations. Each individual sugar determination occurs in a separate FIA manifold developed specifically for that sugar. Thus, it is possible for each sugar determination to be independently designed and optimized with respect to selectivity, sensitivity, stability, and sample throughput. I In achieving these goals, the first consideration was immobilization of the various enzymes required for the sugar methods. Support enzyme loadings have a direct affect on sensitivity of the sugar determinations. Investigations were conducted to improve enzyme immobilization efficiencies over those initially obtained. These studies are discussed in Chapter IV. In addition to the enzyme loading efficiency resulting from the immobilization procedure itself, the type of support as well as the immobilized enzyme reactor (IMER) configuration are also important considerations with respect to sensitivity and sample throughput. Evaluation of an IMER depends on both the enzyme activity and the dispersion inherent in the reactor system. Controlled-pore glass (CPG) and non-porous glass beads were investigated as enzyme supports. Four different enzyme reactor configurations containing the CPG and/or €11,011 €11,011 H O OH O O OH CHZOHO H H H 011 11 OH H H O 11 OH H on OH H B—D-glucose B—D-galactose B-D-fructose CH20H H OH H H O O -—C CHZOH H OH OH H sucrose CHZOH O CHZOH OH H H 0 OH OH. H H OH H H H H H OH H OH lactose €11,011 €11,011 H O H H 0 OH H H OH H OH H O H H OH H OH maltose FIGURE 1-1. Molecular structures for the six nutritionally important sugars of interest. 6 non-porous glass beads were also compared for each sugar determination. These studies are presented in Chapter V. System factors which influence enzyme kinetics (e.g. buffer composition, pH, reactor residence time, enzyme activators, and reagent concentrations) are extremely important to consider in developing the sugar methodologies. These factors will ultimately affect the selectivity, sensitivity, stability, and sample throughput. Reaction-rate dependent parameters were investigated and optimized for each sugar determination. A multivariate approach was employed in the optimizations to account for interaction of system variables. A composite modified Simplex routine was used as the method of optimization. These optimization studies are detailed in Chapter V. Optimal system performance was accomplished by combining the results from the support evaluations, enzyme reactor configuration comparisons, and the Simplex optimizations. Analytical figures of merit such as dynamic range, detection limit, and precision are presented for each sugar determination in Chapter V. Selectivities of the sugar determinations, as a consequence of enzyme specificity, are also covered in Chapter V. Background information pertinent to the development of the sugar methodologies is presented in Chapter 11 along with results from previous studies in our laboratory relevant to the work described in this dissertation. A general introduction to the various studies performed in this research and the instrumentation involved is given in Chapter 111. Finally, future perspectives of this work are discussed in Chapter VI. CHAPTER II HISTORICAL BACKGROUND Background information pertaining to three unique yet related fields is important in developing the methodologies for the sugar determinations described in this dissertation. These fields are: - Carbohydrate Determinations in Food Analysis - Immobilized Enzymes in Analytical Chemistry - Flow Injection Analysis The following sections present background information in these areas upon which this dissertation builds. A section describing previous, relevant studies in our laboratory is also included. A) Carbohydrate Determinations in Food Analysis Carbohydrates make up a primary class of compounds present in foods along with lipids and proteins. Due to the importance of carbohydrates as a principal matabolic source of energy, their identification and determination within various foodstuffs is essential to many areas of food science (e.g., food design, formulation, preparation, and processing). Traditionally, carbohydrates were determined by difference after the determination of other food components. This methodology provided analysis results representing the total carbohydrate composition and did not allow for individual sugar identification. The traditional approach also suffered from significant analysis errors if the determinations of non- sugar components were inaccurate. With the growth and sophistication of 7 8 food science and technology, more stringent requirements have been placed on the food industry regarding nutrition, composition, and shelf life. The traditional "difference" methods do not meet current analytical demands. Numerous analytical methods are presently employed for the identification and determination of food carbohydrates. These methods are principally categorized as specific and non-specific techniques. 1. Non-specific Methods Non-specific techniques include both physical and chemical methods of analysis (1,2). These methods are classified as "non-specific" because they are generally incapable of distinguishing between individual sugars present in a sample. Physical methods determine some overall feature of the carbohydrate(s) and include such techniques as polarimetry, refractometry, and hydrometry. Physical methods are primarily suitable if only one or two sugars are present in the sample or if a total sugar determination is desired. Chemical methods take advantage of specific functionalities within carbohydrates, exploiting their reducing properties or their ability to undergo condensation and substitution reactions (3,4). Non-specific methods tend to be very time consuming and suffer considerably from various interferences. The interference can result from sugars other than the sugar(s) of interest or from non-sugar components present in the sample matrix. For example, other optically active sample components such as amino acids and glycosides affect polarimetric measurements (1). In addition, acids and inorganic ions interfere with the specific rotation of sugars. Chemical methods generally exploit the reactivity of the carbonyl functional group present in sugars; however, 9 other sample components possessing carbonyl groups can potentially interfere. Thus, for valid use of non-specific analytical methods it is crucial that the number and types of sugars present in the sample be known as well as the composition of the sample matrix. Despite the limitations of non-specific methods, many official analytical methods utilize these procedures including those of the Association of Official Analytical Chemists (AOAC) (5) and the International Commission of Uniform Methods of Sugar Analysis (ICUMSA). 2. Specific Methods In contrast to non-specific methods, specific methods allow for the identification and determination of individual sugars within a mixture. a. Separation Methods Traditionally, specific methods for carbohydrate determinations have employed separation techniques such as paper chromatography (PC), thin layer chromatography (TLC), gas chromatography (GO), or high- performance liquid chromatography (HPLC). Earlier methods that utilized PC and TLC have largely been replaced today by GC and HPLC. The popularity of GC and HPLC methods for carbohydrate determinations is reflected in the abundance of literature that has been published in these areas (610); however, no one method seems to predominate. A primary disadvantage of chromatographic techniques in general is that they often require relatively complicated pre-separation clean-up procedures. These procedures can involve such steps as extraction, precipitation, addition of clearing agents, and ion-exchange. The clean-up procedures are designed to eliminate interferences and also to protect the 10 chromatographic column from compounds that deteriorate column performance. Carbohydrate determinations by GC require a tedious pre- derivatization of the sugars to produce stable, volatile analogs. Among the most common derivatives utilized are acetates, methyl esters, and trimethylsilyl ethers (TMS) (3,7). The TMS derivative is the most popular due to its stability toward thermal degradation. Once derivatized, the sugars are readily separated and the sensitivity can be quite good with flame ionization detection (FID) or mass spectrometry (MS). Separation methods employing HPLC are more popular than GC; however, they too have their drawbacks. Reported HPLC methods utilize a variety of separation mechanisms as well as a variety of detection methods (4,6,10). The most common separation mechanisms are ion-exchange (ll-13), partition (14-16), and adsorption (17,18). Separation by ion- exchange and partition chromatography is readily applied following appropriate clean-up procedures, whereas applications utilizing adsorption chromatography frequently involve pre-derivatization. The most common derivatives contain nitrated aromatics such as nitrobenzoates (17) or p-nitrobenzyloximes (15,19) which absorb in the UV region. The primary weakness of HPLC methods lies in the detection process. Sugars do not lend themselves to direct absorption detection in the visible or UV regions. They absorb weakly in the short wavelength UV region (approximately 190 nm); however, selectivity is poor in this region due to other absorbing components present in the sample matrix. Refractive index (RI) measurement has traditionally been the most common method of detection for carbohydrates. The major limitation of RI 11 detection is low sensitivity, which can require large injection volumes. In addition, refractive index is a bulk physical pr0perty that changes with temperature, pressure, composition, and dissolved gases. Thus, careful regulation of these parameters is necessary. To overcome these detection limitations, pre-column derivatization techniques may be employed to yield products easily monitored by UV-visible absorption or fluorescence. Post-column reactors have also been reported (13,20,21) and are usually based on reactions which yield products that can be colorimetrically or fluorometrically detected. The derivatization techniques improve selectivity and sensitivity over RI and UV (190 nm) detection. However, the pre-column derivatizations can be tedious, and post-column reactions tend to destroy the chromatographic resolution and lengthen analysis times. In recent years, direct electrochemical detection of carbohydrates in HPLC has advanced significantly. The two most common electrochemical techniques employed are pulsed amperometric detection (PAD), usually at gold or platinum electrodes (22-26), and constant potential amperometric detection at chemically modified electrodes (CMEs) (27,28). In CMEs, oxidizable metals or surface bound redox mediators are coated on carbon electrodes to minimize problems with slow heterogeneous kinetics. These electrochemical detection methods have demonstrated superior sensitivity and detection limits when compared to RI and absorption methods. b. Selective Chemical Methods "Specific" techniques that do not require a separation are based on selective chemistries with the sugar(s) of interest. These techniques are generally referred to as biochemical methods since the sugar selectivity is usually the result of a specific biochemical reaction. A majority of 12 biochemical methods utilize enzymes as the selective reagents. Enzymatic methods for carbohydrate determinations in food analysis have been widely published (1,29) and can be found as official methods for several sugars (5). Although enzymatic methods boast great "specificity", the resulting selectivity depends on enzyme purity and the type and number of substrates with which the enzyme can react. Generally, enzymatic methods are selective, sensitive, rapid, and reproducible. Due to the enzyme selectivity, many of the elaborate clean-up procedures required in other methods can be omitted. Metal ions can be problematic for some enzymatic reactions due to enzyme inhibition. Carbohydrate determinations by enzymatic methods are primarily accomplished by determining an equilibrated reaction product directly (or indirectly) or by monitoring the initial reaction rate which is proportional to substrate concentration. Most common are the equilibrium methods that determine a reaction product directly or indirectly. Two popular enzymatic pathways for glucose determination are illustrated in Figure 2-1. Figure 2-1a shows the oxidation of glucose by oxygen in the presence of glucose oxidase to yield gluconic acid and hydrogen peroxide (reaction 1). Glucose determination can be achieved by electrochemically monitoring the disappearance of oxygen or the production of hydrogen peroxide, since both species are electroactive. Alternatively, an indirect method of detection is shown in reaction 2 where the hydrogen peroxide from reaction 1 is used in a second enzymatic reaction with peroxidase and a reduced dye (leuco-dye) to yield a colored product that is determined by absorption. Examples of some popular leuco-dyes are o-dianisidine, benzidine, leucomalachite green, 13 3 A3 A8 3 $5536.— coo—Sucking mlolu .3..anqu 3 R8252“ 0:36.39 .ousEHo 0.83. As 53353.83“. enough you pasta-A owe—unsung .ulm 550E Mama—<2 + oaaaooguoamnoamlc eta-8.21:3 muons .222 + oaoanaoanluloaoogu ouanmmoamlolomoogu A Sada—anon 92 + oaoogu cum + .8982: 1 2.5.3 “.3153; + «can «cum + Boa canoes—u A «c + cocoa—u o-ovio 0-895 3 14 2,4-dichlorophenol, and 4-methoxy-1-napthol (30). The second pathway shown in Figure 2-1b illustrates the reactions catalyzed by hexokinase and glucose-6-phosphate dehydrogenase (3 and 4). In this pathway, glucose and adenosine triphosphate (ATP) yield glucose-6-phosphate which subsequently reacts with nicotinamide adenine dinucleiotide phosphate (NADP). The NADP is reduced to NADPH which can be monitored by direct absorption at 340 nm or by fluorescence at 460 nm with 340 nm excitation. Similarly, other sugars can be determined by enzymatic conversion to glucose followed by subsequent detection via one of the pathways described above. B) Immobilized Enzymes in Analytical Chemistry With the availability of highly purified, active enzyme preparations, enzymatic methods of analysis continue to proliferate and new applications are often reported (31). However, the use of enzymes in their soluble form as analytical reagents has declined due to several limitations. Free enzymes are prone to denaturation and they cannot be recovered from solution or their activity regenerated. These characteristics result in consumption of large quantities of enzyme which can be very costly. Some of these difficulties have been eliminated or minimized with the development of enzyme immobilization methods. Nelson and Griffin (32) were the first to discover the possible advantages of immobilizing enzymes onto an inert support. They observed that following adsorption of invertase onto charcoal, the enzyme retained its activity and could be reused over a long period of time. Since the work of Nelson and Griffin, it has become generally known that the immobilization of enzymes offers three advantages over their use 15 in soluble form (33,34). First, once the enzyme has been immobilized onto a support or carrier, it can be easily introduced into and removed from a reaction mixture, allowing successive analyses to be performed without replenishing the enzyme. This reusability leads to a significant economic advantage in many cases. The second advantage is greater enzyme stability. Immobilized enzymes often exhibit increased stability over wider temperature and pH ranges. This enhanced stability is attributed to the immobilization environment being similar to that found naturally. Although the immobilization process may lead to a loss in enzyme activity, this reduction is generally small. The resulting activity is quite dependent on the choice of support and immobilization procedure; therefore, these are important considerations. Finally, immobilized enzymes appear to be less susceptible to the activators and inhibitors that affect the free forms, resulting in fewer interferences. 1. Enzyme Immobilization Methods The choice of enzyme support and immobilization method is largely dependent on the intended application. Some of the factors that should be considered when choosing a support include protein binding capacity, ease of activation, mechanical and chemical stability, and the interaction of the support with the analyte or sample matrix. Supports generally fall into one of three classes: inorganic matrices, natural polymers, and synthetic polymers (35-37). Each of these support classes has its advantages and disadvantages. Typically, organic polymers are porous and have many sites available for binding. However, they can be subject to microbial attack and can change configuration with pH and solvent conditions. Inorganic matrices do not suffer from microbial attack and are more rigid 16 than organic polymers but many are non-porous and possess a limited number of active sites. Like the numerous supports available, there are also a variety of enzyme immobilization methods. Finding the support-coupling procedure that results in the highest enzyme activity and stability is essentially a trial and error process. There has been a great deal of pioneering work in this area with various enzymes and, fortunately, these studies are well documented (33,38-43). Four methods of enzyme immobilization are conventionally used: physical entrapment, adsorption, cross-linking, and covalent binding. Entrapment methods involve enclosure of the enzyme within an insoluble polymer lattice or encapsulation within a semipermeable membrane. These techniques can be tedious and there may be difficulties with diffusion of the substrate to the enzyme inside the entrapment material or with the enzyme leaching from the lattice. Adsorption is the simplest of the immobilization methods and usually requires the mildest conditions. However, the binding is easily reversed and care must be taken not to desorb the enzyme during analysis. Cross-linking employs a low molecular weight multifunctional reagent to produce intermolecular covalent bonds between the enzyme and reagent. To create an insoluble matrix, it is necessary to polymerize the multifunctional reagent or bind it to an insoluble surface. Cross-linking techniques typically suffer from poor reagent selectivity which results in undesirable intramolecular binding. The most extensively used technique is covalent attachment of the enzyme to an inert support. This type of immobilization eliminates or reduces many of the limitations associated with the other methods. The enzyme insolubilization is not easily reversed by pH, ionic strength, 17 substrate, solvents, or temperature. Also, activity is generally high and stability with time is usually quite good. Among the covalent binding methods, there are numerous choices for supports and binding procedures (33,36,44,46). Regardless of the support choice, the steps for covalent binding are essentially the same. First, the support may need to be activated to provide sites for enzyme attachment. Second, the enzyme must be coupled to the support. It is important that the binding occur at functional groups not essential for the enzyme's catalytic activity. Thus, a favorable immobilization procedure would minimize reactions at the active site. To compensate for the somewhat rigid environment encountered with covalent binding, spacer molecules can be used to increase separation of the enzyme from the support. This allows the enzyme more freedom for conformational changes which may be needed for substrate interaction. 2. Immobilized Enzyme Applications The continued use of immobilized enzymes in analytical methods is reflected in the abundance of publications over the past several years (46,47). Immobilized enzymes are usually employed in one of three classifications for analytical applications: enzyme probes, enzyme reactors, and membranes. a) Enzyme Probes Enzyme probes are devices in which the enzyme is directly affixed to part of the sensing system. Probes in the form of enzyme electrodes are the most prominent type and have been used in both potentiometric and amperometric applications (48-51). Fiber optic probes (optrodes) are a more recent application with immobilized enzymes (52-55). These sensors 18 optically monitor the enzymatic reaction either directly or indirectly, typically via chemiluminescence detection. b) Enzyme Reactors Immobilized enzyme reactors (IMERs) are very popular due to their compatibility with flow systems and virtually any type of detection system (e.g., spectrophotometric, electrochemical, chemiluminescence, and thermal). The most common reactor configurations are open tubular reactors (OTRs), packed bed reactors (PBRs), and single-bead string reactors (SBSRs). In an OTR, the enzyme is immobilized onto the wall of the reactor tube. Common tube materials such as nylon and glass make suitable supports for immobilization. A somewhat different approach to the OTR, termed an embedded reactor, was recently reported (56) where controlled-pore glass (CPG) was physically embedded into the walls of Teflon or Tygon tubing and the enzyme was subsequently immobilized onto the CPG. A packed bed reactor is a column that has been packed with small particles which are the support material for the enzyme. These reactors typically utilize CPG as the support and are similar to chromatographic columns in their design and implementation. The SBSR contains a packing of spherical support material, typically glass beads, which has a diameter 60 to 80% of the tube diameter (57). This criterion results in a single-file packing structure within the tube. Advantages of PBRs and SBSRs over OTRs include increased surface area available for enzyme attachment and also enhanced mixing of the substrate with the enzyme. Glass is the most popular support material employed in PBRs and SBSRs, particularly for use in flowing systems, due to its rigidity and favorable flow characteristics. 19 c) Enzyme Membranes Enzyme membranes are commonly employed as the transducer interface in the enzyme electrode probes discussed above. Materials such as polyacrylamide, starch, polycarbonate, polyvinylalcohol, nylon, and rayon are frequently used as membrane supports for enzyme immobilization. Enzyme membranes can also be used in conjunction with optical detection in methods such as solid surface fluorescence (SSF) (49). In SSF the enzymes and other reagents necessary to catalyze formation of a fluorescent product are lyophilized onto a silicone rubber pad. Following reconstitution, the sample is placed on the pad and the product monitored fluorimetrically. Enzymes have also been immobilized onto various surfaces to form enzyme stirrers and reactor-separators (58). Reactor- separators integrate the aspects of immobilized enzymes and separation membranes. Gas diffusion and ion-exchange are among the most popular separation membranes used in analytical applications. A unique reactor design recently reported combined the characteristics of enzyme membranes and packed bed reactors (59). The enzyme was entrapped within a microporous hollow fiber membrane that was folded and placed into a section of polyvinylchloride tubing. This reactor design was described as having significantly higher enzyme activity per unit reactor volume compared to other configurations. Specific applications employing immobilized enzymes in carbohydrate determinations are presented in the following section. 20 C) Flow Injection Analysis There exists an ever-increasing need for rapid, precise analytical methodology in modern industrial and academic laboratories. Flow injection analysis (FIA) has proven to be a highly versatile continuous flow analysis technique that meets many of todays analytical demands (60). First reported by Stewart (61) and Ruzicka and Hansen (62), the desirable characteristics of FIA include fast sample throughput, low reagent and sample consumption, and improved precision over manual methods. In addition, FIA methods are readily automated from the sample preparation and manipulation to the in-line detection. The basic principles of FIA involve the reproducible injection of a fixed sample volume into a carrier stream, generally propelled by a peristaltic pump. As the sample zone is transported to a detector, many chemical and/or physical operations may take place within the FIA manifold. Most important analytically, is the controlled dispersion that occurs as the sample zone progresses downstream. The dispersion processes are highly reproducible with constant flow parameters. The degree of dispersion is affected by several factors such as tube diameter, tube length, flow rate, and flow characteristics. Several types of channel geometries have been employed for the control of dispersion processes in FIA. The simplest channel geometry is a straight OTR which results in characteristic laminar flow. More popular are coiled OTRs where dispersion is decreased due to secondary flow (radial mixing) induced by coiling (63). Reactors packed with inert particles such as glass also break up laminar flow. Packed bed reactors and SBSRs (57) are common configurations found in many FIA manifolds. The dispersion and flow theory for these reactor designs have been 21 investigated and compared (64-68). Recently, some unique and interesting reactors designs have been reported where the reactor tubing is woven or braided (69,70). The braiding creates secondary flow patterns and leads to decreased dispersion. In addition to the various manifold designs investigated, alternative approaches to traditional FIA flow patterns have been explored. Two such alternatives are flow reversal and flow recycle techniques (71-74). These techniques allow the sample zone, or portions thereof, to pass through the detector several times by reversing the flow or by recycling the zone. This multidetection of the sample zone provides significantly more chemical and physical information than a single pass while still requiring only one detector. Another alternative flow mode in FIA is merging zones (75). In merging zone FIA, the sample and reagent are both injected as zones into the carrier and merged together downstream. This leads to less sample and reagent consumption than normal mode FIA, where reagents are introduced continuously. Flow injection has evolved into an extremely versatile technique, capable of carrying out many in-line operations for sample pretreatment, manipulation, and analysis (76,77). The technique is also powerful due to its compatibility with many types of detection (78,79). The number of FIA publications has grown exponentially since its introduction (80). Table 2-1 shows areas of application where FIA has become popular and also summarizes several of the possible in-line operations and compatible methods of detection. Chemical reactions and derivatizations are among the most common operations carried out in FIA. They are accomplished by means of soluble (homogeneous) and immobilized (heterogeneous) reagents (81). 22 Eon—3033 Eo§A15nu Bog—.35 EE\.€§3E Brogan as E§1> 3033900 BAPBOHSOQ £53.58 55088384 dog 25 m2 .151:— 1 noggin 133033.15" anon-5.5025 and 18.323. gfiafinoodoooum 13355 E canouolnou "5:059:25 cadence—madman". aofioufingQ-doflodom gugog 30.59an “—21.55" 1358 Fafifiofiohoonm nonna— 3.3335 25.502 33.1.3.5 324 5.3032“ o—Aflcnaoo FOE dofiuoznn< .4: mo hamaanuob and unofldofi—Q: .ulN Baa. 23 Immobilized reagents are generally used in the form of reactors placed in the FIA manifold and can carry out operations such as reductions and ion- exchange. Applications utilizing immobilized bioreagents have become very popular due to the inherent selectivity these reagents provide (82). Enzymes, bacteria, and immunological reagents have all been used in reactor columns. Applications employing IMERs are by far the most common of the bioreagent type. Various configurations of IMERs in conjunction with FIA have been characterized with respect to enzyme kinetics and dispersion (83-87). With the advantages that FIA and immobilized enzymes provide, there have been many methods developed for carbohydrate determinations employing a variety of detection techniques. Generally, three FIA configurations are most often reported for carbohydrate methods. The first configuration employs IMERs in the FIA manifold to carry out the desired enzymatic reaction(s) coupled with optical detection. The second type also uses IMERs in the manifold but is coupled with electrochemical detection. The third configuration uses enzyme electrodes where the enzymatic reaction(s) and detection occur at the electrode surface. Slight variations on this third type may incorporate the addition of IMERs or the use of multi-enzyme electrodes. There have been many reported carbohydrate applications, primarily in the clinical and food analysis areas. Table 2-2 lists several of the applications published over the last few years. 