Anal. Chem. 1997, 69, 863-867
High-Performance Fiber-Optic pH Microsensors for Practical Physiological Measurements Using a Dual-Emission Sensitive Dye Antonius Song, Stephen Parus, and Raoul Kopelman*
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109-1055
A fast and durable ratiometric pH microoptode that is highly accurate, precise, sensitive, reversible, and reproducible over the physiological ranges of pH, ionic strength, and temperature has been developed. The sensing site consists of 5 (and 6)-carboxynaphthofluorescein (CNF) entrapped in a polyacrylamide gel matrix via photopolymerization at the silanized end of an optical fiber with a diameter of 2 (pulled) or 125 µm (unpulled). The optode’s precision for the pH 6.3-8.4 range in rat embryos, sera, or physiological (Earle’s and Tyrode’s) buffers was found to be better than (0.03 pH unit. The pulled and unpulled optodes have respective upper limit response times of 1 and 400 ms for 1-pH-unit change. Over a 7-week period, they retain sensitivity for 600 and 10 000 measurements, respectively. Ratiometric measurements are made using a pH-sensitive emission peak on each side of an isosbestic point. The CNF microoptode is most suitable for biological applications because of its essentially linear response over the pH 7-8 range, its high sensitivity (slope about 2), and its almost perfect correlation with a pH macroelectrode. Furthermore, errors introduced by photobleaching, leaching, quenching, optode movement, and excitation source fluctuations are minimal. Fiber-optic sensors1-10 have been developed to measure extraor intracellular pH. Important biological phenomena, including calcium regulation,11-13 chemotaxis,14-16 neuronal activity,17,18 cell (1) Zhang, S.; Tanaka, S.; Wickramasinghe, Y. A. B. D.; Rolfe, P. Med. Biol. Eng., Comput. 1995, 33, 152-156. (2) Jordan, D. M.; Walt, D. R.; Milanovich, F. P. Anal. Chem. 1987, 59, 437439. (3) Tan, W.; Shi, Z.-Y.; Kopelman, R. Sens. Actuators 1995, B28, 157-163. (4) Zhang, Z.; Seitz, W. R. Anal. Chim. Acta 1984, 160, 47-55. (5) Holobar, A.; Benes, R.; Weigl, B. H.; O’Leary P.; Raspor, P.; Wolfbeis, O. S. Anal. Methods Instrum. 1995, 2, 92-100. (6) Seitz, W. R. CRC Crit. Rev. Anal. Chem. 1988, 19, 135. (7) Dager, S. R.; Yim, J. B.; Khalil, G. E.; Artru, A. A.; Bowden, D. M.; Kenny, M. A. Neuropsychopharmacology 1995, 12, 307-313. (8) Weigl, B. H.; Holobar, A.; Trettnak, W.; Klimant, I.; Kraus, H.; Oleary, P.; Wolfbeis, O. S. J. Biotechnol. 1994, 32, 127-138.. (9) van de Merwe, S. A.; van den Berg, A. P.; van der Zee, J.; Reinhold, H. S. Int. J. Radiat. Oncol. Biol. Phys. 1990, 18, 51-57. (10) Tait, G. A.; Young, R. B.; Wilson, G. J. Am. J. Physiol. 1982, 243, H1027H1031. (11) Malayev, A.; Nelson, D. J. J. Membr. Biol. 1995, 146, 101-111. (12) Korge, P.; Campbell, K. B. Cardiovasc. Res. 1995, 29, 512-519. (13) Villalba, M.; Ferrari, D.; Bozza, A.; DelSenno, L.; DiVirgilio, F. Biochem. J. 1995, 311, 1033-1038. (14) Gow, N. A. Trends Microbiol. 1993, 1, 45-50. (15) Shapiro, B. M.; Cook, S.; Quest, A. F.; Oberdorf, J.; Wothe, D. J. Reprod. Fertil. Suppl. 1990, 42, 3-8. S0003-2700(96)00917-1 CCC: $14.00
