polarization-based sensing with a self-referenced sample

Volume 53, Number 9, 1999 APPLIED SPECTROSCOPY 11490003-7028 / 99 / 5309-1149$2.00 / 0

q 1999 Society for Applied Spectroscopy

Polarization-Based Sensing with a Self-Referenced Sample

JOSEPH R. LAKOWICZ,* IGNACY GRYCZYNSKI, ZYGMUNT GRYCZYNSKI,LEAH TOLOSA, JONATHAN D. DATTELBAUM, and GOVIND RAODepartment of Biochemistry and Molecular Biology, Center for Fluorescence Spectroscopy, University of Maryland School of

Medicine, 725 West Lombard Street, Baltimore, Maryland 21201 (J.R.L., I.G., Z.G., L.T., J.D.D.); and Medical Biotechnology

Center and Department of Chemical and Biochemical Engineering, University of Maryland Biotechnology Institute, 725 West

Lombard Street, Baltimore, Maryland 21201 (G.R.)

We describe a new method of ¯ uorescence sensing based on ¯ uo-

rescence polarization. The sensor consists of two compartments,

both of which contain the sensing ¯ uorophore. One side of the sen-sor contains a constant concentration of analyte, and the other con-

tains the unknown concentration. Emission from both sides is ob-

served through polarizers, with the polarization from the samplebeing rotated 90 8 from that of the reference. Changes in the ¯ uo-

rescence intensity of the sample result in changes in the measured

polarization for the combined emission. We show that this approachcan be used to measure glucose and calcium using ¯ uorophores

which show analyte-dependent intensity changes, and no change in

the spectral shape. Only a single ¯ uorophore is required, this beingthe sensing ¯ uorophore in both sides of the sensor. We also show

that polarization sensing of glucose and calcium can be performed

with visual detection of the polarization. In this case the only elec-tronic component is the light source. These simple schemes can be

used with a variety of analytes. The only requirement is a change

in ¯ uorescence intensity in response to the analyte.

Index Headings: Fluorescence sensing; Glucose; Calcium; Proteinsensors; Polarization; Anistropy.

INTRODUCTION

During the past decade there have been continued ad-vances in the methodology for ¯ uorescence sensing. 1±5

Numerous new sensing ¯ uorophores have been devel-oped that provide responses to speci® c analytes, espe-cially to cations and anions.6±10 A signi® cant problem in¯ uorescence sensing is the dif® culty of performing inten-sity measurements in a real-world situation. Intensitymeasurements are typically unreliable or require frequentre-calibration because of a variety of chemical and in-strumental factors. Hence, there has been a considerableeffort to develop sensing ¯ uorophores that display spec-tral shifts in response to analyte binding, the so-calledwavelength-ratiometric probes. Such probes are availablefor Ca2 1 ,11±13 Mg 2 1 ,14,15 and pH.16 However, it has notbeen possible to develop wavelength-ratiometric probesfor all analytes, particularly analytes such as chloride 17,18

and oxygen,19±23 which act as collisional quenchers.Methods have been developed to avoid the dependence

on intensity measurements. One approach is to use the¯ uorescence lifetime instead of the intensity, which iscalled lifetime-based sensing. 24,25 In this case, one selectsa ¯ uorophore that displays a change in lifetime in re-sponse to the analyte. Lifetime-based sensing has beenreported for a wide range of analytes, including colli-sional quenchers. 26,27

Another method to avoid the need for absolute inten-

Received 22 January 1999; accep ted 19 April 1999.* Author to whom correspondence should be sent.

sity measurements is to use ¯ uorescent internal standards.In this case the sensor contains two ¯ uorophores, one ofwhich responds to changes in the analyte concentration.This approach has been used with phase-modulation ¯ uo-rometry to develop sensors for pH and pCO2,

28,29 glu-cose,30 and calcium.31 These approaches require the useof an amplitude-modulated light source, which is avail-able from simple light-emitting diode light sources.32±35

