A fluorescent chemosensor for sodium based on photoinduced electron transfer

Article abstract of DOI:10.1021/ac0205107

A new optical sensor suitable for practical measurement of sodium in serum and whole blood samples is described. The optical sensor is based on a novel PET (photoinduced eletron transfer) fluoroionophore immobilized in a hydrophilic polymer layer. The des

Full text of DOI:10.1021/ac0205107

Anal. Chem. 2003, 75, 549-555
A Fluorescent Chemosensor for Sodium Based on
Photoinduced Electron Transfer
Huarui He,*,† Mark A. Mortellaro,† Marc J. P. Leiner,‡ Susanne T. Young,† Robert J. Fraatz,† and
James K. Tusa†
Roche Diagnostic Corporation, 235 Hembree Park Drive, Roswell, Georgia 30076, and Roche Diagnostic GmbH,
1 Hans-List-Platz, A-8020 Graz, Austria
sensitive indicator dye3 or interaction of a potential-sensitive dye
with a neutral ion carrier.4,5 Most of these measuring methods
suffer from inherent pH dependencies (if based on pH indicators
or proton exchange), or instabilities associated with leaching of
critical components (if based on an ionophore to extract the
desired cation into a lipophilic polymer membrane). The instabili-
ties arise when an ionophore, or so-called ion carrier, is mobile
and separated from the indicator dye; this leads to instabilities
associated with differences in leaching rate and lipophilicity, which
in turn are aggravated in measurements of fatty bodily fluids, such
as plasma or whole blood.
A new optical sensor suitable for practical measurement
of sodium in serum and whole blood samples is de-
scribed. The optical sensor is based on a novel P ET
(photoinduced eletron transfer) fluoroionophore immo-
bilized in a hydrophilic polymer layer. The design concept
of the fluoroionophore follows the receptor-spacer-fluoro-
phore approach to sensor design using intramolecular
P ET-based signal transduction. Key to the development
of this sensor is the identification of a nitrogen-containing,
sodium-binding ionophore, coupled with a fluorophore
having the correct spectral and electron-accepting proper-
ties. The slope of the sensor is ∼0 .5 %/ mM in the typical
clinically significant range of 1 2 0 -1 6 0 mM. This sensor
has been implemented into a disposable cartridge, used
for a commercially available critical care analyzer (Roche
OP TI CCA) with precision better than (1 mM (1 SD). The
sensor displays excellent stability against hydrolysis and
oxidation, leading to slope changes <5 % after 9 months
wet storage at 3 0 °C. On the basis of this design concept,
fluoroionophores for other cations such as potassium,
calcium and magnesium can be prepared by substitution
of the ionophore.
A number of direct sensing schemes have been described
using colorimetric and fluorescent chemosensor molecules (chromo-
ionophores and fluoroionophores) whose optical properties change
upon direct binding of the cation. For example, chromoionophores
have been prepared by covalently linking colored dyes to simple
crown ethers.5,6,29 Unfortunately, the resulting indicators are
(3) Morf, W. E.; Seiler, K.; Lehmann, B.; Behringer, C.; Hartmann, C.; Simon,
W. Pure Appl. Chem. 1 9 8 9 , 61, 1613-1620.
(4) Wolfbeis, O. S. Sens. Actuators 1 9 9 5 , 29B, 140-147.
(5) Wolfbeis, O. S.; Schaffar, B. P. H. Anal. Chim. Acta 1 9 8 7 , 198, 1-7.
(6) Takagi, M.; Nakamura, H.; Ueno, K. Anal. Lett. 1 9 7 7 , 10, 1115-1122.
(7) Dix, J. P.; Vo¨ gtle, F. Chem. Ber. 1 9 8 0 , 113, 457-470.
(8) Minta, A.; Tsien, R. J. Biol. Chem. 1 9 8 9 , 264, 19449-19457.
(9) Leiner, M. J. P. Anal. Chim. Acta 1 9 9 1 , 255, 209-222.
(10) During preparation of this manuscript, a paper describing a new fluoro-
ionophore and a chromoionophore for measurement of Na+ at millimolar
concentrations appeared. In that paper, the authors construct their sensors
with the sodium ionophore 3 described in this paper as well as in ref 1.
Their fluoroionophore uses the UV excitable anthracene as fluorophore.
See: Gunnlaugsson, T.; Nieuwenhuyzen, M.; Richard, L.; Thoss, V. J. Chem.
Soc., Perkin Trans. 2 2 0 0 2 , 141-150.
The accurate measurement of physiologic cations, such as
sodium, potassium, calcium, and magnesium, is essential in clinical
diagnosis. Traditionally, these analytes were determined in plasma
or serum using ion-selective electrodes; however, with the
explosive growth of near-patient devices used at the hospital
bedside, there is increasing demand for portable systems utilizing
small disposable sensors capable of whole-blood measurements.
Consequently, the development of practical and inexpensive optical
sensors and systems for the clinical determination of these
analytes in whole blood remains an important area of research.1,2
Numerous indirect sensing schemes involving multiple types
of molecules have been described, such as ion exchange between
the measured ion and a proton measured with a lipophilic pH-
(11) Davidson, S. R. Adv. Phys. Org. Chem. 1 9 8 3 , 19, 1-130.