24 .O I 8 a2 .3: .wflesdwuflo. 38¢ mam u 93.385 :93» N3 0.1533036 mam noon 030382—54 Ada-am . I 380A 2: man as: 9 man 2.30 84 Sufi» 0.80 2: nlnlh .3! $5 .8 mum coon 030830534 0.30 0803 no ow ovohooa as 138 2.3038233 0-0098 8 manual. gen .3: man 1 «3.68884 .3525 .8 on don—8 mum 6.5m «33829522.. 8823 8 cm .8 mam I oruofloaonnohoonm 0.32% 8 co Eb dam vac—m 01A! ounce-03:30.8 080:8 3 «a: .8 man .85.. v88 “.3333; «80E.» 8 8-3838 I u .8 mam .595 83.516 93.88834 none...» 8 .3. .3.“ .8 a o «a .o.$i¢oi§ >3 mam a 0388335 Nashua 0303 8 5.— .nnmlouc .uom mum coon «3.58:: «3&2 . 0-3 8 .oaesoommcyowia «mam e8.— ofioasoaaohoonm “gm.“ 8 2 .3 an: .5. man u 933338.588 «unma— uo man—lulu .Nom mam conga 31.101 03.830153 . cocoa—u REP—01m 1% 05 55.23.: .3303.5 .196 ~323— .aofldfla.~30§ 3935:2110 canon—loan“; vouaaoa§}oflooq5 FOE .NIN Hams— 25 33hwt§mriawuam 0000.00000Io0 0050050. I «DI 00000003 I 0..— 0003uou0u I 00: 0000—00 00000.0 I mm 00000300— 000020 I 0000000000400 000008 I ha 5802.508 31300:.886 u Emu-I0 00002 0030109 I Hal—00 .3080200I0I8893 u .7...- . ”Na 8 00000009 I 0303000034 00003». NM. mac 00000003 000.0 0.0.5 0503000034 000055 3. 0.7070 ~30 0300 0303000034 0030000 on. 00 .00 00000009 I 0303000034 000008 0. 0 am 00obm00Jm 330m 0003 0303000034 00000.“. 0: mac 00000009 I 0303000034 00009.0 s... 00 00038.0 0000 00503000034 000030 - 0.0 8 000.3009 I 0303000034 000008 a: 09 00000008 I 0303000034 000003 I 0: ca .00! .00— 0000000." I 0303000034 0020M a. 0 .00 .8 00000009 300m 000.0 0303000034 000098 N0 0 8 0000530! 00.50 I 0303000034 00006 . . 00000 0 u. 8 00: >3 000 3300 tom 0303000034 00008 3 0 00 000 0300.0 000.0 00503000034 00000.!0 00. 000 00.. I 030880080 080%» 00. on .00 000 I 000003000034 0030000 8.03.5.3: o 00 n.6— Wfl Ifipuglg o‘choo—fl a 30300 84 000%” 80 an: .8 0005003 I 000003000034 000005 I 00. 00 000.3005 00—300 80m .0000... 0503000034 0000M“. 3. 0.. .00....5 00.. a... 65.5 3.0068384 ohm“ 003000008 0300.5 0000. 003002004 0030300 0096 030030 .7303 3033030300 300000000000 0002703030 0005403330300?! F05 .NIN H.540 26 D) Previous Work in Our Laboratory Previous studies in our laboratory conducted by Thompson (124) and Stults (125) were relevant to the design and development of the carbohydrate methods described in this dissertation. Their investigations resulted in methodologies for covalent immobilization of glucose oxidase and enzymatic determination of glucose with subsequent optimizations. 1. Glucose Oxidase Immobilization Initially, Thompson immobilized glucose oxidase onto nylon tubing in the form of an OTR which was inserted in a continuous flow system for gluwse determination. Various immobilization methods were investigated for covalent attachment of glucose oxidase to nylon, and linkage via glutaraldehyde provided favorable results. Glutaraldehyde, a bifunctional molecule, not only acts as the binding reagent, but also functions as a spacer between the enzyme and support, thereby allowing more conformational freedom for enzyme activity. Preliminary studies were also carried out in the immobilization of glucose oxidase onto non-porous glass beads. Enzyme attachment via glutaraldehyde following surface modification with 3-aminopropyltriethoxysilane (3-APTS) was very successful for glucose oxidase. This procedure is also advantageous due to the mild reaction conditions required when compared to other common immobilization methods such as cyanogen bromide and diazotization. Both the cyanogen bromide and diazotization methods involve hazardous chemicals, harsh conditions, and can result in poor enzyme activity. Enzyme reactor lifetimes have also been quite good with the 27 3-APTS/glutaraldehyde immobilization method, on the order of several months in many cases (43). The work of Thompson (124) was continued by Stults (125) with the immobilization of glucose oxidase on non-porous glass beads in the form of a SBSR for flow injection determination of glucose. The immobilization procedure developed by Thompson was re-evaluated and optimized by Stults. Conditions were investigated for optimum surface area and glass derivatization with 3-APTS, factors that eventually limit the number of sites available for enzyme binding and subsequent activity. Preliminary studies were also conducted for galactose oxidase immobilization onto the non-porous beads for flow injection determination of galactose. 2. Glucose Determination Both Thompson and Stults developed methods for enzymatic determination of glucose via glucose oxidase. Thompson employed a glucose oxidase OTR in an air-segmented continuous flow system (126) and Stults used a glucose oxidase SBSR in a flow injection system (96). The same method of detection, shown in Figure 2-2, was used in both systems. Hydrogen peroxide produced by the glucose oxidase reaction was colorimetrically determined in the presence of peroxidase by the Trinder reaction (127). In this reaction, 4-aminoantipyrine (AAP) and 3,5-dichloro- 2-hydroxyphenyl sulfonic acid (DCPS) are oxidized and coupled to form a quinoneimine dye. The reaction occurs rapidly with the resulting dye possessing a molar absorptivity of approximately 2.2 x 104 at 510 nm (128). The high molar absorptivity, stability, and non-toxic nature of the Trinder reaction species made it a favorable choice over other oxygen acceptors such as benzidine, o-toluidine, and o-dianisidine. The Trinder defies?" pov£\ omeNo $836 .N-N mgUE 3 $8 ”no #2 8 a 01 .r m z a a AWomuadluREon / O + O + 0 mm . - m . o:ofia—.m6:8:—m-fl smog—mAHA 9mm «mo ..Omam + oumm + ‘12\ 8 2 30 m Noam + cam + so + m 09:58 swoon—m mo moan 29 reaction performed very well under the instrumental conditions described by Thompson and Stults; however, limitations were encountered in this researcher's investigations and are discussed in more detail in Chapter IV. 3. Optimizations In an effort to attain maximum sensitivity and sample throughput, optimization of the glucose FIA manifold was performed by Stults (96). In addition to univariate studies, a composite modified Simplex (CMS) routine was used for the optimization of seven system parameters. The parameters included variables affecting the glucose oxidase reaction as well as the Trinder reaction, namely, carrier flow rate, carrier pH, temperature, peroxidase concentration, AAP concentration, DCPS concentration, and reagent pH. The CMS optimization resulted in an improvement in response by a factor of 22.5 over initial conditions. Multivariate optimizations provided by CMS are also used in the methods developed and described in this dissertation. A more detailed discussion of the principles and background of CMS is found in Chapter V. CHAPTER III INSTRUMENTATION AND ANALYZER MANIFOLD ' DEVELOPMENT A brief discussion of the general instrumentation used and also the considerations necessary for development of the sugar determination methodologies is presented in this chapter. Specific instrumentation and developmental studies are detailed in later chapters. A) Introduction and Instrumentation The determination of each individual sugar is accomplished in a separate FIA manifold developed specifically for that sugar. It is therefore possible for the sugar determinations to be independently optimized with respect to sensitivity, stability, and sample throughput. The enzymatic approach for the six sugars of interest is shown in Figure 3-1. The sugars B-D-glucose and D-galactose yield hydrogen peroxide directly upon oxidation reactions catalyzed by glucose oxidase and galactose oxidase, respectively. Sucrose, lactose, and maltose are all disaccharides that, upon hydrolysis with the appropriate enzyme, yield at least one D-glucose unit. Fructose undergoes an isomerization reaction via glucose isomerase to give equilibrium concentrations of D-glucose and D-fi'uctose. The D-glucose formed by these enzymatic reactions can subsequently undergo oxidation with glucose oxidase for production of hydrogen peroxide. The hydrogen peroxide formed as a result of the enzymatic reaction schemes is then utilized as a means of indirect sugar detection. 30 31 «cum + once—«Gonoalsoflauln .7 .38vo. E523 snag—nu .uln 55¢: ocean—ml: ii cacao—rain cocoa—ulalu + cocoa—uln Agnesooua nu can + 3333 uncaoauaUInln + swoon—ml: lea—.3383... n M can + 330s— omouosuula + manna—ml: A 23839: A u a cam + 3225 cum + so + unsoaiunn 2.3.5 oat-1o «cum + £3 0382?: A 2.38 885 can + so + «confininln 32 In the course of this work, two detection reactions were investigated, the Trinder reaction (Figure 3-2a) and the malachite green reaction (Figure 3-2b). Initially, the Trinder reaction was employed as the detection method due to its favorable characteristics and previous use in our laboratory (discussed in Chapter II section D). However, limitations were encountered with the Trinder reaction, and the malachite green reaction was investigated as an alternative method of detection. Studies were conducted with these two detection reactions, and the results are presented in detail in Chapter IV. A single channel FIA system was used in developing the methods for individual sugars. Figure 3-3 shows the general instrumentation employed for determination of the six sugars. A 12-channel peristaltic pump (Ismatec) propels the carrier and reagent streams. The sugar sample is injected into the carrier stream by means of a pneumatically actuated 6-port valve (Rheodyne) with a 30 ul sample loop. As the sample zone is transported downstream, the appropriate enzymatic reactions take place to yield hydrogen peroxide. Following the formation of hydrogen peroxide, the reagent stream is introduced into the carrier, and the detection reaction takes place in a plain SBSR. After dye formation, the stream enters a flow-through filter calorimeter designed by Patton and Crouch (129). Data acquisition and sample injection are controlled by an IBM PC compatible computer. In an effort to minimize dispersion and also to enhance reagent mixing, a plain SBSR is used as the "reaction coil" for the detection reaction. By minimizing dispersion, sensitivity and sample throughput are improved. The SBSR configuration also enhances mixing of the reagents. Dispersion comparison studies, as shown is Figure 3-4, were 33 ..flofloaom “3300qu .Nln guru noonu 333180092 n 05 N c m + 82o 3381: A Sauna: £2 28...? squasaumnumuotoaoannd u m8: ca. Afiflsoflaalw u 3 Cum + can ogoaoflna A; Swanson «cum + 2.: “aofioaom 5.95 3303.5 3 ~on + mac: + .3. "aofioaom nova; Ad 34 9cm [—‘—'1 1r , w M E f m" Bosctorb) SBSR “J, ‘ Pump ‘ Detector FIGURE 3-3. General instrumentation for sugar determinations. 35 conducted for a staight OTR, a coiled OTR, and a SBSR reaction coil. Phenol red dye (16 uM) was injected into an FIA manifold where 50 cm lengths of each reactor configuration were placed in line and monitored at 540 nm. The carrier and sample diluent was borate buffer adjusted to pH 9.5. No chemical reaction took place; thus, the resulting dispersion is due solely to physical processes occurring during sample transport. The SBSR configuration shows superior dispersion characteristics over the OTR and coiled OTR due to disruption of the laminar flow profile. The specific FIA manifolds will differ for the six sugars due to the various enzymes required. I. Glucose and Galactose Determinations Flow injection manifolds constructed for glucose and galactose determinations are very similar in that only one enzyme is needed for production of hydrogen peroxide. Therefore, only a single enzyme reactor, glucose oxidase for glucose and galactose oxidase for galactose, is needed in the FIA channel. The instrumentation used for these two sugar determinations is shown in Figure 3-5a. 2. Sucrose, Lactose, Maltose, and Fructose Determinations For the other four sugar determinations, the manifolds are essentially the same as that for glucose except the appropriate glucose- forming reactor must be placed in line prior to the glucose oxidase reactor, as shown in Figure 3-5b. For example, lactose determination requires a B-galactosidase reactor inserted prior to the glucose oxidase reactor for the hydrolysis of 'lactose into D-glucose and B—D-galactose. Absorbance Absorbance Absorbance FIGURE 3-4. 0.24 - 0.20 4 0.16 :1 0.12 L 0.08 J 0.04 :1 36 —888R 0.00 J 0.24 - 0.20 ~ 0.10 l 0.12 i 0.011 1 0.04 : 0.00 d 0024 - 0.20 d 0.1 6 . 1 0.12 - 0.08 .. .1 0.04 s 0.00 d I ' I ' l ' r I l t l ' 1 ' l ' r ' I 'fi 5 25 45 as as 105 125 145 105 105 205 Time (sec) Dispersion comparisons for a single-bead string reactor (SBSR), a coiled open tubular reactor (OTC), and a straight open tubular reactor (OTR). 37 PC/X'l‘ Ff.) l Glucose Oxidase m 111.1.“ Gelactoe: 021011011 P1151513 hm 1‘- 1188th 4J0] hen! 1 Pump 0 Detector PC/XT U 1[ ll ‘ ‘ Ina-em Glucose , 01 0211111111 P111111 :4. 33""- :m ”Em m... . hull ' ‘ Pmp V b 0111110101 FIGURE 3-5. Instrumentation for glucose and galactose determination (a); and for sucrose, lactose, maltose, and fructose determination (b). 38 B) Manifold Development In the design and optimization of the FIA manifolds for determining the six sugars, desirable characteristics such as high selectivity, sensitivity, and sample throughput were set as goals in developing the methods. There are a number of system factors which ultimately influence the attainment of these goals, and those investigations can be separated into the following areas: Enzyme Immobilization Reactor Configurations Enzymatic Reaction-Rate Parameters Interferences 1. Enzyme Immobilization Immobilization of several different enzymes is required in determining the six sugars of interest. Although an immobilization procedure developed previously by Stults (125) worked well for glucose oxidase on non-porous glass beads, activities were poor when this procedure was used for other enzymes. Thus, the immobilization procedure was investigated further and modifications were made to improve immobilization efficiencies with glass supports. Optimal conditions for immobilization were evaluated based on support pretreatment, support silylation solvent, and silylation atmosphere. Difl'erent types of glass supports were also investigated for their potential enzyme loading capacities. Various dimensions of controlled-pore glass and non-porous glass beads were compared. These studies are detailed in Chapter IV. 39 2. Reactor Configurations The design of the immobilized enzyme reactors (IMERs) is also an important consideration in developing the FIA manifolds. Different reactor configurations result in different chemical and physical processes within the FIA manifold. Packed bed reactors, SBSRs, and embedded reactors were investigated and compared for each enzyme. Reactor characteristics were studied for systems with and without enzymatic reaction to gain insight into the dispersion contributions from both chemical and physical processes. Reactors were evaluated based on their resulting activity and the dispersion introduced. The optimal design is determined by sensitivity and throughput considerations. The experiments and results of these comparisons are discussed in Chapter V. 3. Enzymatic Reaction-Rate Parameters There are several factors that influence the kinetics of enzyme catalyzed reactions. These include substrate concentration, enzyme concentration, the presence of activators or inhibitors, temperature, pH, buffer composition, and ionic strength. An enzyme will exhibit optimal activity at either a specific value or over a range of values for each of the above variables. These factors must be considered in developing the sugar methodologies since they will ultimately affect the selectivity, sensitivity, stability, and sample throughput. Each FIA manifold was optimized independently for such system variables as pH, residence time, activator concentrations, and reagent concentrations. A composite modified Simplex procedure was employed as a means of multivariate optimization for these variables. The Simplex routine utilizes a defined response function where system factors are 4O weighted and expressed in a mathematical equation. System factors considered in optimizing the sugar determinations were sensitivity, precision, and sample throughput. A description of the Simplex optimization and results from these investigations are presented in Chapter V. 4. Interferences Unfortunately not all enzymes are specific for one substrate. Interferences caused by enzyme non-specificity were investigated. Only those substrates that eventually yield hydrogen peroxide will exhibit interferences in the sugar determinations. Evaluation of interferences resulting from enzyme non-specificity are presented in Chapter V. CHAPTER IV OPTIMIZATION OF ENZYME IMMOBILIZATION AND SUPPORT CONSIDERATIONS Immobilization of several enzymes is required in developing the analytical methods for the six sugars of interest. As previously stated, the 3-aminopropyltriethoxysilane (3-APTS) / glutaraldehyde method of enzyme immobilization has been successful with several enzymes, particularly on controlled-pore glass supports. However, this method of immobilization has seen limited use for applications with non-porous glass supports. Thompson (124) and Stults (125) used the method successfully to immobilize glucose oxidase onto non-porous glass beads. Although the immobilization procedure developed through the investigations of Thompson and Stults worked well for glucose oxidase, activities were poor when this procedure was used for other enzymes on non-porous glass beads. In an effort to improve immobilization efficiencies for enzymes other than glucose oxidase, the various steps of the immobilization process were investigated further and modifications were made to improve enzyme loadings. Optimal conditions for immobilization were evaluated based on support pretreatment, support silylation solvent, silylation atmosphere, and glutaraldehyde solvent. As a result of the improved immobilization efficiencies, limitations of the Trinder reaction were encountered with glucose oxidase. The malachite green reaction was investigated as an alternative detection reaction and compared to the Trinder Reaction. 41 42 Consideration was also given to supports other than non-porous glass beads. Controlled-pore glass (CPG) was examined as an alternative enzyme support, used in a packed bed reactor (PBR) configuration. Various types of CPG were compared with respect to particle size and mean pore diameter as well as various reactor dimensions. A) Enzyme Immobilization The general steps involved in the immobilization of enzymes onto glass supports via 3-APTS and glutaraldehyde are illustrated in Figure 4-1. The first step, prior to actual immobilization steps, may include procedures to clean the support surface for removal of impurities. Also, support pretreatments can include methods to increase the surface area and initial reactivity. For glass supports, the surface reactivity is increased as the concentration of silanol groups increases relative to siloxane groups. Following any pretreatment steps, the support is derivatized via 3-APTS, yielding reactive amino groups on the glass surface. Glutaraldehyde, a bifunctional aldehyde, is then used as a link between the alkylamine modified surface and the enzyme. One aldehyde group reacts with the amino group on the support surface while the other reacts with various amino acid functionalities within the protein structure of the enzyme such as the amine groups of lysine and histidine and the phenol group of tyrosine (130). The enzyme is thus covalently bound to the glass support, and the glutaraldehyde carbon skeleton provides conformational freedom for substrate interaction. The immobilization procedure developed by Stults (125) for attaching glucose oxidase onto non-porous glass beads is shown in Table 4-1. Several details of this procedure are noteworthy. In an effort to 43 1) Pretreatmut: - 01111111111; - increase Surface Area - Increase Surface Reactivity (silanol groups) 2) Activate Surface: Convert surface silanol groups to active poops for enzyme attachment. Silylation: (S-APTS) °CH20H3 l—OH + (CH3CH20)3Si(CH2)3NH2 —-> l—O —Si(CH2)3NHZ (1)012043 Glutaraldehyde: . l l . t )—o— s|i(¢H2)3NHz + H (CH2)3 H ——> 1-0- 5|i(CH2)3N=CH(CH2)3 3) Enzyme Attachment 1 I— O — 5:1(C1'12)3N=CH(CH2)3E'1 + HzN—E 'V I— 0— S|i(CH2)3N=Cl-I(CH2)3CH=N—E FIGURE 4-1. General procedural“. for 3- rLglutaraltlehyde immobilization of enzymes onto glass suppo 44 TABLE 4-1. Initial enzyme immobilization procedure. Step Reagent Time Temperature(°C) Cleaning alcoholic 30 min 23 ............. KOH rinse 320/ acetone 23 dry N2 23 Etching saturated (5% w v) 1 hr 23 (muss/1100‘ dry N2 ' 23 heat 3 hr 450 Silylation 1% (v/v) so min 23 3-AP'l‘S/ acetone curing 15 hr 70 rinse acetone/1120 w 23 Glutaraldehyde 1% WV) 3 hr 23 glutaraldehyde/ buffer pH 3.00 rinse buffer pH 6.35 23 Glucose Oxidase 445 U/ 5.0 ml 24 hr 4 buffer pH 6.35 45 increase surface area, the beads were etched by means of a procedure initially developed by Onuska (131) for etching the inner surface of glass open-tubular capillary columns. Onuska's procedure was altered very little for etching the non-porous beads. As part of the etching process, the beads were sealed in a glass tube prior to heating (450°C). Stults obtained the best results for silylation when the 3-APTS solution was allowed to evaporate completely from the support in a hood at room temperature before curing at 70°C. Following silylation, however, the beads were stuck to the bottom of the glass container. Through a trial and error process, it was discovered that the beads could be released by rinsing with acetone and gradually adding small quantities of water. 1. Glucose Oxidase Immobilization The immobilization procedure developed by Stults, summarized in Table 4-1, was initially used to immobilize glucose oxidase onto non-porous glass beads (0.6 mm diameter) for employment in a single-bead string reactor (SBSR). The SBSR, illustrated in Figure 4-2, was constructed by aspirating glucose oxidase immobilized beads into a 10.0 cm length of Teflon tubing with an inner diameter (id) of 0.81 mm. The tube ends were then crimped to contain the beads. A typical glucose calibration curve employing a 10.0 cm glucose oxidase SBSR is shown in Figure 4-3. Glucose standards (0.5 - 6.6 mM) were prepared fi'om a 0.01 M glucose stock solution that also contained 0.004 M benzoic acid as a preservative. Three replicate injections (30111) were made for each glucose standard. The sample diluent and carrier was 0.05 M phosphate buffer (pH 6.85). The reagent solution for the Trinder reaction consisted of 1.0 ml aliquots of both a 0.01 M stock 46 0.6 mm d 0.81 mm id FIGURE 4-2. Single-bead string reactor. 47 0.30 0.25 .1 0.20 - 0.15-4 Absorbance 0.10- 0.05 4 0.00 .,-,- I ,fir 0.0 1.0 2.0 310 1 410 510 6.0 7.0 Glucose (mM) FIGURE 4-3. Glucose calibration curve with Trinder reaction detection. 48 solution of 4-aminoantipyrine (AAP) and a 0.01 M stock solution of 3,5-dichloro-2-hydroxyphenyl sulfonic acid (DCPS). To the AAP and DCPS, 8.0 mg of horseradish peroxidase (EC 1.11.1.7, Sigma Type II, 220 U/mg) was added and diluted to 10.0 ml with 0.05 M phosphate buffer (pH 6.85). The detection reaction took place in a 40 cm plain SBSR followed by absorption measurement at 510 nm. Carrier and reagent flow rates were 0.50 and 0.05 ml/min respectively. Standard error bars are also shown for the glucose calibration curve. Relative standard deviations (RSDs) ranged from 1.34% to 3.58% over the glucose concentrations used in this study. The non-linearity observed in the glucose working curve was also observed by Stults and is discussed in more detail later. 2. Galactose Oxidase Immobilization The immobilization procedure used for glucose oxidase was also investigated for galactose oxidase. The procedure, shown in Table 4-1, was followed up to the enzyme immobilization step where 450 U of galactose oxidase (EC 1.1.3.9, Sigma) was reacted with 0.6 g of beads in 5.0 ml of 0.05 M phosphate buffer (pH 6.05). Preliminary investigations by Stults (125) for galactose oxidase showed a substantial reduction in activity relative to glucose oxidase. Thus, a 50.0 cm galactose oxidase SBSR was prepared for evaluation instead of a 10.0 cm reactor. Before discussing the galactose oxidase immobilization results, a few details of the enzymatic reaction mechanism and its characteristics should be presented. The implementation of immobilized galactose on'dase for galactose determination in various applications has proven to be somewhat complicated with respect to enzyme activity and stability (97,132). In these reports, galactose oxidase reactors were observed to lose significant 49 activity over a relatively short period of time, consequently limiting their usefulness. This activity loss is thought to be a result of the enzymatic reaction mechanism for galactose, 02, and galactose oxidase. Each molecule of enzyme contains one copper atom which plays an integral role in the catalytic action. Mechanistic investigations suggest that the active enzyme begins with Cu(III) which is subsequently reduced to Cu(I) during the catalytic cycle (133). Ideally the enzyme returns to the Cu(III) state in the final step of the cycle; however, a side reaction can occur resulting in the formation of Cu(H), which is believed to be inactive. To help eliminate this enzyme inactivation with time, a redox couple such as hexacyanoferrate(III) / hexacyanoferrate(II) can be used to mediate the copper oxidation states and maintain enzyme activity (97,133). A solution of 50 uM hexacyanoferrate(III) in 0.05 M phosphate buffer (pH 6.85) was used to activate the galactose oxidase SBSR prior to evaluation by pumping the solution through the reactor for 30 min. This activation step has been previously reported for galactose determinations with galactose oxidase (97,132). The carrier solution contained the redox couple hexacyanoferrate(III) / hexacyanoferrate(II) in a ratio of 10:1 with a total concentration of 4.4 uM in 0.05 M phosphate buffer (pH 6.85). These carrier conditions were reported to yield adequate enzyme activity and stability for galactose oxidase immobilized onto CPG and employed in a PBR configuration (97). Three galactose working solutions (2.0, 3.0, and 4.0 mM) were prepared from a 0.01 M stock solution and diluted with 0.05 M phosphate buffer (pH 6.85). All other experimental parameters were the same as those described above for glucose determination. The results from the galactose oxidase SBSR evaluation are shown in Figure 4-4. Enzyme activity was very poor for galactose oxidase when 50 0.05 0.04 - 0.03 - 0.02 - Absorbance 0.01 4 0.00 . , . , , r 0.0 1.0 2.0 3.0 4.0 5.0 Galactose (mM) FIGURE 4-4. Galactose calibration curve with Trinder reaction detection. 51 compared to glucose oxidase, particularly considering the galactose oxidase SBSR was five times longer. Standard error bars are again given for the three galactose solutions. Precision values for the galactose measurements were extremely poor, with RSDs ranging from 3.15% to 19.0%. Significant loss of enzyme activity was also observed, despite the presence of the redox couple in the carrier solution. B) Immobilization Optimization The various steps involved in the enzyme immobilization procedure, Table 4-1, were examined individually for modifications in an effort to improve immobilization efficiencies on non-porous glass beads. These studies are presented in the order they were conducted, not in the order dictated by the procedure itself. 1. Etching In the process of immobilizing glucose oxidase on a particular batch of non-porous beads, the glass tube employed in the etching process was not completely sealed prior to heating to 450°C. The activity exhibited by this batch of glucose oxidase beads was significantly higher than fiom any previous immobilizations. A comparison was then made for beads sealed and not sealed in the etching process. Following the etching step, a direct microscope observation showed a marked difference in "whisker" formation on the surface. The beads that were not sealed appeared to have a much rougher surface. After glucose oxidase attachment and evaluation, a 2.2 mM glucose solution gave an absorbance of 0.15 for a SBSR (10.0 cm) containing beads from the sealed tube while a SBSR containing beads from the open tube gave an absorbance of 0.30. This 52 increase in etching may be explained by considering the reaction equilibrium: (NH4F)HF 41—.» ZHF + NH3 The gaseous products are contained within the sealed tube while in the open tube they can escape, shifting the equilibrium in favor of the products. This results in more silica whiskers being formed on the glass surface. Another consideration is that support surface water is removed under the open tube conditions whereas it is trapped in the sealed tube. The presence of surface water may inhibit the etching process. An open glass tube was therefore employed in the etching step for all subsequent enzyme immobilizations. 