Anal. Chem. 2003, 75, 6807-6812
Real-Time Determination of Picomolar Free Cu(II) in Seawater Using a Fluorescence-Based Fiber Optic Biosensor Hui-Hui Zeng,† Richard B. Thompson,*,†,‡ Badri P. Maliwal,†,‡ Gary R. Fones,§,| James W. Moffett,§ and Carol A. Fierke⊥
Department of Biochemistry and Molecular Biology, University of Maryland School of Medicine, 108 North Greene Street, Baltimore, Maryland 21201, Center for Fluorescence Spectroscopy, University of Maryland, 725 West Lombard Street, Baltimore, Maryland 21201, Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, and Departments of Chemistry and Biochemistry, University of Michigan, Ann Arbor, Michigan 48109-1055
We report real-time, in situ determination of free copper ion at picomolar levels in seawater using a fluorescencebased fiber optic biosensor. The sensor transducer is a protein molecule, site-specifically labeled with a fluorophore that is attached to the distal end of an optical fiber, which binds free Cu(II) with high affinity and selectivity. The transducer reports the metal’s concentration as a change in fluorescence intensity or lifetime, using a frequency domain approach. The transducer’s response time is diffusion-limited, with a typical measurement requiring 30 s. The sensor demonstrates a detection limit of 0.1 pM free Cu(II) in a seawater model. Accuracy and precision of the sensor were at least comparable to cathodic ligand exchange/adsorptive cathodic stripping voltammetry. Measurements of tidal flushing of a coppercontaminated inlet are shown. Determination of trace metal ions such as Cu(II) in natural waters remains an important task in marine chemistry and environmental monitoring. The importance of Cu(II) as a pollutant is well established and is a source of worldwide concern. Existing methods used for trace metal determination in seawater are sensitive, selective, and accurate. These methods include stripping voltammetry, inductively coupled plasma-mass spectrometry, graphite furnace atomic absorption spectroscopy, atomic emission spectroscopy, flow injection analysis, and ion-selective electrodes.1-3 Except for the last two methods, however, these methods require collection and processing of the sample prior to analysis, steps that often require more time and labor than the analysis itself. The labor-intensive nature of these methods and the need to * To whom correspondence should be addressed: (phone) (410) 706-7142; (fax) (410) 706-7122; (e-mail)
[email protected]. † University of Maryland School of Medicine. ‡ University of Maryland. § Woods Hole Oceanographic Institution. | Current address: School of Ocean and Earth Science, Southampton Oceanography Centre SO14 3ZH United Kingdom. ⊥ University of Michigan. (1) Bruland, K. W. Limnol. Oceanogr. 1989, 34, 269-285. (2) Bruland, K. W.; Rue, E. L.; Donat, J. R.; Skrabal, S. A.; Moffat, J. W. Anal. Chim. Acta 1999, 405, 99-113. (3) Belli, S. L.; Zirino, A. Anal. Chem. 1993, 65, 2583-2589. 10.1021/ac0345401 CCC: $25.00 Published on Web 11/13/2003
© 2003 American Chemical Society
process samples immediately to avoid deterioration of stored samples makes it difficult to make measurements frequently over periods of days without multiple operators working in shifts. Particularly on shipboard, numbers of operators may be limited. Moreover, in view of the well-known risk of contamination during the sampling process, it is desirable to avoid sampling altogether. In marine chemistry, it is often useful to be able to determine analytes such as metal ions rapidly and continuously (or quasicontinuously). Certainly the temporal variation of analyte levels is a key parameter in understanding their sources, sinks, and transport kinetics. In addition, it is typically important to be able to correlate the chemical properties of the ocean with physical properties such as temperature, depth, density, or current velocity. The slowness of retrieval of samples from any significant depth (tens of minutes for depths in the 100-m range) makes it difficult to perform temporal correlation of physical measurements with the analytical determinations for any but the slowest processes. For all these reasons, it is desirable to use a sensor: an instrument capable of frequent, real-time measurements, ideally made in situ. We (and others) have pursued the development of such sensors. Some workers have described flow injection analyzers using a stream of seawater collected from overside that provide continuous readouts, but at the current state of the art, they are limited to shallow depths in the water column unless they are self-contained. The need for rapid determination would seem to require that no separation process such as chromatography or electrophoresis take place; in short, there must be a recognition molecule or transducer that interacts with the analyte to produce a signal that can be related to its presence or concentration. For some time, many workers have been developing metal ion sensors where the transducing molecule is of biological origin and may be termed biosensors.4,5 In our case, the transducing molecules are variants of the enzyme human carbonic anhydrase II. This protein binds certain divalent cations in its active site with high affinity and specificity; in vivo Zn(II) is found there and is required for catalysis. The carbonic anhydrase binding site also binds Cu(II) (4) Thompson, R. B.; Walt, D. R. Naval Res. Rev. 1994, 46, 19-29. (5) Wolfbeis, O. S., Ed. Fiber Optic Chemical Sensors and Biosensors; CRC Press: Boca Raton, FL, 1991.