(12) Electron Transfer in Chemistry; Balzani, V., Ed.; Wiley-VCH: Weinheim, 2001.
(13) Burdette, S. C.; Walkup, G. K.; Spingler, B.; Tsien, R. Y.; Lippard, S. J. J.
Am. Chem. Soc. 2 0 0 1 , 123, 7831-7841.
(14) Yoshida, K.; Mori, T.; Watanabe, S.; Kawai, H.; Nagamura, T. J. Chem. Soc.,
Perkin Trans 2 1 9 9 3 , 393-398.
(15) Bissell, R. A.; de Silva, A. P.; Gunaratne, H. Q. N.; Lynch, M. P. L.; Maguire,
G. E. M.; Sandanayake, K. R. A. S. Chem. Soc. Rev. 1 9 9 2 , 21, 187-195.
(16) Bryan, A. J.; de Silva, A. P.; Rupasinghe, R. A. D. D.; Sandanayake, K. R. A.
S. Biosensors 1 9 8 9 , 4, 169-179.
* Corresponding author. Phone: (770) 576-5000. Fax: (770) 576-5020.
E-mail: huarui.he@roche.com.
(17) de Silva, A. P.; Gunaratne, H. Q. N.; Gunnlaugsson, T.; Huxley, A. J. M.;
McCoy, C. P.; Rademacher, J. T.; Rice, T. E. Chem. Rev. 1 9 9 7 , 97, 1515-
1566.
(18) Presented in part at the 218th National ACS Meeting, New Orleans, LA,
August 22-26, 1999.
† Roche Diagnostic Corporation.
‡ Roche Diagnostic GmbH.
(1) Desvergne, J.-P., Czarnik, A. W., Eds. Chemosensors of Ion and Molecule
Recognition; NATO ASI Series; Kluwer Academic Publishers: Dordrecht,
The Netherlands, 1996.
(19) Schultz, R. A.; White, B. D.; Dishong, D. M.; Arnold, K. A.; Gokel, G. W. J.
Am. Chem. Soc. 1 9 8 5 , 107, 6659-6668.
(2) Fiber Optic Chemical Sensors and Biosensors; Wolfbeis, O. S., Ed.; CRC
Press: Boca Raton, FL, 1991; Vol. II.
(20) Zagrebel’nyi, S. N.; Yasnetskaaya, S. M.; Vasil’eva, L. A. Okrytica, Izobret.
Prom. Obraztsu, Tovarnye Znaki 1 9 8 3 , 26, 76. (Chem. Abstr. 99: 177723h).
10.1021/ac0205107 CCC: $25.00 © 2003 American Chemical Society
Published on Web 12/24/2002
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003 549

Figure 1. Synthetic scheme for the preparation of fluoroionophore 11. Abbreviations: DMF, N,N′-dimethylformamide; AMBA, 4-(aminomethyl)-
benzoic acid; CDI, 1,1′-carbonyldiimidazole; DBU, 1,8-diazabicyclo[5.4.0]undec-7-ene; DIPEA, diisopropylethylamine; NMP, N-methylpyrolidinone;
TFA, trifluoroacetic acid.
usually placed in organic phases or lipophilic membranes because
of the weak aqueous ion binding properties of the crown ethers.
Simple crown ethers and open-chain podands generally cannot
bind aqueous cations at clinically important concentrations (0.3-
300 mM); more complex ionophores are required. Minta and
Tsien8 reported the first fluoroionophores suitable for the meas-
urement of sodium at millimolar concentrations in aqueous
solutions. However, these indicators were designed for the
measurement of intracellular sodium concentrations (∼30 mM);
their low dissociation constants (Kd’s < 50 mM) preclude their
use for measurement of blood sodium, the extracellular concentra-
tion of which is near 140 mM. Furthermore, the fluoroionophores
require excitation with UV light, leading to practical problems
associated with variable background fluorescence from optical
components, carrier polymers, and the intrinsic fluorescence of
serum.9 Fluorescent dyes that can be excited with inexpensive
visible light sources, such as blue LEDs (light-emitting diodes)
are strongly preferred for the development of practical optical
sensing devices.10
Photo-induced electron transfer (PET)11,12-type fluoroiono-
phores have proven highly successful as direct fluorescent cation-
sensing molecules, utilizing the switchable intramolecular self-
quenching mechanism associated with PET.13-17 The general
design of a PET-type fluoroionophore is as follows: A fluorescent
moiety (fluorophore) is covalently linked to an ion receptor
(ionophore) by means of a nonconjugating spacer group. Typically,
the ionophore will contain a tertiary amine the electrons of which
can ligate the cation. In the absence of a bound cation, the HOMO
(highest occupied molecular orbital) of the unbound receptor has
a higher energy than the half-filled HOMO of the excited
fluorophore. This energy difference, among other factors, drives
rapid electron transfer from the receptor to the excited-state
fluorophore, thus quenching, or “switching off”, the fluorescence.
However, when the ionophore is bound to a cation, the energy
level of the receptor’s electron pair is lower than that of the HOMO
of the excited fluorophore; the electron transfer is not energetically
favored, and thus, fluorescence is “switched on”. Excitation of a
macroscopic ensemble of such molecules leads to an intensity-
based sensor governed by reversible ion binding.
(21) Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals,
6th ed.; Molecular Probes: Eugene, OR, 1996.