2. Silylation Solvent Comparison The silylation step in the immobilization procedure is extremely important with respect to eventual enzyme loadings. The extent of silylation or surface activation ultimately controls the number of sites available for glutaraldehyde linkage and subsequent enzyme attachment. Due to the importance of this step, support silylation conditions were investigated in an efi‘ort to improve the efiiciency of this reaction. An abundance of literature is available on the preparation of bonded stationary phases for chromatography in which glass supports, usually CPG, are modified with various organosilanes (134-137). An investigation of this literature revealed three common solvents utilized for CPG silylation: acetone, tetrahydrofuran (THF), and toluene. A comparison study was conducted for these three silylation solvents with 3-AP'I‘S. It 53 was thought that less reactive solvents such as toluene and THF may provide a better reaction environment for silylation than acetone. Non-porous beads (0.6 g) were cleaned and etched as described above. Prior to silylation, the beads were divided into three 0.3 g quantities and placed in small glass vials. Each 0.3 g quantity was then reacted with 50 pl of 3-APTS in 5.0 ml of acetone, toluene, or THF under a N 2 atmosphere. An inert atmosphere such as N2 was thought to provide more consistent reaction conditions between solvents. The solvents were allowed to evaporate completely under N2, followed by curing at 70°C for 16.5 hr. Afier curing, the beads were rinsed with their appropriate solvent and loosened from the vials by a methanol/H20 mixture. Glutaraldehyde linkage was accomplished by reacting each 0.3 g quantity with 200 pl glutaraldehyde in 5.0 ml of 0.05 M phosphate buffer (pH 8.00) for 3.5 hours at room temperature. The beads were then rinsed with 0.05 M phosphate buffer (pH 6.85) and reacted with 25 mg glucose oxidase (17.8 U/mg) in 5.0 ml of pH 6.85 buffer for 24 hours at 4°C. Finally the beads were rinsed with buffer and packed into three 10.0 cm SBSRs. The reactors were evaluated in the FIA system by injecting standard glucose solutions (30 ul) ranging in concentration fi'om 1.1 to 4.4 mM. All other experimental conditions were held constant. The carrier and sample diluent was 0.05 M phosphate buffer (pH 6.85). Flow rates for the carrier and reagent were 0.50 and 0.05 ml/min, respectively. The Trinder reagent solution was prepared as described previously in section A1. Figure 4-5 shows partial calibration curves obtained for the three reactors. Beads prepared from the acetone silylation solvent exhibited higher immobilization efficiencies relative to toluene and THF, Absorbance 0.20 0.15 0.10 0.05 0.00 54 e Acetone a Toluene ‘ - ms .1 ' 1 1 l ' r ‘ I r 0.0 1.0 2.0 3.0 4.0 5.0 Glucose (mM) FIGURE 4-5. Silylation solvent comparison for non-porous beads. 55 as indicated by the glucose sensitivities. The less reactive solvents did not appear to improve surface activation with 3-APTS. 3. Pretreatment, Silylation, and Glutaraldehyde Activation Studies investigating support pretreatment, silylation atmosphere, post-silylation support rinse, and glutaraldehyde activation were conducted simultaneously. These four aspects of the immobilization procedure are briefly discussed followed by the experimental methods and results. Support Pretreatment: As previously stated, any support pretreatment that will increase the surface activity will ultimately improve enzyme loadings. Activity of glass supports can be improved by increasing the concentration of surface silanol groups. This can be accomplished by washing the glass in an aqueous acid solution, as shown in Figure 4-6. Under acidic conditions, surface siloxane groups are converted (hydroxylation) to silanol groups; this process is reversed (dehydroxylation) via dehydration at elevated temperatures (above 200°C). Effects of acid washing the non-porous beads were examined. An activity comparison was made for beads that were acid washed in concentrated HCl following the etching step versus beads that were not washed. Silylation Atmosphere: A complex sequence of reactions is possible when reacting glass surfaces with 3-APTS (138). Which reactions predominate depends on the availability of protons in the silylation solvent (essentially H2O). The most common bonding models for trialkoxysilanes with surface silanol groups 56 .Eslfimsuo actsEnougnoEcofiwaoémn suntan mama .w-v Ema _oco=m mcoxo=m Io Ia 008m / a \o/ 1l__m|ol__m|. 4 I o I + |.__mlo|__ml. 57 are shown in Figure 4-7. For conditions where the solvent is free of protic impurities, the top two bonding models are possible. In these cases, no hydrolysis takes place between neighboring ethoxy groups, preventing polymerization of the silane. This result is very desirable in the preparation of bonded-phase chromatography supports such as octadecylsilane (ODS) supports. When protic impurities are present, the bottom two bonding models predominate. ' Here the ethoxy groups are hydrolyzed and extensive polymerization of the 3-APTS can occur. The polymerization should result in higher yields of active amino groups when compared to conditions where polymerization is prevented. Thus, protic impurities may be beneficial in surface activations for enzyme immobilization applications. However, there is a limit to the extent of desirable polymerization. It is essential that the polymerization take place on the support and not in solution. Investigations of the effects of protic impurities were accomplished by reacting the beads with 3-APTS in ambient air and under N2. Dry acetone was used as the silylation solvent in both cases to minimize the solvent itself as a proton source. Post-silylation Support Rinse: In the original immobilization procedure, the silylation solvent was completely evaporated in a hood prior to curing at 70°C. The resulting beads adhered to the glass container after curing. Before glutaraldehyde activation, it was necessary to release the beads by rinsing with acetone and gradually introducing small quantities of H20. This loosening process was imprecise and slight variations resulted in poor reproducibility for enzyme immobilization efficiencies. 58 Aprotic Solvents: I ?CH2CH3 -—Sli—OH + 3-APTS ——> —Si-O-Si(CH2)3NH2 I . ' CHZCHg — i—OH — i 3 + 3—APTS ——> 3m0>3i (5 O‘Si/(CH2)3NH2 —s'i-0H —S'i/O/ \9 | I —Si(CH2)5NH2 FIGURE 4-7. Common bonding models for 3-APTS. 59 In an effort to improve silylation reproducibility, the beads were loosened after solvent evaporation and thus remained loose after curing. The original silylation procedure required a rinse to free the beads; however, the modification that now prevents the beads from sticking does not require a rinse. Enzyme activities for beads with rinses of acetone and ethanol were compared to no rinse prior to glutaraldehyde linking. Glutaraldehyde Solvent: Thus far, the glutaraldehyde reaction had been conducted in 0.05 M phosphate buffer (pH 8.00) and appeared to work quite well. Other workers have reported glutaraldehyde activation in an ethanol medium for glass supports (92). A comparison was made between phosphate buffer (pH 8.00) and absolute ethanol as the glutaraldehyde reaction solvent. Experimental Methods and Results: Experimental conditions were designed such that investigations of the four immobilization parameters could be conducted simultaneously: support pretreatment (surface activity), silylation atmosphere, post- silylation support rinse, and glutaraldehyde solvent. Figure 4-8 illustrates the immobilization scheme used for simultaneous investigation. The beads were initially cleaned in concentrated HN03 rather than alcoholic KOH. Concentrated HN03 was observed to be superior for cleaning; the beads appeared almost white following HN 03 wash, while they were slightly yellow after alcoholic KOH. A 1.5% (v/v) solution of 3-APTS in dry acetone was used for each aliquot of beads. The acetone was allowed to evaporate completely under ambient air or' N2 over a period of five hours. Following solvent evaporation, the beads were loosened from the glass containers and cured at 70°C for approximame 15 hours. After curing, the beads were rinsed as 60 .36 o.» ..3 85m 2334 _ Jung—cu ovhfiogaggu was .35." anon—man 6.343383 down—ham 35.53593 “.8998 NO GOBGUMHQOPA: nfioflsvaamn ROM OEOAOD aOngmmmfloag .QI* EDGE .36 .36 0.6 .3 o.» 3 35m 3.5. oz 2.223 _ N2 .2 _ $573 8.33.6 g _ GI 02 35.3... 3:..sz Ana—coon; m 021 43—0 #30 .3—0 $3.0 0.6 .3 .66 6.6 .3 od .3 35m 35m 35m 35m oc0+oo< ICE 2.2004 105 ...4 N2 _ 3:73 3.33.6. _ Lac; .0: L 61 indicated in the immobilization scheme. Glutaraldehyde linkage was accomplished by reacting a 13.0% (v/v) solution of glutaraldehyde in either 0.05 M phosphate buffer (pH 8.00) or absolute ethanol. Glucose oxidase (89 U/ml) was finally reacted with the beads in 0.05 M phosphate buffer (pH 6.85) Single-bead string reactors (10.0 cm) were prepared from beads for each of the seven experimental conditions and evaluated by injecting standard glucose solutions (30 ul) ranging in concentration from 1.0 to 5.0 mM. All other FIA parameters were held constant. The carrier and sample diluent was 0.05 M phosphate buffer (pH 6.85) with a carrier flow rate of 0.50 mllmin. The Trinder reagent was prepared as described previously (section A1) with a flow rate of 0.05 mllmin. Figure 4-9 shows partial glucose calibration curves for the seven reactors. The most notable effect on enzyme loadings was observed for the acid wash pretreatment. Reactor activities were substantially higher for those beads that had been acid washed prior to silylation. Thus, there is a great advantage in maximizing the surface silanol groups prior to surface activation with 3-APTS. The activation atmosphere did not appear to play a critical role in the immobilization efficiencies. The use of dry acetone versus non-dried seemed to be more important in the silylation reaction than an ambient versus inert atmosphere. It is also important to realize that silylation under N2 was not completely free from ambient air since the reactions were not conducted in a sealed apparatus such as a glove box. Support rinse comparisons showed that an acetone rinse yielded better enzyme loadings than an ethanol rinse, but no rinse at all appeared to be superior with respect to reactor activities. The reactor with the 62 :9“. :9“. ..z .6: o... .3 :06 ..z .6: o... :3 3332 :2 .6: 32 od :3 2332 ..z .6: 02 o... .3 .85: oz ..2 .6: 32 od :3 2332 ..z .6: od :3 6332 c3 .6: ‘IOOO-l-D .3828 6352332» 98 .35» 98.33 .239883 "832.36 35.53563 €233 you 3865388 .8682 mg... 33-23% 6.6 Ema Canny omoogu o.m 0+ own o.~ oh, 1 m6 5.0 aounqiosqv 63 poorest activity was that which used ethanol as the glutaraldehyde solvent. Therefore, phosphate buffer (pH 8.00) was significantly better for the glutaraldehyde reaction. Substrate Conversion: As a measure of the improvement in glucose oxidase immobilization on the non-porous beads, substrate conversion from glucose to hydrogen peroxide was evaluated for the reactor prepared under initial immobilization conditions and that prepared under the optimized conditions. A plain 10.0 cm SBSR was inserted in the FIA manifold and a calibration curve was constructed for injections of standard hydrogen peroxide solutions. With injections of hydrogen peroxide, only the Trinder reaction occurs in the manifold. Hydrogen peroxide working solutions in the range of 0.1 to 4.0 mM were prepared from a 0.0600 M stock solution (standardized against KMnO4) and diluted with 0.05 M phosphate buffer (pH 6.85). A plot of hydrogen peroxide concentration versus peak area was used as a means of calculating substrate conversion for glucose. The glucose oxidase reactors were then placed in the FIA manifold and glucose solutions (2.0 - 4.0 mM) were injected under the same experimental conditions as for hydrogen peroxide. Areas of the glucose peaks were calculated and compared to the equivalent peak area for hydrogen peroxide. Percent conversions for the reactors were 10% for the initial immobilization procedure and 46% for the procedure under optimized conditions. 64 4. Optimized Immobilization Procedure The optimized conditions resulting from the immobilization investigations are shown in Table 4-2. Cleaning: . The initial support cleaning was modified to concentrated HN03 from alcoholic KOH due to the superiority of HNO3. Rinsing and drying steps after support cleaning remained the same as in the initial procedure. Etching: This procedure remained essentially the same with respect to solutions, reaction times, and temperatures. The only modification was to use an open tube in heating the beads to 450°C, which resulted in improved surface etching. Surface Activity: . This step, which did not exist in the initial procedure, was added to enhance the concentration of surface silanol groups via acid washing. After one hour of acid wash, the beads were rinsed with distilled water, acetone, and dried under N2. Silylation: Several modifications were made to this step. A 1.5% (v/v) 3-APTS in dry acetone was used in the investigations described; however, later studies suggested that a concentration of 2.0% (v/v) in dry acetone improved surface silylation of non-porous beads. Although studies indicate that silylation can take place in ambient air or under N2, the inert N 2 atmosphere was usually used for subsequent enzyme immobilizations. The acetone evaporation was allowed to take place over a period of five hours rather than 30 minutes. The beads were then loosened from the 65 TABLE 4-2. Optimized enzyme immobilization procedure. Step Reagent Time Temperature(°C) Cleaning HN03 30 min 23 rinse 320/ acetone 23 dry N2 23 Etching saturated (5731 1 hr 23 (mm) )mI Neat) dry N2 23 heat (open tube) 3 hr 450 Surface Activation HCl 1 hr 23 Silylation 2% (v/v) 3-AP'l‘S/ 5 hr 23 ............... dry acetone curing 15 hr 70 Glutaraldehyde 13% (v/v) 3 hr 23 glutaraldeh e/ 15 hr 4 buffer pH .00 rinse buffer pH 6.85 23 Enzyme appropriate 24 hr 4 enzyme 66 glass container prior to curing at 70°C. After curing, the beads were not rinsed. Glutaraldehyde: A solution of 13% (v/v) glutaraldehyde in 0.05 M phosphate buffer (pH 8.00) was used for linking the enzyme to the support. Although the immobilization studies allowed this reaction to proceed for three hours at room temperature, in later studies an additional period of approximately 15 hours at 4°C yielded improved activities. Thus, three hours may not be enough time for complete reaction. A specific investigation of glutaraldehyde reaction yield versus reaction time was not conducted. An additional fifteen hours is probably not necessary for complete reaction; however, it was usually convenient to allow the reaction to proceed overnight at 4°C. The entire modified immobilization procedure takes four days to complete whereas the initial procedure required three days. The substantially longer glutaraldehyde reaction time is primarily responsible for the additional day. C) Trinder Reaction versus Malachite Green Reaction As a result of the improved immobilization efficiency for glucose oxidase, a significant deviation from linearity was observed for glucose calibration curves (Figure 4-9). This non-linearity was also observed prior to the immobilization procedure investigations, although much less pronounced. Kinetic investigations of the Trinder reaction (139) show that the initial rate becomes a nonlinear function of hydrogen peroxide concentration at moderate concentrations of hydrogen peroxide 67 (approximately 0.2 mM) for a peroxidase concentration of 0.8 mg/ml (1760 U). There is also a strong dependence of the reaction rate on pH for the Trinder reaction. The slope is quite steep for initial rate versus pH in the pH range of 6.00 to 7.50 with a maximum at approximately pH 7.75. Unfortunately, the strong pH dependence occurs within the typical pH values used for the sugar determinations (6.00 to 7.00). Due to the limitations of the Trinder reaction which were more pronounced after increasing the immobilization efficiencies, an alternative detection reaction was sought. The malachite green (MG) reaction shown in Figure 4-10 was investigated as a means of detection. In this reaction, leucomalachite green (LMG) is oxidized by hydrogen peroxide in the presence of peroxidase to malachite green which has an absorption maximum at 620 nm. In contrast to the Trinder reaction (Figure 2-2), the reaction stoichiometry is 1:1 for LMG and hydrogen peroxide where the Trinder reaction stoichiometry is 1:1:2 for AAP, DCPS, and hydrogen peroxide respectively. The MG and Trinder detection reactions were investigated by comparing calibration curves for both hydrogen peroxide and glucose. A stock solution was prepared by dissolving LMG (0.009 M) in 6.5 N acetic acid. The reagent solution was then prepared by combining 8.0 mg peroxidase (1760 U), 0.5 ml LMG stock, and diluting to 10.0 ml with 0.05 M phosphate buffer (pH 6.85). The Trinder reagent solution was prepared as described previously (section A1). Hydrogen peroxide calibration curves were constructed by injecting 30 ul of standard peroxide solutions (0.05 - 1.0 mM) into the FIA manifold. A plain 10.0 cm SBSR was employed in the manifold in place of an IMER. Absorbance values were obtained for the peroxide standards under MG and Trinder detection at 620 and 510 nm, 68 3.332 :36 33332 .2... "EDGE 86.5 3.5332 56.5 33338834 «AmmUvJfi + NAmmUvz __ can“ + Qolglzfimov Al N a. 33383 .6 + o m + 0 23.3.8. 69 respectively. The carrier and sample diluent was 0.05 M phosphate buffer (pH 6.85) with a carrier flow rate of 0.50 mllmin. Detection reagent flow rate was 0.05 mllmin. Hydrogen peroxide calibration curves employing Trinder and MG detection are shown in Figure 4-11a. Malachite green detection is substantially more sensitive for hydrogen peroxide relative to the Trinder reaction. The hydrogen peroxide calibration slope for MG detection was over three times that for Trinder detection, and MG detection also displayed linearity to approximately 2.0 absorbance units. The negative deviation from linearity which occurs around 2.0 absorbance units is primarily due to stray light limitations from the detector. Calibration curves for glucose standards were obtained under the same experimental conditions as hydrogen peroxide, except that a glucose oxidase SBSR (10.0 cm) was used in the manifold. Figure 4-11b shows glucose calibration curves for the Trinder reagent and LMG reagent. Again, the malachite green detection exhibits linearity to approximately 2.0 absorbance units. Glucose sensitivity was approximately eight times greater for MG detection over Trinder detection. As a result of these initial investigations, the malachite green reaction was selected as the detection reaction for further studies. D) Preliminary Support Investigations Following the modifications to the immobilization procedure described above, several studies were conducted with respect to the enzyme support characteristics. Reproducibility of glucose oxidase immobilization on the non-porous beads was investigated utilizing the new procedure. In addition to non-porous beads, CPG was also examined as a support for enzyme immobilization. Various aspects of the CPG 70 2.8 J I MG 242 O Trinder 2.03 1.6~ 1.2- Absorbance 0.8 -‘ --. 0.4- 0.0 . , . , . , . l . T . 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Hydrogen Peroxide (mM) 2.0 I MG 0 Trinder 1.6- 1.2-1 0.8 - Absorbance 0.4-1 0.0 1.0 2fo ' 3T0 ' 4.0 i 5.0 6.0 Glucose (mM) FIGURE 4-11. Comparison of Trinder and malachite green (MG) detection for hydrogen peromde (a) and glucose (b). 71 support such as particle size and mean pore diameter were investigated and compared. The CPG was employed in a packed-bed reactor (PBR) configuration in the FIA system. Various reactor dimensions were also compared for glucose determination with glucose oxidase. 1. Immobilization Reproducibility In order for enzyme activity to be reproducible from one reactor to another, where the reactors contain beads from separate immobilizations, the immobilization procedure must be reproducible with respect to reaction yields at each step. The silylation step in both the initial and modified immobilization procedures is the least reproducible from a reaction yield standpoint. The silylation reaction yields were particularly inconsistent when the acetone/H20 rinse was required to release the beads from the glass container. This step was eliminated in the modified procedure and therefore, the silylation reaction yield should be more reproducible. An investigation of silylation reproducibility was conducted by comparing three sets of beads prepared simultaneously. The beads were treated in the same reaction vessels up to the silylation step where 0.33 g quantities were placed in separate vials. The modified immobilization procedure conditions (Table 4-2) were used for silylation and glutaraldehyde linkage. Glucose oxidase (140 U/ml) was reacted with each 0.33 g quantity of beads for 24 hours at 4°C. The beads were packed in 10.0 cm SBSRs and evaluated by injecting glucose standards. Calibration curves for the three sets of beads are shown in Figure 4-12. Two of the reactors exhibited nearly identical activities, while the activity for the third reactor was slightly less. The relative standard deviation (RSD) for 72 1.0 ' I Slope - 0.66 A/mM I Slope - 0.65 A/mM A Slope = 0.56 A/mM 0.8 - Absorbance 0.2 4 0.0 r l V I I l l U 0.0 , 0.4 0.8 1.2 1.6 2.0 Glucose (mM) FIGURE 4-12. Enzyme immobilization reproducibility for glucose oxidase. 73 the three calibration slopes was 8.8% for the modified immobilization procedure, whereas the initial immobilization procedure gave a RSD of 45.2%. The modified silylation step appears to be more reproducible than the initial procedure as indicated by the resulting enzyme activities. 2. Investigations using Controlled-pore Glass Although the SBSR configuration has several advantages when employed in a FIA system (e.g. low pressure and reduced dispersion), the surface area available for enzyme attachment is quite low for non-porous beads. The etching process helps to increase the surface area; however, the beads remain non-porous. A substantial gain in surface area is obtained if a porous support, such as CPG, is used for enzyme immobilization. A packed bed reactor configuration was investigated as an alternative reactor design. Two enzymes, glucose oxidase and B-galactosidase (lactose determination), were immobilized onto CPG and non-porous beads to compare the PBR and SBSR configurations. In immobilizing enzymes onto CPG, the support particle size and pore diameter are important considerations. Generally, as the particle size decreases, the dispersion in the PBR also decreases. However, the pressure created across the reactor increases as particle size decreases. Thus, for low pressure systems such as in FIA, there is a compromise in particle size considerations between dispersion and pressure. As the pore diameter decreases, the specific surface area increases, theoretically yielding higher enzyme loadings. However, as the pore size decreases, steric hindrance may prevent the immobilized group (enzyme) from entering the pore. Therefore, the minimal pore size that can be utilized depends on the size of the enzyme itself. Typically, the enzyme molecular 74 weight is used as an indicator of molecule size for appropriate pore size selection. Several CPG particle sizes were available in our laboratory with two pore sizes. An intermediate particle size of 120/200 mesh (74 - 125 um) was selected for glucose oxidase and B-galactosidase immobilizations. This particle size was available with 327 A and 544 A pore diameters. The molecular weights for glucose oxidase and B-galactosidase are approximately 152,000 and 540,000, respectively. According to the manufacturer, both the 327 A and 544 A pore sizes should accommodate the enzymes (140). The immobilization procedure used for both enzymes onto the CPG varied slightly from the modified procedure used for non-porous beads. Primarily, the changes are due to the substantial increase in surface area available on CPG. It is also possible to filter CPG as a means of separating the support from reaction solutions. Thus, filtering steps were incorporated where appropriate. Cleaning: The CPG was initially cleaned in concentrated HN03. The acid also functioned to increase the surface silanol groups. The support was then rinsed with distilled water followed by acetone and then filtered. In an effort to release surface water, the support was heated to 150°C for approximately 15 hours. Silylation: The 3-APTS concentration was increased to a 10% (v/v) in dry acetone for CPG, whereas a 2% (v/v) solution was used for non—porous beads. The silylation reaction proceeded for five hours under a N2 atmosphere at ambient temperatures. The activated CPG was then rinsed 75 with dry acetone, filtered, and dried. Like the non-porous beads, the CPG was cured at 70°C for 15 hours. Glutaraldehyde Activation: A 10% (v/v) solution of glutaraldehyde in 0.05 M phosphate buffer (pH 8.00) was reacted with the CPG for 4.5 hours followed by rinsing in the appropriate buffer for enzyme immobilization. Details for each enzyme attachment and results of the reactor comparisons are presented below in separate sections for glucose oxidase and B-galactosidase. a. Glucose Oxidase Glucose oxidase was immobilized onto 120/200 mesh CPG with 327 A and 544 A pore diameters (0.5 g each) using the procedure described above. For enzyme attachment, 3150 U glucose oxidase in 5.0 ml of 0.05 M phosphate buffer (pH 6.85) was reacted with the two types of CPG for 24 hours at 4°C. The PBRs were constructed by packing the CPG, via syringe, into Teflon tubing (0.81 mm inner diameter). The CPG was contained within the reactor by a polycarbonate membrane placed between two low- pressure fittings, as shown in Figure 4-13. Initially, PBR support lengths of 5.0 cm and 10.0 cm were employed in the FIA system with a carrier flow rate of 0.50 mllmin. However, the pressures created with these lengths were too great for the peristaltic pump employed. A support length of 2.6 cm was found to be the practical limit. This corresponds to a support volume of 13.4 pl. - Glucose oxidase reactor comparisons were made between two 2.6 cm PBRs (327 and 544 A) and a 14.0 cm SBSR. A SBSR of 14.0 cm was chosen as an intermediate reactor length that would provide adequate substrate conversion with a residence time comparable to the PBRs. Total manifold 76 .338 338.— .3; 3338 .674 go: 2.63503 coin 29.59323. l I I l / 9.300.. «053.... 9.0 2:39... 33 + * .o 2 EE 5 30C 77 lengths and other system parameters were held constant for the comparisons. The carrier and sample diluent was 0.05 M phosphate buffer (pH 6.85) with a carrier flow rate of 0.50 mllmin. The detection reagent was prepared as described previously (section C) and was propelled at 0.05 mllmin. Glucose standards ranging in concentration from 0.05 to 0.30 mM were injected (30 pl) into the FIA system. Figure 4-14 shows sensitivity and dispersion comparison results for the glucose oxidase PBRs and SBSR. Glucose calibration curves are shown in Figure 4-14a for the three reactors. The PBR containing 327 A CPG shows the greatest glucose sensitivity followed by the 544 A CPG and lastly the SBSR. The superior sensitivities of the PBRs over the SBSR is not surprising due to the greater surface area provided by the CPG supports. Difl‘erences in reactor sensitivities were not substantial despite the very different support characteristics; particularly comparing the non-porous beads to the CPG. Sensitivity for the 327 A PBR was 1.4 times that for the SBSR while the 544 A PBR was 1.2 times more sensitive. Figure 4-14b illustrates the dispersion characteristics for the reactor configurations. Absorbance and variance values are given below for the three reactors (Glucose 0.30 mM). Absorbance Variance (32) PBR 327 A 0.984 216 PBR 544 A 0.831 240 SBSR 0.690 163 While differing in absorbance (sensitivity) for 0.30 mM glucose injections, both PBRs exhibit similar peak shape characteristics which is evident in 78 1.2 - PER 327 {x 0 PER 544 A A SBSR 1J3- 013- cl 0J5— J 0u4— q 052- I DID V T I r T 1 T l 0.00 0.05 0.10 0'15 ' 0'20 0'25 0.30 ' 0.35 Glucose (mM) Absorbance 1-0 " PBR 327 A Absorbance I I T I I T 5 25 45 65 85 105 125 145 165 185 205 Time (sec) FIGURE 4-14. Reactor sensitivity (a) and dispersion (b) comparisons for a SBSR and two PBRs (327 A ,544 A) containing glucose oxidase. 