ANALYST AUGUST 1986 VOL. 11 I 865 Enzymatic Determination of Urea in Water and Serum by Optosensing Flow Injection Analysis Tracy D. Yerian Gary D. Christian* and Jaromir RdLiEkat Center for Process Analytical Chemistry Department of Chemistry BG- 70 University of Washington, Seattle WA 98795 USA Reflectance spectrophotometry was applied to the flow injection measurement of pH in order to perform an assay of urea in water and serum samples via the enzymatic (urease) degradation of urea to ammonia and hydrogen carbonate. Detection is based on commercial non-bleeding acid - base indicator papers situated in the flow stream of an integrated micro-conduit at the tip of a fibre-optic bundle. The sample and reagent are injected via a split-loop injection technique. The serum is analysed by a stopped-flow kinetic measurement to avoid errors in measurements due to the variability in the background sample pH.Keywords 1 Urea assay; optosensing; reflectance flow injection analysis; enzymatic degradation Two general types of optical fibre sensors have been devel-oped for the analysis of chemical species.’ Fibres can be used simply as light carriers with no transducers at the end. More recently sensors using the immobilisation of reagents and the integration of optical fibres have been designed. When a single fibre is used it is necessary to distinguish between the incident and detected radiation either temporally or by wavelength. The main advantage of using a single fibre in a “probe” sensor is that an extremely small probe size becomes possible and this has advantages in bioanalytical techniques.* Other potential advantages of using optical fibres for in vivo sensing include their mechanical flexibility the absence of risk of electric shocks inexpensive construction transmission of only low-energy radiation and relatively good biocompatibility .3 Single fibre systems are also easily applicable to fluorescence measurements as the wavelengths of the detected and incident radiation are different. A pH sensor based on the fluorescence of cellulose-immobilised fluoresceinamine has been reported4; non-fluorescent reagents that form fluores-cent complexes with metals have also been incorporated into fibre-optic sensors.536 In bifurcated optical sensors separate fibres transmit incident and detected radiation and only that part of the reagent phase that falls within both the cone of incidence and the cone of detected radiation is observed.Chemical sensors based on fibre optics may be classified as reversible in which the reagent phase is not consumed by its interaction with the analyte or non-reversible in which the reagent phase is consumed. Optical sensors that are based on the use of reagents immobilised on a solid support have been developed for a number of chemical species. When solid support matrices are employed in sensors it is difficult to measure the transmitted light. In these instances the intensity of the reflected light can be used as a measure of the change in colour of an immobilised reagent phase. The feasibility of using solid-phase immobilised dyes as reversible reagents has been demonstrated by Goldfinch and Lowe in their optoelec-tronic sensors for serum albumin urea penicillin and glu-cose.7,8 pH probes have been developed based on reflectance measurements from indicator dyes immobilised on micro-spheres or polymer supportsslJ; this same type of sensor design has also been adapted to O2 and glucose measure-ment.1 2 ~ 3 * To whom correspondence should be addressed. i Present address Chemistry Department A Building 209, Technical University of Denmark DK-2800 Lyngby Denmark. The fibre-optic sensor coupled to an immobilised chemical reagent has been studied by Kirkbright et al. in a flow system. 14 RfiiiZka and Hansen have incorporated reflectance spectrophotometry via optical fibres into a flow injection analysis (FIA) scheme “optosensing.”ls The solid support at the end of the fibre is situated in a flow path and this flow cell is incorporated into the integrated microconduits described by RfiiiCka in earlier work.16 In this study the integrated microconduit design is used with some improvements on the first microconduit design that incorporated the solid support interfaced to a fibre-optic bundle.The flow of solutions through the cell serves two purposes to transport the sample zone through the detector in the FIA mode and to renew the reagents which cannot be or intentionally are not immobi-lised within the optosensor. This illustrates a basic advantage of the integrated microconduit over a probe design irrevers-ible chemistry can be performed without resorting to “dispos-ables”; also rather than a continuous signal there is a base line avoiding any problems resulting from base-line drift and allowing for easy calibration; and sample pre-treatment can be integrated into the system.For most clinical measure-ments ex vivo techniques in which a single compact portable system can be used to monitor many patients are far more cost effective than in vivo sensors.17 An ex vivo analyser such as ours can also be connected to an invasive conduit serving as a single-patient dedicated instrument. From the viewpoint of process chemistries the incorpora-tion of the fibre allows sampling to be remote from the instrumentation without incorporating a long process time or dispersion of the signal.The solutions have been separated from the expensive instrumentation; a plastic fibre bundle and PVC microconduit are inexpensive and easily replaced. The solid support or “active surface,” implemented in our system is the ColorpHast (non-bleeding) indicator strips. These are cellulose fibres on which acid - base indicators have been covalently immobilised. They are Merck products; the different strips available collectively span the pH range 0.0-13.0 and each individual cellulose pad will change colour over a range of 2.0-3.5 pH units. A detection system developed around a pH change is very non-specific. Enzymes have been recognised in analytical chemistry as potential reagents for over 50 years because of their extreme specificity in a complex matrix.J8.l9 Conve-niently many enzymes also have a pH change