Mikrochim.

Acta 131, 129±135 (1999)

Development of an Optical Fiber Lactate Sensor

Xiaojing Liu and Weihong Tan

Department of Chemistry and UF Brain Institute University of Florida Gainesville, FL 32611, USA

Abstract. Lactate analysis is important in clinical because lactate in¯uences their ¯avor, stability and
diagnostics and the food industry. An ultrasensitive quality [3, 4]. The major techniques for quantitative
optical ®ber lactate sensor with rapid response time determination of lactate are spectrophotometric,
and 50 mm size has been developed. Lactate dehydro- ¯uorometric, electroanalysis and biosensors. Recently,
genase (LDH) has been directly immobilized onto an different types of L-lactate sensors have been
optical ®ber probe surface through covalent binding developed for in vivo application in real time lactate
mechanisms. The optical ®ber surface is initially acti- monitoring [5±7], metabolite determination in body
vated by silanization, which adds amine groups (±NH2) ¯uids, such as blood [8, 9], saliva [10], sweat [11, 12]
to the surface. Aldehyde functional groups (±CHO) are and food analysis [4, 13, 14]. Most L-lactate sensors
then af®xed to the optical ®ber surface by employing a use an electrochemical detection method. These
bifunctional cross-linking agent, glutaraldehyde. The electrochemical measurements usually employ large
amino acids of LDH enzyme molecules readily attach overpotentials and therefore suffer from interference
to these free -CHO groups on the ®ber surface. Optimal by co-oxidation of reducing substances present in
immobilization of LDH occurs between 19 and 23 biological media [15]. The ®ber-optic sensor techni-
hours of exposure in the enzyme solution. The immo- que, however, can avoid these problems because no
bilized LDH enzyme molecules on the ®ber surface electric potential is involved during the detection. In
show high enzymatic activity. The lactate sensor is able addition, optical ®ber sensors have advantages in
to detect lactate with a concentration detection limit of small size, high sensitivity, ¯exible geometry, remote
0.5 mM and the absolute mass detection limit is 8.75 detection capability, and the lack of electrical
attomoles. Moreover, the sensor rapidly responds to interference. All these make an optical sensor method
lactate changes and exhibits good reproducibility. The an attractive technique for the determination of L-
lactate sensor is extremely selective. This immobilized lactate in a variety of samples.
enzyme sensor has been applied to accurately deter- Optical lactate sensors are mainly based on lactate
mine the lactate content in food samples. oxidase [16, 17] and lactate dehydrogenase [18]. A
®ber-optic lactate biosensor with an oxygen optrode
Key words: L-lactate, lactate dehydrogenase; enzyme immobil-
ization; optical ®ber sensor. as transducer was presented by Wolfbeis' group [16].
Lactate mono-oxygenase was immobilized covalently
on nylon membranes. Oxygen was consumed as a
The determination of L-lactate is important in the result of the oxidation of lactate by the enzyme, and
diagnosis of respiratory insuf®ciencies, heart disease the decrease in its partial pressure was indicated by an
and in sports medicine [1, 2]. In the food industry, L- oxygen-sensitive ¯uorescent dye. For two types of
lactate measurements also play an important role in sensors (with different nylon membranes and different
the control of wine, cider, beer and milk products thickness of the indicator layer) the analytical ranges
were 2±50 mM and 0.3±0.6 mM, respectively. Lactate
 To whom correspondence should be addressed dehydrogenase (LDH) has also been used for lactate
130 X. Liu and W. Tan