We now describe a new method of internal referencingthat can be accomplished with only steady-state mea-surements and can be used whenever the ¯ uorescenceintensity changes in response to the analyte. This self-reference method is shown in Scheme I. The sensor con-sists of two parts. One side contains the sensing ¯ uoro-phore and a constant concentration of analyte, and is re-garded as the reference (R). The other side contains theunknown concentration of analyte, and is regarded as thesample (S). The emission from each side passes througha ® lm polarizer prior to reaching the detector. The ori-entation of the sample polarizer (P^) is rotated 90 8 rela-tive to the reference polarizer (P \). The emission fromboth sides of the sensor is observed through an analyzerpolarizer, which allows measurement of the parallel (I\

R)and perpendicular (I^

S ) components of the combinedemission. The parallel (\) orientation is taken as the lab-oratory vertical axis. The combined emission is then ob-served through an analyzer polarizer. Because of theemission polarizers on the sample, the parallel measure-ment reveals the intensity of the reference (I\

R), and theperpendicular measurement reveals the intensity of thesample (I^

S ). These values are used to calculate the po-larization of the combined emission. Changes in the in-tensity from the sample (I^

S ) result in changes in the po-larization values.

The approach shown in Scheme I promises to be bothsimple and reliable. Only a single sensing ¯ uorophore isneeded, and it is used on both sides of the sensor. Fur-thermore, it seems likely that factors which affect theintensity of the sensing ¯ uorophore will be similar forboth sides of the sensor, which should result in goodlong-term stability of the calibration curve.

In the present report we describe the operating prin-ciples of the polarization sensor (Scheme I). We thenshow how this approach can be used to measure glucoseand calcium, in both cases using probes which changeintensity in response to the analyte but do not show spec-tral shifts. We also show how the sensor shown inScheme I can be adapted for use with visual detection ofthe polarization.

1150 Volume 53, Number 9, 1999

SCHEME I. Experimental con® guration for polarization-based sensing.

SCHEME II. Polarization sensing with visual detection. The values ofa is zero degrees when the analyzer polarizer is oriented vertically.

THEORY

Polarization-Based Sensing. The theory describingthe polarization values from the combined sensor can bedeveloped with a few simple considerations. The polari-zation of the combined emission from the sample andreference is given by

T TI 2 I\ ^P 5 (1)

T TI 1 I\ ^

where the superscript T indicates the sum of the inten-sities from the sample and reference. The subscripts in-dicate the parallel (\) and perpendicular (^) componentsof the emission as measured through the analyzer polar-izer. Because of the polarizers in front of the referenceand sample sides of the sensor, the polarized intensitiesare given by

T RI 5 I (2)\ \

T SI 5 I (3)^ ^

where I\R and I^

S represent the intensities from the ref-erence and sample, respectively. In the present cases, andin many situations, the emission from the sample andreference is unpolarized, so that the polarized intensitiesare proportional to the total intensity from each side ofthe sensor.

The intensities from the sample and reference dependon a number of instrumental factors, including the exci-tation intensity, the probe concentration, the ® lter trans-mission, the polarizer ef® ciency, the quantum yield, theobservation wavelength, and the intensity of each lightsource. For simplicity, we chose not to explicitly indicatethese factors.

Substituting Eqs. 2 and 3 into Eq. 1 yields

R SI 2 I\ ^P 5 . (4)

R SI 1 I\ ^

The measured polarization depends on the intensity ofthe sample relative to the reference. If the sample inten-sity is very low, then the polarization approaches 1.0. Ifthe emission from the sample dominates, then the polar-ization approaches 2 1.0. Hence a wide range of polari-zation values is available, resulting in a wide dynamicrange for the sensor.

It is important to notice that polarization-based sensingcan be accomplished without a change in the polarizationof the sample. This counterintuitive result is obtained be-cause the polarizers on the emission side of the sensorprovide polarized light from sample and reference.

Polarization-Based Sensing with Visual Detection.Polarization-based sensing can also be accomplished withvisual detection using the apparatus in Scheme II. In thiscase the emission sides of the reference and sample areagain covered with ® lm polarizers in the vertical and hor-izontal orientations, respectively. The emission is ob-served visually through a long-pass ® lter and a secondanalyzer polarizer. The intensities on the vertical and hor-izontal sides of the image are given by

V R 2I 5 I cos a (5)\

S 2I 5 I sin a (6)H ^

where a is the angular displacement of the analyzer fromthe vertical position. For visual measurements, the ana-lyzer is rotated until the intensities from the reference andsample are perceived to be equal. For this condition onehas

R 2 S 2I cos a 5 I sin a (7)\ ^

and

RI\2tan a 5 . (8)

SI^

Changes in the intensity from the sample result in chang-es of the analyzer angle needed to equalize the intensities.We call this difference the compensation angle, D a .