associated with the reaction that they catalyse. Enzymatic reactions have been coupled to glass electrode detection systems over the last 20 years in many different ways.20-*3 However optical systems have some inherent advantages over electrodes the signal is not subject to electrical interference; the reagent phase doe 866 ANALYST AUGUST 1986 VOL. 111 not have to contact the fibre optic physically (so it is a simple matter to change the reagent phase); an optical sensor can offer cost advantage over electrodes especially if a single spectrophotometer is used with several sensors; the optical system can take advantage of multi-wavelength and temporal information; no “reference” is required; liquid junction potentials and streaming potentials are avoided; the response time is faster especially in low buffer capacity solutions; an optical system is less sensitive to temperature pressure and large background interferences such as a high sodium ion concentration; and sensors can be developed for which electrodes are not available.24 Urease is an extremely specific enzyme reacting to only one substrate urea (one source reacts to a slight extent with hydroxyurea) with the following stoicheiometry (at pH 7.4): urease NH2CONH2 + H20- 2NH3 + CO2 2NH3 + H20 - 2NH4+ + 20H-C02 + H20- HC03- + H+ Urea has many industrial uses but the bulk of urea analyses are performed on human body fluids as adjunct to the investigation and monitoring of renal and hepatic disease or in the crude determination of nitrogen balance in patients fed on elemental diets.Non-enzymatic determinations exist for urea; in fact a large number of clinical laboratories use direct, spectrophotometric techniques.25 However most spectro-photometric techniques suffer from interferences and the preferred most selective procedure involves the use of unpleasant volatile reagents high temperature and stabilisers. Enzymes provide an appealing alternative as the conditions are necessarily mild. The determination of urea in serum has been performed using flow injection analysis with a glass pH electrode.26 In this work urea was determined in water and serum using flow injection analysis and the “optosensing” detection system. Experimental Instrumentation All the experiments were carried out with ii Bifok-Tecator FIAstar 5020 flow injection analyser.Manifolds with inte-grated flow-through detectors and miniaturised injection valves were incorporated into the FIAstar system replacing the original injection valve. The manifold (described below) was made from a black PVC block; the light source was a relatively weak tungsten lamp and ambient light had to be eliminated. Optical communications were made with plastic optical fibres (Du Pont Crofon fibre 1610; Optronics Cam-bridge UK) consisting of 64 individual acrylic strands randomly bifurcated at the remote end. The fibre bundles were protected from ambient illumination by black PVC sleeving. The optical fibres were glued in place at all ends with opaque black epoxy adhesive F156 (Tra-Con Medford, MA USA) rather than a clear fibre-optic adhesive.The black adhesive eliminated the large blank signal due to the source illumination being piped directly to the detector via a clear epoxy - fibre channel. At the remote end the fibres were fixed perpendicular to the flow channel with the black epoxy hence eliminating any problems of alignment and the movement of individual fibres and also minimising any stray light contribu-tion. The fibre epoxy interface was sanded and polished to allow the fibre-optic bundle to serve as a cell window. Reflectance measurements were performed with the Bifok-Tecator FIAstar 5023 scanning spectrophotometer which was modified in the following manner the 40&700 nm blue filter was removed; the flow-through cuvette was replaced by a black plastic block that held the bifurcated ends of the optical fibre bundle one end in front of the internal light source the signal is weaker than the conventionally measured absor-bance the signal of the non-attenuated reference beam was balanced electronically with built-in potentiometers.The reflectance measurements (A,.) were registered auto-matically on the FIAstar 5023 spectrophotometer and digitally displayed and printed on the FIAstar 5020 spectropho-tometer. The results were concurrently fed to a recorder (Radiometer Servograph REC-61 furnished with an REA-112 high-sensitivity interface). Reagents All chemicals used were of analytical-reagent grade; all water was doubly de-ionised. The carrier solution for the buffer pH determinations was 5 x 10-4 M HC1.The carrier solution for urea determinations was a dilute buffer solution containing 1 X lop3 M Tris - HCl in 0.140 M NaCl adjusted to pH 7.0 for the preliminary aqueous urea analyses and pH 7.4 for the serum analyses. The carrier was prepared from a stock buffer solution by SO-fold dilution with 0.140 M NaCl. The stock buffer solution was made by mixing SO ml of 0.1 M Tris (12.114 g 1 - I ) with 4.0 ml of 1.0 M HCl and diluting to 100 ml with water. The urease (Sigma U-2125) contained 71000 p~ units g-l and was dissolved in the carrier solution at a concentration of 60 mg per 100 ml. Sigma unit definition is that one micromolar unit will liberate 1.0 pmol of ammonia from urea per minute at pH 7.0 at 25 “C under Sigma assay conditions.Volumes of 25 pl of this solution were injected per sample corresponding to 1.08 units per analysis. Bovine serum albumin (Sigma A 6003) solutions were prepared by dissolving albumin in carrier adding stock urea diluting to volume and adjusting the pH directly before use. The urea standards were prepared by dilution of a stock urea solution with the carrier. Stock urea solution (100 mM) was prepared by dissolving 0.600 g of urea (Sigma) in 100 ml of de-ionised water. Serum samples were prepared by diluting 1 + 20 with the carrier solution (100 p1 of serum and 2.00 ml of carrier solution) immediately before use. Microconduit The design of the microconduit is illustrated in Fig. 1. A schematic flow system is shown in Fig.2. The black PVC blocks measure 70 x 45 x 10 mm and channel patterns were I Fig. 1. Integrated microconduit for the measurement of pH. consisting