sensor application. The concept of this type of bilization of LDH molecules. Lactate determination is
biosensor has been demonstrated for lactate and achieved by detecting the ¯uorescent NADH produced
pyruvate with a ®ber-optic biosensor constructed by from the reaction
immobilizing LDH on a nylon membrane [18]. The
Lactate ‡ NAD‡ ! Pyruvate ‡ NADH ‡ H‡
greatest attraction of using dehydrogenases is the
ability to design a wide variety of sensors for a large The lactate sensor has shown excellent detection
array of analytes, especially those in biological capability with fast response, good reproducibility and
systems. The lactate sensor probe consisted of two high selectivity.
separate ®ber bundles with an overall probe diameter
of 8 mm at the common end. The biosensor had a
detection limit of 2 mM for L-lactate. Sensor response Experimental
times ranged from 5 to 12 min. More recently, a multi-
Apparatus
enzyme-based bioluminescent ®ber-optic sensor for
the determination of lactate was described [19]. It was The experimental set-up for characterization of the LDH enzyme
probe was based on an inverted Olympus ¯uorescence microscope
based on a reaction sequence catalyzed by three (model IX708F) [20]. In this apparatus, the ¯uorescence micro-
different enzymes: luciferase, NAD(P)H:FMN oxi- scope was connected to an avalanche photodiode (APD) (SPCM-
doreductase and lactate dehydrogenase. The sensor's 200 CD 2027, EG&G, Quebec, Canada) through the side port of
the ¯uorescent microscope to eliminate stray light interference.
detection limit for lactate was 0.2 mM, but the The source for the excitation radiation was a 350-nm laser beam
preparation and use of the multi-enzyme sensor were from an Innova 90 Ar‡ laser (Coherent Laser, Santa Clara, CA). A
more complicated than for single-enzyme sensors. high precision single mode ®ber coupler from Newport Research
Corp. was used to direct the laser beam into an optical ®ber splice
In most of these ®ber-optic sensors the enzyme is (CamSplice, Siecor). The excitation light source was thus ®xed by
immobilized on a membrane. These membranous coupling an LDH immobilized ®ber to one end of the device. The
layers create a type of cage that traps the working ¯uorescence was collected via a quartz microscope objective with
sensor biomolecules at the surface of the probe. The a 40 magni®cation. The objective lens transmits in the UV range
of 300±400 nm, and has limited auto¯uorescence. The NADH
analyte is then allowed to ¯ow over the surface and luminescence was transmitted by a dichroic mirror through a
into the layers where it reacts with the immobilized 420 nm long-pass ®lter, a 465 nm interference ®ber (Oriel Cor-
biomolecules. This would imply that response times poration, Stratford, CT) and subsequently directed into the APD
system. The signal was ampli®ed by a DC-30 MHz ampli®er
would be diffusion dependent and therefore longer model SR445 (Standard Research Systems, Sunnyvale, CA). An
than those required in in vivo measurement. The EG&G ORTEC MCS-Emulator was used to collect ¯uorescence
recovery time of the sensor is also long, since the intensity data as a function of time. The entire microscope-based
set-up was on a ¯oating laser table to eliminate vibration.
same amount of time would be needed to eliminate
the analyte from the membranous layer. If the
biomolecule is covalently bound directly to the Reagents
surface by a cross-linking arm, the migration time L-Lactate dehydrogenase (LDH) (EC1.1.1.27) from Bovine Heart
of the analyte through the depth of the membrane is (550 units/mg) and L-lactate were purchased from Sigma
reduced as well as the recovery time. The biomole- Chemical Co. (St. Louis, MO). Nicotinamide adenine dinucleotide
(NAD‡ ) and glutaraldehyde (25% aqueous solution) were obtained
cules can also be ef®ciently used for reaction and from Acros Organics (Fairlawn, NJ). Trimethoxysilylpropyldiethy-
biochemical sensing, resulting in excellent analytical lenetriamine (DETA) was purchased from United Chemical
sensitivity, fast response time and feasible miniatur- Technologies (Bristol, PA). All other chemical reagents were of
analytical reagent grade. Distilled, demineralized water used for
ization. However, direct covalent binding of biomo- the preparation of all solutions. Silica optical ®bers with a core
lecules on an optical ®ber surface is a dif®cult task. diameter of 105 and 50 mm were obtained from General Fiber
In this paper, an immobilized LDH optical ®ber Optics.
sensor for the determination of lactate is described.
The construction of the lactate sensor involves covalent Optical Fiber Probe Surface Modi®cation
immobilization of LDH on an optical ®ber surface by A batch of six optical ®ber probes was used in a single immo-
®rst activating the optical ®ber with a silanization bilization cycle. The ®ber tips were ®rst cleaned by immersion in
agent, then adding a cross-linking arm, and ®nally 1:1 v/v concentrated HCl:MeOH mixture for 30 min. They were
covalent binding of LDH molecules directly onto the then rinsed in demineralized water and submerged in concentrated
sulfuric acid for 30 min. Further rinsing and then boiling in
cross-linker arm attached to the optical ®ber surface. demineralized water for 8±10 minutes followed. Silanization of the
There is no gel or membrane involved for the immo- ®bers was performed by immersion in freshly prepared 1% (v/v)
Development of an Optical Fiber Lactate Sensor 131

¯uorescence emission at 465 nn was measured as a function of

time. After each substrate injection, the capillary and the enzyme
probe were rinsed with fresh aliquots of Tris ±HCl buffer to restore
the baseline signal. All experiments were done at room tem-
perature (23  C).