D a 5 a 0 2 a . (9)

In this expression a 0 refers to the angle needed to equal-ize the intensities in the absence of analyte, or some otherchosen initial condition. The value of a 0 will depend onthe intensity of the reference relative to that from thesample with no analyte or at the initial condition.

At ® rst glance one is tempted to believe that the ac-curacy would be poor in visual determination of a . Infact, we found that an accuracy of one to two degreescan be routinely obtained.36 This excellent accuracy isdue to the high sensitivity of the human eye in detectingsmall differences in relative intensity. To be more precise,we found that a 10% difference in intensity is detectableby most individuals, and a 10% difference is adequate todetermine a to within one to two degrees.36

APPLIED SPECTROSCOPY 1151

FIG. 1. Emission spectra of the vertical (V) and horizontal (H) com-ponents from the combined sensor. The vertical polarizer is placed infront of the ANS/HSA solution. The horizontal polarizer is placed infront of the ANS-Q26C GGBP sample. The upper and lower panelsshow the component spectra in the absence and presence of 8 m Mglucose, respectively.

FIG. 2. Wavelength-dependent polarization of the combined emissionfrom ANS-Q26C GGBP and ANS/HSA. The arrow shows the wave-length chosen for measurement of the glucose concentration.

MATERIALS AND METHODS

Human serum albumin (HSA) and 8-anilino-1-naph-thalenesulfonic acid (ANS) were obtained from Sigma,Inc. and used without further puri® cation. The HSA con-centration was 3.3 mg/mL. The ANS concentration was1.2 3 10 2 5 M as calculated from Î (372 nm) 5 7800 M 2 1

cm 2 1. The solution of HSA and ANS was in 0.05 Mphosphate buffer, pH 5 7.

The glucose assay was accomplished by using the glu-cose/galactose binding protein (GGBP) from E. coli. Weused a mutant which contained a single cysteine residueat position 26.30 This protein was labeled with (4 9 -io-doacetamidoanilino)naphthalene-6-sulfonic acid (I-ANS)from Molecular Probes, Inc. A solution containing 2.5mg/mL Q26C GGBP in 20 mM phosphate, 1 mM tris-(2-carboxyethyl)phosphine (TCEP), pH 7.0, was reactedwith 50 m L of a 20 mM solution of I-ANS in tetrahy-drofuran (purchased from Molecular Probes, Inc.). Theresulting labeled protein was separated from the free dyeby passing the solution through a Sephadex G-25 column.The protein-ANS conjugate was puri® ed further on Se-phadex G-100 and dissolved in 20 mM phosphate, pH7.0.

Fluo-3 was obtained from Molecular Probes, Inc. Thecalcium concentration was controlled by using calciumbuffer kits, C-3009, also from Molecular Probes, Inc.

RESULTS

Operating Principle of Self-Referenced PolarizationSensing. To illustrate the principles of polarization sens-ing, we chose to initially present our results using twodifferent ¯ uorophores. One side (V) of the sensor con-tained a constant concentration of HSA with noncova-lently bound ANS. The other side (H) of the sensor con-tained the glucose/galactose binding protein from E. coli.These samples displayed similar but slightly differentemission spectra. The emission maxima for ANS/HSAand ANS-Q26C GGBP (cysteine26 mutant of GGBP la-beled with I-ANS) were 485 and 450 nm, respectively.

The different emission maxima for the two sides of thesensor allow visualization for determining the contribu-tion from each ¯ uorophore to the total emission. Theemission spectra were recorded from the combined sensor(Fig. 1). When the emission polarizer was in the verticalorientation, the emission spectrum was characteristic ofANS/HSA. When the emission polarizer was in the hor-izontal position, the emission spectrum was characteristicof labeled GGBP. These results demonstrate that Eqs. 2and 3 provide an accurate description of the polarizedcomponents of the emission. More speci® cally, the emis-sion spectra observed through a vertically or horizontallyoriented polarizer represent the emission from ANS/HSAand ANS-Q26C GGBP, respectively.