We now report a new and practical fluorescent “sensing
molecule” designed and commercially proven (within the Roche
OPTI CCA) for direct determination of sodium at extracellular
concentrations found in whole blood, serum, or plasma.18 The
respective sensor is constructed by covalent immobilization of a
new fluoroionophore (1 1 in Figure 1) onto an amino-functional-
ized hydrophilic polymer. Selection of an appropriate fluorophore
and ionophore was a key to developing a stable fluoroionophore
compatible with visible LEDs. We chose 4-amino-1,8-naphthalimide
because of its excellent thermal and photochemical stability,
desirable spectral properties (high aqueous quantum yield, LED-
compatible λ-max near 450 nm, high Stokes’ shift > 50 nm), and
(22) Alexiou, M. S.; Tychopoulos, V.; Ghorbanian, S.; Tyman, J. H. P.; Brown, R.
G.; Brittain, P. I. J. Chem. Soc., Perkin Trans. 2 1 9 9 0 , 837-842.
(23) de Silva, A. P.; Gunaratne, H. Q. N.; Habib-Jiwan, J.-L.; McCoy, C. P.; Rice,
T. E.; Soumillion, J.-P. Angew. Chem., Int. Ed. Engl. 1 9 9 5 , 34, 1728-1731.
(24) Valeur, B.; Leray, I. Coord. Chem. Rev. 2 0 0 0 , 205, 3-40.
(25) Unpublished Data.
(26) de Silva, A. P.; Gunaratne, H. Q. N.; Gunnlaugsson, T. Chem. Commun.
1 9 9 6 , 1967-1968.
(27) de Silva, A. P.; Gunaratne, H. Q. N. J. Chem. Soc. Chem. Commun. 1 9 9 0 ,
186-188.
(28) Leiner, M. J. P.; He, H.; Mortellaro, M.; Fraatz, R.; Tusa, J. K. Book of
Abstracts, 5th European Conference on Optical Chemical Sensors and
Biosensors; Lyon-Villeurbanne, France, April 16-19, 2000.
(29) Witulski, B.; Weber, V.; Bergstrasser, U.; Desvergne, J. P.; Bassani, D.;
Bouas-Laurent, H. Org. Lett. 2 0 0 1 , 3, 1467-1470.
550 Analytical Chemistry, Vol. 75, No. 3, February 1, 2003

low sensitivity to pH because of the absence of ionizable functional
groups within the physiological pH range. We chose N-(o-
methoxyphenyl)aza-15-crown-519 as a suitable ionophore because
of its sodium binding in the desired measuring range, minimal
interference from pH, and the presence of a tertiary amine that
acts as a PET fluorescence quencher. The synthesis described
below starts with the bifunctional commercially available starting
material, 4-chloro-1,8-naphthalic anhydride; its two reactive func-
tional groups allow convenient synthesis of an “ionophore-
spacer-fluorophore” type construct.15
D4 hydrogel dispersion (9.5 g) containing 10% solids in 90% w/ w
ethanol/ water. The resulting dispersion was knife-coated onto a
125-µm polyester sheet such that the indicator layer dried to a
thickness of 10 µm. A second hydrogel layer consisting of 3% w/ w
carbon black (Degussa Corporation) in the same hydrogel
dispersion was then knife-coated and allowed to dry overnight. A
sensor disk 25 mm in diameter was then punched out and soaked
in buffer for at least 16 h prior to use.
Syntheses. The synthesis of the sodium fluoroionophore 1 1
from commercially available compounds is illustrated in Figure
1.
EXPERIMENTAL SECTION
o-Anisidine (1 ) was dialkylated with 2-chloroethanol then
reacted with bis[(2-chloro-ethoxy)]ethane. The resultant phenyl-
azacrown ether (3 ) was formylated, and the aldehyde (4 ) was
condensed with nitromethane to afford â-nitrostyrene (5 ). Reduc-
tion of the nitrostyrene with lithium aluminum hydride gave the
phenethylamine derivative (6 ), which was coupled with 4-chloro-
naphthalimide (9 ) (prepared from 4-chloronaphthalic anhydride
in two steps) to form the fluoroionophore (1 0 ). Hydrolysis of the
tert-butyl ester afforded carboxylic acid derivative 1 1 suitable for
immobilization onto a solid support.
Reagents. Solvents and reagents used in the synthesis of
fluoroionophore 1 1 were purchased form Aldrich (Milwaukee,
WI) and used without further purification. Analytical grade buffer
and inorganic salts were purchased from either Fluka AG (Buchs,
Switzerland) or Sigma Co. (St. Louis, MO). The hydrophilic
hydrogel D4 was purchased from Tyndale Plains-Hunter Ltd.
(Ringoes, NJ); the polyester sheets (MELINEX), from Polymer
Supplier (Atlanta, GA).