of a split loop injection valve. a mixing coil and an optosensor ( L ) . S Sample solution; P and P, tubes leading t o Deristaltic Dumtx; W waste other end aligned with the detector; and a; the reflected . f i ANALYST AUGUST 1986 VOL. 111 I 867 I I I I I -D Fig. 2. Manifold for pH measurement consisting of two peristaltic pumps (P and P?) a timer (T) and the microconduit Shown in Fig. 1 (boxed area). S Sample; C carrier stream; M mixing coil (100 PI); D, source plus detector; and W waste. The flow cell communicates with the light source and detector of the spectrophotometer by means of optical fibres impressed or engraved on the underside of the block.The channels are closed by transparent plastic using a pressure-sensitive polymeric glue and this yields the semicircular channels typical of the integrated microconduits. The chan-nels are accessed from the top of the block by perpendicular holes into which the PVC tubing is glued. The flow cell compartment is 3.2 mm in diameter the same diameter as the fibre-optic bundle it faces. A hole is cut into the transparent plastic and tape 3 mm or more in diameter to accommodate the cellulose pad containing a covalently bound acid - base indicator. This pad is 3.2 mm in diameter cut from the Merck non-bleeding indicator strips (ColorpHast cata-logue number 9583). A diffuse reflector (off-white Formica) is glued behind the cellulose pad to seal the channel and reflect any unreacted source illumination.The circular pads are cut 3.2 mm in diameter and if unrestricted swell to a depth of 1.8 mm; in the manifolds the pads are allowed to swell to a depth of only 0.4 mm (or 0.2 mm see Flow Cell Optimisation). For simple pH determinations the injection valve is used in the conventional mode with a variable sample volume. For urea determinations using soluble urease a split loop injection technique is used.15 This technique allows the sample and reagent to be injected simultaneously using only one injection valve and one pump; both volumes can be varied by changing the length of the sample or the reagent loops. The effect is similar to a merging zones configuration but requires a much simpler system and is potentially more sensitive as the dispersion of the sample and reagent before merging is minimised.In the earlier optosensing paper urea was determined via its enzymatic degradation to ammonia. 15 The ammonia was detected by diffusion across a gas diffusion membrane to an acid - base indicator solution. The indicator was renewed after each measuring cycle. In the present design ammonia is detected directly in the reacting solution which is in contact with the cellulose-immobilised acid - base indicator. The system is therefore simplified and the response is no longer dependent on diffusion across a membrane. This should allow for a faster response as the immobilised indicator response is essentially instantaneous and also a greater sensitivity as 100% of the ammonia now has the potential to be detected.To allow for variability in the choice of dispersion values and mixing times the manifold in this system incorporated a length of knotted microline tubing as a mixing coil. For pH 0.4( d 0.20 0 1 1.25 8.98 .79 8.50 I I -Time Fig. 3. pH response of the optosensor with Merck indicator 9583 to standard buffer solutions (0.1 M) injected into a 5 x lop4 M HCI carrier stream. The decrease in reflected signal measured as increase in absorbance (A,) at 610 nm. Negative deflections observed at leading edge of buffer responses were eliminated on decreasing the dead volume of the flow cell and increasing the amount of light incident on the indicator pad measurements the mixing coil was represented by the shortest possible length of tubing (ca.2.5 cm) in order to minimise the dispersion of the sample. Depending on the buffering capacity of the sample to be analysed dispersion can be a critical parameter. For urea determinations with soluble urease the sample and reagent need time to mix adequately while minimising the dispersion of the product. A 100-pl (50 cm) length of tubing provides adequate mixing (peak distortions due to variations in refrative index were eliminated) and the tight knots in the microline tubing minimise dispersion.27 The flow-rate through the microconduit is 1.2 ml min-1. The sample is pumped at 3.0 ml min-1 and the reagent at 0.8 ml min-1 for 8 s before each injection. Sensor Characterisation Reflectance spectrum of Merck 9583 acid - base indicator To characterise the response of the indicator with respect to wavelength a scan of the “yellow” (acidic) form of the pad was performed from 400 to 700 nm and stored by the spectrophotometer.The “blue” (basic) form was then scanned and the yellow spectrum was subtracted from the blue by the spectrophotometer. The resulting spectrum is a map of the change in response from 400 to 700 nm exhibiting a maximum at 608 nm. The band pass of the spectrophotometer is 19 nm at 700 nm; in all subsequent analyses the colour change was monitored at 610 nm. Merck indicator 9583 response to ApH To determine the magnitude of the indicator response in the pH range 6.5-10.0 and the shape of the A vs.pH curve a series of buffers (100 vl) were injected in duplicate into a carrier of 5 x M HCI (Fig. 3). The buffers were prepare 868 ANALYST AUGUST 1986 VOL. 111 according to the recipes in Tables 10.37 (0.1 M boric acid -NaOH) 10.32 (0.1 M Tris - HC1) and 10.25 (0.1 M phosphate) in Perrin and Dempsey’s book.28 The response is linear with pH (as measured by a Beckman glass pH electrode) in the pH range 7.3-9.0 corresponding to an A of 0.088-0.400 (Fig. 4). Optimally the urea calibration graph will be designed to fall within this linear region. To perform these pH determinations the shortest possible distance between the point of injection and the flow cell was utilised (corresponding to approximately a 5-pl volume) to minimise the dispersion and mixing of the sample with the carrier.The change in response due to these factors is strongly dependent on the buffer capacity of the samples injected so any change due to mixing with the carrier stream must be minimised for sample pH determinations. Determination of pK of bound dye To