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003 6807
with high affinity (0.1 pM)6 and selectivity; for a sensor operating in media as complex as seawater, selectivity is as important as and harder to achieve than sensitivity since potential interferences such as Ca(II) and Mg(II) are present at a millionfold higher levels. In fact, carbonic anhydrase II not only exhibits excellent selectivity as the wild type but its structure may be subtly modified to improve its selectivity as well as sensitivity.7,8 Furthermore, the kinetics of metal ion binding may also be improved, which is most important in the case of zinc, for which the association rate constant can be improved 10 thousandfold.9,10 To form an effective sensor, the recognition of the metal ion by the carbonic anhydrase binding site must be transduced as a signal that may be readily measured. Fluorescence has several advantages for sensor transduction, being inherently sensitive and easy to measure. Various investigators have transduced analyte recognition as variations in fluorescence intensity, wavelength ratio, anisotropy (polarization), or lifetime. Since the relative merits of these approaches have been discussed elsewhere, it should suffice that the last three approaches are preferable because of their relative freedom from artifact and facile calibration compared with simple intensity measurements. The measurements described herein include both intensity and lifetime measurements. For determining the composition of seawater, it is particularly desirable that the measurement be made in situ, perhaps through a length of optical fiber; for such applications, changes in lifetime are particularly good, inasmuch as they are facile through a length of optical fiber.11-13 Thus, lifetime-based fiber optic sensors have shown adequate sensitivity and selectivity, demonstrating picomolar detection limits and a rapid response even in seawater model matrixes.14,15 The basis of the transduction in this case is that metal ion binding to a site-specifically fluorescent-labeled apoprotein variant results in partial quenching of the fluorescent label and a concomitant reduction of the lifetime. The protein is attached to the distal end of the fiber optic, whence its fluorescence may be readily excited and measured. MATERIALS AND METHODS The variants N67C and L198C of human carbonic anhydrase II were constructed, expressed, and purified as previously described.16 The L198C-CA was conjugated with Alexa Fluor 660 C2-maleimide (Molecular Probes, Eugene, OR; Catalog No. (6) McCall, K. A. Ph.D. Thesis, Department of Biochemistry, Duke University, Durham, NC, 2000. (7) Hunt, J. A.; Ahmed, M.; Fierke, C. A. Biochemistry 1999, 38, 9054-9060. (8) Ippolito, J. A.; Baird, T. T.; McGee, S. A.; Christianson, D. W.; Fierke, C. A. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 5017-5021. (9) Huang, C.-c.; Lesburg, C. A.; Kiefer, L. L.; Fierke, C. A.; Christianson, D. W. Biochemistry 1996, 35, 3439-3446. (10) Thompson, R. B.; Zeng, H.-H.; Loetz, M.; Fierke, C. A. In SPIE Conference on AdvancedMaterials and Optical Systems for Chemical and Biological Detection; Fallahi, M., Swanson, B. I., Eds.; Society of Photooptical Instrumentation Engineers: Boston, MA, 1999; Vol. 3858, pp 161-166. (11) Betts, T. A.; Bright, F. V.; Catena, G. C.; Huang, J.; Litwiler, K. S.; Paterniti, D. P. In Laser Techniques in Luminescence Spectroscopy, ASTM STP-1006; Vo-Dinh, T., Eastwood, D., Eds.; American Society for Testing and Materials: Philadelphia, 1990; pp 88-95. (12) Thompson, R. B.; Lakowicz, J. R. Anal. Chem. 1993, 65, 853-856. (13) Thompson, R. B. In Topics in Fluorescence Spectroscopy Vol. 2: Principles; Lakowicz, J. R., Ed.; Plenum Press: New York, 1991; Vol. 2, pp 345-365. (14) Thompson, R. B.; Ge, Z.; Patchan, M. W.; Huang, C.-c.; Fierke, C. A. Biosens. Bioelectron. 1996, 11, 557-564. (15) Thompson, R. B.; Zeng, H. H.; Loetz, M.; Fierke, C. In In-vitro Diagnostic Instrumentation; Cohn, G. E., Ed.; SPIE: San Jose, CA, 2000; Vol. 3913, pp 120-127.