© 1997 American Chemical Society
growth and division,19-22 and tissue oxygenation23-26 are highly dependent on pH. Nevertheless, many pH optodes are inadequate and have limited biological applicability because the pH change depends on measuring the intensity change at one emission peak. These single-excitation, single-emission pH optodes are vulnerable to intensity fluctuations in the excitation source or changes in emission collection efficiency. The former (laser power drift, fiber coupling efficiency) can be minimized or corrected for in many cases, such as back-collection. However, the smaller the optode, the less efficient is the back-collection. Changes in emission collection efficiency become most problematic when emission is collected by a secondary lens system. Then optode movement, alignment drift, and changes in sample turbidity will cause the measured signal intensity to vary even if the pH was constant. Furthermore, this pseudo pH change can also be caused by leaching or photobleaching. The latter is an inevitable problem for most dyes, arising as the number of molecules capable of emission diminishes with time. Both leaching and photobleaching may become more severe at smaller sensor sizes. Photobleaching manifests itself particularly in microoptode applications because the excitation light is directed to a small number of dye molecules that are photoexcited significantly more often in a given time span than if they were in solution or in a large bulk-type optode (>200-µm diameter). In a typical operational mode,27 the laser input is kept constant, and the signal decreases linearly with r, the optode radius. As the number of fluorophores decreases with r3 (the volume), the signal/fluorophore increases with r-2. Thus, the photobleaching may also increase with r-2 (i.e., inversely with optode cross section). In reality, this may be compensated for by shorter excitation periods.28 Due to the faster response times (t ∼ r2, using Einstein’s diffusion equation), a judicious reduction of exposure time (as r-2) will make the photobleaching rate per measurement (16) Welch, M.; Margolin, Y.; Caplan, S. R.; Eisenbach, M. Biochim. Biophys. Acta 1995, 1268, 81-87. (17) Taira, T.; Paalasmaa, P.; Voipio, J.; Kaila, K. J. Neurophysiol. 1995, 74, 643649. (18) Chesler, M.; Kaila, K. Trends Neurosci. 1992, 15, 396-402. (19) Gekle, M.; Pollock, C. A.; Silbernagl, S. J. Pharmacol. Exp. Ther. 1995, 275, 397-404. (20) Strazzabosco, M.; Poci, C.; Spirli, C.; Zsembery, A.; Granato, A.; Massimino, M. L.; Crepaldi, G. Hepatology 1995, 22, 588-597. (21) Lengheden, A.; Jansson, L. Eur. J. Oral. Sci. 1995, 103, 148-155. (22) Vicentini, L. M.; Villereal, M. L. Life Sci. 1986, 38, 2269-2276. (23) Ferrazzi, E.; Bellotti, M.; Marconi, A.; Flisi, L.; Barbera, A.; Pardi, G. Obstet. Gynecol. 1995, 85, 663-668. (24) Fiddian-Green, R. G. Br. J. Anaesth. 1995, 74, 591-606. (25) Schlichting, E.; Lyberg, T. Crit. Care Med. 1995, 23, 1703-1710. (26) Clark, C. H.; Gutierrez, G. Am. J. Crit. Care 1992, 1, 53-60. (27) Rosenzweig, Z.; Kopelman, R. Anal. Chem. 1996, 68, 1408-1413. (28) Dourado, S.; Kopelman, R. Proc. SPIE-Int. Soc. Opt. Eng., in press.