GGBP labeled with I-ANS displays only a moderatechange in intensity due to glucose. From other experi-ments we know that the glucose binding constant is near1 m M.30 Addition of 8 m M glucose to labeled GGBP re-sults in an approximately twofold decrease in the inten-sity of the ANS label (Fig. 1). We used this decrease asthe basis of our polarization sensor.

Polarization values were measured across the emissionspectra from the combined sensor (Fig. 2). The polari-zation values increase with increasing wavelength. This


FIG. 3. Glucose-dependent polarization values from the two-part sen-sor.

FIG. 4. Emission spectra (top) and polarization spectra (bottom) for atwo-part sensor consisting of only ANS-Q26C GGBP in both sides ofthe sensor. The glucose concentration was constant on the left (V) sideof the sensor and was varied in the right (H) side of the sensor (seeScheme I).

effect is due to the increasing contribution of ANS/HSAat longer wavelengths, and the vertically oriented polar-izer in front of ANS/HSA. The polarization values arelower and even negative at shorter wavelengths becauseof the shorter wavelength emission of ANS-Q26C GGBPand the horizontal polarizer in front of this sample. Thepolarization values were also found to be dependent onthe glucose concentration. Increasing amounts of glucoseresult in larger polarization values. This change occursbecause binding of glucose to labeled GGBP decreasesits intensity. Hence the polarization shifts towards thevalue of 1 1.0Ð characteristic of the ANS/HSA side ofthe sensor.

We used the glucose-dependent polarization values todevelop a calibration curve for glucose (Fig. 3). It is sur-prising that the polarization increases substantially from0.06 to 0.36. This range is larger than that encounteredfor typical polarization assays in which the polarizationchange is due to changes in the rotational correlation timeof the ¯ uorophore. This larger range of polarization val-ues is the result of using a pair of orthogonally orientedpolarizers as part of the sensor design. The increase inpolarization is complete above 6 m M glucose. This resultoccurs because ANS-Q26C GGBP becomes saturatedwith glucose and the intensity becomes constant.

It is informative to consider the potential accuracy ofthese glucose measurements. Since polarization measure-ments are easily accurate to 6 0.01, we estimate the ac-curacy in glucose concentrations to be about 6 0.1 m M.In the future one can expect glucose binding proteins thatdisplay larger changes in intensity than shown by ANS-Q26C GGBP. In these cases the polarization change willbe larger, and the accuracy in the glucose concentrationwill be higher. Of course the maximum accuracy will befound near the midpoint of the glucose-GGBP bindingcurve. The accuracy will decrease as the glucose concen-tration becomes much smaller or much larger than theglucose binding constant.

Glucose Assay with a Self-Reference. We next ex-amined a sensor that contained ANS-Q26C GGBP onboth sides of the sensor. The reference solution was ob-served through a vertical polarizer, and the solution withvarious glucose concentrations was observed through ahorizontal polarizer (Fig. 4, top). When the analyze po-larizer was in the vertical orientation, the emission spec-trum was independent of glucose concentration ( ).This result occurs because the vertical analyzer polarizerselects for emission from the reference side of the sensor.When the analyzer polarizer is in the horizontal orienta-tion, the emission spectra display decreasing intensitywith increasing glucose concentration (± ± ±). This resultoccurs because the horizontal analyzer selects the emis-sion from the side of the sensor that contains the variableglucose concentration.

We measured the polarization spectra for the combinedemission from the sensor (Fig. 4, bottom). In this casethe polarization values are independent of wavelength be-cause both sides of the sensor display the same emissionspectra. This is an advantage of using the sensor as thereference: there is no dependence of the polarization val-ues on the observation wavelength. As the glucose con-


FIG. 5. Glucose-dependent polarization sensing using only ANS-Q26CGGBP in both sides of the sensor.