determine the pK of the bound dye on the cellulose pad it is assumed that the change in the reflected signal is equivalent to an absorbance change. According to the equations derived by Bishop,*y the ratio of the undissociated and dissociated indicator can be determined and used to calculate the pK, of the indicator without knowing the concentration of the indicator. The transition range the transition point and the conditional formation constant for the indicator can then be calculated from the pK and [H+] if applicable.The ratio of the undissociated and dissociated forms of the indicator immobilised on the pad is determined from spectrophoto-metric data using the equation [In-l/[HInl = [Amix - AHInl/[AIn- - Am,,] * * . * (1) where AHIn is the absorbance of the undissociated indicator, AI is the absorbance of the dissociated form of the indicator and Amix is the absorbance of the undissociated and disso-ciated forms. Absorbance values were measured after pump-ing solutions of the appropriate pH through the flow cell until a steady-state signal was reached. The pH values of the solutions used were obtained potentiometrically (using a Radiometer Research pH meter); absorbance values from eight different intermediate pH buffers (pH 7.29-8.25) were obtained and the pK given as calculated from the following equation is an average of the eight values: The pK of the indicator immobilised on ColorpHast 9538 is determined to be 8.15 F 0.12.. . . . . . pK = pH - log[In-]/[HIn] * . (2) Lifetime Study In order to determine the stability of a single indicator pad, 1200 replicate injections were performed two sets of 300 on two consecutive days (Table 1). Buffers (50 pl) of pH 8.0,9.0 and 7.0 were injected through a simple manifold (no split loop) 99 times each. Standard deviations and relative standard 0.40 d 0.20 0 1 I 1 I I I 7.5 8.0 8.5 9.0 PH Fig. 4. Peak height vs. pH linear region (pH 7.29-8.98) deviations were calculated on each set of 99 injections. Before the data were collected each day 99 injections were per-formed to stabilise the instrument (about 40 min).Whereas the average signals showed some large changes (up to 20%) from one run to the next the relative standard deviations for the fourth set of injections were actually slightly better than the first set of injections. This indicates that the pads are stable for at the very least 1400 injections but the system should be calibrated before each set of measurements. Changes in flow-rate due to changes in pump tubing behaviour on ageing, changes in temperature slight changes in pad position and changes in source intensity are some of the possible contribu-tors to the changes in average signal intensity. Results and Discussion Dispersion and Retention Data The initial data on dispersion were collected by two different methods using the 0.4 mm flow cell with an indicator pad 2.9 mm in diameter.To study solution dispersion bromothymol blue (BTB) dye was injected into a borax carrier. ColorpHast indicator number 9590 a pad that remained yellow at the pH of the carrier (9.0) was placed in the flow cell. A 50-1.1 volume of BTB was first injected through the manifold containing a 5-1.11 mixing coil (the design used for buffer pH determina-tions) and then into the same manifold containing a 100-yl mixing coil (used for sample analysis) to determine the effect of the mixing coil on dispersion. The simple injection system was then changed to the split loop configuration with a volume of 50 1.11 in the first loop (the sample loop) and 25 p1 in the second loop (the reagent loop).BTB was injected through the 50-yl loop with the carrier in the 25-y1 loop and then through the 25-1-11 loop with the carrier in the 50-1.11 loop. The change in the dispersion value for a 5O-pl sample in the split loop injection system compared with the simpler injection system indicates an increase in the sample dispersion of 0.056 due to the split configuration and another 0.087 increase can be attributed to the mixing coil (Table 2). The reagent loop shows a higher dispersion (0.60 larger) due to the smaller Table 1. Lifetime data on a single indicator pad Relative Average Standard standard signal deviation deviation O/O 0.242 0.442 0.064 0.214 0.398 0.062 0.221 0.399 0.055 0.196 0.369 0.055 6.03 7.77 2.08 5.80 5.98 2.43 7.62 5.07 1.52 2.92 3.61 1.90 2.5 1.7 3.2 2.7 1.5 3.9 3.5 1.3 2.8 1.5 0.98 3.4 * ( l ) (2) (3) and (4) refer to the four sets of 99 consecutive injections of each pH buffer; sets (1) and (2) were collected during one day and sets (3) and (4) the following day.Table 2. Dispersion data for bromothymol blue Conditions 50 pl BTB 5-pl coil . . . . . . . . . . 50 pl BTB 100-pl coil . . . . . . . . 50 p1 BTB 100-pl coil split loop . . . . 25 p1 BTB 100-pl coil split loop . . . . 50 pl BTB 100-pl coil split loop 3.2-mm pad 25 pl BTB 100-pl coil split loop 3.2-mm pad 50 pl NH3 100-p1 coil split loop 3.2-mrn pad 25 p1 NH3 100-pI coil split loop 3.2-mm pad D . . . . . . 1.15 .. . . . . 1.25 . . . . . . 1.32 . . . . . . 2.11 . . . . . . 1.45 . . . . . . 2.89 . . . . . . 1.22 . . . . . . 2.3 ANALYST AUGUST 1986 VOL. 111 869 volume and the greater distance travelled. The indicator pad was then replaced by a pad 3.2 mm in diameter and the dispersion of the 50- and 25-p1 volumes in the split loop system increased by 0.098 and 0.37 respectively. Retention times, defined as the time elapsed between the sample injection and the peak maximum,30 for the 50- and 25-pl volume BTB are 7.6 and 9.0 s respectively. Similarly dispersion data were collected by the injection of ammonia into a dilute acid carrier ( 5 x 10-4 M HCI) using a Merck indicator pad number 9583, in order to compare the pad response data with BTB solution reflectance data (Fig.5). The retention time data for ammonia were the same within the error of day-to-day measurements. The BTB shows a higher dispersion some distortion and greater peak widths than ammonia probably owing to the larger BTB molecule within the pad taking longer to flush out than the smaller NH4+ ion. However BTB response has the advantage of being independent of the buffer capacity of the injected sample hence simplifying the interpretation of