Absorbance and Fluorescence Titrations. Absorption meas-
urements were performed with a Shimadzu UV2101PC spectro-
photometer. Titration of ionophore 3 was carried out in the
following manner: A methanolic solution of ionophore was diluted
with deionized water in a volumetric flask, the required amount
of solid sodium chloride was added, and the solution’s absorption
spectrum was measured. Fluorescence measurements were
performed with an ISS PC1 photon-counting spectrofluorometer
with a xenon lamp as excitation source. A sensor disk (25 mm,
see below) was placed into a custom-made flow-through cell fitted
into the fluorometer. N-(2-Hydroxyethyl)piperazine-N′-(ethane-
sulfonic acid) (HEPES) buffer solutions of the appropriate pH and
ion concentrations were then pumped through the cell prior to
collection of emission and excitation spectra. Buffer pH values
were measured at room temperature (∼22 °C) and 37 °C with a
Corning pH meter 125 and AVL 987 analyzer, respectively. All
NMR data were collected at room temperature by QE 300
(Nicolet/ GE), 300 MHz, with 0.1% TMS as standard. Elemental
analyses were performed by Galbraith Laboratories, Inc., Knox-
ville, TN.
Immobilization of Sodium Fluoroionophore (1 1 ) onto
Aminocellulose. Aminocellulose (5 g)20 was suspended in 50 mL
of 2.5% aqueous sodium carbonate for 30 min, filtered, resus-
pended in 50 mL DMF for 30 min, filtered, and then washed twice
more with DMF in order to replace trapped water. The washed
cellulose was then transferred into a flask containing 1 1 (0.21 g,
0.3 mmol), N,N-dicyclohexyl-1,3-carbodiimide (DCC, 0.62 g, 3
mmol), and N-hydroxysuccinimide (NHS, 0.35 g, 3 mmol) in
anhydrous DMF (20 mL), and the suspension was stirred at room
temperature for 20 h. The yellow cellulose fiber was filtered,
washed with DMF (5 × 50 mL) until the filtrate became colorless,
and then washed with water (50 mL), 0.2 N HCl (2 × 50 mL),
water (50 mL), 0.2 N NaOH (2 × 50 mL), water (10 × 50 mL),
acetone (2 × 50 mL), and ether (2 × 50 mL), after which it was
then dried at room temperature for 16 h. The resulting cellulose
powder was yellow-colored, and the exact amount of the im-
mobilized dye was not measured.
N,N-bis(2-Hydroxylethyl)-2-methoxyaniline (2). A solution
of 452 g (4 mol) of 1 in 2-chloro-ethanol (1932 g, 24 mol) and 350
mL of water was heated to 80 °C for 15 min. K2CO3 (829 g, 6
mol) was slowly added such that the temperature of this
exothermic reaction was kept below 110 °C. The mixture was
heated at 95 °C for 22 h, cooled, and ∼800 mL of unreacted
2-chloroethanol was removed under vacuum. The residue was
diluted with water (2 L) and extracted with CHCl3 (2 L). The
CHCl3 solution was dried over K2CO3, and the solvent was
evaporated to afford 802 g (98%) of brown oil. 1H NMR (300 MHz,
CDCl3) δ ) 3.18 (t, 4H, N CH2), 3.50 (t, 4H, O CH2), 3.60 (br, 2H,
O H), 3.82 (s, 3H, Ar O CH3), 6.90-7.19 (m, 4H, Ar H). Anal.
Calcd for C11H17NO3: C, 62.54; H, 8.11; N, 6.63. Found: C, 61.33;
H, 8.28; N, 6.43.
2 -Methoxyphenylaza-1 5 -crown-5 (3 ). Compound 2 (403 g,
1.91 mol) was dissolved in dioxane (2.21 L) and heated at 80 °C
for 20 min. Powdered NaOH (168 g, 4.20 mol) was added slowly
within about 3 h. The temperature was then increased to 95 °C,
bis(2-chloroethanoxyethane) (300 mL, 1.93 mol) was added in one
portion, and the mixture was kept at 95 °C for 30 h. The
suspension was then filtered hot, the solvent was evaporated, and
the residue was treated with a solution of NaClO4 (234 g, 1.91
mol) in methanol (640 mL). The mixture was stirred at 60 °C for
30 min and concentrated to ∼300 mL. Ethyl acetate (860 mL) was
added and the mixture was stirred at room temperature for 20
min, then allowed to stand at room temperature for 2 h. The
resulting precipitate was filtered, washed with ethyl acetate (2 ×
200 mL), and dried at room temperature for 30 min to give 199 g
of azacrown-sodium perchlorate complex as a soft white powder.
This powder was dissolved in a mixture of CH2Cl2 (600 mL) and
water (600 mL), the layers were separated, and the aqueous phase
was extracted with CH2Cl2 (400 mL). The organic solutions were
combined, washed with water (8 × 600 mL), dried over Na2SO4,
then evaporated to afford 100.4 g (16%) of pale yellow oil. 1H NMR
(300 MHz, CDCl3) δ ) 3.49 (t, 4H, N CH2), 3.68 (m, 16H, O CH2),
3.82 (s, 3H, Ar O CH3), 6.88-7.12 (m, 4H, Ar H). Anal. Calcd for
P reparation of Sensor Disk. Sieved (25-µm) indicator-
immobilized amino-cellulose fiber (0.5 g) was stirred 16 h into a
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003 551

C17H27NO5: C, 62.70; H, 8.36; N, 4.30. Found: C, 61.63; H, 8.44;
N, 4.26.
1H NMR (300 MHz, DMSO-D6) δ ) 5.30 (s, 2H, Ar CH2), 7.45-
8.60 (m, 9H, Ar H).
t-Butyl 4 -Chloro-1 ,8 -naphthalimidylmethyl Benzoate (9 ).