FIG. 6. Polarization sensing of calcium using Fluo-3. The top panelshows the emission spectra of the vertical component (Scheme I, R)and of the horizontal component (Scheme I, S). The calcium concen-tration is constant at 1.35 m M in the left side (R) of the sensor. Thecalcium concentration is variable in the right side of the sensor (S). Thelower panel shows the polarization across the emission spectra.

centration increases, the polarization increases. This in-crease occurs because the intensity of ANS-labeledGGBP decreases in the presence of glucose, so that thevertically polarized reference emission contributes a larg-er fraction to the total emission.

The polarization values were again used to develop aglucose calibration curve (Fig. 5). In this case the polar-ization changes from near zero to over 0.3. This changein polarization occurs for only a twofold change in sam-ple intensity. Since the reference and sample are the same¯ uorophore, one expects both sides of the sensor to dis-play similar photobleaching or dependence on tempera-ture. Hence the glucose calibration curve should be stablefor extended periods of time.

The reader may question the usefulness of a glucosesensor with micromolar sensitivity when the concentra-tion of glucose in the blood is near 5 mM. High glucoseaf® nity is useful for two reasons. First, one can use aminimum volume of blood, which is then diluted into thesample in contact with the sensor. Second, there is in-creasing interest in the use of extracted interstitial ¯ uidto monitor blood glucose. In this sample the glucose con-centration is often in the micromolar range.37

Polarization Sensing of Calcium. Fluorescence sens-ing methods are needed for a variety of cations and an-ions, particularly for blood gases and electrolytes. Hencewe examined how our polarization sensor could be usedto measure calcium. For these experiments we chose thecalcium-sensitive ¯ uorophore Fluo-3.38 This ¯ uorophoredisplays a dramatic increase in ¯ uorescence upon bindingcalcium, approximately 100-fold.39 While the intensitychange of Fluo-3 is dramatic, there is no spectral shiftupon calcium binding. Hence, wavelength-ratiometricmeasurements are not possible. Furthermore, Fluo-3 isnot useful with lifetime-based sensing. This is becauseFluo-3 is non¯ uorescent in the absence of calcium. If thelifetime is measured, one observes only the emission

from the calcium-bound form. Hence, the lifetimes areindependent of calcium concentration.

While Fluo-3 is not useful as a wavelength-ratiometricof lifetime sensor, its large change in intensity makes itwell suited for use in polarization sensing. The sensorwas con® gured with Fluo-3 in both sides. The calciumconcentration was constant at 1.35 m M in the reference(vertical) side and was variable in the sample (horizontal)side of the sensor. Emission spectra were recordedthrough the analyzer polarizer (Fig. 6, top). The spectrumwas constant with the analyzer in the vertical orientationbecause of the constant signal from the vertical side ofthe sensor. The emission spectrum seen with the analyzerin the horizontal position increases with the calcium con-centration because of the intensity increase of Fluo-3.

Since the sensor contained Fluo-3 on both sides, thepolarization was independent of wavelength (Fig. 6, bot-tom). The polarization decreased dramatically from 0.8to 0.0 with increasing calcium concentrations. The de-crease in polarization occurs because the intensity fromthe horizontally polarized side of the sensor increases asthe calcium concentration increases. The calibrationcurve for the polarization values (Fig. 7) shows a dra-matic dependence on the calcium concentration. In thiscase the calcium concentrations are expected to be ac-curate to 6 0.01 m M. Once again, since the same ¯ uoro-


FIG. 7. Calcium-dependent polarization values using only Fluo-3 inboth sides of the sensor.

FIG. 8. Images seen through the analyzer for different concentrations of glucose (top) and calcium (bottom). In each case the initial angleof the analyzer was adjusted to yield equal intensities in the absence of glucose on calcium. The images were then recorded at this sam eanalyzer angle. The values under the images are the analyzer angles needed to equalize the intensities, not the angle needed to equalize theintensities.

phore is present on both sides of the sensor, one expectssimilar changes in the emission due to temperature orphotobleaching. Hence one can anticipate a stable calci-um calibration curve. If desired, the polarization could

increase with higher calcium concentrations by reversingthe sample and reference in Scheme I. The calcium-sen-sitive range is from 0 to 400 nM Ca 2 1 , which is the typ-ical range for intracellular calcium.