79 the variance values. In comparison, the SBSR shows superior dispersion characteristics over the PBRs. b. B-Galactosidase Reactor comparison studies for B-galactosidase were conducted similarly to those for glucose oxidase. The immobilization procedure was the same as described above, using 0.5 g of each CPG (120/200 mesh, 327 A and 544 A pore diameter). B-Galactosidase (100 U/ml; EC 3.2.1.23, Sigma Grade VI) was reacted with each CPG type in 0.05 M phosphate buffer (pH 7.3) for 24 hours at 4°C. Two B-galactosidase PBRs (2.6 cm) were constructed, one containing 327 A pore CPG and the other containing 544 A pore CPG. The reactors were then evaluated by injecting (30 pl) a 1.0 mM lactose solution. A glucose oxidase SBSR (14.0 cm) was inserted following the B-galactosidase reactor for production of hydrogen peroxide. Prior to evaluating the influence of pore size on B-galactosidase activity for the two PBRs, carrier conditions were investigated with respect to buffer pH and the presence of Mg”. B-Galactosidase is reported to be activated by Mg+2 ions (141). The LMG detection reagent was prepared as described previously (section C) with a flow rate of 0.05 mllmin. Carrier flow rate was 0.46 mllmin for the each of the conditions investigated. Activity results for the 2.6 cm PBR (544 A pore) are shown in Figure 4-15 for three carrier conditions. The poorest activity resulted when 0.05 M phosphate buffer (pH 7.3) was used as the carrier. Activity improved by a factor of 1.4 when the buffer pH was reduced to 6.6. A further increase in activity was observed for a carrier pH of 6.6 with a 0.01 M concentration of Mg"2 (prepared from a 1.0 M MgClz stock solution). The resulting activity was 2.3 times that for the pH 7.3 carrier without 80 Absorbance 0-0 _ I I I I I I r 5 25 45 65 85 105 125 145 165 185 205 225 Time (sec) FIGURE 4-15. B—Galactosidase activity comparison for three different carrier conditions. 81 Mg”. This study emphasizes the importance of optimizing the enzymatic reaction-rate dependent parameters such as carrier pH and activator concentrations. These optimizations are presented in detail in Chapter V. A comparison study for the two CPG pore diameters (327 A and 544 A) with B-galactosidase exhibited no significant difference in enzyme activity. A 1.0 mM lactose solution gave an average absorbance of 1.307 for the 544 A pore CPG while that for the 327 A pore CPG was 1.292. The essentially equivalent activities for the two pore sizes was not expected; however, it is not surprizing that the difference for B-galactosidase would be less than for glucose oxidase. The molecular weight for B-galactosidase is nearly five times that of glucose oxidase, and the dependence of enzyme loading on pore size would be expected to be less for B-galactosidase. A more detailed investigation of CPG characteristics with respect to particle size and pore diameter is presented in the next section. Figure 4-16 shows sensitivity and dispersion comparisons for a 2.6 cm PBR (327 A pore diameter) and a 14.0 cm SBSR for B—galactosidase. Experimental conditions were similar to those given above for the glucose oxidase reactor comparisons. However, for the B-galactosidase comparisons, the carrier contained 0.01 M MgClz in 0.05 M phosphate buffer (pH 6.60) and had a flow rate of 0.57 ml/min. Sensitivities for the PBR and SBSR are shown in the calibration curves of Figure 4- 16a. The PBR configuration exhibited a significantly greater enzyme activity with a sensitivity 1.9 times that of the SBSR configuration. Dispersion characteristics for the two reactors are shown in Figure 4-16b. Specific absorbance and variance values are given below for a 2.0 mM lactose injection. 82 - PBR 327 A i I SBSR Absorbance ‘1 , . 0.0 0.5 1f0 135 2.0 2.5 Lactose (mM) Absorbance O m l 0.4-4 04) _ I I I I 1' r 5 25 45 65 85 105 125 145 165 185 Time (sec) FIGURE 4-16. Reactor sensitivity (a) and dispersion (b) comparisons for a PBR (327 A) and SBSR containing B-galactosidase. 83 Absorbance Variance (92) PBR 327 A 1.348 129 SBSR 0.707 120 Variances for the PBR and SBSR configurations are very comparable. Thus, dispersion characteristics for these two B-galactosidase reactors are not significantly different. However, the PBR configuration exhibits much greater lactose sensitivity relative to the SBSR. 3. Reactive Amino Group Determinations As a measure of support enzyme loading capacity, the determination of reactive amino groups following support activation with 3-APTS was explored. The extent of support silylation has a direct effect on resultant enzyme loadings. Therefore, a quantitative evaluation of surface amino groups may aid in the interpretation of apparent enzyme activities observed for various silylation conditions and also various glass supports. a. Determination Method The method employed for amino group determination following support silylation was originally developed by Mottola and Snelling (142). This method involves a non-destructive "on/oft" chemical procedure where a chromophoric probe (p-dimethylaminocinnamaldehyde) is attached to the silylated support surface and then detached under different experimental conditions. Spectrophotometric measurement of the detached probe leads to quantitative evaluation of the aminated glass surface. p-Dimethylaminocinnamaldehyde (DACA) was chosen as the chromophoric probe because its reactivity with surface amino groups is 84 similar to the reactivity of glutaraldehyde, which is used for enzyme attachment in the immobilization process. The procedure developed by Mottola and Snelling was followed in the amino group determinations and is outlined below. Attachment of Probe: Aminopropyl activated glass (50.0 mg) was combined with excess DACA (40.0 mg) in 20 ml of a reaction solvent consisting of 1x10'3 M piperidine in anhydrous ethanol. After one hour of contact, the unreacted DACA was aspirated off and the support was washed several times with 10 ml portions of anhydrous ethanol. The support was then dried under N2 flow. Detachment of Probe: Immediately after drying, the reacted glass (10.00 mg) was combined with 25.0 ml of a hydrolysis solution (95% ethanol). Hydrolysis of the DACA was allowed to proceed for one hour at 40°C, with periodic shaking. Measurement of Detached Probe: Absorption of the detached DACA was measured at 390 nm. A 1.00 mM stock solution of DACA in 95% ethanol was used for further preparation of working solutions ranging in concentration from 1.0 to 75.0 pM. The working solutions were used to construct a calibration curve at 390 nm for determining detached DACA. By appropriate calculations the concentration of active amine groups is determined yielding mmol NHg/g support. b. Support Silylation Investigations Three separate studies were conducted for investigating various aspects of the silylation procedure and also support characteristics. These 85 three studies are first described and then results are presented and discussed. Study 1: Support Pretreatment As reported previously, support surface reactivity is improved by increasing the concentration of surface silanol groups via acid washing prior to silylation. This was evident in the results obtained from support pretreatment investigations presented in section B3 (Figure 4-9). The effects of acid washing, discussed earlier, were evaluated based on the final apparent enzyme activity for glucose oxidase. This final evaluation is not as direct as investigating immediately following support silylation. Using the DACA probe directly after silylation, acid wash effects on resulting surface amino groups were examined. In addition to support reactivity (acid washing), support surface hydration was also investigated for the silylation reaction. Comparisons were made for supports where surface water had been removed by heating (150°C) and for supports which were hydrated (not heated) prior to silylation. The support hydration studies were conducted to investigate the effects of surface water on the silylation reaction in which 3-APTS (in dry acetone) reacts with surface silanol groups. In the support pretreatment studies, four 0.25 g quantities of 120/200 mesh CPG (327 A pore diameter) were prepared as follows: Surface Surface B |° 'I H l |° 0.25 g Acid Wash 25°C (hydrated) 0.25 g Acid Wash 150°C (dehydrated) 0.25 g No Wash 25°C (hydrated) 0.25 g No Wash 150°C (dehydrated) 86 Each 0.25 g quantity of CPG was activated with a 10% (v/v) solution of 3-APTS in dry acetone following the pretreatment steps. After support curing, the procedure described above was used for amino group determination. Study 2: Various Support Investigations A variety of glass supports were available in the laboratory for enzyme immobilization. Throughout previous studies, only three support varieties had been utilized: CPG (120/200 mesh, 327 A), CPG (120/200 mesh, 544 A), and non-porous glass beads. Table 4-3 lists the assortment of supports available and their characteristics. A comparison was made for these supports with respect to particle size and pore diameter. The amino group determinations were therefore a measure of the effects of support dimension on the extent of surface silylation and subsequent enzyme loadings. Two of the available supports had already been activated with 3-APTS when purchased. Thus, a comparison could be made between commercially activated glass and laboratory activated glass to determine if any advantage was gained in purchasing pre-activated supports. For the unactivated supports, 0.25 g quantities were pretreated by acid washing and were also hydrated prior to reaction with 3-APTS. Support silylation was conducted as described previously using the procedures developed for CPG (page 74) and non-porous beads (Table 4-2). Study 3: Various Organo-silane Comparison All support activations had thus far employed 3-APTS as the silylation reagent. With the availability of protons, the bonding characteristics for 3-APTS (Figure 4-7) allow for extensive polymerization, which is believed to be advantageous in enzyme immobilization 87 A0333; «2: <2 I mad I I can seem unouomIaoz Auoaoo-ovaoboeflv Hum... {38: co 8.. S «a 33 8768 3683583..." :68 om«\cc H mm A 3w «z 4 <2 I . 8 65.516 2 can mm“ 2. £608 ocm\cm— «am. a m. I A-oudS—oaaohoofiv an m 36 vs. .5 A68 oc¢\ccm flu. a pm a. 0 .vv 1 Audion—oaseboofiv h 3.. z. 326 6838 No.5 . .v I 786. m an N. awn mud .2. £608 ccm\cmu Au\aaov Au\m8v AN 3 A3 3.50863 3:: 98996 o§_o> do: aoflsflbna ouom 803 68:83 noes 0.8m 385m 0.6m o—ofiuem 2833:3288“ panda you 3563.305: 3.89:5 .mlw mama. 88 applications. The effects of silane polymerization were investigated for CPG by comparing the following silanes: 3-aminopropyltfiethmsilane (Sigma) 3-aminopropylmethyldiejhggysilane (Petrarch Systems) 3-aminopropyldimethylethgxysilane (Petrarch Systems) These silanes provide three, two, and one reactive ethoxy groups respectively for silylation. The triethoxysilane would be expected to exhibit the most extensive polymerization while the diethoxysilane could exhibit slight polymerization, and the ethoxysilane should yield no polymerization. The support employed in this study was 120/200 mesh CPG (327 A pore diameter) which had been acid washed and was hydrated prior to silylation. Three 0.20 g quantities of CPG were reacted with a 10% solution of the appropriate silane in dry acetone and subsequently cured at 70°C for 15 hours. 0. Results and Discussion Amino group determinations for the three silylation and support investigations are shown in Table 4—4. Before discussing the individual results, a comment should be made about the precision of the determinations. The amino group determination variance is a reflection of both the reproducibility of the silylation reaction and the reproducibility of the determination method. A precision study was not specifically conducted for the amino determination method, although Mottola and Snelling reported good reproducibility (RSD of 2%). As an approximate indication of precision, the support and experimental conditions marked by an asterisk (*) in Table 4-4 were constant in all three investigations. 89 TABLE 4-4. Amine functional group concentrations on glass support surfaces. Support Pretreatment: (mo/zoo mesh. 327 i. 3-ms) pmol/g CPG ‘ Acid. 25°C 170 Acid. 150°C 164 No Acid. 25°C 156 No Acid. 150°C 150 Various Supports: (3-AP'I‘S) umol/g Support I 120/200 mesh 327i 153 120/200 mesh 544i 129 zoo/400 mesh 544i 119 120/200 mesh 350 i(3-sminopmpy1) 155 80/ 120 mesh 5441(3-aminopropyl) 46 Non-porous Beads . 14 Various sashes: (120/200 mesh. 327 i) umol/g CPG ‘ 3-aminopropyltriethoxysilane 189 3-aminopropylmethyldiethoxysilane 108 3-aminopropyldimethylethoxysilane 112 90 The amino group concentrations range from 153 to 170 umng CPG in these three experiments with a standard deviation of 9.5 umol/g (RSD of 6%). Although this discrepancy in amino group determination is not great for the common experimental conditions (*), a direct comparison and interpretation of determination results between studies should be made with caution. However, comparisons can be made within each study and general trends interpreted. The results of the first study (support pretreatment) indicate that an acid wash does indeed improve the surface reactivity of glass supports by increasing surface silanol groups. Both CPG samples which were acid washed exhibited greater amino group concentrations than those that were not washed. The extent of surface hydration on the glass support may possibly have a small effect on the silylation reaction. Glass which was hydrated (25°C) prior to silylation, resulted in slightly higher amino concentrations than dehydrated (150°C) surfaces. However, the differences observed in amino group concentrations for hydrated versus dehyrated surfaces was less than the standard deviation calculated from the common experimental conditions (*). Without further investigation, it cannot be concluded that surface hydration enhances silylation. Results of the second study (various supports) essentially reflect the dependence of amino group concentration on support surface area. The two supports with the greatest surface areas (120/200 mesh 327 A and 350 A) exhibited the highest amino group concentrations with 153 and 155 umng support, respectively. Also noteworthy was the agreement in amino group concentration between the commercially activated support (120/200 mesh 350 A) and the laboratory activated support (120/200 mesh 327 A). Although there is a slight difference in pore diameter, this should 91 not make a significant difference in the amino group concentration. Surface areas for the 544 A pore diameter supports (120/200 and ZOO/400 mesh) are identical and results show comparable amino group concentrations. As expected, the non-porous beads exhibited a substantially lower amino group concentration than the CPG supports. This is due to the much smaller surface area available on the non-porous beads relative to CPG. The third study (various silanes) reveals that the trifunctional triethoxysilane does result in extensive polymerization as reflected in the amino group concentration. The di- and mono-functional silanes both gave similar determination results and thus exhibited no polymerization. Although the di-functional silane has two reactive sites available, the most likely bonding characteristics would only yield a monolayer coverage of the support surface (see Figure 4-7). It is possible to form a dimer, but this condition is self terminating and would not lead to polymerization. 4. Reactor Dimension Comparisons: Glucose Oxidase Although the packed bed reactor configuration performed very well in initial investigations for glucose oxidase and B-galactosidase, the pressure associated with this design limited the maximum practical reactor length. As previously stated, a reactor length limit of 2.6 cm (13.4 1.11) was observed for 0.81 mm id tubing with 120/200 mesh CPG before leak problems occurred with the low pressure connections and peristaltic pump. Alternative reactor dimensions were considered in an effort to gain larger reactor volumes and improve substrate conversion efficiencies. 92 Flow in packed bed configurations is described by the Darcy equation (143): , u = linear flow velocit ELLE—PL) B0 = permeability coefficient (Bler) “L B = specific permeability or = total column porosity P, = inlet pressure F = nrzeru P0 = outlet pressure L = column length _ “1‘23 (P, ' Po ) n = viscosity (poise) — 11L F = volumetric flow rate r = column radius Rearranging, the pressure gradient can be expressed in terms of various column parameters. FnL 1rr2B (Pi 'P0) = AP The specific permeability term (B) is proportional to the square of particle size and also dependent on the interparticle porosity. Since the CPG support dimensions remain constant, term B is essentially constant. All other terms, 1], F, L, and r, can be considered in an effort to decrease the pressure drop, AP. Lowering the solution viscosity (n) is impractical and thus was not attempted. Decreasing the volumetric flow rate (F) would result in lower pressures; however, sample throughput would suffer as a consequence. Column length (L) reduction was also not practical since the goal was to increase reactor volumes. Finally, the column pressure is inversely proportional to the square of the column radius (r). Thus, by increasing the column radius, the pressure across the reactor decreases 93 for a fixed reactor volume. This is, therefore, a practical way to increase the reactor volume, yet maintain a system-compatible pressure. Teflon tubing with inner diameters of 1.50 and 0.81 mm were compared for packed bed designs. Thus far, the 0.81 mm tubing had been used in constructing all reactors. Glucose oxidase was immobilized onto 120/200 mesh CPG (327 A pore diameter) using the previously described procedure. The support was then packed into four reactors with the following dimensions: hhineJd Beam Baamflolnme 0.81 mm 2.6 cm 13.4 1.11 1.50 mm 0.8 cm 13.4 ul 1.50 mm 2.6 cm 45.9 [.11 1.50 mm 3.0 cm 53.0 1.11 The results for reactors of different dimensions are . shown in Figure 4-17 where glucose calibration curves were obtained for each reactor. The larger reactor volumes, 45.9 and 53.0 ul, in the 1.50 mm tubing exhibited the highest substrate conversions as reflected in glucose sensitivity. However, these conversions were not significantly better (less than 3%) than the 13.4 1.11 reactor with 0.81 mm tubing. Reactors which had the same volume (13.4 ul) showed a slight difference in substrate conversions, with the 1.50 mm tubing exhibiting an 8% lower glucose sensitivity. The reactor dimension results indicate that no significant advantage was gained by employing wider diameter tubing in packed bed reactor construction. Although pressure was decreased with the 1.50 mm tubing, substrate conversion was not substantially improved despite 94 larger reactor volumes. Due to the comparable substrate conversions and reduced support requirements, the 0.81 mm tubing was retained for packed bed reactor construction throughout further studies. 95 0.25 1.6 o 1.5 mm id. 2.6 cm I 1.5 mm id. 3.0 cm ‘ a 0.81 mm id. 2.6 cm 0 1.5 mm id. 0.76 cm 1.2-4 0 . E g o 84 0 U) 4‘3 . < 0.4- 000 T V I if I U l I 0.00 0.05 0.10 0.15 0.20 Glucose (mM) FIGURE 4-17. Packed bed reactor dimension comparison for glucose oxidase. CHAPTER V MANIFOLD OPTHVIIZATION S A) Introduction Achieving the sugar determination methodology goals (e.g. high sensitivity, selectivity, and sample throughput) requires consideration of aspects such as enzyme immobilization, FIA manifold designs, and system optimizations. . Enzyme immobilization investigations and optimizations were presented in Chapter IV in addition to preliminary support evaluations. These studies indicate that the choice of enzyme support and reactor configuration is important with respect to sensitivity and sample throughput. Different reactor designs result in different chemical and physical processes within the FIA manifold. Ideally the reactor should provide high substrate conversion with minimal dispersion. A more thorough investigation of reactor configurations was conducted by comparing four designs. The reactors were evaluated based on the observed activity and also the dispersion introduced. Optimization of system factors that influence enzyme kinetics (e.g. buffer composition, pH, residence time, enzyme activators, and reagent concentrations) is also important in developing the sugar methodologies. These factors ultimately influence the selectivity, sensitivity, stability, and sample throughput. Reaction-rate dependent parameters were investigated and optimized for each sugar determination manifold. A composite modified Simplex (CMS) routine was employed as a means of multivariate optimization for system parameters. A multivariate 96 97 approach to optimization was selected over univariate methods to account for interaction of system variables. Possible interferences in the sugar determinations due to enzyme non-specificity were investigated for nine sugars: the six nutritionally important sugars of interest and three additional sugars that were thought to be potential interferents. Selectivities for the nine sugars were measured for each manifold and compared to the sugar of interest. The reactor configuration comparisons, Simplex optimizations, and interference evaluations are briefly introduced in the next sections followed by specific results and discussions in separate sections for each sugar. 1. Reactor Configurations a. Designs and Dimensions Four reactor design configurations were investigated for their kinetic and flow characteristics: a packed bed reactor (PBR), a single-bead string reactor (SBSR), a CPG embedded reactor (EMBR), and a combined CPG embedded and single-bead string reactor (EMSB). These configurations are illustrated in Figure 5-1. In each design, the enzyme was covalently bound to the glass support via the immobilization procedures described previously (Chapter IV) for non-porous beads and CPG. The packed bed and embedded configurations employed CPG support with a particle size of 120/200 mesh (74-125 um) and a pore diameter of 327 A. Non-porous beads (0.6 mm diameter) were used as the support in the SBSRs (traditional and embedded). Teflon tubing (0.81 mm inner diameter) was used in preparing all of the reactors. Construction of PBRs and SBSRs was discussed previously in Chapter IV. The EMBR was Single-Bead String Reactor ...n-a-s-ns-suou-u-se ...s...o . a tit-Iluinollol- 10010 an -s_:-- ...- . a. n. g‘ ‘- ..lmWI CPG Embedded Reactor / Single-Bead String Reactor FIGURE 5-1. Immobilized enzyme reactor configurations. 99 prepared by embedding CPG into the walls of Teflon tubing by means of a procedure originally reported by Mottola et al. (56). Generally, the procedure involves packing a piece of tubing with CPG, sealing the ends, and heating to approximately 350°C. The melting point of Teflon is 327°C. Following the application of heat, the excess CPG is flushed from the reactor. The enzyme is subsequently immobilized onto the CPG by flowing the reagents through the reactor. Total manifold volumes and experimental conditions were held constant for the reactor comparison studies. Reactor lengths ranged from 13 to 15 cm for the SBSR, EMBR, and EMSB configurations (corresponding to reactor volumes ranging from 67.0 to 77.3 III), while the PBR length was 2.6 cm (corresponding to a reactor volume of 13.4 pl). The much reduced support volume in the PBR is a consequence of the substantial pressure drop created in the packed bed design as the length of the CPG support increases (Chapter IV, section D4). A length of 2.6 cm was observed to be the practical limit for implementation in this system. Specific details and results are presented in the appropriate sugar sections below. b. Flow Characteristics The observed dispersion characteristics for each reactor configuration are a combination of the chemical processes (enzymatic reaction) taking place as well as the physical processes. Other workers have assessed the chemical kinetic and physical contributions to dispersion for various systems (68,144-146). In an effort to gain insight into the chemical and physical dispersion processes occurring within each reactor configuration, sensitivity and dispersion characteristics were evaluated for each design without an enzymatic reaction taking place. 100 For reactor evaluations without enzymatic reaction, a reactor of each configuration was prepared with unmodified glass support. Reactor lengths of 14 cm (72.1 ul volume) were used for the SBSR, EMSB, and EMBR, while the PBR was 2.6 cm (13.4 ul volume). These reactor dimensions were typical for the sugar determinations and were therefore used in this investigation. The dispersion comparison was conducted by injecting hydrogen peroxide (30 ul) into a manifold containing the unmodified reactor. In this way, only the detection reaction takes place following dispersion of the sample plug through the reactor. A 0.10 mM hydrogen perom'de working solution was prepared fi'om a 0.008 M stock solution (standardized against KMnO4) and diluted with 0.10 M phosphate buffer (pH 6.00). The detection reagent was prepared by combining 1980 U of horseradish peroxidase (EC 1.11.1.7) with 2.4 ml of 0.10 M phosphate buffer (pH 5.9) and 0.6 ml of leucomalachite green stock solution (0.003 M leucomalachite green in 30% acetic acid). The mixture was then diluted to 10.0 ml with 0.05 M acetate buffer (pH 4.01). The composition of the reagent solution resulted from an optimization investigation conducted by P. Aspris (147) and was used in all subsequent experiments. The carrier solution was 0.10 M phosphate buffer (pH 6.00). Flow rates for the carrier and reagent solutions were 0.34 and 0.05 mllmin, respectively. The detection reaction was conducted in a 33.0 cm plain SBSR. Reactor dispersion results are shown in Figure 5-2. Absorbance, variance, and sample throughput values are given below for the four reactor configurations. Sample throughput (samples/hr) was calculated as follows, S = 3600/(4ot), where at is the time standard deviation of the absorbance peak (60). 101 0.9 . —— sssa 0.5- ruse . PBR 0% EMBR 0 0.6- o . S 0.5- .o . 5 .3 0.4- 4 0.3- o.2~ 0.1- .g 0-0‘ IFI'I'I'I'I'I'I'F'I'I' 25 35 45 55 55 75 55 95 105115125135145155 Time (sec) FIGURE 5—2. Reactor configuration comparison without enzymatic reaction. 102 Throughput Absorbance Variance (82) (samples/hr) EMBR 0.471 528 39 SBSR 0.805 2 1 1 62 EMSB 0.738 229 60 PBR 0.673 264 55 For physical dispersion processes alone, the SBSR configuration exhibited the greatest sensitivity (absorbance) and sample throughput followed by the EMSB, the PBR, and lastly the EMBR. Dispersion characteristics, as a measure of sample throughput, for the SBSR and EMSB are expected to be similar and appear quite close in this study. The PBR characteristics are comparable to the SBSR and EMSB, exhibiting only slightly more dispersion. The poor characteristics of the EMBR are not surprising due to its similarity to an open tubular reactor where laminar flow predominates. From this investigation, results indicate that the SBSR or EMSB is the best choice for reactor design. However, the chemical kinetic aspects of the reactors are extremely important and were found to be the deciding factor in optimal reactor selection for each manifold. 2. Enzymatic Reaction-Rate Parameters a. Introduction to Simplex Optimization System parameters considered in optimizing the sugar determination manifolds included the carrier pH, carrier flow rate, enzyme activator concentrations, and detection reagent flow rate. The enzymes employed for galactose, lactose, and fructose determinations all 103 require specific activating reagents to maintain enzyme activity; therefore, the concentrations of these reagents were optimized in addition to the pH and flow rate parameters. A composite modified Simplex (CMS) procedure was utilized as a means of multivariate optimization of the system parameters (148,149). A multivariate approach to optimization was chosen over univariate methods to account for possible interdependency of the variables. This approach also minimizes the number of experiments required for optimization compared to univariate techniques. The CMS routine has proven to be successful in optimizing several FIA-based determinations (150,151) and has been previously reported in the optimization of a flow injection system for the enzymatic determination of glucose (96). Application of the Simplex procedure requires the definition of a response function which represents the system attribute(s) desired in optimization. These attributes could be as simplistic as peak height or peak area or could be more complex, incorporating several system factors. In the more sophisticated cases, the relevant system factors are generally weighted and expressed in a mathematical equation. Initiating the Simplex routine also requires specification of the variables to be optimized and their boundary conditions. Boundary conditions are values that represent the minimum (forward boundary) and maximum (reverse boundary) levels allowed for each variable. An estimate of the precision of each variable is also necessary to begin the Simplex. b. General Response Function Three measures of system performance were considered in optimizing the sugar determinations: sensitivity, sample throughput, and precision. These factors were weighted and mathematically expressed in 104 the equation shown in Figure 5—3. The response function is comprised of three terms: the first (Aexp/Abase) is related to sensitivity, the second (1/tp + 0.1) to sample throughput, and the third [1/(sA + 0.1)] to precision. Baseline (initial) experimental conditions were used for each sugar determination optimization. Five replicate injections of the appropriate sugar were made under baseline conditions at the start of every set of experiments. The average absorbance for baseline conditions was then calculated (Abase)- Subsequent experimental absorbances (Aexp) were ratioed to the baseline condition absorbance in order to compensate for day-to-day variations in the system. The times of the peak maxima were also averaged for the five injections (tp). The value for sA was obtained by calculating the standard deviation of the five absorbance measurments. The constants, (0.10 in the second and third terms), are weighting factors that were empirically chosen from simulated and real data. Prior to assigning specific values, the constants were designated as "a" and "b" where "a" was the throughput term constant and "b" was the precision term constant. Response (R) values were calculated for a range of weighting factors using representative data that varied in sensitivity, throughput, and precision. Glucose determinations were conducted at nine different carrier flow rates. Three absorbance measurements were taken at each flow rate. Values were then calculated for the average absorbance, average time of peak maximum, and absorbance standard deviation. Over the course of the experiments, absorbance (sensitivity) increased, as the flow rate decreased, while sample throughput decreased. The precision values oscillated between high and low values for consecutive experiments. The appropriate weighting factors were selected so that all three terms contributed to the response function, with an 105 Aexp - absorbance of peak maximum Am - absorbance of peak maximum (initial conditions) tp - time of peak maximum 5;; - standard deviation of Aexp FIGURE 5-3. General response function used in CMS optimizations. 