A suspension of 8 (29.2 g, 80 mmol) in DMF (320 mL) was stirred
at 40 °C under a stream of nitrogen, and then 1,1′-carbonyldi-
imidazole (52.0 g, 320 mmol) was added slowly over 20 min. The
suspension clarified and then became turbid again in 15 min. tert-
Butyl alcohol (52 mL, 1600 mmol) and 1,8-diazabicyclo[5.4.0]-
undec-7-ene (DBU, 48 mL, 320 mmol) were added, and the
mixture was heated to 60 °C for 18 h. The mixture was then cooled
and poured into ice-cold 1 N HCl (2 L) with vigorous stirring.
The resulting precipitate was filtered, washed with 1 N HCl (2 ×
300 mL), and dried in a desiccator with P2O5 for 18 h to afford
28.5 g of crude product. Purification by silica gel column chro-
matography (1:1 CHCl3/ cyclohexane) afforded 12.0 g of white
4-Formyl-2-methoxyphenylaza-15-crown-5 (4). Compound
3 (100 g, 308 mmol) was dissolved in DMF (145 mL, 1850 mmol)
in a 500-mL three-neck flask and cooled to -5 °C. POCl3 (57.4
mL, 616 mmol) was added dropwise via an addition funnel such
that the solution temperature did not exceed 5 °C. After stirring
at room temperature for 16 h, the solution was heated to 60 °C
for 1 h, cooled, and poured into 500 g of ice; the flask was rinsed
with 70 mL of water; and the combined aqueous solutions were
adjusted to pH 7 (by pH paper) with saturated K2CO3. The solution
was extracted with CHCl3 (2 × 500 mL), the CHCl3 phase washed
with water (2 × 500 mL) then dried over MgSO4 (100 g) for 1 h.
Evaporation of the solvent afforded 85 g of light yellow oil that
crystallized upon standing overnight. Recrystallization from ethyl
acetate/ hexane (1:4) afforded 56 g (51%) of light orange crystals.
1H NMR (300 MHz, CDCl3) δ ) 3.68 (t, 16H, O CH2), 3.78 (t, 4H,
N CH2), 3.82 (s, 3H, Ar O CH3), 7.05-7.28 (m, 3H, Ar H), 9.78 (s,
1H, -CHO). Anal. Calcd for C18H27NO6: C, 61.17; H, 7.70; N, 3.96.
Found: C, 61.05; H, 8.01; N, 4.04.
1
powder (36%). H NMR (300 MHz, CDCl3) δ ) 1.50 (s, 9H, t-Bu
H), 5.30 (s, 2H, Ar CH2), 7.45-8.60 (m, 9H, Ar H). Anal. Calcd
for C24H21ClNO4: C, 68.33; H, 4.78; N, 3.32. Found: C, 67.79; H,
5.01; N, 3.24.
t-Butyl 4 -{4 ′-[4 ′′-C-(aza-1 5 -crown-5 )-3 ′′-Methoxyphenyl-
ethylamino]-1′,8′-naphthalimidylmethyl} Benzoate (10). Com-
pound 9 (2.11 g, 5 mmol), 6 (2.62 g, 10 mmol), and diisopropyl-
ethylamine (1.63 g, 12.5 mmol) were suspended in N-methyl-
pyrrolidinone (12.5 mL) and heated at 100 °C for 15 h. The mixture
was poured into water (238 mL), and the precipitate was collected
by filtration, washed with water (2 × 50 mL), and dried in a
desiccator with P2O5 for 18 h to afford 3.8 g brownish-yellow solid.
This solid was purified on a silica gel column eluted with a CHCl3/
4-Nitroethylenyl-2-methoxyphenylaza-15-crown-5 (5). Com-
pound 4 (18.8 g, 50 mmol) and ammonium acetate (38.5 g, 500
mmol) were suspended in acetic acid (100 mL) and stirred at room
temperature for 10 min. Nitromethane (59.4 mL, 1100 mmol) was
added, and the mixture was heated at 60 °C for 5 h and then
poured into ice water. The resulting crystals were collected by
filtration, washed with water, then dried in a desiccator with P2O5
to afford 11.9 g (60%) dark red needles. 1H NMR (300 MHz,
CDCl3) δ ) 3.68 (t, 16H, O CH2), 3.78 (t, 4H, N CH2), 3.82 (s, 3H,
Ar O CH3), 6.88-7.08 (m, 3H, Ar H), 7.45 (d, 1H, CdC H), 7.85
(d, 1H, NO2CdC H). Anal. Calcd for C19H28N2O6: C, 57.56; H,
7.12; N, 7.07. Found: C, 56.91; H, 7.42; N, 7.12.
1
ethyl acetate gradient, affording 2.34 g (62%) of orange gum. H
NMR (300 MHz, CDCl3) δ ) 1.55 (s, 9H, t-Bu H), 3.05 (t, 2H, Ar
CH2), 3.45 (t, 2H, HN CH2), 3.50 (t, 4H, N CH2), 3.65 (t, 16H, O
CH2), 3.78 (s, 3H, Ar O CH3), 5.25 (s, 2H, Ar CH2), 6.78 (m, 3H),
7.05-8.55 (m, 12H, Ar H). FABMS (70 eV, NBA/ Li matrix): 754
(100%), (M + H); 313 (64%) (phenylazacrown + Li). Anal. Calcd
for C43H51N3O9: C, 65.51; H, 6.82; N, 5.57. Found: C, 65.82; H,
7.00; N, 5.71.