the results. Flow Cell Optimisation The flow cell first designed by RGiiCka and Hansenls involved placing the optical fibres against a window of the same plastic as used in construction of the fibres. This window was glued into the manifold above the flow channel and the indicator pad was placed into the channel with the reflector glued behind.To decrease the path length that the light had to travel thereby increasing the reflectance signal the window was removed for the present studies. The fibres were glued together (and to the PVC manifold) with the same black epoxy adhesive used at the bifurcated end. The fibres were brought to the base of the manifold rather than the top of the flow channel for two reasons the fibres are now accessible for polishing and a flow cell with a much smaller path length can be constructed. The indicator pad is placed directly against the fibre bundle (insert Fig. 1). Steady state T A A, - T Fig. 5. Peaks resulting from the injection of NH into a carrier stream of 5 X 10-4 M HCI through the split loop injection system; peak A is from a 5O-pl sample volume and peak B is from a 25 pI sample volume.Steady-state response to the NH3 (approximately 2 mM) is superimposed near the peaks; arrow indicates time of injection In the first construction of this flow cell design two layers of plastic and adhesive were built up around the pad with the adhesive and the diffuse reflector behind the indicator pad. This resulted in a flow cell approximately 0.4 mm in depth. In an improved design one layer of plastic and adhesive was removed decreasing the flow cell thickness to 0.2 mm. With this modification the average signals of the three buffers showed an increase of 28% (pH 8.0) 21% (pH 9.0) and 25% (pH 7.0) with correspondingly improved relative standard deviations 1.13% (pH 8.0) 1.08% (pH 9.0) and 2.04% (pH 7.0).The speed of response was also improved the sampling rate of the first flow cell was 126 samples h-1 (50 pl of pH 8.0 buffer) compared with 144 samples h-1 for the second flow cell under the same conditions. Not only is the dispersion reduced but the pad also has less room for movement in the flow cell contributing to the improved relative standard deviation for the buffer measurements. In principle reflectance by an opaque layer is independent of its thickness d provided that a minimum d value has been reached.31 This is one of the conditions under which the reflected signal can be equated with an absorbance signal i.e., over a relatively short range of concentrations the Beer -Lambert law will be obeyed. The two thicknesses of the opaque layer investigated in these experiments were 0.4 and 0.2 mm.To determine the efficiency of the reflecting system, four flow cell designs were compared (1) a cell 0.4 mm in depth with one indicator pad; (2) a cell 0.4 mm in depth with two indicator pads; (3) a cell 0.4 mm in depth with one indicator pad and one pad of white filter-paper; and (4) a cell 0.2 mm in depth with one indicator pad. To compare these systems three buffers of pH 7.29 8.25 and 9.25 were injected in triplicate. As was observed above, one pad in a 0.2-mm cell shows approximately a 20% increase in signal for all three buffers (Table 3). However one pad, with filter-paper behind it shows an even larger signal increase (28-44%) with the following two phenomena probably contributing to the increased reflectance signal for a given ApH; the indicator pad is forced closer to the end of the fibre-optic bundle resulting in increased collection efficiency, and the filter-paper is itself a good diffuse reflector increasing the amount of light that is available to interact with the bound dye molecules.The response to two pads (c) in Table 3 showed an increase in reflected signal of 90% for pH 7.29 66% for pH 8.25 and 28% for pH 9.25. While the signals are higher the slope of the calibration graph is actually decreased owing to the non-linear change in response. The reflectance signal as described by the Kubelka - Munk equation is more complex than an absorbance signal; a scattering coefficient is included, which is non-linearly affected by the changes in optical path length.31 When systems ( b ) and (c) were compared using the urea-soluble urease systems no increase in signal was observed for the NH3 produced in contrast to the measure-ment of buffers.It can be concluded that for urea determina-tions using soluble urease one pad is sufficient. For buffer determinations depending on the range of ApH of the samples there may be an advantage to using a second indicator pad. Both systems may show an improved signal using the larger flow cell and incorporating filter-paper as a diffuse reflector but the sampling rate will be decreased. Table 3. Effect of flow cell design on reflected signal 4 0.4 mm (d), PH one pad one pad two pads filter-paper ?.29 0.132 0.113 0.214 0.155 8.25 0.395 0.322 0.536 0.465 9.25 0.678 0.589 0.756 0.766 0.2 mm (a) 0.4 mm (b) 0.4 mm ( c ) one pad 870 Optimisation of carrier p H In order to determine the optimum carrier pH the following must be considered the enzyme activity the indicator region of colour change the sample pH and the sensor response to NH3.The indicator has been demonstrated to respond from pH 6.5 to 10.0. The carrier is 1 x 10-3 M Tris 140 mM NaCl; the pH was adjusted to 6.7,7.0,7.3,7.5 and 7.7 with 0.1 M HCI and 0.1 M NaOH. A 2 mM NH3 solution was prepared by dilution of concentrated ammonia solution with de-ionised water and 50 pl were injected into each carrier. The maximum response was observed at pH 6.7 with a 2% decrease at pH 7.0 a 10% decrease at pH 7.3 a 19% decrease at pH 7.5 and a 22% decrease at pH 7.7.NH3 equilibrium favours a larger signal with increasing pH but the higher the carrier pH the greater the background colour of the pad so the smaller the response to the pH change with the injection of Enzyme solutions were prepared from the different carriers by 1 + 9 dilution of a stock enzyme solution (60 mg of enzyme in 10 ml of de-ionised water). Calibration graphs of 0-1 mM urea were generated (Fig. 6) to establish the pH of maximum apparent enzyme activity. At pH 6.7 and 7.0 the enzyme behaviour is essentially the