Analytical Chemistry, Vol. 69, No. 5, March 1, 1997 863
independent of probe size. Even so, the lifetime of a small sensor is affected by photobleaching since there are fewer fluorophores to excite. Our results show that it is still possible to make precise pH measurements using dual-emission wavelength ratios, even when photobleaching causes drastic reductions in emission intensity. In addition to photobleaching, dye washout or leaching from the optode is another source of pseudo pH change that usually cannot be accounted for in single-emission optodes. While it can be minimized by presoaking and by entrapment or covalent attachment of the dye,2,3,27 its effects are also minimized by the ratiometric approach. The ratiometric method thus gives the optode stability, reproducibility, and durability, which are indispensable for practical applications. In contrast to single-emission, dual-excitation optodes, we have developed a single-excitation, dual-emission optode using 5 (and 6)-carboxynaphthofluorescein (CNF). Preliminarily, CNF was chosen because its pKa of 7.6 is near that of biological fluids. CNF is a self-referencing, dual-emission dye similar to seminaphthofluorescein (SNAFL)29 and seminaphthorhodafluor (SNARF),29,30 and it has been previously characterized as a single-emission, LEDcompatible, pH-sensitive dye.31 Ratiometric pH optodes have been prepared before for use in the physiological range.30,32-34 We note that those employing dual excitations require dual data acquisition for a measurement. We demonstrate below a high-performance pH microoptode in two different sizes with fast response times, high sensitivity, good reversibility, and very good reproducibility and durability. Furthermore, both sizes operate well routinely in physiological buffers and samples, over wide ranges of pH, temperature, and ionic strength. EXPERIMENTAL SECTION Materials. 5 (and 6)-Carboxynaphthofluorescein (CNF) was purchased from Molecular Probes, Inc. Acrylamide, N,N-methylenebis[acrylamide], 3-(trimethoxysilyl)propyl methacrylate, and triethylamine were purchased from Aldrich. Tyrode’s salts was purchased from Sigma. Concentrated 10× Earle’s balanced salt solution (EBSS) was purchased from Life Technologies. All chemicals were used as received. Multimode optical fiber (o.d. 250 µm, cladding 125 µm, core 100 µm) was purchased from General Fiber Optics, Inc. Human and rat sera and gestational day (GD) 10 rat embryos removed from a pregnant rat were obtained from the University of Michigan School of Public Health. Embryos were kept alive inside an incubator (Precision Scientific 6M). Preparation of Solutions. Phosphate buffers were made from the appropriate potassium phosphates. Unless specified otherwise, they had an ionic strength (IS) of 150 mM. Tyrode’s salts solutions and EBSS were made according to their enclosed instructions. The pH values of the test solutions were measured with an electrode connected to a digital readout pH meter (Jenco 672) that was two-point calibrated with Aldrich pH standards of (29) Whitaker, J. E.; Haugland, R. P.; Prendergast, F. G. Anal. Biochem. 1991, 194, 330-344. (30) Parker, J. W.; Laksin, O.; Yu, C.; Lau, M.-L.; Klima, S.; Fisher, R.; Scott, I.; Atwater, B. W. Anal. Chem. 1993, 65, 2329-2334. (31) Wolfbeis, O. S. Mikrochim. Acta 1992, 108, 133-141. (32) Tan, W.; Shi, Z.-Y.; Smith, S.; Birnbaum, D.; Kopelman, R. Science 1992, 258, 778-781. (33) McCarthy, J. F. Ph.D. Thesis, University of Illinois, Urbana, IL, 1989. (34) Saari, L.; Seitz, W. R. Anal. Chem. 1982, 54, 821-823.
864
Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
7.00 and 10.00. The polymerization solution consisted of 0.4 mM CNF, 27% acrylamide, and 3% N,N-methylenebis[acrylamide] in 0.1 M phosphate buffer, pH 6.5. Preparation of CNF Optode. Each end of a 1 m length of the multimode optical fiber had 2 cm of buffer jacket removed by dipping in dichloromethane. For pulled optodes, a customized micropipet puller (Sutter Instrument Co. P-87) was used to taper the distal end to 2 µm. The fiber region up to 1 cm from the distal tip was enclosed in a borosilicate glass capillary tube. The distal tip was silanized for 75 min in a 2% solution of 3-(trimethoxysilyl)propyl methacrylate in pH 3.45 water and then dried in a nitrogen atmosphere. The silanized end was positioned into a small Petri dish containing 1 mL of polymerization solution and 40 µL of triethylamine on the stage of an inverted microscope (Olympus IMT-2). The proximal end was coupled to an air-cooled argon ion laser (Ion Laser Technology 5500AWC-00) exciting at 488 nm. The rate and size of polymer formation was controlled by visually monitoring the distal tip through the microscope. The fabricated sensor was air-dried and stored in air, buffer, or water. pH Data Collection. The excitation source was either the argon ion laser exciting at 488 or 514.5 nm, a diode-pumped Nd: YAG laser (ADLAS DPY305II) exciting at 532 nm, or a He/Ne laser (Spectra Physics 136) exciting