Glucose and Calcium Sensing with Visual Detec-tion. In another report we described polarization sensingwith visual detection.36 The basic idea is to visually ob-serve the emission from both sides of the sensor throughan analyzer polarizer (Scheme II). The analyzer is rotateduntil the adjacent intensities from the vertically and hor-izontally polarized emission are equalized. The angle ofthe analyzer polarizer is used to determine the analyteconcentration.

We applied this method to sensing of glucose and cal-cium, using the apparatus shown in Scheme II. The im-ages seen through the analyzer polarizer are shown inFig. 8. For these images the initial angular rotation ofthe analyzer ( a 0) was adjusted to match the intensitiesseen from both sides of the sensor. The images werethen recorded for various analyte concentrations, withthe analyzer left in the same angular position ( a 0). Asthe glucose concentration increases, the intensity fromthe horizontal (right) side of the sensor decreases, as canbe seen in the top panel of Fig. 8. As the calcium con-centration increases, the intensity in the horizontal(right) side of the sensor increases, as can be seen in thelower panel.

The angular position of the analyzer can be adjustedto yield visually equivalent intensities. These angles areshown under the images in Fig. 8. For glucose the angles


FIG . 9. Calibration curve for glucose sensing with visual detection.

FIG . 10. Calibration curve for calcium sensing with visual detection.

of the analyzer must be increased to equalize the inten-sities. This effect is due to the decreased horizontal in-tensity, as well as the need to increase the horizontal in-tensity to visually match both sides of the sensor. Forincreasing concentrations of calcium, the angle of the an-alyzer must be decreased to equalize the intensities. Thisdirection of change is needed because the increased in-tensity from the horizontal component must be attenuatedto yield visually equivalent intensities.

Figures 9 and 10 show the calibration curves for thecompensation angles for glucose and calcium, respec-tively. In another report we found that the compensationangles are typically accurate to one or two degrees.36 Anaccuracy of 1 8 in the compensation angle results in anaccuracy of 6 0.5 m M in glucose and 6 0.05 m M in cal-cium, as found for the most sensitive part of the curve.While the accuracy is somewhat less than that availablewith electronic detection, the accuracy may be adequatefor some clinical or analytical applications, particularlywhere a yes/no answer is adequate.

DISCUSSION

What are the advantages of the polarization sensingschemes described above? One advantage is the needfor just one ¯ uorophore, which is used in both sides ofthe sensor. This arrangement should result in stable cal-ibration curves that are insensitive to many instrumentalor chemical changes in the sample. Another advantageis the wide dynamic range of polarization values, from1 1.0 to 2 1.0. Importantly, this range of values is avail-able without any change in the actual polarization of theemission of the sensing ¯ uorophores. The polarizationvalues of 1 1.0 and 2 1.0 are imposed on the sensor bythe use of polarizers through which the emission is ob-served.

Perhaps the most important characteristic of polariza-tion sensing is that the approach is generic and can be

used in any situation where an intensity change occurs.For instance, intensity changes occur for the sodium andpotassium probes SBFI and PBFS,40 but the spectralshifts are rather small. As shown for the glucose sensor,a twofold change in intensity is easily adequate for apolarization sensor. Also, polarization sensing can beused for collisional-quenched ¯ uorophores that do notdisplay spectral shifts. Hence, one can imagine polari-zation sensors for Na 1 , K 1 , Mg 2 1 , Cl {, pH, and O2, toname a few.

It is also important to recognize the increasing avail-ability of intensity-based ¯ uorophores for glucose. Theseprobes contain boron and depend on binding of the glu-cose diol to the boron, resulting in a change in probeintensity.41±45 Such glucose-speci® c ¯ uorophores could beused with polarization sensing to develop low-cost de-vices for measuring glucose in blood.

In summary, polarization sensing provides a simplemethod to convert any change in intensity into a changein polarization. The use of polarizers within the sensordesign results in a wide range of polarization values.Polarization sensors can most probably be designedwith solid-state light sources and detectors, makingclinical measurements available for point-of-care ap-plications.

ACKNOWLEDGMENT

This work was supported by a grant from the National Institutes ofHealth National Center for Research Resources, RR-08119, RR-10955,and RR-14170.