4-Aminoethyl-2-methoxyphenylaza-15-crown-5 (6). LiAlH4
(3.8 g, 100 mmol) was slowly added into THF (200 mL) in a 500-
mL three-neck flask. A solution of 5 (4.0 g, 10 mmol) in THF (50
mL) was then added dropwise within 2 h. The mixture was heated
to reflux for 4 h, cooled in an ice bath, then quenched with 6 N
KOH. The slurry was filtered, the solvent was evaporated, and
the residue was dissolved in CHCl3 (150 mL). The CHCl3 was
washed with water (2 × 150 mL), dried over K2CO3 (15 g), and
evaporated to afford 3.7 g (101%) orange oil. TLC appeared about
90% purity, and the major spot turned bluish violet when heated
with ninhydrin solution. This amine was used directly without
4 -[4 ′-[4 ′′-C-(aza-1 5 -crown-5 )-3 ′′-Methoxyphenylethyl-
amino]-1 ′,8 ′-naphthalimidylmeth-yl] Benzoic Acid (1 1 ). To
a solution of 10 (1.88 g, 2.5 mmol) in CH2Cl2 (40 mL) was added
10 mL trifluoroacetic acid. The orange solution was stirred at room
temperature for 50 min. The solution was diluted with CHCl3 (50
mL) and evaporated to dryness. The residue was dissolved in
CHCl3/ CH3OH (9/ 1, v/ v) and evaporated to dryness again. This
process was repeated twice more to remove any remaining
trifluoroacetic acid, and 1.70 g of yellow gum was isolated. This
powder was dissolved in 50 mL of CHCl3/ CH3OH (9/ 1, v/ v) and
washed with 50 mL of 5% TFA. The organic layer was dried over
K2SO4. Solvent was evaporated to give 1.75 g (98%) of yellow foam.
1H NMR (300 MHz, CDCl3) δ ) 2.90 (t, 2H, Ar CH2), 3.25 (t, 4H,
HN CH2), 3.50 (t, 16H, O CH2), 3.60 (t, 4H, N CH2), 3.75 (s, 3H,
Ar O CH3), 5.25 (s, 2H, Ar CH2), 6.78-8.75 (m, 12H, Ar H).
FABMS (70 eV, NBA/ Na matrix): 720 (60%), (M + Na); 607
(100%), (debenzylated + 2Na+ - H+). Anal. Calcd for C43H51N3O9
+ CF3COOH + 3H2O: C, 56.87; H, 5.82; N, 4.85. Found: C, 56.87;
H, 5.56; N, 4.59.
1
further purification. H NMR (300 MHz, CDCl3) δ ) 2,62 (t, 2H,
Ar CH2), 2.95 (t, 2H, H2N CH2), 3.68 (t, 16H, O CH2), 3.78 (t, 4H,
N CH2), 3.82 (s, 3H, Ar O CH3), 6.68-6.98 (m, 3H, Ar H). Anal.
Calcd for C19H32N2O5: C, 61.93; H, 8.75; N, 7.60. Found: C, 60.98;
H, 8.97; N, 6.67.
4 -Ch lor o-N -(4 -car boxyp h en ylm eth yl)-1 ,8 -n ap h th al-
imide (8 ). 4-Chloro-1,8-naphthalic anhydride (7 , 46.4 g, 200
mmol), 4-(aminomethyl)benzoic acid (30.2 g, 200 mmol), and
K2CO3 (13.8 g, 100 mmol) were suspended in DMF (2 L), stirred
at room temperature for 16 h, then at 60 °C for 6 h. The mixture
was poured into water (4 L), the solution was adjusted to pH 4
with 6 N HCl, and the resulting precipitate was collected by
filtration. Drying the solid at 60 °C for 18 h afforded 36 g (51%) of
an off-white powder. Product showed ∼90% purity by thin-layer
chromatography (TLC) and was used directly for the next step.
RESULTS AND DISCUSSION
Ionophore Selection. Selection of a suitable ionophore was
driven by several design criteria. First, the ionophore must contain
552 Analytical Chemistry, Vol. 75, No. 3, February 1, 2003

Figure 3. Excitation and emission spectra of a sensor disk exposed
to different sodium chloride solutions (left: excitation spectra, fixed
emission at 560 nm; right: emission spectra, fixed excitation at 470
nm).
Figure 2. Absorption spectra of sodium ionophore 3 in 95/5 (v/v)
water/methanol.
a tertiary nitrogen that can act as an electron donor and will also
interact with a bound sodium cation. Additionally, the ionophore’s
binding properties should be insensitive to pH changes in the
clinically important range of 6.6-7.8 so as to minimize undesirable
pH interference to the measurement of sodium. Finally, the
ionophore should preferentially bind sodium with an aqueous
complex dissociation constant (Kd) near the desired measuring
range of 100-180 mM while in the presence of typical blood
concentrations of potassium (4.5 mM) and lithium (<2 mM).