106 emphasis on sensitivity. Figure 5-4 shows response trends for a and b values of various orders of magnitude. In Figure 5-4A, where a and b have a value of 1.0, the response trend appears to place too much weight on the sensitivity term. This is indicated by the relatively featureless line that only reflects the increase in absorbance from experiment one through nine. In contrast, Figure 5-4B illustrates a response trend that appears to reflect all three system parameters. A general increase in response is observed over the first six experiments until the sample throughput term becomes significant and a decline in response occurs over experiments six to nine. The oscillatory behavior of the precision term is also observed in the response trend. Figure 5-4C shows a response trend that emphasizes sample throughput over sensitivity. Finally, the response trend in Figure 5-4D dramatically reflects the oscillating precision term and more subtly the throughput term. From these order of magnitude comparisons for constants a and b, Figure 5-4B (a=0.1, b=0.1) shows the best performance of the response function. In a similar manner, comparisons were conducted for more precise values of a and b ranging between 0.05 and 0.5. Mixed values (a not equal to b) were also investigated. Despite all the additional trials, the best response performance was observed when a and b were both assigned values of 0.10. These constant values were therefore employed in the response function for all the sugar optimizations. 3. Selectivity The methodologies developed and optimized are ultimately intended for determination of nutritionally important sugars in various food samples. Therefore, an evaluation of interferences resulting from enzyme non-specificity is essential. In general, only those substrates that 1.5 A a=1.b=l Am 5 51.3% 912+ 0 or. 1.11 1.0 . . r . . 1 1 3 4 5 0 7 0 9 Experiment C 6.0 .1 a=0.01. b=0.01 $5.“ 8 4 54.01 E' .. D “3.04 4 2". I I 5 I I I I 107 1.62 1.604 E 1.5m 1.5a 1.64- B‘ 1.5% 1.504 1.404 1.46 a=0.1, b=0.1 $0.001. b=0.001 FIGURE 5-4. Response trends as a function of various order of magnitude values for constants a and b. 108 eventually yield hydrogen peroxide will exhibit an interference in the sugar determinations. Listed below are the sugars investigated for possible interference in each determination. Glucose (D-glucopyranose) Galactose (D- galactopyranose) Sucrose (a-D- HglucoFyranosyl --[1 2]- B-D- fructofuranose) Lactose (O- B-D-ga actopyranosyl- [1- 4] --D glucopyranose) Maltose (O- a-D- glucopyranosyl- [1- 4] --D glucopyranose) Fructose (D- fructofuranose) Melezitose (O- a-D- glucopyranosyl- [1- 3]- O-B—D- fructofuranosyl- [2-1]--o: D- glucopyranose) Melibiose (6- O- a-D- galactopyranosyl- D- glucopyranose) Raffinose (6- O- oc-D- galactopyranosyl-a- D- glucopyranosyl- [1- 2]- B—D- fructofuranose) The six nutritionally important sugars of interest (Figure 1-1) were investigated for their possible interference with each other as well as three additional sugars which were thought to be potential interferents due to their compositions. Structures for the three additional sugars investigated are shown in Figure 5-5. An interference that is anticipated to be quite problematic in the determination of sucrose, lactose, maltose, and fructose is that caused by the presence of glucose. These four manifolds all contain a glucose oxidase reactor in addition to the glucose-forming reactor. Any glucose present will undergo oxidation in the glucose oxidase reactor and yield hydrogen permdde which is subsequently detected. Glucose was therefore expected to exhibit signifcant responses in the interference evaluations for the sucrose, lactose, maltose, and fructose manifolds. Specific relative responses are given in the appropriate selectivity sections below. Possible solutions for minimizing or eliminating the glucose interference are discussed in Chapter VI, section B. melibiose 109 raffmose CH20H H H CHZOHO H H OH H OH OH 0 CHZOH H OH H CH20H H H OH H OH O H OH melezitose FIGURE 5-5. Molecular structures for ramnose, melibiose, and melezitose. 1 10 B) Manifolds The following sections present the reactor configuration and Simplex optimization evaluations for the sugar determination manifolds. In order to determine if comparing the four reactor configurations for each enzyme was worthwhile, reactor studies were conducted prior to Simplex optimizations for glucose oxidase (glucose), invertase (sucrose), B—galactosidase (lactose), and a-glucosidase (maltose). Initially, enzyme activities for galactose oxidase (galactose) and glucose isomerase (fructose) were not adequate for reactor configuration comparisons. Therefore, Simplex optimizations were conducted first in an effort to improve activities for these enzymes. Optimal system performance is attained by combining the results from the reactor design comparisons and the Simplex optimizations. Each sugar determination manifold is characterized for the principal analytical figures of merit (e.g. working range, detection limit, precision, and sample throughput). Finally, selectivities of the sugar determinations, as a consequence of enzyme specificity, are presented. 1. Glucose (Glucose Oxidase) The flow injection manifold used for glucose determination is shown in Figure 5-6. Oxidation of B-D-glucose to D-gluconic acid takes place in the glucose oxidase reactor with the production of hydrogen peroxn'de. Glam Oxidase B-D-glucose + 02 ————-> D-gluconic acid + H202 Prior to reactor configuration evaluations and Simplex optimization, several initial investigations had been conducted with glucose oxidase. Preliminary reactor comparisons were made between the PBR and SBSR 111 FIGURE 5-6. Flow injection manifold for glucose determination. 112 designs with the PBR exhibiting greater sensitivity (Chapter IV, section D2a). a. Reactor Configuration Evaluations Glucose oxidase was immobilized onto non-porous beads and 120/200 mesh CPG (327 A pore) using the procedures described previously for the two supports (Chapter IV). Enzyme concentrations of 874 U/0.25 g support and 2185 U/0.20 g support were used for the non-porous beads and CPG, respectively. A reactor volume of 13.4 ul (corresponding to a length of 2.6 cm) was observed to be the practical limit for the PBR design due to the low pressure instrumentation employed. The other three reactor designs (EMSB, SBSR, and EMBR) do not suffer fi'om high pressure drops. Thus, a larger reactor volume of 72.1 ul (corresponding to a length of 14.0 cm) was chosen for these low pressure reactors as an intermediate volume where a compromise is made between residence time and sample throughput. Glucose calibration curves were obtained for each of the four reactor designs (30 ul glucose injections). Glucose standards were prepared from a 0.01 M stock solution that contained benzoic acid (0.01 M) as a preservative. Working solutions were diluted with 0.10 M phosphate buffer (pH 6.00). The carrier was also phosphate buffer (pH 6.00) and was propelled at a flow rate of 0.34 mllmin. The detection reagent flow rate was 0.05 mllmin. Reactor configuration results for glucose oxidase are shown in Figure 5-7. Sensitivity comparisons are illustrated in Figure 5-7a while an indication of dispersion can be seen in Figure 5-7b (glucose 0.20 mM). Absorbance, variance, and sample throughput values are tabulated below. 1.6 Absorbance .0 .0 7‘ ¥ on N l m L s l s 0).. mmm zng manna) 3021(1) \\ D , . 0.00 0.05 0.10 o.'1 5 0.20 0.25 Glucose (mM) 1 6 b ' . ._ —- PBR / \ 1.4.. I \\ -- EMSB ‘ ,’ \ — SBSR 1.2- o J / ,\\\ --- EMBR O I / \\ a 1.0- l / \\ III . I / \\ f 0.84 I’ / u a 3 I I ‘\ a 0.6 I! <1 ' 1 / \ 0.4- II I \ . ’ / 0.2- ............. ‘ x/ll ,,,,,, 1.339. 0.0- ----- I IT I I . I I I . I I I I I I I . I I I 25 35 45 55 65 75 85 95 105115125135145155 Time (sec) FIGURE 5-7. Reactor configuration comparisons for glucose oxidase, sensitivity (a), dispersion (b). 114 Throughput Absorbance Variance (s2) (samples/hr) EMBR 0.213 861 3 1 SBSR 0.707 324 50 EMSB 1. 160 380 46 PBR 1.483 349 48 Results show that the PBR design provides the greatest glucose sensitivity followed by the EMSB and SBSR. However, the difference in sensitivity between the PBR and EMSB in not substantial (a factor of 1.3). Not surprising, the EMBR exhibited the poorest glucose sensitivity. The SBSR, PBR, and EMSB configurations all show comparable dispersion characteristics which leads to similar values for sample throughput with the SBSR yielding a slightly higher value of 50 samples per hour. Sample throughput for the EMBR was significantly lower at 31 samples per hour due the open tubular nature of this reactor. As expected, the EMSB proved to be a favorable reactor design. Dispersion characteristics for this reactor are very similar to those of the SBSR including a low pressure drop, yet the EMSB generally provides higher substrate conversion efficiencies. b. Simplex Optimization Three rate-dependent parameters were optimized for glucose determinations utilizing the Simplex procedure: carrier pH, carrier flow rate, and reagent flow rate. The carrier pH, and carrier flow rate influence the enzyme kinetics and enzyme stability, while the reagent flow rate influences the characteristics of the detection reaction. 115 Originally, five parameters were attempted for optimization. In addition to the three parameters stated above, a carrier pH modifier and modifier flow rate were also included as variables in the Simplex optimization. This created problems with the Simplex routine due to the similarity of the variables. There were essentially two pH variables and three flow rate variables. The flow rate variables also indirectly influence the pH. As the relative flow rates between carrier and reagent change, so does the pH. With so many like variables, it is possible to have several combinations that would lead to local maxima, making it difficult to find the global maximum. Consequently, the variables were reduced to three less interdependent parameters. The Simplex optimization was initiated by supplying the baseline or initial experimental conditions and boundary conditions for each parameter. This information is summarized in Table 5-1. The initial (baseline) conditions selected were values typically employed in previous investigations with glucose oxidase. Absorbances were measured for five replicate injections (30 ul) of a 0.10 mM glucose standard (pH 6.00) for each set of experimental conditions. The configuration of the glucose oxidase reactor was a packed bed design with a support volume of 13.4 ul (2.6 cm). The results of the Simplex optimization are shown in Table 5-2. Because the CMS is a search procedure, it is possible to obtain several sets of "optimal" conditions near the response surface maximum, depending on the surface topology. In this application, these optimal conditions are expressed as the top four responses obtained throughout the optimization. . Since the optimization was conducted for three variables, a complete Simplex is comprised of four points in four dimensional space. Averages of the top four responses and the corresponding parameter values were 116 mm: 5—1. man d boun ' op intion. dary upu'imental conditions for glucose Experimental Forward Reverse Initial Variable Boundary Boundary Conditions Carrier pH 5.00 8.00 6.60 5:173:13" 3‘“ 0.10 1.00 0.42 wt ’1" 3“” 0.03 0.00 0.03 (ml/min) 117 and N96 N06 8.: “0365: 6.3ng e no: on: 86 and Nu .u chord cm; 3.: and 86 3 «NA 86 0N6 3.0 an em; cod mad Mud Nu Av 99: was 8.: and 2... a 1808 3 .u 8.: Nod 86 u 733 Ana—>0: Ana—\HEV :5 8am to: Sam .53 an .52 050.50% anomaom ago ago .5 50305533. 0.005% new 30335.0 anon—tonne 1ng can 33 .Nlm an. 118 calculated to obtain conditions for each of the three variables. Over the course of the optimization, the response function improved by a factor of 1.2 from initial conditions. A scatter diagram of the progress of the Simplex optimization is shown in Figure 5-8. From this diagram it can be seen that the response did not improve significantly from the initial value (experiment 1). This behavior indicates that the initial conditions were nearly optimal. The Simplex routine located slightly better conditions within the first few experiments and then appeared to oscillate around the response surface maximum. Individual parameter trends can be interpreted from the scatter diagrams shown in Figure 5-9. These diagrams should not be viewed as univariate data since all variables were changed simultaneously. However, general parameter trends are observed. The carrier pH decreased over the course of the Simplex procedure from the initial value of 6.60 to approximately 6.12. This trend is illustrated in Figure 5-9a, where the optimum responses center between pH 6.00 and 6.20. From the standpoint of enzyme kinetics, the response function initially increased as the carrier flow rate decreased (Figure 5-9b). This is expected due to the longer residence times provided by lower flow rates. There is, however, a compromise that is eventually reached when flow rates become so low that the sample throughput starts to suffer significantly. Since sample throughput is also a consideration reflected in the response function (Figure 5-3), exceedingly long residence times are prohibited. The optimal reagent flow rate increased fiom initial conditions (Figure 5-9c). This can be primarily explained by sample throughput considerations. Due to the loss in throughput from the reduction in carrier flow rate, an increase in reagent flow rate helps compensate for that loss. The increase in reagent 119 1.6 1.2-1 0 . 0 ° osJ Response (R) 0.4- 000 fir ' V I V I I i I l I l fi I U I r 0 2 4 6 8 10 1 2 1 4 1 6 16 Experiment Number FIGURE 5-8. Response function progress of the Simplex optimization for glucose. 120 a b 10 1.0 . , 1 . .N 6 o .2. 0 0 A 1.2-4 0 o A 1.2- o 0 5 o . 0 ° 8 ‘ o 0 o . g ' o E 0.0« a 00 e 1 3' I: 0.4.1 I3 0.4. 000 ' t ' r f I f 0.0 f I ' f ' I ' I f 5.0 5.5 0.0 0.5 7.0 0.0 0.1 0.2 0.3 0.4 0.: Carrier pH Can-tar Flow Rate (ml/min) c 1.11 3:: , a ” . .3 . C B 0.0 o g1 1 g 0.41 0.0 - , - , - r f 0.00 0.05 0.10 0.15 0.20 Reagent Flow Rate (ml/min) FIGURE 5-9. Scatter diagrams for individual glucose optimization variables, carrier pH (a), carrier flow rate (b), reagent flow rate (0). 121 flow rate also lowers the pH for the detection reaction, which appears to exhibit greater sensitivity and stability at more acidic pH (147). Prior to final manifold characterizations for glucose determination, an additional reactor configuration comparison was made between the PBR and EMSB designs under the optimal conditions obtained from the Simplex procedure. These two configurations had exhibited similar glucose sensitivities in earlier studies. Thus, a final comparison was made to ascertain whether or not the PBR design remained the optimal configuration based on sensitivity considerations. Each reactor was inserted in the FIA manifold and evaluated by injecting a standard glucose solution (0.05 mM). The PBR (2.6 cm) gave an absorbance of 0.203, while the EMSB (15.0 cm) yielded an absorbance of 0.188. The observed reactivities were extremely close for the two designs under optimized conditions. Thus, either design could be utilized for the characterization studies. The PBR configuration was selected for characterization due to its relatively simple preparation and implementation compared to the EMSB. c. Characterization By combining the results of the Simplex optimization and reactor comparison evaluations, optimal system performance is attained for glucose determination. Calibration curves obtained for initial and optimal experimental conditions are shown in Figure 5-10 for a 2.6 cm glucose oxidase PBR. The glucose sensitivities are nearly identical under initial and optimal conditions with only a factor of 1.2 difference. The linear dynamic range obtained for glucose was 1.12x10'5 M to 3.06x10'4 M under optimal conditions while that for the initial conditions was 1.81x10°6 M to 3.58x10'4 M. The optimized conditions provide a slightly lower detection 122 0 Optimized (58 samples/hr) 1.4- I Initial (64 samples/hr) Absorbance 000 A r r I I' l I ‘ 0.00 0.05 0.10 0.15 0.120 0.25 0.30 Glucose (mM) FIGURE 5-10. Glucose calibration curves for initial and optimized manifold conditions. 123 limit, relative to initial conditions, at 11.2 uM (2.0 ppm). The difference in sample throughput is also small, initial conditions yield approximately 6 more samples per hour over optimal conditions. System stability was not influenced by the optimized conditions. Over the glucose concentrations employed, relative standard deviations (RSDs) of absorbance measurements ranged fi'om 0.4% to 1.3% and 0.3% to 1.2% (n=3) for the initial and optimized conditions, respectively. Because the initial and optimized conditions do not yield substantially different manifold characteristics, the choice of which conditions to employ is somewhat dependent on the particular application and the importance of low detection limit versus sample throughput. If glucose concentrations are low (1.1M range) and accurate quantitation is desired, a calibration curve at the lowest possible levels is probably of greater concern than sample throughput. Alternatively, if glucose is present at higher levels, sample throughput can be optimized by sacrificing a lower detection limit. d. Selectivity Glucose oxidase (B-D-glucose: oxygen 1-oxidoreductase; EC 1.1.3.4) is fairly specific for B-D-glucose. There is an absolute requirement for a hydroxyl group at the C(l) position of the substrate, preferably the B anomer. Modifications in positions C(2) through C(6) greatly reduce the activity of the enzyme (152). Eight sugars were evaluated for their relative responses compared to glucose, employing the optimized conditions discussed above. Standard solutions were prepared for each sugar at 0.20 mM in 0.10 M phosphate buffer (pH 6.00). Table 5-3 presents the responses of the interferent ratioed to the glucose response. Relative responses designated by "ND" 124 TABLE 5-3. Glucose manifold selectivity. Glucose 1.0 Galactose 0.0024 Sucrose ND lactose ND Maltose 0.0052 Fructose ND Melezitose ND Melibiose 0.029 Raffinose ND N0 = not detected 125 indicate that any signal was indistinguishable from the baseline. Three of the eight sugars investigated exhibited a small interference response with glucose oxidase. It is unclear whether glucose oxidase accepts these compounds as substrates or whether impurities present in the enzyme reagent caused the positive responses. The enzyme was obtained from Sigma as a lyophilized powder containing 80% protein. Unfortunately, several other enzymes are listed as impurities by the manufacturer. Both galactose oxidase and maltase (OI-glucosidase) are listed as impurities at 0.90% and 1.35% of the glucose oxidase activity, respectively. Their presence would explain the positive responses to galactose and maltose and also the magnitude of the relative responses. However, the positive response to melibiose (6-O-01-D-galactopyranosyl-D-glucose) may very well be due to glucose oxidase activity. This disaccharide possesses a [1-61-01 linkage between galactose and D-glucose. a-Glucosidase does exhibit a small activity towards melibiose, but this does not explain the greater response observed for melibiose compared to maltose. It is possible that these interferences could be eliminated if a purer form of glucose oxidase was obtained. Melibiose poses the most significant selectivity interference with glucose oxidase with a signal that is approximately 3% of that of glucose. Fortunately, melibiose is a relatively uncommon sugar present in food samples. 2. Galactose (Galactose Oxidase) Similar to glucose, determination of galactose only requires one enzyme, galactose oxidase, for hydrogen peroxide production. Galactac Oxidase D-galactose + 02 ——> galactohexodialdose -I- H202 126 The flow injection manifold employed for galactose determination is shown in Figure 5-11. As previously discussed in Chapter IV, section A2, galactose oxidase requires a redox couple, hexacyanoferrate (III) / hexacyanoferrate (II), to mediate the copper cofactor oxidation state in order to maintain enzyme activity. Preliminary investigations employing the enzyme in a SBSR configuration were disappointing due to poor enzyme activity and reproducibility. Because of the poor performance of the SBSR, conducting a reactor configuration comparison prior to optimization would be futile. Therefore, the CMS optimization was performed first in an effort to improve the enzyme activity and stability. a. Simplex Optimization Five rate-dependent parameters, (carrier pH, carrier flow rate, reagent flow rate, hexacyanoferrate (III)/(II) ratio, and total hexacyanoferrate concentration), were optimized utilizing the CMS procedure. Of these parameters, the carrier pH, carrier flow rate, and hexacyanoferrate variables all influence the enzyme kinetics and enzyme stability, while the reagent flow rate influences the characteristics of the detection reaction. Initial (baseline) conditions and boundary conditions supplied to the Simplex routine are shown in Table 5-4. Conditions (initial and boundary) selected for carrier pH, carrier flow rate, and reagent flow rate were the same as those employed in the glucose Simplex optimization. Initial redox couple conditions, hexacyanoferrate (III)/(II) ratio and total concentration, were obtained from published methods for galactose determinations with immobilized galactose oxidase (97,132). These conditions were reported to yield adequate enzyme activity and stability. Boundary conditions for the hexacyanoferrate variables were given a broad range since the optimal 127 PC/XT £—#I LL _' . m m" Galactose Oxidase Phin m 5]“ .1... ... has! ' Pump Detector FIGURE 5-11. Flow injection manifold for galactose determination. 128 TABLE 5-4. Initial and boundary experimental conditions for galactose optimization. Experimental Forward Reverse Initial Variable Boundary Boundary Conditions Carrier pH 5.00 8.00 6.60 Carrier Flow Rate (ml /min) 0.10 1.00 0.42 PACK)“ (III/II) Ratio 0.01 100.00 10.00 Il'e(CN)o (pl!) 1.0 50.0 4.4 wt 1'1"" 3“" 0.03 0.00 0.00 (ml/min) 129 values could not be predicted. Absorbance measurements were collected for five replicate 30 11] injections of a 1.0 mM galactose standard in 0.10 M phosphate buffer (pH 6.60) for each set of experimental conditions. Due to the poor performance of the galactose oxidase SBSR (discussed above), a packed bed design (2.6 cm) was used for the Simplex optimization. The reactor was prepared by adding a mixture of 450 U galactose oxidase in 2.0 ml of 0.10 M phosphate buffer (pH 6.60) to 0.20 g activated CPG (120/200 mesh, 327 A pore). Prior to the Simplex experiments, a solution of 50 um hexacyanoferrate (III) in 0.10 M phosphate buffer (pH 6.60) was flushed through the reactor to activate the enzyme. This activation step was previously reported to be effective by other workers (97,132). When the reactor was not in use, it was stored at 4°C in a phosphate buffer (pH 6.60) that contained 0.10 mM EDTA and 2.0 mM CuSO4. This storage buffer preserved the enzyme activity between experiments. Results from the Simplex optimization are given in Table 5-5. Optimal conditions are expressed as the top six responses (R) obtained throughout the Optimization. Averages of the top six responses and corresponding parameter values were calculated to obtain optimized conditions for each of the five variables. Over the course of the optimization, the response function improved by a factor of 5.1 from initial conditions. A scatter diagram of the progress of the Simplex is shown in Figure 5-12. The response improved rapidly in the first several experiments with a gradual leveling off as the search approached the maximum. The optimization was terminated after 43 experiments when the Simplex appeared to oscillate around the response surface maximum. Several trends were observed for the five variables as the Simplex proceeded. Scatter diagrams for the five variables are shown in 130 8... 8.5 3 2s. 3.: mm... sonar»: 2356 .... .3. sum 2.: v... 3.9.. 2... he use: «a.» I... S. 8.3 3.5 mm.» 3 «mm 3... ..o 3.6. 2... 8s 8 «a.» an... o... 8.8 2.: m3. 8 as 8.: 2: 8.8 3... «he 8 on m 2 a e a an 3 2 a 8 s S a. as: 2...... 2.: as 8.3 an... «as on 7:58 3.: 8... vs. 8.3 ms... 8.: _ 1:3 Ann—{any 63:. Aggy :3 3e»— 30E e33 SEE 30m Form an .oz «385. anon-em 980.. :5?— 050 sound .3. defining—e 0.3001. you afloat—:60 Bus-Enema 2:5qu and :53 .mln an. Response (R) FIGURE 5-12. Response function progress of the Simplex optimization 2J4 131 24)- 1.64 1.2-4 (L84 CL4~ 01) .1 I r l ' l ' l ' I ' l ' I ' I ' I V 8 12 16 20 24 28 32 36 40 44 Experiment Number for galactose. 132 Figure 5-13. The carrier pH increased from the initial value of 6.60 to approximately 7.17 at the optimum (Figure 5-13a). Similar to the trend observed in glucose optimization, the response function initially increased as the carrier flow rate decreased until a point was reach where the sample throughput component of the response function suffered significantly. Figure 5-13b indicates that the carrier flow rate optimum is located between 0.10 and 0.20 mllmin. Enzyme stability was greatly affected by the hexacyanoferrate parameters, both the Fe(III/II) ratio and the total concentration. The optimal hexacyanoferrate (III/II) ratio (Figure 5-13d) was substantially larger (nearly 14 times) than the initial ratio. This indicates that a higher relative concentration of Fe(III) results in greater enzyme stability and subsequently improved precision. Eventually, a level was reached for Fe(III) where any further increase did not significantly improve the stability. The total hexacyanoferrate concentration under optimized conditions was approximately double the initial value, as illustrated in Figure 5-13e. Finally, the optimized reagent flow rate increased from initial conditions due to the improved throughput component of the response function and also the greater sensitivity and stability of the detection reaction at higher pH (Figure 5- 13c). b. Reactor Configuration Evaluations As a result of the CMS optimization, galactose sensitivity and enzyme stability were greatly improved. Galactose sensitivity under optimized conditions was 6.8 times- that for initial conditions. Thus, a reactor comparison study was conducted employing the optimized system parameters. Galactose oxidase was immobilized onto non-porous beads (0.25 g) and CPG (0.20 g) by dissolving 450 U in 2.0 ml of 0.10 M phosphate buffer 133 a b 2.4 . 2.4 ‘ . .0... 1 0' O 2.0. .‘ 0 2.0.1 3 o o A ‘ .0 0.0 A . . 5 1.0- fl. ' . 5 1.0-1 ' ' e ‘ o . 0 ‘ 5 . 0 ° 5 L2. .: o. . g 1.21: : e O . : I»: " . In: . 4 I 0.4« ' 0.1. I 1 e 5 1 . . o M ' j ' I :U ' I ' U ' Ifi I r M ' 1' ' V f T" I ' I ' U‘f V ' uuaoasuuaoasao 0.001030504050001” Can-tor pl! Carrier Flow Rate (ml/min) 2.4 o. 2.4 d. . 1 . 5 e . . ' .0: ”J 0 . 2.0-I o . A . t . . A . s 5 ‘0.“ O ; . 5 ".q C :1 o w : : ° . ° 1.2. : . I 0 . o . 0.04 g . 0.04 0 0.4- ° 0.“ I O 0 ‘ o 0 ‘ . o 0'0 #IA'Y' I'I'I' 00° 'fivrvr' f1 'l' 0.000050.100.150.200.250..10 11020304'ITsTso'I00000 Reagent flow Rate (ml/min) Ratio (Fa mm o 2.4 . . 