Results and Discussion

Direct Covalent Immobilization of LDH

on an Optical Fiber Surface
LDH enzyme molecules can be directly immobilized
onto an optical ®ber surface by covalent binding. The
Fig. 1. Procedure for immobilization of L-lactate dehydrogenase immobilized LDH molecules have excellent enzy-
(LDH) on a ®ber-optic surface matic activities. This immobilization mechanism is a
novel combination of several surface modi®cation
technologies [20±23] speci®cally designed for our
solution of DETA in 1 mM acetic acid for 20 minutes at room sensor development. Proper surface preparation is
temperature (23  C). The DETA-modi®ed ®ber tips were thor- essential for optimal LDH immobilization. For
oughly rinsed with demineralized water to remove excess of
DETA. The silanized ®ber probes were dried under nitrogen and
silanization, DETA was found to be the most reliable
®xed by heating in a 120  C oven for 5 min. of the silanization agents we have tested. Conjugation
Conjugation of the tips was done by using the bifunctional of LDH is accomplished through the use of glutar-
cross-linker glutaradehyde. A condensation reaction of glutaralde-
hyde aldehyde groups with the amino groups on the fiber's surface
aldehyde, where one aldehyde group reacts with an
formed carboamides. The silanized tips were kept in a 12.5% (v/v) amine group from the aminosilane layer on the ®ber,
glutaraldehyde solution in 0.1 M potassium phosphate buffer and the other aldehyde group reacts with the amine-
solution (PBS) at pH 6.8 for 14 h at room temperature. In the containing biomolecules. Once the probes had been
next step, it was particularly critical to remove all excess of
glutaraldehyde before placing the tips in the enzyme solution for properly treated with silanization and cross-linking,
immobilization. Therefore, meticulous rinsing with demineralized they were inserted into an LDH buffer solution for
water and complete drying under nitrogen followed the conjuga- LDH immobilization. We maintained the LDH
tion. The unreacted aldehyde termini of the bifunctional cross-
linker were left for linking to the primary amines on the target concentrations at 10ÿ5 M for all ®bers. We have
enzyme, as illustrated in Fig. 1. tested the effect of the duration of immobilization of
LDH molecules on the treated optical ®ber probes. As
Enzyme Immobilization
shown in Fig. 2, the maximum activity of the probe
occurs at between 19 and 23 h of immobilization of
The LDH enzyme with a solution concentration of 110ÿ5 M was
prepared in 0.1 M PBS at pH 6.8. The tips of the ®bers were
the LDH, but even after 7.5 h of immobilization there
immersed in the LDH/PBS solution for 23 h at room temperature
to allow the enzyme to be immobilized on the surface. The tips
were subsequently washed with buffer, and either used immedi-
ately or stored in PBS at 4  C when not in use.

Sensor Response Analysis

Measurement of the LDH activity was carried out in the con®nes
of ultra-small capillary tubes (300 mm inner diameter, Polymicro
Technologies, Inc., Phoenix, AZ) called microscopic enzymatic
reactors. The enzyme probe is inserted into a clean capillary and
stabilized on the microscope stage so that the very tip of the LDH
immobilized optical ®ber is in focus at the objective lens. It is
rinsed once by injection of 20 mM Tris-HCl buffer (pH 8.5)
through the opposite hole of the tube. A substrate solution of a
known amount of lactate and NAD‡ in Tris ±HCl buffer (pH 8.5)
was injected into the capillary, and its optical intensity inside the
capillary, excited by the LDH sensor probe, was recorded as the Fig. 2. Effect of immobilization time on the activity of lactate
baseline signal. The reaction was allowed to take place and sensor at pH 8.5 (Tris±HCl buffer) in the presence of 0.5 mM
monitored by the APD-based detection system. The NADH lactate and 1 mM NAD‡
132 X. Liu and W. Tan

is measurable NADH production. After 23 h, the

activity of the immobilized enzyme decreases. The
reason for the decline in enzyme activity after 23 h in
LDH solution is not clear. A possible reason is that the
longer immobilization time may cause more enzyme
molecules to bind on the optical ®ber surface, which
probably leads to the blocking of enzyme active sites,
thus limiting the substrate's access to the immobilized
enzyme. The immobilization time used in all sub-
sequent experiments was 23 h.