at 632.8 nm. Appropriate bandpass and/or neutral-density filters were used. The laser excitation was coupled into the proximal end held in a fiber positioner (Newport Corp. F-915 Series). The distal, sensing end was held by a three-axis positioner (Newport Corp. 460A Series) on the microscope stage above the objective lens. The unpulled optode was immersed in 1 mL of buffer or serum held in an optically clear chamber. For embryos, the pulled optode was noninvasively inserted about 175 µm through the visceral yolk sac into the extraembryonic fluid. The live embryo was discarded after taking a total of seven pH measurements during a 30-min period. After collection by the objective, excitation light was removed with a dichroic mirror (488-532 nm) or a holographic notch filter (632.8 nm). The emission was directed to an imaging spectrograph (Acton Research Corp. 150) and ultimately to a charge-coupled device (CCD) detector (Princeton Instruments, Inc. LN/CCD-1024E/1). An experimental illustration is given elsewhere.35 Output Power Measurement. The laser light passing through the distal end was measured by a power meter (Newport Corp. 815) calibrated for the appropriate excitation wavelength. No correction for light absorbed by the dye was made. Photobleaching and Lifetime Tests. A pulled or unpulled optode was placed in buffer and continuously illuminated at 514.5 nm, while emission spectra were recorded every 2-10 min for photobleaching tests and every 0.5-0.75 s for lifetime tests. The tests were repeated using 632.8-nm excitation and with different optodes, buffers, and pH values. Response Time Measurements. An unpulled optode in a clear, fused quartz flow cell was positioned over the microscope objective and continuously excited at 514.5 nm. At fixed time intervals, either pH 7 or pH 8 phosphate buffer was guided through the quartz cell using a peristaltic pump (Rainin Instrument Co., Inc., rabbit model). Intensity ratios were collected with the CCD every 0.1 s using an integration time of 0.05 s. Leaching. The distal regions from six unpulled optodes were broken off and placed in a vial containing 2 mL of pH 7.4 phosphate buffer. The emission spectra of the solution using (35) Rosenzweig, Z.; Kopelman, R. Anal. Chem. 1995, 67, 2650-2654.
a
b
c
Figure 1. Emission spectra of the CNF optode in phosphate buffers using (a) 488-nm, (b) 514.5-nm, and (c) 632.8-nm excitation wavelengths. The pH values from top to bottom at 690 nm are 8.44, 7.95, 7.61, 7.38, 7.15, 6.86, and 6.31.
various excitation wavelengths were collected with a spectrofluorophotometer (Shimadzu RF-5000U) over 30 days. Ionic Strength Effects. NaCl solutions with ISs ranging from 110 to 20 mM were prepared in 100 mM IS pH 7.4 phosphate buffer. The pH values were measured with the digital readout pH meter and compared to those of a calibrated optode. Temperature Effects. The pH values of pH 7.4 phosphate buffer and rat serum were measured with the optode and electrode at 20, 25, 30, 35, and 40 °C. The temperature in the small optode testing chamber was measured with a type K digital thermometer (Cole-Parmer 8533-40). RESULTS AND DISCUSSION A polymer thickness of 2-5 µm was sufficient to produce fluorescence spectra of several hundred CCD counts from 0.5-s integrations. The pulled optode had a 2-µm hemispherical sensing site at its conical distal end. The unpulled optode had a hemiellipsoid-shaped sensing site that was less than 5 µm thick and 50 µm in diameter inside the 100-µm-wide fiber core. Photographs of the gel layer using standard illuminationabsorption, as well as dye fluorescence, showed that the layer remained 50 µm wide, independent of polymer thickness and, hence, photopolymerization time. When excited at 488, 514.5, or 532 nm, the emission spectra of CNF showed an isosbestic point near 640 nm that was independent of excitation wavelength. On each side of the isosbestic point is a peak of a fluorescent tautomer. As pH increases, the intensity at the lower wavelength acid-tautomer decreases, while the intensity at the higher wavelength basetautomer increases (Figure 1). The opposite occurs when pH decreases. Thus, the two peaks exhibit concurrent, pH-dependent intensity changes that allow pH values to be determined from an intensity ratio of two wavelengths. The intensities at 660 and 595 nm from 514.5-nm excitation resulted in both large ratio changes with pH and small standard deviations from duplicate calibrations. For the physiologically important pH range of 6.9-7.9, the measured signal change with pH was found to be quite large (slope ∼2), with an intensity ratio standard deviation of under 1.5%. A second-order polynomial calibration curve fitting of intensity ratios vs pH for the pH 6.3-8.4 range results in a correlation coefficient (r) value greater than 0.999. A linear calibration fit for the pH 7-8 range results in an r value greater than 0.99. Both types of fits were found to be reproducible for a week without significant change.