4 3 o 2.0-1 0.. . A 1 0 .0 so g 5 ".1 . ee' o 1.2-: ; .. 0.a« . 0 0.44 . O 0.0 ° ' 4. antio' 4105030' (0'10.on Concentration (ml) [Fe(CNM FIGURE 5-13. Scatter diagrams for individual galactose optimization variables, carrier pH (a), carrier flow rate (b), reagent flow rate (0), ratio Fe III/II ((1), concentration Fe(CN)6. 134 (pH 6.60) and adding the mixture to the activated support. Reactor lengths of 14.0 cm were used for the SBSR, EMBR, and EMSB configurations, while the PBR was 2.6 cm. Galactose standards were prepared from a 0.10 M stock solution and diluted with 0.10 M phosphate buffer (pH 6.60). Reactor configuration results for galactose oxidase are shown in Figure 5-14, where calibration curves were constructed for each of the four designs. Absorbance, variance, and sample throughput values are given below for the four reactors. Throughput Absorbance Variance (32) (samples/hr) EMBR 0.263 996 29 SBSR 0.141 696 34 EMSB 0.402 613 36 PBR 0.493 586 37 The PBR configuration was by far superior with a galactose sensitivity over five times that of the EMSB which exhibited the second best performance (Figure 5-14a). The SBSR showed the poorest galactose sensitivity while the EMBR was slightly higher. These sensitivity considerations indicate that the CPG support yields higher galactose oxidase immobilization efficiencies than non-porous beads. Dispersion characteristics for the reactors are illustrated in Figure 5-14b. Like previous observations, the PBR, EMSB, and SBSR all exhibit comparable sample throughput. Although the EMBR was not the least sensitive configuration in this case, it still had the lowest throughput. Considering both sensitivity and sample throughput, the PBR is clearly the optimal 135 n 1.20 . o PBR I EMSB ‘°°°‘ . EMBR 0 ‘ o SBSR a 0.80"4 {3 (160-1 0 . .3 4 0.404 0.20. 0000 U r I l I l I I I I I I I I U 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Galactose (mM) b 0.60 . — PER 050- Egg: ‘ SBSR g 0.40.. a 0.30- O . .3 <# OJNJA 0.10- 0.00 N'“ s 30 55 8'0 '105'130r1és'1éo'205'230'255 Time (sec) FIGURE 5- 14. Reactor configuration comparisons for galactose oxidase, sensitivity (a), dispersion (b). 136 reactor configuration for galactose determinations with galactose oxidase under optimized conditions. c. Characterization System characterization for galactose determinations was conducted utilizing a galactose oxidase PBR (2.6 cm). Figure 5-15 shows galactose calibration curves obtained under optimized and initial conditions. Galactose sensitivity was 6.8 times greater under optimal conditions with a sample throughput of 41 samples per hour. Even more significant was the improvement in the system stability. Initial conditions resulted in a relative standard deviation (RSD) for of 14.3% for a 1.0 mM galactose solution, while optimal conditions exhibited a RSD of 1.2%. The working range obtained for galactose was 8.00x10'5 M to 1.75x1043 M for a 30 ul sample volume with RSDs ranging from 0.02% to 2.45% (n=3) over the galactose concentrations employed. The detection limit was approximately 80 uM or 14 ppm galactose. The non-linear relationship between absorbance and galactose concentration is most likely due to galactose behaving as an activator for galactose oxidase (153). d. Selectivity Galactose oxidase (D-galactose: oxygen 6-oxidoreductase; EC 1.1.3.9) catalyzes the oxidation of D-galactose and several other galactose containing compounds (154). The position at C(4) on the substrate is essential; thus, any alterations at that position yield no oxidation. The C(1) position does not have to be free from substitution. The nine sugars were evaluated for their ability to act as substrates for galactose oxidase. Similar to the glucose manifold, only those substrates that eventually yield hydrogen peroxide will interfere with galactose determinations. Standard solutions of each sugar were prepared Absorbance 137 1.8 0 Optimized (41 samples/hr) 1.6-‘ I Initial (45 samples/hr) 1.4- 1 1.2-4 J 1.0- 0.84 0.6-4 //. I j I I I l ' l ' j fi l ' I 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Galactose (mM) FIGURE 5-15. Galactose calibration curves for initial and optimized manifold conditions. 138 at 1.0 mM in 0.10 M phosphate buffer (pH 7.00). Table 5-6 shows the interference responses ratioed to the galactose response. Relative responses designated by "ND" indicate that the signal obtained was indistinguishable from the baseline. The three sugars that exhibited interference responses are di- and trisaccharides that contain at least one galactose unit. Their ability to act as galactose oxidase substrates is expected and has been documented elsewhere (155). The order of magnitude for the relative values agrees with literature values, although absolute interference values were lower for this system. Lactose is the most probable sugar to cause interference problems with galactose determinations in food samples. However, the response is only 0.36% of that of galactose. 3. Sucrose (Invertase/Mutarotase) Sucrose determination requires three enzymes. The disaccharide is first hydrolyzed by invertase to yield a-D-glucose and B-D-fructose. Hydrolysis is followed by equilibration (mutarotation) of the on and [3 glucose anomers via mutarotase. Finally, glucose oxidase is used for the production of hydrogen peroxide, which is detected as described previously. lnvertnoe sucrose + H20 —> a-D-glucose + B-D-fructose Mutant-Io a-D-glucose <———-> [S-D-glucose Glucose Oxidase B-D-glucose + 02 ———> D-gluconic acid + H202 The FIA manifold used for sucrose determination is illustrated in Figure 5-16. 139 TABLE 5-6. Galactose manifold selectivity. Sugar RReeslpime Galactose 1.0 Glucose ND Sucrose ND Lactose 0.0036 Maltose ND Fructose ND Melezitose ND Melibiose 0.54 Raffinose 1.0 N0 = not detected 140 PC/X'l‘ fl IF [I w Glucose ma v... Invertuelnteroteseondaee Plain a"! 1 Reactor Reactor Reactor sass H he! ‘Pump' Detector FIGURE 5-16. Flow injection manifold for sucrose determination. 141 a. Reactor Configuration Evaluations Because three enzymes are necessary for sucrose determination, evaluation of reactor configurations for each enzyme should be considered. Reactor comparisons for glucose oxidase were conducted previously, with the PBR and EMSB designs exhibiting superior performance over the SBSR and EMBR. Thus, either the PBR or EMSB reactor configuration could be employed in the sucrose manifold for the glucose oxidation reaction. A comparison of a SBSR and PBR configuraton was initially conducted for mutarotase to determine if an evaluation of all four designs was worthwhile for this enzyme. Mutarotase was immobilized onto non- porous beads and CPG by reacting 8516 U of the enzyme with 0.25 g and 0.10 g activated support, respectively. A SBSR (14.0 cm) and PBR (2.6 cm) were then constructed as described previously. Invertase and glucose oxidase reactor configurations were a 2.6 cm PBR and a 14.0 cm SBSR, respectively. Absorbance values were compared for 30 pl injections of a 0.15 mM sucrose standard into the sucrose determination manifold, first employing the mutarotase SBSR followed by the mutarotase PBR. Resulting absorbances were 0.388 for the mutarotase PBR and 0.181 for the SBSR. This is a considerable difference in activity for the two reactor designs. Although an EMSB design would yield a higher apparent enzyme activity relative to the SBSR, it was thought that the PBR configuration would still provide superior a-D-glucose mutarotation efficiencies. Therefore, the PBR design was employed in all subsequent sucrose manifold experiments. Reactor configuration evaluations were conducted for invertase with SBSR, EMBR, and EMSB lengths of 13.0 cm and a PBR length of 142 2.6 cm. Invertase concentrations of 7300 U/0.14 g and 7300 U/0.25 g support were used for the CPG and non-porous beads, respectively. Calibration curves were obtained for each reactor with sucrose standards, ranging in concentration from 0.1 to 0.7 mM, prepared in 0.10 M phosphate buffer (pH 6.00). The carrier, 0.10 M phosphate buffer (pH 6.00), was propelled at 0.34 mllmin while the reagent flow rate was 0.05 mllmin. Results for the reactor configuration comparisons are shown in Figure 5-17. The calibrations curves in Figure 5-l7a indicate that the EMSB configuration yields slightly enhanced sucrose sensitivity over the PBR followed by the SBSR and EMBR. Figure 5-17b shows the dispersion characteristics for each of the four reactor designs. Absorbance, variance, and sample throughput values are tabulated below (sucrose 0.50 mM). Throughput Absorbance Variance (82) (samples/hr) EMBR 0.244 1296 25 SBSR 0.846 338 49 EMSB 1.358 320 50 PBR 1. 162 443 43 Sample throughput was comparable for the EMSB, PBR, and SBSR, with the single-bead designs yielding slightly higher values. Again, the EMBR exhibited the poorest sensitivity and dispersion characteristics. Although sucrose hydrolysis appeared slightly greater with the EMSB configuration, the PBR was very close in activity. Due to the ease of preparation of the PBR compared to the EMSB, the PBR design was used in the Simplex optimization studies. 143 2.0 a - EMSB o PBR 1.6~ A SBSR 0 . o EMBR S 12 c ' I .o ‘6 .2 0.8~ <: . 0.4- // 0-0 I r r I I ' l ' 1 ' I ft 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Sucrose (mM) 1.. g . If \ —- EMSB 1.2- _ \ -- PBR . V \ \\ — sass a, 1.04 ’l \ \ --- EMBR o . I 5 0.8- .0 . ‘6 06 .2 ' 7 4 0.4- 1 0.2- 0.0 I I I 25 40 Time (sec) FIGURE 5-17. Reactor configuration comparisons for invertase, sensitivity (a), dispersion (b). 144 b. Simplex Optimization For sugar determinations that require more than one enzyme, the kinetic characteristics of each enzyme should be considered in manifold optimizations. Ideally, this means that individual reactor residence times and individual reaction pH values should be optimized independently. However, as mentioned previously, the Simplex routine has difficulty finding a global maximum when the parameters selected for optimization are too similar. Thus, for enzymes with similar optimum pH ranges and similar reaction-rates, individual optimizations are not practical. Since sucrose determination requires three enzymes, individual pH optimizations were considered. Reported pH range optima for glucose oxidase and invertase in solution are approximately 5.0 to 6.5 (156) and 4.5 to 6.0 (157), respectively. These ranges are too similar for separate optimization; however, the reported pH optimum for mutarotase is somewhat higher at approximately 7.4 (158). Initial investigations were, therefore, conducted to determine if a separate pH optimization for mutarotase was practical. A reagent tee was inserted in the FIA manifold between the invertase and mutarotase reactors so that a pH modifier stream could be introduced to the carrier stream (pH 6.0). Two pH modifier values were compared, pH 6.0 and pH 8.0, for the mutarotase reaction. Phosphate bufi‘er (0.10 M) at these pH values were introduced at a flow rate of 0.05 mllmin. The other instrumental parameters were the same as those employed above. The absorbance observed for a 0.15 mM sucrose standard with the pH 6.0 modifier was 0.398, while that for the pH 8.0 modifier was 0.354. Raising the pH for the mutarotase reaction did not improve sucrose 145 sensitivity and therefore was not attempted in the Simplex optimization for sucrose determination. The rate-dependent parameters optimized for sucrose determination were carrier pH and carrier flow rate. Because the optimized reagent flow rate values obtained from glucose and galactose optimizations were comparable at 0.08 and 0.17 mllmin, an intermediate value of 0.10 mllmin was selected as a constant manifold parameter in the sucrose Simplex optimization. Initial (baseline) and boundary conditions supplied to the Simplex routine are shown in Table 5-7. An initial carrier pH of 6.00 was chosen due to success in previous sucrose determination experiments with this value. Carrier flow rate values were the same as those used in glucose and galactose optimizations. The Simplex optimization studies were conducted with an invertase PBR (2.6 cm), a mutarotase PBR (2.6 cm), and a glucose oxidase SBSR (14.0 cm). Five replicate 30 [41 injections of a 0.15 mM sucrose standard were made for each experimental condition. Results from the Simplex optimization are given in Table 5-8, where optimal conditions are expressed as the top three responses. Averaging the corresponding parameter values yields optimized conditions for sucrose determination, with a carrier pH of 5.41 and carrier flow rate of 0.18 mllmin. Figure 5-18 shows the progress of the Simplex optimization as monitored by the response function. The response improved only slightly (30 %) from the initial conditions, and the Simplex routine was terminated after 11 experiments. Individual parameter trends are observed in the scatter diagrams shown in Figure 5-19. Carrier pH decreased from an initial value of 6.00 to approximately 5.14. The scatter diagram (Figure 5-19a) exhibits a wide nan: 5-7. Inn 146 Lal. and boundary experimental conditions for sucrose optimisation. Experimental Forward Reverse Initial Variable Boundary Boundary Conditions Carrier pH 5.00 8.00 6.00 cm" n" R‘“ 0.10 1.00 0.42 (ml/min) 147 3.0 3.: and flofissn Bus—33m a no: on.“ 2... :6 985 an; 3... . 3.... a m... a an a m— a v 8 no: 8.. 2.: 9...... n 158 84 mo... 8... a 125 3335 :3 3am to: an 6: 0309.83 .0an .8380 .5 .doflaflauouov cannon—e new 233.3900 ado-Such”. manna—no one 133 dim. an: 148 1.6 O . O 1.2‘1 . O . A O O . E", o 31 :1 0.84 o 0 Q: m g 0.4d 1 0.0 I l r l I T I l I I I 0 2 4 6 8 10 12 Experiment Number FIGURE 5-18. Response function progress of the Simplex optimization for sucrose. 149 I 1.6 4 0 . . A 1.2- ‘ ea . t . . a l a 0.8- O o no . '1 8 0.4. 0.0 .,.,.,.,.,.,. 4.8 5.0 5.2 5.4 5.0 5.8 6.0 6.2 Carrier pH b 1.6 O . . A 1.2-1 0 ’ 0 85 J I . , 8 a 0.e~ O o 9- . n O a: 0.4- 0.0 r . , . , . , . 0.0 0.1 0.2 0.3 0.4 0.5 Carrier Flow Rate (ml/min) FIGURE 5-19. Scatter diagrams for individual sucrose optimization variables, carrier pH (a), carrier flow rate (b). 150 pH optimum range, which is not surprising since the combined effects of pH on three enzymatic reactions are reflected in the overall response. Slight pH variations do not substantially affect sucrose sensitivity, until the pH approaches higher values (> 5.8). Similar to carrier pH, the scatter diagram for carrier flow rate (Figure 5-19b) also indicates a wide optimum range. The longer reactor residence times provided by decreased flow rates do not strongly influence the substrate conversion. Prior to manifold characterization for sucrose determination, a comparison was made between the PBR and EMSB configurations for invertase under the optimized conditions obtained from the Simplex procedure. These two designs had exhibited very similar sensitivities in the previous reactor configuration evaluations. Each reactor was inserted in the FIA manifold and evaluated by injecting a standard sucrose solution (0.15 mM). The PBR (2.6 cm) gave an absorbance of 0.472, while the EMSB (15.0 cm) absorbance was 0.454. Since the two configuration exhibited very similar sensitivities, either reactor design could be used for final system characterization. c. Characterization System characterization for sucrose determination was conducted utilizing an invertase PBR (2.6 cm), a mutarotase PBR (2.6 cm), and a glucose oxidase EMSB (15.0 cm). An EMSB configuration was selected for glucose oxidase over the SBSR configuration, used in the Simplex optimization, due to the enhanced substrate (glucose) conversion observed with the EMSB. The EMSB and SBSR designs were observed to possess nearly identical dispersion characteristics for glucose oxidase as shown in the table on page 114 and in Figure 5-7. Thus, employing a glucose oxidase EMSB would only result in improving. glucose conversion, while not 151 substantially altering the dispersion characteristics of the optimized sucrose manifold. Calibration curves were obtained for sucrose determinations under initial and optimal conditions, shown in Figure 5-20. In contrast to the Simplex results, the initial conditions were observed to yield a slightly higher sucrose sensitivity relative to the optimized conditions. The most likely explanation for this behavior is revealed by considering the overall conversion efficiency of the sucrose manifold. Substrate conversion efficiencies for the sucrose manifold under initial and optimized conditions are both nearly 100% when a glucose oxidase EMSB configuration is employed. This observation is discussed in more detail in the next section (Selectivity). Therefore, sucrose sensitivity is not going to improve with increased reactor residence times (optimized conditions) if nearly complete conversion is already obtained at shorter residence times (initial conditions). Sucrose sensitivities were very close with only a factor of 1.14 difference for initial and optimized conditions. The linear dynamic range ' for sucrose determination is approximately 2.0x10'5 to 5.4x10'4 M for a 30 01 sample volume with absorbance RSDs ranging from 0.4% to 2.1% (n=3) over the sucrose concentrations used. The detection limit for sucrose was approximately 20 11M or 7.0 ppm. Sample throughput values for initial and optimized conditions were significantly different with values of 61 and 29 samples per hour, respectively. This large difference is due to the considerably longer manifold residence time that results under optimized conditions. Considering both sensitivity and sample throughput, the initial conditions obviously result in superior manifold characteristics and are therefore the conditions of preference for sucrose determination. 152 1.0 I Initial (61 samples/hr) I Optimized (29 samples/hr) 0.8 ~ 0.6-1 Absorbance 0.00 I 0.05 ' 0.10 ' 0.15 ‘ 0.20 0.25 0.30 Sucrose (mM) FIGURE 5-20. Sucrose calibration curves for initial and optimized manifold conditions. 153 d. Selectivity Invertase (B-D-fructofuranoside fructohydrolase, EC 3.2.1.26) hydrolyzes the bond between C(2) and the glycosidic oxygen on the B-fructose moiety of sucrose. The enzyme can also hydrolyze other sugars that contain an unmodified B-D-fructose (157). Standard solutions of the nine sugars were prepared at 0.20 mM in 0.10 M phosphate bufi‘er (pH 6.00). Table 5-9 shows the relative responses of the nine sugars compared to sucrose when injected into the sucrose manifold. Responses designated by "ND" were indistinguishable from the baseline signal. Due to the presence of glucose oxidase in the manifold, interferences from sugars acting as glucose oxidase substrates will also be observed as interferences in the determination of sucrose. The sugars that exhibit interferences in the sucrose manifold are primarily the same sugars that interfere in the glucose manifold. Therefore, the observed interferences appear to be due to glucose oxidase selectivity, not invertase or mutarotase. As discussed previously, the glucose oxidase enzyme preparation contains galactose oxidase and a-glucosidase (among other enzymes) as impurities. These impurities would explain the positive interference responses of galactose and maltose. The magnitude of the interference for these two sugars is greater in the sucrose manifold when compared to the glucose manifold. However, the reactor residence times are longer in the sucrose manifold which employs a 15.0 cm glucose oxidase EMSB versus a 2.6 cm PBR used in the glucose manifold. The smaller relative interference of galactose compared to maltose is observed in both manifolds. Melibiose also interferes in sucrose determination, similar to glucose determination. These interferences may be eliminated if a purer 154 TABLE 6-9. Sucrose manifold selectivity. Sugar Relative Response Sucrose 1.0 Glucose 1.05 Galactose 0.001 Lactoee ND Maltose 0.002 Fructose N1) Helentose ND Melibiose 0.021 Reffinose 0.0013 ND=notdetected 155 source of glucose oxidase is obtained. The very small response of raffinose is actually due to non-specificity of invertase. Raffinose is a trisaccharide comprised of a-D-galactose, a-D-glucose, and B-D-fructose. Invertase exhibits a very low catalytic rate for hydroysis of raffinose at the B-D-fructose moiety. By comparing the responses of the sucrose and glucose injections, an indication of percent conversion of sucrose is revealed. The concentrations of the two sugars were the same; therefore, the responses should be identical if 100% sucrose hydrolysis was taking place. The observed responses were very close and correspond to a 95% conversion in the invertase and mutarotase reactors. 4. Lactose (B- Galactosidase) Determination of lactose requires two enzymes, B-galactosidase and glucose oxidase. The disaccharide is hydrolyzed by B-galactosidase into B-D-galactose and D-glucose, followed by B-D-glucose oxidation via glucose oxidase. ‘ B-Galactosidaee lactose + H20 ——> B-D-galactose + D-glucose Glucose (hidase B-D-glucose + 02 ——> D-gluconic acid + H202 The FIA manifold used for lactose determination is illustrated in Figure 5-21. Results from preliminary investigations for lactose determinations were discussed in Chapter IV, section D2b. Initial comparisons of B-galactosidase immobilized onto CPG and non-porous beads were presented as well as a comparison of three carrier conditions. The enzyme 156 PC/X'l‘ fl ' l L m Reactor Reactor SBSR wk» Pimp Detector FIGURE 5-21. Flow injection manifold for lactose determination. 157 exhibited greater substrate conversions at a lower pH value (6.60) and also in the presence of Mg2 ions, which enhances activity of the enzyme. a. Reactor Configuration Evaluations Reactor configuration comparisons were conducted for B-galactosidase with SBSR, EMBR, and EMSB lengths of 13.0 cm and a PBR length of 2.6 cm. The enzyme was immobilized by reacting 390 U with 0.14 g and 0.25 g of activated CPG and non-porous beads, respectively. Experimental conditions were the same as those used in glucose (section 1a) and sucrose (section 3a) reactor comparisons with the exception of the carrier and sample pH which was 6.60 and the carrier buffer included Mg+2 at a concentration of 0.01 M. The glucose oxidase reactor was a 14.0 cm SBSR. Figure 5-22 shows the reactor configuration comparison results for B-galactosidase. Calibration curves were constructed for each reactor design (Figure 5-22a) by injecting 30 pl of standard lactose solutions, ranging in concentration from 0.25 - 0.75 mM. Dispersion characteristic are shown in Figure 5-22b. Absorbance, variance, and sample throughput values are given below for the four reactors. (lactose 0.50 mM) Throughput Absorbance Variance (s2) (samples/hr) EMBR 0.149 1176 26 SBSR 0.845 404 45 EMSB 1. 192 385 46 PBR 1. 173 44 1 43 Absorbance Absorbance 2.0 158 1.6- 1.2‘ 0.8 - 0.4J OD.- EMSB PBR SBSR EMBR // 0.0 0.00 l 0.25 ‘ 0'50 0'75 Lactose (mM) 1.00 1.4 1.21 1.04 0.8 4 —- EMSB \ - - PBR — SBSR EMBR --- .... 'O C " O O —- 1 j l ' l ' l ' l 1 l 1 l 70 85 100 115 130 145 160 175 Time (sec) FIGURE 5-22. Reactor configuration comparisons for B—galactosidase, sensitivity (a), dispersion (b). 159 The PBR and EMSB configurations show comparable sensitivity (absorbance) and throughput with the EMSB exhibiting slightly better results in both characteristics. The SBSR exhibits reasonable sensitivity and the throughput compares well with the EMSB and PBR values. Low sensitivity and throughput is observed with the EMBR configuration, similar to previous results with other enzymes. b. Simplex Optimization Since lactose determination requires two enzymes, consideration was given to independent optimization of the carrier pH. However, the optimum pH ranges for both enzymes are too similar for separate optimization. The range for B-galactosidase is approximately 6.0 to 7.0 (in the presence of Na+ ions) (159), while glucose oxidase exhibits maximum activity at pH 5.0 to 6.5 (156). The reaction-rate dependent parameters optimized for lactose determination via CMS were carrier pH, carrier flow rate, and Mg+2 concentration. Sodium ions also enhance B-galactosidase activity (159); however, they are already present in the phosphate buffer employed as the carrier solution. Additional NaCl added to the carrier did not improve enzyme activity. Initial (baseline) and boundary conditions used in lactose optimization are shown in Table 5-10. The initial carrier pH selected (6.60) was reported as an optimum for B-galactosidase (159) and had provided favorable enzyme activity in preliminary studies. Carrier flow rate values were the same as those employed in the previous optimizations. Magnesium ion concentrations were given a broad range since an estimate of the optimum concentration was not predicted. An intermediate Mg+2 concentration of 0.01 M was selected as a starting 160 TABIE 5-10. Initial and boundary experimental conditions for lactose optimnation. Experimental Forward Reverse Initial Variable Boundary Boundary Conditions Carrier pH 5.00 6.00 6.60 mgmfiw 3‘“ 0.10 1.00 0.42 [n+2] (I) 0.00 0.10 0.01 161 point. This concentration had produced adequate enzyme activity in previous investigations. Reactors used in the Simplex studies (were a 2.6 cm PBR for B-galactosidase and a 14.0 cm SBSR for glucose oxidase. Five replicate injections (30 pl) of a 0.15 mM lactose solution were made for each experimental condition. The detection reagent flow rate was held constant at 0.10 mllmin based on previous optimization results. Table 5-11 shows the Simplex optimization results for lactose determination. The parameter values yielding the top four responses are given along with averaged values. The response function under optimized conditions improved by a factor of 2.2 compared with that under initial conditions. Progress of the Simplex optimization is illustrated in Figure 5-23. The response improved rapidly in the first five experiments and then leveled off; the Simplex routine was terminated after 15 experiments. Scatter diagrams for the individual parameters are shown in Figure 5-24. In Figure 5-24a, the pH optimum appears to occur between 6.0 and 6.2, decreasing from the initial value of 6.6. As expected, the carrier flow rate also decreased from initial conditions and this trend is clear in Figure 5-24b. Optimized Mg+2 concentration also decreased from the initial value, although a trend is not apparent in Figure 5-24c. For enzymes that require specific activators, these reagents usually only need to be present in slight excess, unless they become inhibitory. Thus, the Simplex procedure need only find the threshold value. Similar to glucose and sucrose, 8 second B—galactosidase reactor comparison was made between an EMSB and PBR configuration under optimized conditions. Previous reactor evaluation results showed that these two designs exhibited essentially identical activities. A 0.15 mM lactose standard was injected (30 pl) into the optimized manifold first 162 8.: 38.: 36 8.: doze: BIB—Ian c 3.: owfi n58... :6 .56 4 anfi 886 :6 36 3 3..” aged :6 and 3 Se 38 a a. a 8 e S 3 as. aafi «08.: :6 and a 133.5 :4 93.: «ed and u 33 A5056. 6 3 so. .5: an oz sag—Ion Fab-n— ear-Han horns“. .E flags-v 333— new afloat—=30 "Ian—caravans saga—o eds 133 . u a In an. 163 Response (R) I l T I I I T 17 I I l r I 1 O 2 4 6 8 10 12 1 4 1 6 Experiment Number FIGURE 5-23. Response function progress of the Simplex optimization for lactose. 1614 2.0 1 . 2.4. 8 . 0.0 1 E 2.04 1 A A 5 5 I a . . E . . g ‘02. g 1.2‘ 0.0- 0.0- 0.4: 0 0.4: o . 1 Mae'efe' 010' 012' 014' 010' eIe'7.0 M0.0 r 031* 012 ' 03 ' 014V 0.: Carrier pH Carrier Flow Rate (ml/min) ‘s.‘ . 1 '024 1 0.0 4 Response (R) 0.4 J) ( 0'0 'IrI'I'If 'I' V 0.0 2.0 4.0 0.0 0.0 10.012.014.011.0 “I“(mll) FIGURE 5-24. Scatter diagrams for individual lactose optimization variables, carrier pH (a), carrier flow rate (b), MgI‘2 concentration (0). 165 containing a B-galactosidase EMSB and then a PBR. The absorbance values for the EMSB and PBR were 0.756 and 0.737, respectively. Again, the two configurations exhibited nearly identical activities. Thus, either design could be employed for system characterization. c. Characterization System characterization for lactose determination was conducted utilizing a B-galactosidase PBR (2.6 cm) and a glucose oxidase EMSB (15.0 cm). Similar to the sucrose manifold characterization, a glucose oxidase EMSB was chosen due to the improved substrate (glucose) conversion over that of the SBSR. Calibration curves, shown in Figure 5-25, were obtained for lactose standards under initial and optimized conditions. Despite the relative response function improvement (factor of 2.2) over initial conditions, lactose sensitivities were not substantially different between optimized and initial conditions. Optimal conditions yielded a sensitivity 1.5 times that of initial conditions, with throughputs of 31 and 55 samples per hour, respectively. Since the response function is a measure of three system parameters, sensitivity only being one of the three, relative sensitivity improvements may be quite different than relative response improvements. The linear dynamic range for lactose determination under optimized conditions was approximately 2.6x10'5 to 6.2x10-4 M for a 30 pl sample volume with absorbance RSD values ranging from 0.2% to 2.2% (n=3) over the concentrations used. The detection limit for lactose was approm'mately 26 pM or 8.9 ppm. Selecting the overall superior conditions for lactose determination is somewhat dependent on the particular application and the importance of low analyte detection versus sample throughput. For determination of samples that possess low (pM) levels of lactose, the optimized conditions 166 1.0 I Optimized (31 samples/hr) I Initial (55 samples/hr) 0.8 -1 0.6- 0.4-1 I Absorbance 0.2 J 000 I 1 j I I r 0.00 0.05 0.10 0.15 r l 0.20 Lactose (mM) FIGURE 5-25. Lactose calibration curves for initial and optimized manifold conditions. 