APPENDIX

In the case of glucose sensing, the labeled protein pro-vided only a modest twofold change in intensity in re-sponse to glucose. When one is using visual detectionwith such sensors, it is important to select initial condi-tions that result in the largest overall change in a . Sup-pose the initial (0) polarized intensities of the sample and


FIG. 11. Effect of the initial conditions on the total change in anglefor visual sensing.

reference are I0S and I0

R, where we have dropped the po-larization subscript for simplicity. According to Eq. 8 theinitial angle is given by

RI02tan a 5 5 n (A1)0 SI0

where n equals the initial intensity ratio. Hence

a 0 5 a tan Ï n. (A2)

Now suppose the addition of analyte changes the inten-sity on the sample side of the sensor by a factor a . Thesample intensity is then given by

I S 5 dI0S. (A3)

The compensation angle needed to equalize the intensi-ties is then given by

a 5 a tan Ï n /d. (A4)

Hence, the change in compensation angle for a factor ofd change in sample intensity is given by

D a 5 a 2 a 0 5 a tan Ï n /d 2 a tan Ï n. (A5)

We used Eq. A5 to calculate the values of n needed toobtain the maximum changes in a . These simulations areshown in Fig. 11. The calcium sensor displays a largeincrease in intensity upon binding calcium; that is, d islarger than 10. For this situation, it is preferable to chosethe initial conditions so that n is near 3; that is, the ref-erence is about threefold more intense than the sample.For the glucose sensor, the intensity decreases upon glu-cose binding, and d is near 0.5. In this case it is desirableto begin the experiment with n near 0.5, which is the

case where the sensor is about twofold more ¯ uorescentthan the reference.

1. U. E. Spichiger-Keller, Chemical Sensors and Biosensors for Med-ical and Biological Applications (Wiley±VCH, New York, 1998),p. 413.

2. Fiber Optic Chemical Sensors and Biosensors , O. S. Wolfbeis, Ed.(CRC Press, Boca Raton, Florida, 1991), Vol. II, p. 358.

3. Sensors and Actuators B , Proceedings of the 3rd European Con-ference on Optical Chemical Sensors and Biosensors, Part I: Ple-nary and Parallel Sessions; Part II: Poster Sessions, R. E. Kunz,Ed. (Elsevier, New York, 1996).

4. Topics in Fluorescence Spectroscopy , Vol. 4, Probe Design andChemical Sensing, J. R. Lakowicz, Ed. (Plenum Press, New York,1994), p. 501.

5. Advances in Fluorescence Sensing Technology III , SPIE Proc. Vol.2980, R. B. Thompson, Ed. (SPIE, Bellingham, Washington, 1997),p. 582.

6. L. Fabbrizzi, M. Licchelli, L. Parodi, A. Poggi, and A. Taglietti, J.Fluoresc. 8, 263 (1998).

7. A. D. Roshal, A. V. Grigorovich, A. O. Doroshenko, V. G. Pivo-varenko, and A. P. Demchenko, J. Phys. Chem. A 102, 5907 (1998).

8. L. Prodi, F. Bolletta, N. Zaccheroni, C. I. F. Watt, and N. J. Mooney,Chem. Eur. J. 4, 1090 (1998).

9. C. R. Chenthamarakshan and A. Ajayaghosh, Tetrahedron Letts. 39,1795 (1998).

10. A. Metzger and E. V. Anslyn, Angew. Chem. Int. Ed. 37, 649(1998).

11. J. P. Y. Kao, Methods in Cell Biology (Academic Press, New York,1994), Vol. 40, pp. 155±181.

12. R. Y. Tsien, Methods in Cell Biology (Academic Press, New York,1989), pp. 127±156.

13. E. U. Akkaya and J. R. Lakowicz, Anal. Biochem. 213, 285 (1993).14. H. Illner, J. A. S. McGuigan, and D. LuÈ thi, P¯ ugers Arch 422, 179

(1992).15. B. Raju, E. Murphy, L. A. Levy, R. D. Hall, and R. E. London,

Am. J. Physiol. 256, C540 (1989).16. J. E. Whitaker, R. P. Haugland, and G. Predengast, Anal. Biochem.