Simple crown ethers, including azacrowns, bind sodium in
aqueous solutions with dissociation constants in the molar range;
that is, ∼2 M for phenylaza-15-crown-5.8 Minta and Tsien8
recommended that an additional binding site was needed to
increase the binding strength of a simple crown ether. Thus,
bis[N,N-(o-methoxyphenyl)diaza-15-crown-5] and its correspond-
ing fluoroionophore (SBFI) were prepared on the basis of this
concept. For intracellular sodium concentrations of ∼25 mM, SBFI
with an effective Kd of 17 mM is an excellent indicator and has
been commercialized for several years.21 However, this fluoro-
ionophore saturates above 100 mM sodium and is obviously not
suitable for the measurement of extracellular sodium concentra-
tions that are near 140 mM. Conceptually, the binding strength
of the SBFI ionophore can be decreased to bring up the sodium
Kd toward the desired 140 mM by decreasing the number of ligand
atoms.
emission centered near 460 nm serves as the excitation source
while a beam splitter/ band-pass filter pair limits the detection
wavelengths to 510-670 nm. 4-Aminonaphthalimides are fluores-
cent compounds with excitation maxima typically near 460 nm,
broad emission around 540 nm, and high quantum yields (∼0.7
in ethanol).22 Additionally, their practical use as fluorophores in
PET-based systems has already been demonstrated by de Silva
and co-workers.23 Finally, a less obvious but no less important
reason for choosing 4-aminonaphthalimide is the presence of two
sites for reactivity: the imide and the 4-amino group. The imide
functionality allows for immobilization to a solid support, while
the 4-amino position allows for coupling to the ionophore. A
benzoic acid moiety linked to the imide was found suitable for
covalent immobilization to amino-containing polymers.
Spectral Response of the Sensor with Sodium Chloride
Solutions. Fluorescence emission and excitation measurements
on 25-mm sensor disks were carried out in pH 7.4, 100 mM
HEPES buffer and at room temperature. Figure 3 shows fluores-
cence emission and excitation spectra of a sensor disk exposed
to different sodium chloride solutions. The fluorescence intensity
increases substantially with increasing concentrations of sodium.
Binding of the cation to the azacrown ether inhibits fluorescence
quenching by the anisidine donor, as expected. The fluorophore
does not directly interact sterically or electronically with the
sodium cation. As a result, the excitation and emission maximums
are nearly invariant with changing sodium concentrations. This
is characteristic of a PET indicator and unlike internal charge
transfer (ICT) probes in which the excitation maximums of a
fluorophore change upon binding of an analyte.24
N-(o-Methoxyphenyl)aza-15-crown-5 (3) was reported by Gokel
and co-workers19 to bind sodium in methanol with a log Ka of 3.86
(Kd of 0.14 mM); however, no binding data in aqueous solution
was reported. To determine the binding strength in aqueous
media, compound 3 was prepared according to a convenient
literature procedure7 and titrated with sodium chloride. Figure 2
shows the absorption spectra of 3 in 95/ 5 (v/ v) water/ methanol
with different concentrations of sodium chloride. From these data,
the dissociation constant is estimated at ∼80 mM, close to the
desired 100-180 mM range.
Dynamic Response of the Sensor to Sodium and pH
Changes. Figure 4 shows the dynamic response of a sensor disk
to varying sodium ion concentrations in pH 7.4 HEPES buffer (0-
1800 s) and to changes in pH at 150 mM sodium chloride (1800-
3000 s). Stable signals are generally obtained in <1 min, and the
fluorescence intensity returns to its initial value at the end of the
experiment. The slope of the sensor, defined as percentage signal
change per millimolar change in sodium, is ∼0.5%/ mM in the
typical clinically significant range of 120 to 160 mM. This allows
the Roche OPTI CCA analyzer to provide measurement precisions
Fluorophore Selection. The choice of the fluorophore was
constrained by the established spectral properties of our whole
blood analyzer platform, the Roche OPTI CCA, for which this
sensor was developed. In this system, a blue LED with broad
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003 553

Figure 6. Response at three extremes: (A) 0 mM NaCl, pH 12;
(B) 6 M NaCl, pH 12; (C) 0 mM NaCl, pH 2. Excitation at 470 nm,
emission at 540 nm.
Figure 4. Dynamic response of a sensor disk to various sodium
chloride concentrations at pH 7.40 HEPES buffer. Excitation at 470
nm, emission at 540 nm.
response parameters throughout their 6-9 month wet-storage
period at variable temperature (2-30 °C). Although this topic is
beyond the scope of this paper, we note the sensor displays
excellent stability against hydrolysis and oxidation, leading to slope
changes of <5% after 9 months wet storage at 30 °C.25
P roposed Signal Transduction Mechanism. To better
understand the quenching mechanism, we measured the response
of the immobilized fluoroionophore at extreme concentrations of
sodium and hydrogen ions, namely, zero and 6 M sodium chloride
at pH 2 and 12 (using 0.01 M HCl and tetramethylammonium
hydroxide, respectively). The response curve shown in Figure 6
clearly shows the nearly equivalent response of the fluoroiono-
phore to proton (pH 2) and sodium (6 M at pH 12). For this
ionophore, these conditions represent three different extremes:
(A) At zero sodium and pH 12, the ionophore is not bound to a
proton or sodium and is able to effectively quench fluorescence
via PET; that is, it is “switched off”. (B) At saturated (∼6 M)
sodium and pH 12, the ionophore is saturated with sodium but
not protons, resulting in an 18-fold increase in fluorescence
attributed to sodium binding, leading to cessation of PET quench-
ing, that is, it is “switched on”. (C) At zero sodium and pH 2, the
large fluorescence increase is attributed to protonation of the
trigger nitrogen, similarly leading to cessation of PET quenching.