Fig. 4. Effect of pH on the activity of LDH. Experimental

NAD‡ Effect on Lactate Sensor Performance conditions: 20 mM Na2HPO4±NaH2PO4 (pH ˆ 5.7±6.8), 20 mM
Tris±HCl buffer (pH ˆ 7.5±8.9), 20 mM Na2CO3±NaHCO3 buffer
Because NAD‡ is required for the reaction, the effect (pH ˆ 9.4±9.8), 0.5 mM lactate and 1 mM NAD‡
of NAD‡ concentration on LDH sensor activity has
been established. Sensor response was monitored for
0.5 mM lactate solutions with various concentrations so a high pH is advantageous to the forward reaction.
of NAD‡ . As shown in Fig. 3, when the concentration However, too high a pH may result in the decom-
of NAD‡ was increased, the sensor responded in an position of NAD‡ and instability to the enzyme. That
asymptotic approach to maximum enzyme activity. its optimum response is over the physiological pH
The LDH activity on the probe is in essence indepen- range is an advantage for this sensor. All subsequent
dent of the co-substrate when the concentration of measurements were performed at pH 8.5, obtained by
NAD‡ is above 1 mM. The concentration of NAD‡ using a Tris±HCl buffer.
used in all subsequent experiments was 1 mM.
LDH Probe for Sensitive Lactate Detection
In¯uence of pH on the Lactate Sensor
The lactate response curve obtained under optimal
The environment in which a biosensor is used may conditions is presented in Fig. 5. We used initial
have a marked effect on its response. We have reaction rates to characterize the sensor's sensitivity in
therefore investigated the in¯uence of pH on the monitoring lactate changes. At constant enzyme
lactate sensor performance. The results are shown in concentration, the initial rate is directly proportional
Fig. 4. The sensor has its highest activity in the pH to the substrate concentration when the concentration
range 6.8±8.9. The reaction of lactate and NAD‡ of substrate is much less than KM [24] (KM, the
under catalysis by LDH is a process of releasing H‡ , Michaelis constant, is equal to the substrate concen-

Fig. 3. Effect of NAD‡ concentration on the activity of lactate

sensor at pH 8.5 (Tris±HCl buffer) in the presence of 0.5 mM
lactate Fig. 5. Calibration curve for L-lactate under optimal conditions
Development of an Optical Fiber Lactate Sensor 133