Figure 2. Photobleaching results for unpulled optodes in pH 6.52 phosphate buffer (ex ) 514.5 nm) (×), in pH 7.78 Tyrode’s salts solution (ex ) 514.5 nm) (O), and in pH 7.22 Tyrode’s salts solution (ex ) 632.8 nm) (0). Respective logarithmic fit r values are 0.9943, 0.9983, and 0.9789.
An acquisition time of 0.5 s was sufficient to record an intensity of several hundred counts on the CCD detector when 15 µW of excitation power passes through the unpulled distal tip. Hence, an individual measurement caused little photobleaching. Upon continuous excitation or many measurements, the overall emission intensity inevitably decreased with time. Intensity ratios, however, allowed pH measurements to be made despite photobleaching. For instance, the intensity of an unpulled optode diminished by 75% of its original value after 80 min of continuous excitation at 514.5 nm (Figure 2), but the maximum ratio change corresponded to only 0.18 pH unit. The biggest ratio fluctuations occurred during the first 20 min (2400 measurements) of illumination. Then there was a period of great stability over which 5000 pH measurements have been taken with a precision of better than (0.03 pH unit (Figure 3). For practical purposes, 20-25 min of preillumination stabilized the intensity ratios and eliminated errors caused by their fluctuation. The pulled optode showed less fluctuations, with an illumination lifetime of 5 min, over which 600 measurements have been taken with a precision of better than (0.03 pH unit. Overall, the photobleaching influence was very minor, as even large decreases in emission intensities resulted in only small changes in the intensity ratios, as mentioned above. When emission intensity vs time was normalized for a given excitation wavelength, the type of buffer or pH did not affect the photobleaching rate (Figure 2). The photodegradation rates using Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
865
Figure 3. Lifetime, precision, and reproducibility demonstrations of the unpulled CNF optode in pH 7.5 Earle’s balanced salt solution.
a
b
Figure 4. Response times of the unpulled CNF optode (a) going up from pH 7 to pH 8 and (b) going down from pH 8 to pH 7.
514.5-nm excitation were found to be nearly identical for different optodes in different buffers and pH. The excitation wavelength, however, caused different photobleaching rates. Excitation at 632.8 nm caused more rapid photodegradation than that at 514.5 nm (Figure 2) when the same laser power was used, presumably due to higher absorbance. However, due to the change in optical filter setups, no meaningful comparison was attempted. The response times to pH changes of the unpulled optode are about 300 ms per pH unit. Only 200 ms is required for 10-90% response changes from pH 7 to pH 8 (Figure 4a); 400 ms is required for the reverse direction (Figure 4b). These figures also demonstrate the optode’s good reversibility. The 10-90% response times of a 3-µm pulled optode with a different pH-sensitive dye36 but the same polyacrylamide gel matrix are under 1 ms for 866 Analytical Chemistry, Vol. 69, No. 5, March 1, 1997
Figure 5. Comparison of ionic strength (IS) effects on the pH electrode (×) and CNF optode (O). The optode is calibrated (a) with 100 mM IS pH 7.4 buffer and (b, inset) over the physiologically relevant IS range. Here the line represents a linear fit of the optode points in this region and should be compared with the electrode data points (×).