167 should be employed. Otherwise, lower detection limits can be sacrificed in order to gain sample throughput by using the initial conditions (or faster flow rates). By choosing conditions that lower the sensitivity, the linear range is also extended to higher lactose levels (> 6.2x10'4 M). d. Selectivity B-Galactosidase (B-D-galactoside galactohydrolase, EC 3.2.1.23) catalyzes the hydrolysis of B-D-galactosides (159). The best substrate for B-galactosidase is o-nitrophenyl-B-D-galactoside with a hydrolysis rate 300 times that of lactose. However, this substrate and ones like it are not an interference problem with lactose determination since D-glucose is not a hydrolysis product. Only substrates that yield glucose as a hydrolysis product are potential interferents in this system. Standard solutions of the nine sugars were prepared at 0.20 mM in 0.10 M phosphate buffer (pH 6.00). The responses relative to lactose are shown in Table 5-12. Responses designated by "ND" were indistinguishable from the baseline. Again, the interferences observed in lactose determination were those observed in glucose determination and, thus, result from the non-specificity of glucose oxidase rather than B-galactosidase. The magnitude of the interference responses for galactose, maltose, and melibiose are comparable to those obtained for the sucrose manifold. Comparing the response of glucose to lactose gives an indication of the percent lactose hydrolysis taking place. The responses correspond to a 93% reactor efficiency for B-galactosidase. 168 TABLE 5-12. Lactose manifold selectivity. Lactose 1.0 Glucose 1.00 Galactose 0.041 Sucrose ND Maltose 0.047 Fructose ND Melezitose ND. Melibiose 0.0060 Raifinose ND ND = not detected 169 5. Maltose (a-Glucosidase) Similar to the lactose manifold, maltose determination requires two enzymes, a-glucosidase and glucose oxidase. The disaccharide, composed of two glucose moieties, is hydrolyzed by a-glucosidase into a-D-glucose and D-glucose. D-Glucose (B-anomer) is subsequently oxidized via glucose oxidase to yield hydrogen peroxide. orGlucosidaee maltose + H20 ——’ a-D-glucose + D-glucose Glucose Oxidase B-D-glucose + 02 ——> D-gluconic acid + H202 The FIA manifold used for maltose determination is illustrated in Figure 5-26. a. Reactor Configuration Evaluations a-Glucosidase was immobilized onto activated CPG and non-porous beads by reacting 115 U with 0.14 g and 0.25 g support, respectively. Reactor lengths of 14.0 cm were constructed for the SBSR, EMBR, and EMSB configurations, while the PBR length was 2.6 cm. Experimental conditions were the same as those used for the glucose and sucrose reactor comparisons with a carrier and sample pH of 6.00. The glucose oxidase reactor was a 14.0 cm SBSR. Results from the reactor configuration comparisons are shown in Figure 5-27. Calibration curves were obtained for each reactor design (Figure 5-27a) by injecting 30 pl standard maltose solutions ranging in concentration from 0.50 to 1.25 mM. The corresponding dispersion characteristics are shown in Figure 5-27b. Absorbance, variance, and sample throughput values are tabulated below for the four reactors (maltose 1.25 mM). 170 PC/X'l‘ U '1 [1 Glucose mason Sam a,“ a-Glucosldese Oxidase Plain m Reactor Reactor SBSR ”L Pump Detector FIGURE 5-26. Flow injection manifold for maltose determination. 171 0.35 . 5058 0.30 - SBSR a J PBR 0.25 - EMBR 0.20 4 0.15 - 0.104 O..- Absorbance 0.05 - 0000 ' r 1 ' t l I r I T l T 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 Maltose (mM) . —- EMSB 0.30- ,\ — sass d / \ "" PBR Absorbance ' I I I I I ' I ' I ' I r I I 25 40 55 70 85 100 1 15 130 145 160 175 Time (sec) FIGURE 5-27. Reactor configuration comparisons for ot-glucosidase, sensitivity (a), dispersion (b). 172 Throughput Absorbance Variance (s2) (samples/hr) EMBR 0.142 1548 23 SBSR 0.253 473 4 1 EMSB 0.294 448 43 PBR 0.203 52 1 39 Maltose sensitivities differed substantially between the reactor configurations (a factor of 1.9 between the EMSB and EMBR). The EMSB and SBSR configurations exhibited the highest maltose sensitivity followed by the PBR and EMBR. Dispersion characteristics for the four reactors were similar to previous comparisons with the EMBR, EMSB, and PBR exhibiting comparable sample dispersion and throughput. b. Simplex Optimization Manifold parameters optimized for maltose determination were carrier pH and carrier flow rate. Optimum pH ranges for a-glucosidase and glucose oxidase are too similar for independent optimization, 6.6 to 6.8 for a-glucosidase (160) and 5.0 to 6.5 for glucose oxidase (156). Reagent flow rate was held constant throughout the Simplex optimization at 0.10 mllmin. Initial (baseline) and boundary conditions utilized in maltose determination optimization are shown in Table 5—13. An initial carrier pH value of 6.60 was chosen based on the reported optimum for the enzyme, while the carrier flow rate conditions were the same as those employed in previous optimizations. The a-glucosidase and glucose oxidase reactors were a 2.6 cm PBR and a 14.0 cm SBSR, respectively. Five replicate 173 TABII 5-13. Initial and boundary experimental conditions for maltose optimisation. Experimental Forward Reverse Initial Variable Boundary Boundary Condflions Carrier pH 5.00 6.00 6.60 “"1" "I” 3‘“ 0.10 1.00 0.42 (ml/min) 174 injections (30 pl) of a 0.75 mM maltose standard (pH 6.00) were made for each experimental condition. The results of the Simplex optimization for maltose determination are shown in Table 5-14. An average of the top three responses and corresponding parameter values were calculated to obtain the optimized conditions. The response function improved by a factor of 5.1 over the course of the optimization. Progress of the Simplex optimization is illustrated by the scatter diagram in Figure 5-28. The response improved rapidly, within the first four experiments, and then appeared to oscillate around the maximum. After 11 experiments the Simplex routine was terminated. Individual parameter trends are shown in Figure 5-29. The optimum carrier pH of approximately 6.0 was observed to decrease from the initial value of 6.6 (Figure 5-29a). This is most likely due to the lower pH optimum for glucose oxidase; a value of 6.0 represents a compromise between the two optimal pH values. The effect of carrier flow rate on sensitivity was significant, with the optimized rate of 0.11 mllmin being substantially lower than the initial rate of 0.42 ml/min. The influence of carrier flow rate on the response function is evident in the scatter diagram shown in Figure 5-29b. Following the Simplex optimization, a-glucosidase reactor comparisons were repeated for the PBR and EMSB configurations under optimized conditions. Previously the EMSB had exhibited superior characteristics with greater maltose sensitivity and sample throughput. A 0.75 mM maltose standard was injected into the manifold for re- evaluation of the two reactors. The PBR and EMSB yielded absorbances of 0.836 and 0.312, respectively. In contrast to earlier findings, the PBR 175 86 «6.: n— 6 dogs; urns—33m k." as, vs... S... 23 .85 9.... «a... 8... a 2.6 3... 8... a a a3 8.... 3... 3.» o hang“ 2.. 3... 8... a 133 36):: A5 3am to: m.— a: canoe»: 8E5 85 .E 05.53.53 8313 .5 358:8 38859.. 1.5.... e...- .53 .2 18 Mia. Response (R) 176 7.0 6.0-“ 5.0: 4.0; 3.0; 2.0.3 1.0 J 0.0 ”—1 I 4 r I 6 r 8 I 10 12 Experiment Number FIGURE 5-28. Response function progress of the Simplex optimization for maltose. Response (R) Response (R) 7.0 177 8.0 i 5.0 - 4.0 - 3.0 a 2.0 - 1.0 d 0.0 O. 7.0 5.4 ' ' I I ' l ' I ' I ' 5.8 5.8 6.0 6.2 6.4 6.6 6.8 Carrier pH 6.01 5.0: 4.0-1 3.0 - 2.0 .. 1.0 - 0.0 0.0 I 1 01.1 0.3 ' 013 014 0.5 Carrier Flow Rate (ml/min) ‘7 FIGURE 5-29. Scatter diagrams for individual maltose optimization variables, carrier pH (a), carrier flow rate (b). 178 design was superior under optimal conditions. Since the residence time for a-glucosidase appears to be very important, the conditions employed in the first reactor comparison study may not have provided adequate residence time for maltose hydrolysis in the PBR configuration, while the longer SBSR and EMSB reactors had correspondingly longer residence times. c. Characterization Manifold characterization for maltose determination was conducted using an a-glucosidase PBR (2.6 cm) and a glucose oxidase EMSB (15.0 cm). As in previous manifold characterizations, the EMSB configuration was chosen for glucose oxidase due to its greater glucose conversion efficiency over the SBSR. Calibration curves were obtained for both initial and optimized conditions (Figure 5-30). Sensitivity for maltose under optimized conditions was 3.9 times that under initial conditions, with sample throughput values of 23 and 48 samples per hour, respectively. The linear dynamic range obtained for maltose employing optimized conditions was approximately 4.9x10'i5 to 2.1x1043 M for a 30 pl sample volume with absorbance RSD values ranging from 1.1% to 2.9% (n=3) over the maltose concentrations employed. The detection limit for maltose determination was 49 pM or 16.8 ppm. (1. Selectivity a-Glucosidase (a-D-glucoside glucohydrolase, EC 3.2.1.20) hydrolyzes (IL-glucosidic linkages, preferably a-1,4-Linkages (161). a-1,6- Linkages are attacked at a slower rate while B-glucosidic linkages are not attacked. The linkage is split at the glucosidic C(1)-O-a position. Interference responses relative to maltose were obtained for the nine sugars prepared at concentrations of 0.20 mM in phosphate buffer 179 1.0 I Optimized (23 samples/hr) I Initial (48 samples/hr) 0.8- 0) o a 0.6- 13 . 8 to 0.4- .a 4 0.2“ / 0.0 ..,.,.,.,. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Maltose (mM) FIGURE 5-30. Maltose calibration curves for initial and optimized manifold conditions. 180 (pH 6.00). The results are shown in Table 5-15. Responses designated by "ND" were indistinguishable from the baseline signal. As noted earler, galactose and melibiose responses are a result of interference with glucose oxidase. The increase in relative responses observed for galactose and melibiose over those responses observed in the sucrose and lactose manifolds is most likely due to the slower carrier flow rate employed for maltose determination, allowing longer reactor residence times. Interferences resulting from a-glucosidase non-specificity are observed in the sucrose response. Sucrose is a disaccharide composed of an oc-D-glucose moiety and a B-D-fructose moiety. Sucrose acts as an a-glucosidase substrate; however, the a-D-glucose hydrolysis product is not as reactive with glucose oxidase as the B-D-glucose. Thus, the sucrose interference is fairly small despite the production of free glucose. The conversion efficiency of the a-glucosidase reactor is substantially lower than the efficiencies observed for invertase and B-galactosidase. This is also apparent in the maltose calibration curve, which exhibits lower maltose sensitivities relative to the curves obtained for sucrose and lactose. The maltose response corresponds to only a 37% conversion. Due to the lower substrate conversion, the relative response for glucose is quite high for the maltose manifold. 6. Fructose (Glucose Isomerase) The flow injection manifold used for fructose determination is shown in Figure 5-31. D-Fructose undergoes an isomerization reaction, catalyzed by glucose isomerase, to yield equilibrium products of D-glucose and D-fructose. The glucose formed in the isomerization reaction is then 181 TABLE 5-15. Maltose manifold selectivity. Maltose 1.0 Glucose 2.7 Galactose 0.34 Sucrose 0.071 Lactose ND Fructose ND Melezitose ND Melibiose 0.098 Raffinose ND ND = not detected 182 m isomerase . — Era—W 1111!.- \_7 heat Pump Detector FIGURE 5-31. Flow injection manifold for fructose determination. 183 determined by the glucose oxidase pathway, with hydrogen peroxide production. Glucose leomeraee D-fructose <——> D-glucose Glucose Oxidase B-D-glucose + 02 —> D-gluconic acid + H202 Prior to the reactor evaluations and Simplex optimization, preliminary investigations were conducted in an effort to determine satisfactory manifold conditions for such variables as carrier pH and activator concentrations. A literature survey for glucose isomerase revealed that the enzyme can exhibit optimal reaction rates anywhere from pH 6.0 to pH 8.5 depending on the microbial source (162,163). Glucose isomerase employed for fructose determination was obtained from Genencor International and was purified from Streptomyces rubiginosis. Although specific information could not be found for S. rubiginosis, general information on Streptomyces was obtained. Typically, glucose isomerase recovered from Streptomyces species has an optimal pH range from 8.0 to 8.5 (162). The enzyme also exhibits enhanced reaction rates when Mg+2 ions are present as an activator. Due to the relatively large difference between the optimal pH ranges for glucose isomerase and glucose oxidase, pH 5.0 to 6.5 (156) for glucose oxidase, a pH modifier stream was explored as a means of lowering the pH between the two enzymatic reactions. Initial experiments were conducted employing 0.10 M phosphate buffer carrier streams (0.42 mllmin) at pH 7.0 and 8.0 with and without 5.0 mM Mg”. A 30 mM fructose standard was injected into the manifold containing a glucose isomerase PBR (2.6 cm) and a glucose oxidase SBSR (14.0 cm). The highest responses were obtained with a carrier pH of 8.0 184 and with Mg’r2 ions present. In addition, a pH modifier stream was inserted after the glucose isomerase reactor to evaluate the effects of lowering the carrier pH prior to the glucose oxidase reaction. Phosphate buffer (pH 6.0) was added to the carrier at a flow rate of 0.05 mllmin. This slight pH modification resulted in a 35 percent improvement in response. Therefore, modification of the carrier pH prior to glucose oxidation was considered worthwhile in optimizing the fructose determination manifold. Despite the improved responses obtained from preliminary manifold parameter studies (carrier pH 8.0, Mg”, modifier pH 6.0), the magnitude of the response for a 30 mM fructose solution remained relatively low at 0.0127 absorbance units. Thus, the Simplex optimization studies were conducted before the reactor configuration evaluations in an effort to improve fructose determination levels to a range comparable to the other sugars (0.1 to 10 mM range). a. Simplex Optimization Three rate-dependent parameters were optimized, via CMS, for fructose determination: carrier pH, carrier flow rate, and modifier pH. Carrier pH and carrier flow rate were optimized as in all the previous manifold Optimizations. In addition, a modifier pH variable was included in the Simplex due to favorable results when the pH was lowered prior to the glucose oxidase reaction. Although the presence of Mg"2 ions enhances glucose isomerase activity, this variable was not optimized by Simplex optimization. While conducting the initial investigations with glucose isomerase, difficulties were encountered with the solubility of Mg+2 compounds at higher pH values (pH 7.0 to 8.0). Magnesium sulfate (MgSO4) was found to be more soluble than magnesium chloride (MgClz) at the required pH values, and was therefore used in all further studies. A 185 fixed concentration of 2.5 mM MgSO4 was used in the carrier stream throughout the Simplex experiments. This concentration provided adequate enzyme activity and remained in solution up to a pH of 8.0. Initial (baseline) and boundary conditions supplied to the Simplex routine are given in Table 5-16. Carrier flow rate was given the same initial and boundary values as those in previous Simplex optimizations for other manifolds. Carrier pH values were chosen based on the expected reaction rate characteristics of glucose isomerase, where a relatively high pH was predicted to be optimal. The reverse boundary for carrier pH was limited to pH 8.0, despite the possibility of a higher optimal value, due to dissolution of the silica support at pH values above 8.0 (164). Initial carrier pH was set at 7.5, a value near the expected optimum (8.0), yet not starting at the boundary. Other manifold parameters that remained constant throughout the Simplex included a modifier pH flow rate of 0.05 mllmin, a detection reagent flow rate of 0.10 mllmin, and a carrier MgSO4 concentration of 2.5 mM. The Simplex optimization was conducted with a glucose isomerase PBR (2.6 cm) and a glucose oxidase SBSR (14.0 cm). Five replicate injections (30 pl) of a 40 mM fructose solution (pH 7.5) were made for each experimental condition. Optimization results are shown in Table 5-17 where the top four responses are given with the corresponding parameter values. Averages of the top four responses and parameter values are also given with the standard deviations. Over the course of the optimization, the response function improved by a factor of 26 from initial conditions. A scatter diagram of the progress of the Simplex is shown in Figure 5-32. From the diagram it can be seen that the Simplex routine appeared lost in the first several experiments, not finding any indication of a response surface 186 TABII 5-16. Initial and boundary experimental conditions for fructose op ation. Experimental Forward Reverse Initial Variable Boundary Boundary Conditions Carrier pH 5.00 8.00 7.50 CW” ’1" 3‘“ 0.10 1.00 0.42 (ml/ min) Modifier pH 2.00 8.00 7.50 187 no; on." 8.: «ad .6335: ens-55m e no». 8.8 8.» 2... «as arts: no.3 8d 36 8.9 cm coda on.» 36 86 an 3 am 3 o 2 o 3 a. 8 a. .53 3.5 «N6 3.: v5.5 «N 783.5 .5.— cafi Nod 38 u 133— 3:58. 3 u.— 38 5E ma 6: 033 .8560: .8180 .850 .E don—I332» DIR—curd hem Idoflflaoo inasmuch”. 18390 fine 33 .h— In an. 188 32$) 28.01 24x»: 2011‘ 0 ° 161Le 12.o- . ° Response (R) ELD- 4.0 -‘ . . 0 O . 000 I T I U r Y I r U U I I r I U I 15 20 25 Experiment Number . I. 10 FIGURE 5-32. Response function progress of the Simplex optimization for fructose. 189 maximum. After experiment 15, the Simplex progressed in a new direction and as a result located the response surface "peak". Once headed in the right direction, the routine found the maximum fairly quickly. The reason for the initial confusion is due to the optimization of two pH variables, carrier and modifier. Although the optimized pH values are quite different, both pH values were changed simultaneously throughout the search procedure. This causes problems when two seemingly very different sets of experimental conditions result in similar responses. For example, a carrier pH of 7.5 and a modifier pH of 7.8 yield a similar response to that of a carrier pH of 7.0 and modifier pH of 6.8. In the first set of conditions the carrier pH value is near the optimum for glucose isomerase but the modifier pH is far fi'om optimal for glucose oxidase. Alternatively, the second set of conditions provides a more favorable pH for glucose oxidase, but a poor pH for glucose isomerase. This response behavior makes it difficult for the Simplex procedure to find the variable trends which lead to improved responses and subsequent system optimization. Individual parameter trends are shown in Figure 5-33. The scatter diagrams exhibit obvious trends for all three variables. The carrier pH optimum (Figure 5-33a) is clearly near a value of 8.0 which was the value given as the reverse boundary. Optimized carrier flow rate (Figure 5-33b) was reduced from the initial value of 0.42 mllmin to 0.10 mllmin. Similar to maltose, the substantially reduced flow rate indicates that glucose isomerase requires relatively longer residence times for substrate conversion. Finally, the modifier pH optimum (Figure 5-33c) occurs at lower pH values ranging from approximately 2.0 to 5.0. This relatively 190 a b 32.0 32.0 4 O. i . 24.0 . 24.0 - e , . e . , 3 20.0: .0 3 20.0: g 10.0 - a 10.0. ‘ O ‘ .0 a 1 : g- 12.01 . . . 12.01 . . a: 0.0 - & 0.0~ 4.0: . a 3 . 4.0:: . ; ‘ ° ~ ‘ o ' 3 . : 000 v r ' 1 ' r ' I * ifi 0.0 ' F ' 1 ' l ‘ U ' 6.0 6.5 7.0 7.5 0.0 0.3 9.0 0.0 0.1 0.2 0.3 0.4 0.5 Carrier pH Curler flow Rate (ml/min) c 32.0 ° 0 20.0 ‘ . , 24.04 g . a 20.0: . o g 10.0- ° 0 1 o . 9 12.04 . , é ..o. l o. 4.0 o 0 .1 O . . ‘. . 0'0 'r'I'IfiTff‘r'r' 1.0 2.0 3.0 4.0 5.0 0.0 7.0 0.0 2.0 Hodflierpfi FIGURE 5-33. Scatter diagrams for individual fructose optimization variables, carrier pH (a), carrier flow rate (b), modifier pH (c). 191 broad optimum modifier pH range is most likely due to the broad optimum pH range for glucose oxidase. b. Reactor Configuration Evaluations As a result of the CMS optimization, fructose sensitivity was greatly improved. A reactor comparison study was therefore conducted employing the optimized system parameters. Glucose isomerase was immobilized onto activated CPG and non- porous beads by reacting 10363 U in 1.0 ml phosphate bufl'er (pH 8.0) with 0.2 g and 0.25 g support, respectively. Reactor lengths of 15.0 cm were used for the SBSR, EMBR, and EMSB configurations, while the PBR was 2.6 cm. The glucose oxidase reactor was a 14.0 cm SBSR. Manifold parameters were as follows: carrier pH, 7.92; carrier flow rate, 0.10 mllmin; modifier pH, 3.69; modifier flow rate, 0.05 mllmin; reagent flow rate, 0.10 mllmin; Mgi'2 concentration, 2.5 mM. Reactor configuration results for glucose isomerase are shown in Figure 5-34. Calibration curves (Figure 5-34a) were obtained for each of the four designs by injecting 30 ul of standard fructose solutions in 0.10 M phosphate buffer (pH 7.50). The fructose standards ranged in concentration from 20 to 60 mM. Absorbance, variance, and sample throughput values are given below for the four reactors (fructose 60.0 mM). Throughput Absorbance Variance (82) (samples/hr) EMBR 0.916 3922 14 SBSR 0. 132 1222 26 EMSB 1. 171 1247 26 PBR 1.304 1378 24 Absorbance Absorbance 192 a 1.4 .. o PBR 1.2.1 I EMSB . 0 EMBR 1.0.. A SBSR 0.8- 0.6- 0.4- 0.2- d "T”_ - 000 I r I FI I I l l I I I I I I 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 Fructose (mM) b 1.4 - .4 —- PBR r 1.2d I, \ —- EMSB - x 2:2: 1.0~ I / \ j I l ‘ ' ..... s 0.8- ,’ / \,x I I I, ” ‘\\‘ 0.6“ I, I I” \ \s‘ q I I I, ‘s I ” \, 0.4- I I x J I I ’I o 2- 1’ ’ " ‘ . d I I ’ \‘ I I .’ ~"‘~— . 0.0 I I I l I -_-I’ l’ I l r I I I I I r r 60 90 120 150 180 210 240 270 300 330 360 Time (sec) FIGURE 5-34. Reactor configuration comparisons for glucose isomerase, sensitivity (a), dispersion (b). 193 The PBR design exhibited the greatest fructose sensitivity. However, the EMSB performance was very close to the PBR, yielding a fructose sensitivity within 10% of the PBR. The SBSR showed the poorest fructose sensitivity, while the EMBR sensitivity fell between the SBSR and the EMSB. Similar results were observed in the galactose oxidase reactor evaluations, where the EMBR performance was superior to the SBSR. Glucose isomerase reactor comparison results, like galactose oxidase, indicate that the CPG support yields significantly higher substrate conversion efficiencies than the non-porous beads. The corresponding dispersion characteristics for each reactor design are shown in Figure 5-34b. Again, the PBR, EMSB, and SBSR configurations all exhibit comparable sample dispersion with essentially identical sample throughput values. Although the EMBR design provided greater fi'uctose sensitivity than the SBSR, this reactor still exhibited the poorest sample throughput as in all previous comparisons due to laminar flow. 0. Characterization System characterization for fructose determination was conducted using a glucose isomerase PBR (2.6 cm) and a glucose oxidase EMSB (15.0 cm). Similar to the sucrose, lactose, and maltose manifold characterizations, a glucose oxidase EMSB was chosen due to the improved substrate (glucose) conversion over that of the SBSR. Prior to final manifold characterization, a comparison was made for carrier streams containing 2.5 mM and 5.0 mM MgSO4. This parameter had not been optimized via Simplex optimization due to solubility problems under basic conditions. A concentration of 5.0 mM was observed to be the approximate upper limit for solubility at pH 8.0. Although a Mg+2 concentration of 2.5 mM was used throughout the Simplex procedure, it 194 was thought that an increased Mg+2 level may improve glucose isomerase reaction rates. Under optimized conditions, a 40 mM fructose solution yielded an absorbance of 0.518 and 0.691 with carrier streams containing 2.5 mM and 5.0 mM Mg+2, respectively. Manifold characterization was therefore conducted using a carrier solution containing 5.0 mM Mg+2. Calibration curves obtained for fructose determinations under initial and optimized conditions are shown in Figure 5-35. Fructose sensitivity was 54 times greater under optimized conditions with a sample throughput of 25 samples per hour. The linear dynamic range obtained for fructose was 8.5x10'4 to 6.7x10-2 M for a 30 pl sample volume with absorbance RSD values ranging from 0.3% to 7.5% (n=3) over the range of fructose concentrations used. The high RSD value of 7.5% was obtained for the near detection limit concentration of 1.0 mM fructose. This concentration gave an average absorbance of 0.0094 with a standard deviation of 0.00071. Excluding the 1.0 mM value, the RSD ranged from 0.3% to 2.4%. The detection limit for fructose was approximately 0.8 mM or 144 ppm. (1. Selectivity Glucose isomerase (D-xylose ketol—isomerase; EC 5.3.1.5) catalyzes the isomerization between D-glucose and D-fructose in addition to D-xylose and D-xylulose, as the name implies. The enzyme has traditionally been designated as D-xylose isomerase due to its activity toward D-xylose and D-xylulose, independent of the enzyme source. Not all strains of D-xylose isomerase act on D-glucose (165); however, D-xylose preparations from Streptomyces species do exhibit D-glucose activity and are often referred to as glucose isomerase. 195 2.0 I Optimized (25 samples/hr) 0 Initial (30 samples/hr) 1.6-1 3 5 1.2-1 ..O 6 3 0.8- ‘i l 0.44 I I . . 1 . . m . 0.0 10.0 20.0 36.0 40.0 50.0 60.0 70.0 80.0 Fructose (mM) FIGURE 5-35. Fructose calibration curves for initial and optimized manifold conditions. 196 Interference responses relative to fructose, shown in Table 5-18, were obtained for the nine sugars prepared at concentrations of 1.0 mM in 0.10 M phosphate buffer (pH 7.0). Responses designated by "ND" were indistinguishable from the baseline signal. Although five out of the eight sugars do exhibit fructose determination interference, these interferences result from glucose oxidase not glucose isomerase. The observed interferences agree with those presented above for the sucrose, lactose, and maltose manifolds, which also include a glucose oxidase reactor. The magnitude of the interferences are greater for fructose due to the higher sugar concentrations (1.0 mM versus 0.20 mM); however, the relative orders of magnitude agree with previous results. As stated above, these interferences could possibly be minimized or eliminated by obtaining a purer preparation of glucose oxidase. 