same. At pH 7.3 however the apparent enzyme response to urea is suppressed by an average of 11% (relative to the response at pH 6.7) which reflects the behaviour of the NH3 at this pH. At pH 7.5 and 7.7 however, enzyme behaviour is clearly inhibited (43% suppression of signal at pH 7.5 48% suppression at pH 7.7) compared with the 19 and 22% for suppression of the signal of injected NI€3 solutions.The indicator is non-linear below pH 7.3. Hence in order to maximise the enzyme response to urea and also stay near to the region of linear indicator response the carrier pH used in the aqueous urea experiments was 7.0. NH3. Optimisation of ionic strength In order to maximise the enzyme activity 1 x 10-3 M Tris carrier streams at pH 7.0 with 0.00 17.5 35 70 140,210 and 280 mM NaCl were compared €or urea analysis. Calibration graphs of &2 mM urea were established in each carrier stream (Fig. 7). An NaCl concentration of 140 mM the concentration of NaCl in serum resulted in the steepest line below 1 mM urea.Large changes in the background A result when the ionic strength of NaCl in the carrier is changed (Table 4). This is probably due to some extent to the change in refractive index of the solution in combination with an indicator salt error.29 In conclusion the NaCl concentration should be the same for the sample reagent and carrier. Na+ is reported to 0.5 Concentration of ureairnM 1 .o Fig. 6 . Response graphs for urea (50 PI) at 0.0-1.0 mM with urcase (25 pl 60 mg dl-‘) as a function of pH lines A B. C D and E represent pH 6.7,7.0,7.3,7.5 and 7.7. respectively for the carrier and sample; solutions were 140 mM in NaCl ANALYST AUGUST 1986 VOL. 111 have an inhibitory effect on the urease,-32 but 140 mM NaCl resulted in the fastest rate under the conditions of this experiment.Also to avoid a blank response to differences in the sample and carrier ionic strength the carrier stream and reagent stream should be adjusted to the same ionic strength as the samples being injected. Conditions for Urea Determination Using Soluble Urease Stopped flow The stopped flow approach was selected for enzymatic measurements because of the following advantages noise caused by any vibration of the cellulose pad by the flowing stream would be eliminated; the sample and reagent could be allowed a longer time to react without increasing the dispersion with a long length of mixing c o i P ; a kinetic measurement could be performed as the FIAstar has the option to measure only the change in absorbance generated during the stop time-this will eliminate the need to know the background pH of the system and so the problems due to the variable sample pH will be minimised.The duration of the stop time was somewhat variable (10-30 s) as the system was not thermostated. At the typical Seattle summertime temper-atures of 23-28 “C the stop time used was 20 s. The actual time that the “stop” was initiated had to be calibrated for a given run; it depended primarily on the age of the peristaltic pump tubing and varied from 6 to 8 s for a flow-rate of 1.2 ml min-1. By viewing the output of the recorder the position with respect to maximum NH4+ signal was easily determined. I€ the flow was stopped prior to the position of the signal maximum then when the flow resumed a sharp “spike” characterised the remainder of the signal.The relative magnitude of this spike provides information about how far from the signal maximum the stop position is; a peak that simply trails off when the flow is resumed indicates that the stop is past the largest signal. ~~ ~ ~ Table 4. Effect of ionic strength on background reflected signal [ NaCl]/mM A‘% 280 0.064 210 0.055 140 0.000 70 -0.016 35 -0.037 17.5 - 0.060 0 -0.164 0.3 qL 0.15 0 I 1 1 .o 2.0 Concentration of ureairnw Fig. 7. Response graphs for 0.0-2.0 mM urea with soluble urease (25 PI. 60 mg dl-1) at FH 7.0 as function of [NaCI] in carrier stream and sample solutions. [NaCI] for lines A. B C. D E. Fand G are 0 17,35. 70 140 210 and 280 mM.respectively. The line for 280 r n M is not shown; it shows the same response as 210 mM NaCI ANALYST AUGUST 1986 VOL. 111 87 1 The flow-rate was chosen simply as the fastest that the manifold could accommodate without leaking in order to maximise the turbulence of mixing the speed of product washout and the amount of product formed during the stop time rather than before stopping. At different positions along the sample - reagent plug, different amounts of NH4+ are produced hence unique calibration graphs are generated at each stop time (Fig. 8). For this work the region of the maximum response to low levels of urea was chosen; for other types of analyses for example the determination of urea in urine the leading edge of the plug, where the signal shows a linear response from 2 to 4 mM urea, may be more ideal.Enzyme concentration A series of calibration graphs were generated with 50 pl of &2 mM urea samples and 25 pl of 15,30,45,60 and 75 mg urease per 100 ml of 1 x 10-3 M Tris (pH 6.5) as reagents (Fig. 9). At concentrations of urease above 45 mg dl-1 there is no increase in the signal for 2 mM urea but the signal did increase for 0-1 mM urea. As this is the region of interest a concentration of 60 mg dl-1 was chosen for subsequent analyses. With such a small reagent volume this corresponds to only one unit of enzyme per sample. Sample size The choice of sample size is governed by the desire for a maximum signal for a given sample volume and the preferred use of the smallest volume of serum possible.As 25 pl of enzyme solution are used two volumes of sample were tried, 25 and 50 pl. Although twice the size of the smaller sample the larger volume sample increases the dilution of the enzyme, and the enzyme activity is non-linear with concentration. The actual gain in the signal is 20%. As 50 pl of injected sample correspond to only 2.5 p1 of serum 50 p1 of sample were used for this study. Entrapped (adsorbed) enzyme Once the cellulose pad with covalently bound indicator has been exposed to the enzyme in solution the pad will exhibit enzyme catalysis in the absence of any injected enzyme solution for some time the duration depending on how long the pad has been exposed to