1 pH unit change. These response times show that the two different-sized optodes are roughly related by r2, presumably due to lateral diffusion;28 i.e., (r ) 25 µm)2/(r ) 1.5 µm)2 ≈ 280, compared to 300 ms/1 ms. In addition to the short response times, leaching or dye washout is less than expected. Leaching is expected since the dye molecules are physically entrapped in and not covalently linked to the porous polymer. To determine if leaching occurred, the fluorescence emission intensity from a solution containing unpulled distal tips was measured with a spectrofluorophotometer using 488-, 514.5-, 598-, and 632.8-nm excitation wavelengths. The fluorometer was set to high sensitivity and to the maximum excitation and emission bandwidths. To prevent unwanted photoexcitation from room lights, the solution was kept in the dark as much as possible between measurements. As expected from the absorption spectra of CNF, the highest fluorescence intensity was seen when the sample was excited at 598 nm, followed by 632.8, 488, and 514.5 nm. Virtually all leaching was discovered to occur between the sixth and 14th days. The maximum emission intensity derived from the leaching of an optode corresponded to about 1.5 × 10-19 mol. And, since an unpulled optode was originally determined to contain 1.4 × 10-15 mol, the upper leaching limit is an insignificant 0.011%. Even if more leaching occurred, its influence would still be negligible with the ratiometric technique. In addition, the CNF optode can be stored in air, minimizing leaching effects. For the ionic strength (IS) experiments, a set of NaCl solutions with ISs ranging from 110 to 200 mM in 10 mM IS increments were prepared in 100 mM IS pH 7.4 phosphate buffer. When the CNF optode and pH electrode are calibrated at 100 mM IS, they correlate very well up to 200 mM IS (Figure 5a). A linear fit of the optode values in the physiological IS region (130-170 mM) resulted in even closer correlation with the electrode (Figure 5b). Experimentally, calibration of the optode in EBSS before measuring sera and embryos produced very satisfactory results (Table 1). Our results confirmed that the extraembryonic fluid pH of rat embryos undergoes gradual acidification from GD10 to (36) Shi, Z.-Y.; Kopelman, R. Unpublished work, University of Michigan, 1994.
Table 1. Summary of Optode and Electrode pH Values of Sera and Rat Embryos
rat serum human serum GD10 rat embryo GD11 rat embryo GD12 rat embryo
optode
electrode
7.64 ( 0.027 (n ) 9) 7.75 ( 0.029 (n ) 9) 7.51 ( 0.035 (n ) 7) 7.40 ( 0.026 (n ) 7) 7.31 ( 0.021 (n ) 7)
7.65 ( 0.025 (n ) 9) 7.77 ( 0.024 (n ) 9) n/a n/a n/a
GD12.37,38 All results show that the CNF optode correlates very well with the pH electrode within a reasonably wide ionic strength range in both buffers and biological samples. Temperature changes affected equally the optode and electrode when rat serum or phosphate buffer was measured at 20, 25, 30, 35, and 40 °C. According to the optode measurements, the serum pH changed 0.04 pH unit from 20 to 40 °C, compared to 0.2 pH unit for phosphate buffer over the same temperature range. Since the electrode also gave the same pH changes and values, there is no apparent temperature effect on the optode. The observed pH changes with temperature must be inherent within the serum or buffer. Overall, the CNF microsensors are among the highest performance pH optodes, and they can be used for a variety of routine pH measurements. Our success rates (for both the production and use) of pulled and unpulled CNF optodes are over 75% and (37) Collins, M. D.; Duggan, C. A.; Schreiner, C. M.; Scott, W. J., Jr. Am. J. Physiol. 1989, 257, R542-R549. (38) Thorsrud, B. A. Ph.D. Thesis, University of Michigan, Ann Arbor, MI, 1995.
95%, respectively. A major reason behind the high success rates is the durability of the optodes. For instance, the 2-µm pulled optode can be pushed a distance of 80-100 µm against an aluminum block without breakage; it simply bends. When pulled back, it straightens out and shows no loss of performance. In conclusion, the CNF dual-emission optode correlates with a pH electrode extremely well over the physiological regions of pH, ionic strength, and temperature. The unpulled optode (125 µm) with a 5-µm sensing site manifests an extraordinary combination of precision (