tration at which the reaction rate is half of its maximal Table 1. Relative response from potential interferencesa
value). Reaction rates were calculated for the ®rst Compound % Response Compound % Response
2000 data points (25 ms for each point) in continuous Lactate 100 ascorbic acid 0
monitoring of the production of NADH at different Malonic acid 0 uric acidb 0
lactate concentrations. From this response curve in Glycolic acid 0 L-cysteine 0
Isocitric acid 0 glutathione 0
Fig. 5, linear response of the sensor to lactate was Quinic acid 0 acetamidophenolc 1.9
from 2.0 to 200 mM. The detection limit (S/N ˆ 3) for a
Response from 50 mM tested compound (except as indicated)
lactate was 0.5 mM. The calculated absolute mass
relative to the response of 0.2 mM lactate in the presence of 1 mM
detection limit is 8.75 attomoles. Our lactate sensor NAD‡ ; b6.15 mM uric acid; c10 mM acetamidophenol.
also exhibits reasonable reproducibility (RSD ˆ 4.8%
for 0.1 mM lactate (n ˆ 7)).
Compared with a previously reported optical ®ber
LDH enzyme probe [18], the LDH probe prepared in offers high selectivity for lactate, and this was
this work is smaller in diameter by a factor of 200, demonstrated by testing for any response to other -
and thus smaller by a factor of 4104 in its active area hydroxy-acids and those compounds which most
for sensing. However, the LDH probe is more commonly interfere in the determination of lactate
sensitive than any that has been reported. There are in either physiological or food samples. As shown in
three reasons that contribute to the high sensitivity of Table 1, no responses were observed for the majority
the present lactate sensor. The ®rst is the direct of the compounds tested. Only acetamidophenol
immobilization of LDH molecules on a ®ber surface showed a response, at a concentration of 10 mM, of
without a membrane, gel or any other carrier. The about 1.9% of the response to lactate at a concentra-
substrate has direct access to the LDH molecules. tion of 0.2 mM.
Second, there is a long enough free space between the For most enzyme-based sensors, the time a sensor
®ber surface and the LDH molecules through the long needs to achieve a steady-state response after the
chains of the cross-linker, so the LDH molecules are introduction of a sample is reported as the response
more effective in catalyzing the reaction between time. We de®ne the response time for a lactate sensor
lactate and NAD‡ . The third reason is the high as the time required for the LDH probe to generate
ef®ciency of the detection system built for the sensor suf®cient NADH for detection. The shortest response
application. The avalanche photon diode and laser time of the sensor corresponds to the lowest required
induced ¯uorescence used allow the ultrasensitive optical signal generated by NADH. We used the
detection of NADH down to about 10 nM. standard deviation of the base line to estimate the
response time of the sensor. Our probes demonstrated
response times less than 75 ms for 1 mM lactate. A
Lactate Sensor Stability, Selectivity
great bene®t of the LDH biosensor was its extremely
and Response Times
short recovery time. The LDH biosensors prepared in
The lifetime of sensors, especially those employing this work ready for experimental measurements
biologically active materials such as enzymes or directly after rinsing with Tris±HCl buffer. This fast
antibodies, is a major concern. The lactate sensor was response time of the lactate sensor was due to the
stored in pH 6.8 buffer at 4  C when not in use. The direct immobilization of the enzyme molecules onto
stability of the system was evaluated on the basis of the optical ®ber surface.
hourly response measurements on a 0.2 mM lactate
sample under optimal conditions. The lactate sensor
Analytical Application
showed good stability in the ®rst 12 h of use. An
obvious degradation in the sensor's response was Our lactate sensors were applied for the determination
observed after 24 h. Owing to the fragility of a single of lactate in food samples. Pretreatment of the food
optical ®ber sensor, it was relatively dif®cult to collect samples was simple. The juice samples were ®ltered
data after 24 h of usage. Optimal storage was found to to remove solids, and then diluted with a buffer
be 4  C. solution. Owing to the high selectivity of LDH
LDH itself is highly selective in its reaction for enzyme for its substrates, background on the enzyme
lactate relative to other substrates, so the LDH sensor activity was very low. However, the signal for NADH
134 X. Liu and W. Tan

Conclusion
An ultrasensitive optical ®ber lactate sensor, with a
diameter of 50 mm, has been developed by directly
immobilizing lactate dehydrogenase molecules onto
optical ®ber probe surfaces through covalent binding
mechanisms. The immobilization procedures, includ-
ing tip silanization, conjugation and chemical binding
of lactate dehydrogenase, have been created to
coordinate the enzyme molecules to the optical ®ber
tip. Optimal immobilization of LDH occurred between
Fig. 6. Linear calibration plot for the method of standard additions 19 and 23 h exposure in the enzyme solution. The
with a lactate sensor. The juice samples were diluted 250-fold with lactate sensor is small in size, simple in its prepara-
buffer. Standard solutions of lactate were added to the diluted tion, easy to use and fast in response to lactate changes.
unknown juice samples. The concentration of unknown solution
was determined by extrapolation. () Apple juice sample; (&)
The lactate sensor has shown excellent lactate
orange juice sample detection capability over a broad concentration range.
The sensor rapidly responds to lactate changes and
exhibits good reproducibility. It is also extremely
selective. This immobilized enzyme sensor has been
Table 2. Determination of L-lactate in food samples successfully applied to the determination of lactate
content in juice samples. Further miniaturization [26]
Sample Lactate sensor LDH in solution
(g/L) (g/L)
of our lactate probe will enable real time monitoring
of lactate in an extracellular space with excellent
Apple juice 0.81 0.79
Orange juice 0.43 0.46
spatial and fast temporal resolution.