the enzyme [Fig. lO(a)]. The magnitude of the response will resemble the solution enzyme response and the pad enzyme yields a calibration graph of the same shape as soluble urease.Urea determinations using entrapped enzymes show reliability at least as good as when using soluble urease (Table 5 ) . To characterise this behaviour the responses of three pads were compared (a) a cellulose pad soaked overnight in enzyme solution; (b) a cellulose pad soaked in a vigorously stirred enzyme solution for 1 h; and (c) a cellulose pad placed in the microconduit flow cell with the enzyme solution flowing through the flow cell for 10 min. The pad soaked in the enzyme solution overnight showed negligible enzyme activity in the presence of urea. The pad stirred with the enzyme solution showed a small but stable response and the pad placed in the flow cell showed a signal roughly equivalent to the response obtained when enzyme (60 mg dl-1) is injected with urea [Fig.lO(b)]. These results indicate that the phenomenon is flow-related which supports an “entrapment” hypothesis. The response of the entrapped enzyme to changes in NaCl concentration and changes in pH is very similar to the behaviour of the soluble enzyme. The apparent activities of both the soluble enzyme and the entrapped enzyme diminish slightly if the original enzyme solution is filtered or centri-fuged verifying that the pad is not simply trapping “colloidal junk” from the enzyme solution. Also this entrapment phenomenon is not dependent on the enzyme source; urease obtained from Millipore (51 U mg-1) exhibits similar behav-iour to urease from Sigma.The use of the pad as an entrapment bed for the enzyme could be exploited. It would eliminate the need for the split loop injection system hence simplifying the injection system. It would allow faster sampling rates and smaller sample volumes to be used as no mixing with the reagent would be necessary and no dilution of the sample by the reagent or dispersion in the mixing coil would result. In this system two Table 5. Recovery of urea in aqueous solutions Urea Urea taken/mM f o u n d h 0.0925 0.088 0.375 0.350 0.0315 0.025 0.0925 0.085* 0.187 0.184* 0.375 0.387* * Obtained using adsorbed urea. Recovery, Yo 95 93 89 92 98 103 0.10 4 0.05 I 1 I 1 I 0 1 .o 2.0 3.0 4.0 Concentration of urea/mM Fig. 8.Response raphs for 0-4.0 mM urea with soluble urease (25 @,60 mg per 100 mlf pH 7.0 at different stop points (times) along the sample - reagent plug line A is obtained by stopping 5 s after injection of sample and reagent via the split loop injection system; B is 6 s and C is 7 s after injection 0.40 1 0 0.5 1 .o 1.5 2.1 Concentration of ureairnw Fig. 9. Response graphs for urea at 0 . s 2 . 0 mM as a function of soluble enzyme concentration lines A B C D and E represent 15, 30 45 60 and 75 mg dl-1 respectively. The pH of the carrier and sample is 7.0 and the NaCl concentration is 140 m 872 0.30 d 0.15 ANALYST AUGUST 1986 VOL. 111 -4 min --. B -0. k 0.05 I I I I 0 25 50 75 No. of injections ( b ) 0.15 I No. of injections Fig.10. (a) response of adsorbed urease to 2 mM urea at pH 7.0, atter flowing the enzyme through the flow cell for 5 min (A) and 15 min (B). Each point is the average of five injections. ( b ) Response of adsorbed urease to 2 mM urea after soaking the pad overnight in enzyme (A) stirring with enzyme for 1 h (B) and flowing enzyme through flow cell for 10 min (C) pads should double the amount of enzyme entrapped hence increasing the lifetime of the entrapped signal. Research in the field of solid-phase reagent chemistry has demonstrated that solid phases particularly cellulose have a stabilising effect on labile reagents. The storage stability of some biochemical systems has been prolonged for up to 2 years at room temperature.33 These are fibre-impregnated systems; the enzyme we are accumulating on the active surface placed in the flow cell is an equivalent system.It is probable that this system could be used as a system for “loading” enzymes on to the cellulose pads and that these pads could be used immediately or stored for later analysis. Each pad could be calibrated and then used to perform a limited number of assays; this would be convenient for a laboratory performing a relatively small number of enzymatic determinations on a routine basis. Loading could possibly be increased or the system stability improved by first activating the cellulose with glutaral-dehyde borrowing an established technology from membrane electrode te~hnology.3~ This will initiate cross-linking of the enzyme with the solid support.If the adsorbed enzyme is not to be exploited it is important to realise that it will still be present if the pad has been exposed to enzyme. No “blank” injections of sample without enzyme would be reliable. This enzyme catalysis has been observed at any feasible base-line pH (6.5-7.6) for the indicator. Aqueous urea determination Urea standards and samples were prepared by the dilution of a 100 mM urea stock solution with a carrier of 1 x 10-3 M Tris -HC1 (pH 7.0) and 140 mM in NaCl. These solutions (50 pl) were injected with the enzyme reagent (25 pl) at 60 mg enzyme per 100 ml of carrier (Fig. 11). The response to the standards was plotted vs. urea concentration and the concentrations of the “samples” were calculated from this calibration graph. Recovery of the expected sample concentrations ranged from 89 to 103%.This is fairly good as the graph is non-linear above 0.125 mM urea under these conditions and the recovery data were collected directly from all regions of the graph. O L -r Fig. 11. Response graphs for urea determination at pH 7.0. Replicate injections of 2.00 1.00 0.500 0.250 0.125 0.063 0.031 and 0.015 mM urea with soluble urease (60 mg dl-I). The samples following were injected in triplicate except the first sample which was injected five times to obtain a relative standard deviation of 1.56% 0.40 aJ c (D e 0.20 a v) 0 0.5 1 .o 1.5 2.0 Concentration of ureaimM Fig. 12. Response graphs for 0-2.0 mM urea with soluble urease (25 p1,60 mg dl-I) as a function of [HC03- 1; HC03- is at a concentration of A 0.00; B 1.25; C 2.50; and D 5.00 mM in carrier stream and sample.The