197 TABLE 5-18. Fructose manifold oelecfivity. Sm: 333m Fructose 1.0 Glucose >621 Galactose 19 Sucrose 1.1 lactose ND Neltoee 19 Melezitose ND Melibiose 2.2 Reffinoee ND ND = not detected CHAPTER VI CONCLUSIONS AND FUTURE PROSPECTS The original goals of this research were to develop methodologies for the determination of six nutritionally important sugars. Desirable characteristics such as high sensitivity, selectivity, and sample throughput were pursued in the manifold developments. Through consideration and investigation of various system factors (e.g., enzyme immobilization strategies, manifold designs, and system optimizations), these goals have been attained. However, the work accomplished still leaves many avenues open for further investigations. These avenues include automated multichannel parallel sugar determinations, sugar determination applications, interference minimization, alternative manifold designs, and continued exploration of enzyme immobilization techniques. Investigations in several of these areas have been initiated by other researchers in our laboratory and are introduced in the following sections. In addition, conclusions to the work described in this dissertation are presented in the appropriate sections. A) Automated Multichannel Parallel Sugar Determinations Development of the sugar determination methodologies were conducted with the intent of eventual automated, parallel detection. Each sugar manifold was developed and optimized independently to provide the desired versatility. The composite modified Simplex (CMS) routine proved very successful for individual optimizations. Improvements in the 198 199 Simplex response function and determination sensitivity for the six manifolds are given in Table 61. Sample throughput is also given for each manifold under the optimized conditions. In the case of sucrose determination, initial conditions proved superior in sensitivity and sample throughput, thus, the value for sucrose sample throughput was calculated for initial conditions. Two sugar determinations in particular, galactose and fructose, were significantly improved from initial determination results which suffered from poor selectivity and reproducibility. Sensitivity for galactose and fructose determinations improved by factors of 6.8 and 54 respectively following CMS optimization. Improvements in the other four sugar determinations were not as substantial. Maltose exhibited a marginal increase in sensitivity following CMS optimization (factor of 3.9), while sensitivities for glucose, sucrose, and lactose were nearly equivalent before and after CMS (factors of 1.2, 0.9, 1.5, respectively). These relatively small improvements in sensitivity for optimized versus initial conditions are due to the choice of initial parameter values. Initial conditions selected for glucose, sucrose, and lactose optimizations were very close to the optimized conditions found by the Simplex routine. This is illustrated in the response progress diagrams for the three optimizations (Figures 5-8, 5-18, 5-23). The response trends are essentially flat, with no significant improvement as the Simplex routine progressed. The lack of improvement for these sugars is not too surprising since initial conditions were selected from preliminary investigation results and also from literature values for enzymatic reaction-rate variables. The optimized manifolds developed for each sugar can potentially be employed in a parallel, multichannel fashion for simultaneous 200 TABLE 6-1. Simplex optimization summary. Improvement Improvement Throughput Response (R) Sensitivity (samples/ hr) Glucose 1.2 1.2 58 Galactose 5.1 6.8 41 Sucrose 1.3 0.9 61mm) lactose 2.2 1.5 31 Maltose 5.1 3.9 23 Fructose 26 54 25 201 determinations. A preliminary design of the multichannel analyzer is illustrated in Figure 6-1. In this system, a 12-channel peristaltic pump propels the carrier and reagent streams for the six channels while sample injection is accomplished by means of a dual syringe pump. The syringe pump delivers the sample stream to a 1-to-6 selection valve which subsequently injects a separate sample plug into each manifold. The appropriate enzymatic reactions take place in each manifold followed by parallel detection. This versatile design will allow determination of a single sugar, all six, or any combination in between. These options can be incorporated into the data acquisition software developed for the multichannel detector. The analyzer can also be adapted to provide additional sugar determinations by substituting or adding further manifolds. Automated sample pretreatment processes such as extractions, dilutions, or reagent additions can also be incorporated into the FIA system. This is a great advantage of this type of system over conventional sugar determination methods (e.g., GC, HPLC, chemical methods) where sample preparation is generally a completely separate operation from sample analysis. Substantial progress has been made toward the development of the multichannel analyzer. This dissertation has presented the individual sugar manifold developments and optimizations. Work has also been conducted on the multichannel detector. A flow cell module, complete with source and phototransducer, was designed and developed by E. Castellanos (166). The cell was tested under typical system parameters with the detection of malachite green. Cell performance was comparable to the colorimeter employed in developing the manifolds. 202 FIGURE 6-1. 5' s wanna—Inlet: Q Parallel, multichannel flow injection analyzer. 203 B) Sugar Determination Applications and Interference Minimization Analytical figures of merit are presented in Table 6-2 for each sugar determination manifold. Manifold characteristics are given for determinations under optimized conditions except for those of sucrose, where the initial conditions were observed to be superior. Relative standard deviations (RSDs) were calculated for the mid-range standard used in constructing the calibration curves, shown in Figures 5-10 (glucose), 5-15 (galactose), 5-20 (sucrose), 5-25 (lactose), 5-30 (maltose), and 5-35 (fructose). Preliminary studies have been conducted in the areas of real sample analysis and interference minimization. Two researchers in our laboratory have evaluated the FIA/immobilized enzyme approach to sugar determinations in food analysis and compared the FIA results to official methods. First, S. Karayanni determined glucose, sucrose, and maltose in several food samples (wheat flour, light honey, wine, and a soft drink) (90). For the food samples investigated, the FIA determination results were comparable to the official methods (5); however, the FIA determinations were observed to be more rapid and selective than the official methods. A second researcher, P. Aspris, employed the FIA system for the determination of glucose, sucrose, and fructose in several fruits (oranges, lemons, grapefruit, olives, and cherries) (147). Interference evaluations and minimizations are essential for the sugar determinations. If the sample composition is known, evaluation of interferences is much easier than if the sample composition is not known. Ideally, all possible interferences would be completely eliminated for each determination, thus making the knowledge of the sample matrix less 204 TABLE 6-2. Analytical figures of merit for the determination manifolds. I ' D t cti Thro t at? uiit (Sin as (x). (smpiir) Glucose #1332: i 11 0.27 so Galactose $2332"; in so 0.35 41 Sucrose 2233:: i 20 0.45 61 Lactose 3:33: i 28 0.57 31 Maltose 3233: i 49 1.5 23 Fructose 3:33;; 300 0.27 25 ‘ Calculated for mid-range standard used in calibration curves. 205 crucial. However, a more realistic approach in most analytical determinations is to eliminate or minimize the most common interferences, while still obtaining as much information as possible about the sample. Several interferences are anticipated in determining the six sugars of interest. The most likely interference to be encountered in the sugar determinations will occur when glucose is present in the sample. For those manifolds that contain a glucose oxidase reactor in addition to a glucose-forming reactor (sucrose, lactose, maltose, and fructose), a positive error will result for determinations where glucose is also present. 'l\vo possible approaches can be taken to minimize or eliminate this interference. The simplest approach is to subtract the concentration obtained for glucose in the glucose channel from the values for sucrose, lactose, maltose, and fructose. Although this correction is easily accomplished and can be readily incorporated into the acquisition software, significant determination errors can result for samples where the glucose concentration is substantially larger than the sugar of interest. The second possible approach in eliminating the glucose interference is to destroy any glucose present in the sample prior to the glucose-forming reactor in the sucrose, lactose, maltose, and fructose channels. Destruction of glucose can be accomplished by a series of three enzymatic reactions. Mutarotue a—D-glucose 4———-> flD-glucose Glucose Oxidase B-D-glucose + 02 ———> D-gluconic acid + H202 Catalan 2H202 ——> 2H20 + 02 206 First, mutarotase must be employed as a means of converting a-D-glucose to B-D-glucose since glucose oxidase is selective for the B anomer. The glucose oxidase reaction is subsequently used to convert glucose to gluconic acid with the production of hydrogen peroxide. Finally, catalase is used to destroy the hydrogen perom'de formed from the glucose oxidase reaction. The catalase reaction yields oxygen and water which are not interfering species in the sugar determinations. In addition to glucose interference, another type of interference expected in the sugar determinations is due to enzyme non-specificity, where additional components (sugars) also act as enzyme substrates. These interferences were evaluated for each sugar determination and were presented in Chapter V. The predominant problem due to enzyme non-specificity was encountered with glucose oxidase. The enzyme itself is specific for B-D-glucose; however, the enzyme preparation obtained from Sigma contains small amounts of galactose oxidase and maltase (a-glucosidase) as contaminants. Thus, galactose, maltose, and melibiose all exhibit interference responses in any channel that contains glucose oxidase, which includes all sugar manifolds except that for galactose. As stated earlier, this interference could very likely be eliminated or reduced substantially if a purer form of glucose oxidase is obtained. Other than glucose oxidase, the only other enzyme exhibiting non- specificity interferences was galactose oxidase. Lactose, melibiose, and raffinose all exhibited interferences with galactose determination. Melibiose and raffinose were significant at 54% and 100% of the galactose response, respectively. In this case, the interference is not due to enzyme contamination. Galactose oxidase also accepts these other sugars as 207 substrates and, in the case of rafiinose, can yield higher relative reaction rates than for galactose. A similar non-specificity type of interference is observed with ascorbic acid and the peroxidase-catalyzed detection reaction. Ascorbic acid present in a sample acts as an oxygen acceptor for the perom’de/peroxidase complex (167). This interferes with the detection reaction because the desired oxygen acceptor, leucomalachite green, must compete with the ascorbic acid. Investigations conducted by S. Karayanni showed that the ascorbic acid interference in the Trinder reaction could essentially be eliminated by employing ascorbate oxidase (EC 1.10.3.3). This enzyme oxidizes ascorbic acid, rendering it inactive toward the peroxide/peroxidase complex (168). Ascorbate oxidase treatment, although successful for the Trinder reaction, must be re-evaluated for the leucomalachite green reaction. Other types of interferences that should be evaluated are those from non-sugar components present in the sample which may interfere with the enzymatic reactions or the detection reaction. The most probable interference of this type would result from sample components that inhibit enzyme activity. C) Enzyme Immobilization There are two alternative enzyme immobilization methods that would be very interesting to explore in future work with the sugar determinations. The first method employs immunochemicals for the immobilization of an enzyme onto a solid support. In general, the immunoaffinity immobilization methods bind one half of an antigen/antibody complex to a support surface while the other half is 208 labeled with the enzyme (169). When combined in this manner, a reversible immobilized complex is formed. Because the antigen/antibody complex can be severed under certain chemical conditions, typically by altering the pH and/or ionic strength, it is possible to regenerate the reactor by applying a new aliquot of the enzyme-labeled antigen or antibody. This can be advantageous over irreversible covalent attachment of the enzyme where once the reactor activity is no longer adequate the enzyme and support must be discarded. Several coupling techniques have been investigated for binding antigens and antibodies to various supports (169-171). The antigen or antibody can be directly bound to the support or indirectly bound using coupling reagents such as avidin and biotin (170,171). Comparisons of enzyme activity, stability, and lifetime for the immunochemical immobilization methods versus the covalent attachment method (employed in this work) would be extremely interesting. The second alternative method involves co-immobilization of two or more enzymes onto a support. If the co-immobilization is successful, reaction yields for equilibrium enzymatic reactions can be enhanced. For example, in sucrose determination, three enzymes are required for hydrolysis of sucrose (invertase), mutarotation of a-D-glucose to B-D-glucose (mutarotase), and finally B-D-glucose oxidation (glucose oxidase). If these three enzymes were all present simultaneously, the reaction equilibria would be driven to the product sides, since the products for one reaction are the reactants for the next. Typically in co-immobilization, the enzymes are combined together in a reaction mixture and then exposed to the activated support. An alternative approach to co-immobilization would be to mix the enzyme bound supports together following individual immobilization. Thus, 209 optimal immobilization conditions could be used for each enzyme since they are immobilized separately and only mixed prior to reactor preparation. A primary disadvantage of co-immobilization or mixed enzyme reactors is that a compromise must be found for such reaction parameters as pH, if the enzymes require different conditions for optimal activity. Thus, co-immobilization or mixed bed reactors would be most suitable for enzymes with similar optimal reaction-rate parameters (e.g. buffer, pH, temperature, activators). D) Alternative Manifold Designs Another area where future investigations may prove fruitful is exploration of alternative manifold designs. In particular, the sugar determination manifolds developed and optimized in this work could be scaled down to a microconduit FIA system (172,173)). Several advantages are possible in miniaturizing the FIA-based sugar determinations. First, from just a physical standpoint, the outer dimensions of the manifold are greatly reduced over conventional FIA systems. More important analytically, greater reproducibility can be attained due to the rigid position provided by the manifold block, particularly if the injection and detection components are integrated into the microconduit as well. Lower sample and reagent volumes are required in miniaturized systems, which is advantageous when reagents are expensive and/or hazardous. In the case of the sugar determination manifolds, the quantity of immobilized enzymes required for the reactors will most likely be reduced. This should result in an economic advantage over the larger FIA manifolds. A 210 miniaturized system would also be advantageous for eventual field use, where portable, lightweight analytical instrumentation is preferred. 10. 11. 12. 13. 14. 15. 16. 211 LIST OF REFERENCES M.W. Kearsley in Analysis of Food Carbohydrate, (G.G. Birch, Ed.), Elsevier Applied Science Publishers: New York, 1985. OK. Lee in Developments in Food Analysis Techniques - 1, (RD. King, Ed.), Applied Science Publishers LTD: London, 1978. M.F. Chaplin in Carbohydrate Analysis: A Practical Approach, (M.F. Chaplin and J .F. Kennedy, Eds.), IRL Press: Oxford, 1986. CA White and J .F. Kennedy in Carbohydrate Analysis: A Practical Approach, (M.F. Chaplin and J .F. Kennedy, Eds.), IRL Press: Oxford, 1986. AOAC, Ofi‘icial Methods of Analysis, 14th ed., Association of Official Analytical Chemists, Arlington: 1984. R. Macrae in Analysis of Food Carbohydrate, (G.G. Birch, Ed.), Elsevier Applied Science Publishers: New York, 1985. DJ. Folkes in Analysis of Food Carbohydrate, (G.G. Birch, Ed.), Elsevier Applied Science Publishers, New York: 1985. NR. Williams, Carbohydr. Chem, 17 (1985) 229. SC. Churms, Ed., CRC Handbook of Chromatography: Carbohydrates, Vol. II, CRC Press, Boca Raton: 1990. RE. Shaw, CRC Handbook of Sugar Separations in Foods by HPLC, CRC Press: Boca Baton, 1988. E. Raiakyla and M. Paloposki, J. Chromatogr., 282 (1983) 595. P. Nordin, Anal. Biochem., 131 (1983) 492. P. Vratny, U. Brinkman, R. Frei, Anal. Chem., 57 (1985) 224. CE. Damon and B.C. Pettitt, J. Assoc. Off: Anal. Chem, 63 (1980) 476. CC. Chen and GD. McGinnis, Carbohydr. Res, 122 (1983) 322. CH. Lochmuller and W.B. Hill, Jr., J. Chromatogr., 264 (1983) 215. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 212 F. Nachtmann and K.W. Budna, J. Chromatogr., 136 (1977) 279. C.A. White, S.W. Vass, J.F. Kennedy, and D.G. Large, J. Chromatogr., 264 (1983) 99. MA. Lawson and G.F. Russel, J. Food Sci, 45 (1980) 1256. G.K. Grimble, H.M. Barker, R.H. Taylor, Anal. Biochem., 128 (1983) 422. D. Betteridge, N.G. Courtney, T.J. Sly, D.G. Porter, Analyst, 109 (1984) 91. DC. Johnson, Nature, 321 (1986) 451. CC. Neuburger and DC. Johnson, Anal. Chem, 59 (1987) 150. CC. Neuburger and DC. Johnson, Anal. Chem, 59 (1987) 204. D.S. Bindra and GS. Wilson, Anal. Chem, 61 (1989) 2566. LE. Welch, D.A. Mead, Jr., D.C. Johnson, Anal. Chim. Acta., 204 (1988) 323. S.V. Prabhu and RP. Baldwin, Anal. Chem, 61 (1989) 852. J. Wang and Z Taha, Anal. Chem, 62 (1990) 1413. H.U. Bergmeyer, Ed., Methods of Enzymatic Analysis, 3rd Ed., Vol. 6, Verlag Chemie meH: Weinheim, 1984. P.W. Carr and L.D. Bowers, Immobilized Enzymes in Analytical and Clinical Chemistry, John Wiley & Sons: New York, 1980, pg 131. H.A. Mottola, Analyst, 112 (1987) 719. J .M. Nelson and E.G. Griffin, J. Amer. Chem. Soc., 38, (1916), 1109. H.H. Weetall, Anal. Chem, 46 (1974) 602A. L.D. Bowers, TRAC, 1 (1982) 191. J .F. Kennedy and C.A. White, Enzyme Microb. Technol., 2 (1980) 82. H.H. Weetall in Advances in Experimental Medicine and Biology. Vol. 42, Immobilized Biochemicals and Aflinity Chromatography, (Bruce Dunlap, Ed.), Plenum Press: New York, 1974. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 213 D.L. Eaton in Advances in Experimental Medicine and Biology. Vol. 42, Immobilized Biochemicals and Affinity Chromatography, (Bruce Dunlap, Ed.), Plenum Press: New York, 1974. O. Zaborsky, Immobilized Enzymes, CRC Press: Cleveland, 1973. T.M.S. Chang, Ed., Biocedical Applications of Immoblized Enzymes and Proteins, Vol 1, Plenum Press: New York, 1977, pg 7-35. H.H. Weetall, Biochim. Biophys. Acta, 212 (1970) 1. K. Mosbach, Ed., Methods in Enzymology, Vol. 44, Immobilized Enzymes, Academic Press: New York, 1976. M.D. Trevan, Immobilized Enzymes, John Wiley & Sons: New York, 1980. G. Melrose, Pure Appl. Chem, 21 (1971) 83. H.H. Weetall in Methods in Enzymology, Vol. 44, Immobilized Enzymes, (K. Mosbach, Ed.), Academic Press: New York, 1976, pg 134-148. P.W. Carr and L.D. Bowers, Immobilized Enzymes in Analytical and Clinical Chemistry, John Wiley & Sons: New York, 1980, pg 164-180. G.G. Guilbault, Analytical Uses of Immoblized Enzymes, Marcel Dekker: New York, 1984. L.D. Bowers, Anal. Chem, 58 (1986) 513A. P.W. Carr and L.D. Bowers, Immobilized Enzymes in Analytical and Clinical Chemistry, John Wiley & Sons: New York, 1980, pg 197-310. L.D. Bowers and P.W. Carr, Anal. Chem, 48 (1976) 545A. R.K. Kobos, TRAC, 6 (1987) 6. F. Scheller, D. Pfeifi‘er, R. Hintsche, I. Dransfeld, F. Schubert, U. Wollenberger in Biosensors: Applications in Medicine, Environmental Protection, and Process Control, (R.D. Schmid and F. Scheller, Eds.), VCH Publishers: New York, 1989. B.A.A. Dremel, B.P.H. Schafi‘ar, R.D. Schmid, Anal. Chim. Acta, 225 (1989) 193. H. Gunasingham, C.H. Tan, J .K.L. Seow, Anal. Chem, 62 (1990) 755. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 65. 66. 67. 68. 69. 70. 71. 72. 214 OS. Wolfbeis and E. Koller in Biosensors: Applications in Medicine, Environmental Protection, and Process Control, (R.D. Schmid and F. Scheller, Eds.), VCH Publishers: New York, 1989. W. Trettnak, M.J.P. Leiner, O.S. Wolfbeis, Analyst, 113 (1988) 1519. M.C. Gosnell, R.E. Snelling, H.A. Mottola, Anal. Chem, 58 (1986) 1585. J.M. Reijn, W.E. van der Linden, H. Poppe, Anal. Chim. Acta, 123 (1981) 229. P.W. Carr and L.D. Bowers, Immobilized Enzymes in Analytical and Clinical Chemistry, John Wiley & Sons: New York, 1980, pg 435-441. H. Hwang and PK. Dasgupta, Anal. Chem, 59 (1987) 1356. J. Ruzicka and EH. Hansen, Flow Injection Analysis, 2nd ed., John Wiley & Sons: New York, 1988. K.K. Stewart, G.R. Beecher, and RE. Hare, Fed. Proc., 33 (1974) 1439. J. Ruzicka and EH. Hansen, Anal. Chim. Acta, 78 (1975) 145. R. Tiissen, Anal. Chim. Acta, 114 (1980) 71. J.H.M. van den Berg, R.S. Deelder, and H.G.M. Egberink, Anal. Chim. Acta, 114 (1980) 91. WE. van der Linden, TRAC, 1 (1982) 188. J .M. Reijn, H. Poppe, and WE. van der Linden, Anal. Chim. Acta, 145 (1983) 59. J .M. Reijn, H. Poppe, and W.B. van der Linden, Anal. Chem, 56 (1984) 943. G. Horvai and E. Pungor, CRC Crit. Rev., 17 (1987) 231. CM. Selavka, D.S. Jiao, and LS. Krull, Anal. Chem., 59 (1987) 2221. M. Uihlein and E. Schwab, Chromatographic, 15 (1982) 140. D. Betteridge, P.B. Oates, and AP. Wade, Anal. Chem., 59 (1987) 1236. A. Rios, M.D. Luque de Castro, and M. Valcarcel, Anal. Chem., 60 (1988) 1540. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 215 J. Toei, Analyst, 113 (1988) 475. J. Toei, Talanta, 36 (1989) 1233. BR. Kowalski, J. Ruzicka, and GD. Christian, TRAC, 9 (1990) 8. GE. Pacey and RB. Bubnis, TRAC, 6 (1987) 165. J. Ruzicka, Fres. Z. Anal. Chem., 329 (1988) 653. T. Matsue, A. Aoki, E. Ando, and I. Uchida, Anal. Chem., 62 (1990) 407. G. Christian, I. Krull, and J. Tyson, Anal. Chem., 62 (1990) 455A. J. Ruzicka and EH. Hansen, Anal. Chim. Acta, 179 (1986) 1. J. Ruzicka and EH. Hansen, Anal. Chim. Acta, 214 (1988) 1. P.J. Worsfold, Anal. Chim. Acta, 145 (1983) 117. H.A. Mottola, Anal. Chim. Acta, 145 (1983) 27. G. Johansson, L. Ogren, and B. Olsson, Anal. Chim. Acta, 145 (1983) 71. B. Olsson and L. Ogren, Anal. Chim. Acta, 145 (1983) 87. R.Q. Thompson, H. Kim, and CE. Miller, Anal. Chim. Acta, 198 (1987) 165. R.Q. Thompson and H. Kim, Biochim. Biophys. Acta, 964 (1988) 116. J. Ruz, A. Rios, M.D. Luque de Castro, and M. Valcarcel, Anal. Chim. Acta, 211 (1988) 281. B. Olsson, B. Staalbom, and G. Johansson, Anal. Chim. Acta, 179 (1986) 203. S.M. Tzouwara-Karayanni and SR. Crouch, Food Chem., 35 (1990) 109. P. Linares, M.D. Luque de Castro, and M. Valcarcel, Anal. Chim. Acta, 202 (1987) 199. C.A. Swindlehurst and TA. Nieman, Anal. Chim. Acta, 205 (1988) 195. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 216 T. Dullau, B. Reinhardt, and K. Schuegerl, Anal. Chim. Acta, 225 (1989) 253. M. Akiba and S. Motomizu, Anal. Chim. Acta, 214 (1988) 455. BA. Petersson, Anal. Lett., 22 (1989) 83. C.L.M. Stults, A.P. Wade, and S.R. Crouch, Anal. Chim. Acta, 192 (1987) 155. B. Olsson, H. Lundback, and G. Johansson, Anal. Chim. Acta, 167 (1985) 123. M. Skoog, G. Johansson, B. Olsson, and R. Appelqvist, Mikrochim. Acta, 3 (1988) 131. BA. Petersson, Anal. Chim. Acta, 209 (1988) 231. K. Matsumoto, H. Kamikado, H. Matsubara, and Y. Osajima, Anal. Chem., 60 (1988) 147. E. Dominguez, B. Hahn-Haegerdal, G. Marko-Varga, and L. Gorton, Anal. Chim. Acta, 213 (1988) 139. LA Larew, D.A. Mead, Jr., and DC. Johnson, Anal. Chim. Acta, 204 (1988) 43. J. Emneus and L. Gorton, Anal. Chem., 62 (1990) 263. H. Lundback and B. Olsson, Anal. Lett., 18 (1985) 871. W. Matuszewski and M. Trojanowicz, Analyst, 113 (1988) 735. X. Yang, G. Johansson, and L. Gorton, Mikrochim Acta, 1 (1989) 9. J. Abdul Hamid, G.J. Moody, and J .D.R. Thomas, Analyst, 114 (1989) 1587. L. Olsson and CF. Mandenius, Anal. Chim. Acta, 224 (1989) 31. R. Appelqvist, G. Marko-Varga, and L. Gorton, Anal. Chim. Acta, 169 (1985) 237. M. Masoom and A. Townshend, Anal. Chim. Acta, 166 (1984) 111. M. Masoom and A. Townshend, Anal. Chim. Acta, 171 (1985) 195. KP. Ang, H. Gunasingham, B.T. Tay, V.S. Herath, P.Y.T. Teo, P.C. Thiak, B. Kuah, and KL. Tan, Analyst, 112 (1987) 1433. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 217 G.J. Moody, G.S. Sanghera, and J .D.R. Thomas, Analyst, 112 (1987) 65. J. Abdul Hamid, G.J. Moody, and J .D.R. Thomas, Analyst, 113 (1988) 81. SK. Beh, G.J. Moody, and J .D.R. Thomas, Analyst, 114 (1989) 29. G. Joensson and L. Gorton, Anal. Lett., 20 (1987) 839. G.J. Moody, G.S. Sanghera, and J .D.R. Thomas, Analyst, 111 (1986) 605. G. Marko-Varga, R. Appelqvist, and L. Gorton, Anal. Chim. Acta, 179 (1986) 371. T. Yao, M. Sato, Y. Kobayashi, and T. Wasa, Anal. Chim. Acta, 165 (1984) 291. GS. Sanghera, Anal. Proc., 25 (1988) 108. K. Matsumoto, O. Hamada, H. Ukeda, and Y. Osajima, Anal. Chem., 58 (1986) 2732. T. Buch-Rasmussen, Anal. Chem., 62 (1990) 932. B. Olsson, H. Lundback, G. Hohansson, and F. Scheller, Anal. Chem., 58 (1986) 1046. R.Q. Thompson, Ph.D. Dissertation, Michigan State University, East Lansing, MI, 1982. C.L.M. Stults, Ph.D. Dissertation, Michigan State University, East Lansing, MI, 1987. R.Q. Thompson, C.J. Patton, and S.R. Crouch, Anal. Lett. 16 (1983) 219. P. Trinder, Ann. Clin. Biochem., 6 (1969) 24. D. Barham and P. Trinder, Analyst, 97 (1972) 142. C.J. Patton and S.R. Crouch, Anal. Chim. Acta, 179 (1986) 189. P.W. Carr and L.D. Bowers, Immobilized Enzymes in Analytical and Clinical Chemistry, John Wiley & Sons: New York, 1980, pg 172. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 218 EL Onuska, M.E. Comba, T. Bistricki, and R.J. Wilkinson, J. Chromatogr., 142 (1977 ) 117. S.K. Dahodwala, M.K. Weibel, and AB. Humphrey, Biotechnol. Bioeng., 18 (1976) 1679. GB. Gyrkacz, R.D. Libby, and GA. Hamilton, J. Am. Chem. Soc., 98 (1976) 626. E. Grushka and E.J. Kikta, Jr., Anal. Chem., 49 ( 1977) 1005A. G.J. Little, J .A. Whatley, A.D. Dale, and MB. Evans, J. Chromatogr., 171 (1979) 435. C. Fulcher, M.A. Crowell, R. Bayliss, K.B. Holland, and J .R. J ezorek, Anal. Chim. Acta, 129 (1981) 29. T.G. Waddell, D.E. Leyden, and MT. BeBello, J. Am. Chem. Soc., 103 (1981) 5303. GS. Caravajal, D.E. Leyden, G.R. Quinting, and GE. Maciel, Anal. Chem., 60 (1988) 1776. M.I. Karayannis, unpublished work, Michigan State University, 1987. Electro-Nucleonics, "CPG-10: Controlled-Pore Glass", Electrop- Nucleonics, Inc.: New Jersey. P.D. Boyer, Ed., The Enzymes, Volume VII, 3rd. Ed., Academic Press: New York, 1972, pg 645-648. R.E. Snelling and H.A. Mottola, Anal. Chim. Acta, 200 (1987) 503. C.A. Cramers, J .A. Rijks, and C.P.M. Schutjes, Chromatographia, 14 (1981) 439. D. Betteridge, C.Z. Marczewski, and A.P. Wade, Anal. Chim. Acta, 165 (1984) 227. CC. Painton and H.A. Mottola, Anal. Chim. Acta, 158 (1984) 67. J .M. Hungerford and GD. Christian, Anal. Chim. Acta, 200 (1987) 1. P. Aspris, Masters Thesis, Michigan State University, East Lansing, MI, 1990. D. Betteridge, A.P. Wade, and AG. Howard, Talanta, 32 (1985) 709. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 219 D. Betteridge, A.P. Wade, and AG. Howard, Talanta, 32 (1985) 723. D. Betteridge, T.J. Sly, A.P. Wade, and J .E.W. Tillman, Anal. Chem., 55 (1983) 1292. D. Betteridge, T.J. Sly, A.P. Wade, and D.G. Porter, Anal. Chem., 58 (1986) 2258. J. Whitaker, Principles of Enzymology for the Food Sciences, Marcel Dekker, Inc.: New York, 1972, pg 562. H.U. Bergmeyer, Ed., Methods of Enzymatic Analysis, 3rd Ed., Vol. 6, Verlag Chemie meH: Weinheim, 1984, pg 281-288. P.D. Boyer, Ed., The Enzymes, Volume XII, 3rd. Ed., Academic Press: New York, 197 2, pg 531. RA. Schlegel, C.M. Gerbeck, and R. Montgomery, Carbohyd. Res., 7 (1968) 193. J. Whitaker, Principles of Enzymology for the Food Sciences, Marcel Dekker, Inc.: New York, 1972, pg 561-570. J. Whitaker, Principles of Enzymology for the Food Sciences, Marcel Dekker, Inc.: New York, 1972, pg 459-462. C.A. Koerner and TA. Nieman, Anal. Chem., 58 (1986) 116. J. Whitaker, Principles of Enzymology for the Food Sciences, Marcel Dekker, Inc.: New York, 1972, pg 462-467. H.U. Bergmeyer, Ed., Methods of Enzymatic Analysis, 3rd Ed., Vol. 6, Verlag Chemie meH: Weinheim, 1984, pg 122. C. Bucke in Developments in Food Carbohydrate - 3: Disaccharidases, (C.K. Lee and MG. Lindley, Eds.), Applied Science Publishers: London, 1982. B.K. Hamilton, C.K. Colton, and CL. Cooney, in Immobilized Enzymes in Food and Microbial Processes, (O. Cossey, Ed.), pg 85. C. Bucke in Enzymes and Food Processing (G.G. Birch, N. Blakebrough, K.J. Parker, Eds.), Applied Science Publishers: London, 1981, pg 51. K.K. Unger, Porous Silica: its Properties and Use as Support in Column Liquid Chromatography, Elsevier Scientific Publishing Co.: New York, 1979. 165. 166. 167. 168. 169. 170. 171. 172. 173. 220 P.D. Boyer, Ed., The Enzymes, Volume VI, 3rd. Ed., Academic Press: New York, 1972, pg 349-354. E. Castellanos, unpublished M.S. research, Michigan State University, East Lansing, MI, 1990. B. Chance, J. Biol. Chem., 151 (1943) 553. S. Karayanni, submitted for publication, 1991. W.U. de Alwis, B.S. Hill, B.I. Meiklejoohn, and GS. Wilson, Anal. Chem., 59 (1987) 2688. S. Luo and DR. Walt, Anal. Chem., 61 (1989) 1069. RC. Gunaratna and GS. Wilson, Anal. Chem., 62 (1990) 402. W.E. van der Linden, TRAC, 6 (1987) 37. J. Rusicka and EH. Hansen, Flow Injection Analysis, 2nd ed., John Wiley & Sons: New York, 1988, pg 247—256. “lililllllllllli