Acknowledgments. WT thanks the Whitaker Foundation for a

Biomedical Engineering Award. This work is supported by NSF
Faculty Career Award CHE-9733650 and by Of®ce of Naval
Research Young Investigator Award N00014-98-1-0621.

in a food sample may be covered by a high back-

ground signal due to other impurities in the food References
sample. To decrease the background and eliminate the
[1] M. Mascini, D. Moscone, G. Palleschi, Anal. Chim. Acta 1984,
matrix effect, the juice samples were diluted 250-fold. 157, 45.
The standard addition method was used to determine [2] J. B. Toffaletti, Clin. Chem. News 1989, 9, 14.
the lactate content in the juice samples. A linear [3] G. H. Aschor, J. Weily, J. Chromatogr. 1984, 287, 452.
[4] M. J. Fernandez Villamil, A. J. Miranda Ordieres, P. Tunon-
calibration plot for the standard addition method with Blanco, Anal. Chim. Acta 1997, 345, 37.
a lactate sensor is shown in Fig. 6. The concentration [5] Y. Hu, Y. Zhang, G. S. Wilson, Anal. Chim. Acta 1993, 281,
of the unknown solution was calculated by extrapola- 503.
[6] Y. Hu, G. S. Wilson, J. Neurochem. 1997, 69, 1484.
tion [25]. Lactate determination was also done with
[7] N. F. Shram, L. I. Netchiporouk, C. Martelet, N. Jaffrezic-
LDH enzyme molecules in solution. Table 2 sum- Renault, C. Bonnet, R. Cespuglio, Anal. Chem. 1998, 70, 2618.
marizes the results of food sample analysis by our [8] C. Meyerhoff, F. Bischof, F. J. Mennel, F. Sternberg, J. Bican,
newly developed lactate sensors. Good agreement was E. F. Pfeiffer, Biosens. Bioelectron. 1993, 8, 409.
[9] D. A. Baker, D. A. Gough, Anal. Chem. 1995, 67, 1536.
found between the results from the LDH sensor and [10] G. Palleschi, M. H. Faridnia, G. J. Lubrano, G. G. Guilbault,
those from free LDH solution. The slope of the Anal. Chim. Acta 1991, 245, 151.
calibration plot for the organic juice sample is close to [11] M. H. Faridnia, G. Palleschi, G. J. Lubrano, G. G. Guilbault,
Anal. Chim. Acta 1993, 278, 35.
that of calibration curve in Fig. 5, but the slope of [12] K. Mitsubayashi, M. Suzuki, E. Tamiya, I. Karube, Anal.
calibration plot for the apple juice sample is a bit Chim. Acta 1994, 289, 27.
larger than that of the calibration curve in Fig. 5. The [13] N. Kim, R. Haginoya, I. Karube, J. Food Science 1996, 61, 286.
determination of the lactate content of juice samples [14] M. J. Lobo-Castanon, A. J. Miranda-Ordieres, P. Tunon-
Blanco, Anal. Chim. Acta 1997, 346, 165.
clearly demonstrates the feasibility of practical [15] W. Zhang, H. Chang, G. A. Rechnitz, Anal. Chim. Acta 1997,
applications of the lactate sensors. 350, 59.
Development of an Optical Fiber Lactate Sensor 135

[16] W. Trettnak, O. S. Wolfbeis, Anal. Lett. 1989, 22, 2191. [23] P. A. E. Piunno, U. J. Krull, R. H. E. Hudson, M. J. Damha, H.
[17] W. Trettnak, O. S. Wolfbeis, Fresenius Z. Anal. Chem. 1989, Cohen, Anal. Chem. 1995, 67, 2635.
334, 427. [24] L. Strver, Biochemistry, 3rd, Edn. Freeman, New York, 1988,
[18] J. Wang, M. A. Arnold, Anal. Chem. 1988, 60, 1080. p. 189.
[19] P. E. Michel, S. M. Gautier, L. J. Blum, Anal. Lett. 1996, 29, [25] D. Skoog, F. J. Holler, T. A. Nieman, Principle of
1139. Instrumental Analysis, 5th Edn. Saunders Brace, 1998, pp.
[20] J. Cordek, X. Wang, W. Tan, Anal. Chem. 1999, 71, 1529. 15±18.
[21] L. A. Chrisey, G. U. Lee, C. E. O'Ferrall. Nucl. Acid. Res. [26] W. Tan, Z-Y. Shi, S. Smith, D. Birnbaum, R. Kopelman,
1996, 24, 3031. B. Liu, R. Hu, J. Deng, Anal. Chem. 1997, 69, Science 1992, 258, 778.
2343.
[22] G. F. Bickerstaff, Immobilization of Enzymes and Cells,
Humana Press, New Jersey, 1997, pp. 331±337. Received September 7, 1998. Revision December 10, 1998.