Surprisingly, in this experiment, the total fluorescence enhance-
ment for sodium and hydrogen ions is similar. We note that the
physical properties of the 6 M sodium solution will perturb the
fluorophore’s absorbance and quantum efficiency relative to the
0.01 N HCl solution. Thus, the similar intensities of protonated
and sodium-saturated species may be coincidental.
Figure 5. pH response curve of a sensor disk in the absence of
sodium chloride. Excitation at 470 nm, emission at 540 nm.
better than (1 mM (1 SD), necessary for clinical decision making.
As predicted, only small changes in fluorescence intensity are
observed in response to changes in pH from 7.0 to 7.8. However,
when the sample pH falls below pH 7.0, it is necessary to use a
measured pH to correct the reported sodium correction (see
below). In comparison with 140 mM of sodium, the physiological
concentrations of other cations, such as potassium (4.5 mM),
calcium (1.2 mM), and magnesium (0.5 mM), in blood or serum
are very low. Therefore, the interferences from those ions are
negligible.
pH Response Curve. Figure 5 shows the fluorescent re-
sponse of the sensor disk from pH 1 to 11 in the absence of
sodium chloride. The measured pKa near pH 5.5 is typical of an
N,N-dialkylanisidine.8 Protonation of the trigger nitrogen inhibits
the PET (self-quenching) process, which results in increased
fluorescence. The fluorescence intensity maximum between pH
3.5 and 2 indicates complete protonation of the trigger nitrogen.
The decrease in fluorescence intensity below pH 2 is accompanied
by a shift of absorption spectra (not shown), suggesting proton-
ation of the 4-aminonaphthalimide group.
Intuition might lead one to believe that the positive charge on
the bound sodium ion is shared with all of the ionophore’s
donating atoms. In this case, only one-sixth of the cation’s positive
charge lies at the trigger nitrogen; in fact, this fraction should be
less than one-sixth, because the “hard” sodium cation should
prefer oxygen to nitrogen ligands. In contrast, protonation of the
ionophore results in a full positive charge on the trigger nitrogen.
Following this logic, the total fluorescence enhancement caused
by protonation should be six times higher than that induced by
sodium binding. Yet the data in Figure 6 shows protons and
sodium ions cause similar fluorescence enhancements. A simple
explanation is schematically illustrated in Figure 7. In the absence
of sodium, the aniline nitrogen is conjugated with the phenyl ring,
Sensor Stability. The Roche OPTI CCA system employs a
single-point calibration process and requires extremely stable
sensors. The sensor disks must maintain their factory-barcoded
554 Analytical Chemistry, Vol. 75, No. 3, February 1, 2003

the closer approach of the hard metal to the carboxylate moieties,
thus decreasing the metal-nitrogen interaction. This signaling
mechanism has been proposed for other fluoroionophores.15 In
addition to these, the facts can also be explained by one electron
oxidation of the aniline nitrogen upon binding of sodium if the
nitrogen has a high enough oxidation potential.
CONCLUSION
We have described a new optical sensor suitable for measure-
ment of extracellular sodium. The optical sensor is based on a
novel PET fluoroionophore immobilized in a hydrophilic polymer
layer. The design concept of the fluoroionophore follows the
receptor-spacer-fluorophore approach to sensor design using
intramolecular PET-based signal transduction.15 Key to the de-
velopment of this sensor was the identification of a nitrogen-
containing sodium ionophore, intentionally coupled with a fluoro-
phore having the correct spectral and electron-accepting proper-
ties. On the basis of this design concept, fluoroionophores for
other cations, such as potassium, calcium, and magnesium, can
be prepared by substitution of the ionophore. Potassium and
calcium fluoroionophore based on this family approach have
already been prepared and will be published in more detail in the
future.28
Figure 7. Illustration of C-N bond rotation upon binding of sodium.
and this conjugated system acts as the PET donor (Figure 7A).
Electron transfer across the ethyl spacer effectively quenches the
4-aminonaphthalimide’s fluorescence. In the presence of sodium,
the azacrown moiety is, by necessity, orthogonal to the phenyl
ring and directed toward the o-methoxy in order to bind sodium
tightly.8 The resulting rotated conformation of the C-N bond
decouples the nitrogen lone pair from the aromatic ring and
disrupts the PET process as effectively as protonation of the
nitrogen (Figure 7B). A case employing a structurally similar
receptor offers X-ray structural evidence for the orthogonal
signaling mechanism.26 In that work, the ion-induced fluorescence
enhancements are similar for sodium and lithium and the value
for protonation is not much higher. Nevertheless, a case with a
different amino acid receptor, though with a similar electroactive
unit, shows much larger fluorescence enhancement with protons
than with calcium.27 The latter case may have a contribution from
ACKNOWLEDGMENT
The authors acknowledge the helpful discussions and col-
laboration from Prof. A. P. de Silva (Belfast) and our numerous
co-workers within AVL and Roche.
Received for review August 6, 2002. Accepted November
5, 2002.
AC0205107
Analytical Chemistry, Vol. 75, No. 3, February 1, 2003 555