NaCl concentration is 140 mM and the pH is 7.0 Serum Analysis Effect of hydrogen carbonate The hydrogen carbonate buffer system supplies approximately 95% of the buffering capacity of serum.35 A 20-fold dilution of serum would result in a sample hydrogen carbonate concen-tration of 1 mM. To determine the effect of hydrogen carbonate on the calibration graph Tris carrier streams 0.00, 1.25 2.50 and 5.00 mM in NaHC03 all 140 mM in NaCl (pH 6.5) were used for the analysis. Suppression of the signal at the low end of the calibration graph (normal serum) was slight for 1.25 and 2.50 mM hydrogen carbonate but all three graphs containing hydrogen carbonate established a lower signal for 2 mM urea essentially a new steady-state value for the enzyme (Fig.12). Without hydrogen carbonate in the standard solutions the signals due to serum urea would be lowered for high concentrations of urea. Hence calibration standards should be 1 mM in NaHC03 for serum analysis. Effect of albumin Human serum is approximately 7 g dl-1 in proteins. Although the contribution of albumin to the solution buffering capacity is sma11,36 the effect on the sample viscosity and refractiv ANALYST AUGUST 1986 VOL. 11 I 873 index may be significant and an indicator protein error is another possible source of deviation from the aqueous calibration graph.29 In order to determine the effect of this protein on the calibration graph for urea analysis urea standards of 0.0-2.0 mM were made up of 0.35 g dl-1 in bovine serum albumin 1.0 mM NaHC03 diluted to volume with Tris carrier the pH being adjusted to 7.4.These standards were compared with urea standards with no albumin 1.0 mM NaHC03 and diluted to volume with Tris carrier (pH 7.4). The two resulting calibration graphs differed significantly; the presence of albumin increased the signal resulting from the enzyme reaction. Both calibration graphs have the same zero intercept indicating that this is not a blank response to the albumin. The calibration graph is linear to 2.0 mM urea (correlation coefficient 0.9996); usable data can be obtained to 4 m M urea encompassing any probable serum value. Serum samples Serum samples and glycerol-based BUN standards (Monitrol I and 11) were obtained from the University Hospital Seattle, WA.Hospital urea measurements had been performed on an ASTRA system the previous day and the results were compared with FIA data (Table 6). Agreement is excellent Table 6. Serum values Astra FIA optosensing Urea/mM UreairnM Reported (diluted (diluted BUN sample) A sample) 81 25 41 36 23 5 43 10 13 51 50t 1.38 0.43 0.69 0.61 0.39 0.085 0.75 0.17 0.22 0.085 0.85 0.088 0.026 0.045 0.040 0.026 0.003* 0.048 0.008 0.015 0.003* 0.056 1.44 0.43 0.74 0.66 0.43 0.059 0.79 0.14 0.25 0.059 0.92 * Net signal less than three times the standard deviation of the ?- Monitrol standards. blank. 1 .o a a K I-v) 0 1 .o A, Fig.13. Comparison of results for serum urea FIA optosensing results vs. University Hospital ASTRA results (enzymatic determina-tion with conductimetric detection). The precision of both methods is 3 % when the albumin-containing calibration graph is used to calculate the urea concentrations of samples (Fig. 13). Our results tend to be a few per cent. high probably owing to the elevated pH of the serum samples resulting from bacterial degradation. This will be measurable a few hours after blood drawing; our undiluted serum sample pH values were as high as 8.0 at the time of measurement. The authors thank Tecator Inc. for the loan of the FIAstar system and Winnie Lee University Hospital for graciously providing serum samples. 1. 2. 3.4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. References Narayanaswamy R . Anal. Proc 1985 22 204. Coleman J. T. Eastham J. F. and Sepaniak M. J. Anal. Chem. 1984 56 2246. Smith A. M. Anal. Proc. 1985 22 212. Saari L. A and Seitz W. R. Anal. Chem. 1982 54 821. Saari L. A. and Seitz W. R. Anal. Chem. 1983 55 667. Saari L. A and Seitz W. R. Analyst 1984 109 655. Goldfinch M. L. and Lowe C. R. Anal. Riochem. 1980, 109 216. Goldfinch M. J . and Lowe C. R. Anal. Riochem. 1984,138, 430. Edmonds T. E . and Ross I. D. Anal. Proc. 1985 22 206. Peterson J . I . Goldstein S. R . and Fitzgerald R. V. Anal. Chem. 1980 52 864.Kirkbright G . F. Narayanaswamy R. and Welti N. A . , Analyst 1984 109 1025. Peterson J . I. Fitzgerald R. V. and Buckhold D. K. Anal. Chern. 1984 56 62. Schultz J. S . Mansouri S. and Goldstein I. J. Diabetes Care 1982 5 245. Kirkbright G . F. Narayanaswamy R. and Welti N . A., Analyst 1984 109 15. RGiiEka J . and Hansen E. H. Anal. Chim. Acta 1985 173, 3. RfiiiCka J. Anal. Chem. 1983 55 1040A. McKinley B. A . Houtchens B. A . and Janata J . Zon-Sel. Electrode Rev. 1984 6 173. Marconi W. Bartoli F. Gulinelli S. and Morisi F. Process Biochem. 1974 May 22. Fishman M. M. and Schiff H. F. Anal. Chem. 1972 44, 543R. Guilbault G. G . and Montalvo J. G . Anal. Lett. 1969 2, 283. Johansson G . and Ogren L. Anal. Chim. Acta 1976,84,23. Ripamonti M. Mosca A Rovida E . Luzzana M. Luzi L., Ceriotti F. Cottini F. and Rossi-Bernardi L. Clin. Chem., 1984 30 556. Joseph J. P. Mikrochim. Acta 1984 11 473. Seitz W. R. Anal. Chem. 1984 56 16A. Butler A . R. Trends Anal. Chem. 1982 1 120. Rfiiii-ka J. Hansen E. H. Ghose A. K. and Mottola H. A., Anal. Chem. 1979 51 199. Engelhard H . Neu U. D. Chromatographia 1982 15 403. Perrin D. D. and Dempsey B. “Buffers for pH and Metal Ion Control,” Chapman and Hall London 1974. Bishop E. “Indicators,” International Series of Monographs on Analytical Chemistry Volume 51 Pergamon Press Oxford, New York 1972. Rgiitka J . and Hansen E . “Flow Injection Analysis,” Chemistry and its Applications Volume 62 Wiley New York, 1981. Kubelka P. and Munk F. Z. Tech. Phys. 1931 12 593. Guilbault G . “Handbook of Enzymatic Methods of Analysis, Clinical and Biochemical Analysis,” Volume 4 Marcel Dek-ker New York and Basle 1976. Rocks B. and Sherwood. R. Talanta 1984,31 879. Tran-Minh C. and Broun G. Anal. Chem. 1975 47 1359. Szabo L. L. and Kaplan A. “Clinical Chemistry Interpreta-tions and Techniques.” Third Edition Lea and Febiger, Philadelphia 1983. Paper A6148 Received February 17th 1986 Accepted March 13th 198