194, 330 (1991).17. A. S. Verkman, M. C. Sellers, A. C. Chao, T. Leung, and R. Ket-

cham, Anal. Biochem. 178, 355 (1989).18. J. Biwersi, B. Tulk, and A. S. Verkman, Anal. Biochem. 219, 139

(1994).19. A. Mills, and F. C. Williams, Thin Solid Films 306, 163 (1997).20. P. Hartman and M. J. P. Leiner, Anal. Chem. 6, 88 (1995).21. W. Xu, K. A. Kneas, J. N. Demas, and B. A. DeGraf, Anal. Chem.

68, 2605 (1996).22. I. Klimant, P. Belser, and O. S. Wolfbeis, Talanta 41, 985 (1994).23. H. Chuang and M. A. Arnold, Anal. Chim. Acta 368, 83 (1998).24. H. Szmacinski and J. R. Lakowicz, in Topics in Fluorescence Spec-

troscopy, Vol. 4, Probe Design and Chemical Sensing, J. R. Lak-owicz, Ed. (Plenum Press, New York, 1994), pp. 295±334.

25. H. Szmacinski and J. R. Lakowicz, Sensors and Actuators B 29,30 (1995).

26. M. E. Lippitsch, S. Draxler, and D. Kieslinger, Sensors and Actu-ators B 38± 39, 96 (1997).

27. M. E. Lippitsch, J. Pusterhofer, M. J. P. Leiner, and O. S. Wolfbeis,Anal. Chim. Acta 205, 1 (1988).

28. G. Neurauter, I. Klimant, G. Liebsch, U. Kosch, and O. S. Wolfbeis,Europt(r)ode IV (German Chemical Society, Munich, 1998), pp.231±232.

29. I. Klimant and O. S. Wolfbeis, Europt(r)ode IV (German ChemicalSociety, Munich, 1998), pp. 125±126.

30. L. Tolosa, I. Gryczynski, L. Eichhorn, J. D. Dattelbaum, F. N. Cas-tellano, G. Rao, and J. R. Lakowicz, Anal. Biochem. 267, 114(1999).

31. J. R. Lakowicz, F. N. Castellano, J. D. Dattelbaum, L. Tolosa, G.Rao, and I. Gryczynski, Anal. Chem. 70, 5115 (1998).

32. J. Sipior, G. M. Carter, J. R. Lakowicz, and G. Rao, Rev. Sci.Instrum. 68, 2666 (1997).

33. J. Sipior, G. M. Carter, J. R. Lakowicz, and G. Rao, Rev. Sci.Instrum. 67, 3795 (1996).

34. S. Fantini, M. A. Franceschini, J. B. Fishkin, B. Barbieri, and E.Gratton, Appl. Opt. 33(22), 5204 (1994).


35. L. Randers-Eichhorn, C. R. Albano, J. Sipior, W. E. Bentley, andG. Rao, Biotech. Bioeng. 55, 921 (1997).

36. I. Gryczynski, Z. Gryczynski, and J. R. Lakowicz, Anal. Chem.,paper in press (1999).

37. J. A. Tamada, N. J. V. Bohannon, and R. O. Potts, Nature Medicine2, 1198 (1995).

38. A. Minta, J. P. Y. Kao, and R. Y. Tsien, J. Biol. Chem. 264, 8171(1989).

39. R. P. Haugland, Handbook of Fluorescent Probes and ResearchChemicals (Molecular Probes, Eugene, Oregon, 1996), 6th ed. p.511.

40. A. Minta and R. Y. Tsien, J. Biol. Chem. 264, 19449 (1982).41. M. Takeuchi, T. Imada, and S. Shinkai, J. Am. Chem. Soc. 118,

10658 (1996).42. K. Kataoka, I. Hisamitsu, N. Sayama, T. Okano, and Y. Sakurai, J.

Biochem. 117, 1145 (1995).43. K. R. A. S. Sandanayake, T. D. James, and S. Shinkai, Pure Appl.

Chem. 68, 1207 (1996).44. M. Yamamoto, M. Takeuchi, and S. Shinkai, Tetrahedron 54, 3125

(1998).45. M. Takeshita, K. Uchida, and M. Irie, Chem. Commun. 1807

(1996).

polarization-based sensing with a self-referenced sample

Documents