6808
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
A-20343) and the N67C-CA with Oregon Green C2-iodoacetamide (Molecular Probes; Catalog No. O-6010); the conjugates were immobilized on quartz slides or the distal end of (typically) a 25-m length of commercial telecommunications gradient index multimode optical fiber cable (Optical Cable Corp. AX01-030S-C4DB; 100/140 µm with NA ) 0.29, attenuation and bandwidth distance product specified at 4.0 dB/km and 100 MHz‚km, respectively, at 850 nm) as described.17 Phase and modulation measurements of solutions or labeled proteins on quartz plates were measured in either an ISS K2 or Koala phase fluorometer essentially as previously described18 using Rose Bengal (Aldrich; Catalog No. 19,825-0) in ethanol or a coffee creamer scatterer for Oregon Green- and Alexa Fluor-labeled proteins, respectively, and multifrequency data fit using ISS′ proprietary software. Measurements of phase, modulation, and intensity at fixed frequencies were made through single optical fibers using an apparatus similar to that previously described14 depicted in Figure S-1 (Supporting Information), with modifications. For Oregon Green-labeled protein, excitation was provided by a Spectra-Physics model 2065 Ar ion laser emitting ∼100 mW at 488 nm and modulated at 92 MHz by Pockels’ cell; ∼2 mW was launched into the fiber through a 2-mm hole in the off-axis paraboloid (Janos, Townshend, VT; Catalog No. A8037-112) and a 20 × 0.4 NA ULWD objective (Nikon) with the fluorescence filtered by a potassium dichromate liquid filter.19 For sensors to be used at longer distances, longer lengths of fiber are necessary and it is preferable to use a longer wavelength excitation and emission to minimize attenuation by the fiber.13 Inasmuch as the excitation and emission of Alexa Fluor 660 peak at ∼180 nm longer wavelength than those of Oregon Green 488, use of Alexa Fluor reduces attenuation through the fiber by a factor of 50-100. For the measurements through fibers with Alexa Fluor-labeled protein, excitation was from a Melles Griot 56 DOL 537 660-nm laser diode module powered through a 06 DLD 203 power supply and directly modulated at 200 MHz. While the diode is specified to emit 29 mW at 660 nm, good modulation was achieved at outputs up to 6.6 mW, and satisfactory results were achieved with excitation of only 0.4 mW through 25 m of fiber; measured attenuation of such short lengths of fiber depends primarily on the launching conditions. The laser output was focused by a 100-cm-focal length silica singlet (Newport SPX 034) through the 2-mm hole in the off-axis paraboloid to hit the 4 × 0.1 NA microscope objective and be launched into the fiber. The gold-plated off-axis parabolic mirror was chosen to maximize reflectivity at the Alexa Fluor emission wavelengths. Emission was filtered through a colored glass filter having 50% transmission at 670 nm and a Melles Griot interference filter (03 FIL 051) to ensure that minimal scattered excitation was detected. Metal ion buffers and a seawater model with known concentrations of free metal ions were calculated using the MINEQL program (Environmental Research Software, Hallowell, ME). Fluorescence data were fit to one and two binding site models using Kaleidagraph (Synergy Software, Reading, PA). (16) Thompson, R. B.; Maliwal, B. P.; Fierke, C. A. Anal. Biochem. 1999, 267, 185-195. (17) Bhatia, S. K.; Shriver-Lake, L. C.; Prior, K. J.; Georger, J.; Calvert, J. M.; Bredehorst, R.; Ligler, F. S. Anal. Biochem. 1989, 178, 408-413. (18) Thompson, R. B.; Gratton, E. Anal. Chem. 1988, 60, 670-674. (19) Thompson, R. B.; Levine, M.; Kondracki, L. Appl. Spectrosc. 1990, 44, 117122.
Figure 1. Fluorescence intensity of apo-N67C-Oregon Green as a function of Cu(II) (O), Ni(II) (4), or Co(II) (0) concentration, with the best-fit lines indicated by the curves.
RESULTS Response of Fluorescent-Labeled Carbonic Anhydrases to free Cu(II). We have demonstrated that ions such as Cu(II) can be determined down to picomolar levels using apocarbonic anhydrases having different labels and that apocarbonic anhydrase has high selectivity.16 Oregon Green was chosen as an improved label for copper(II) determination due to its high extinction coefficient, quantum yield, and photostability. Conjugation of apoN67C with Oregon Green provides a fluorescent transducer with excellent response to free Cu(II) ion. In particular, saturation with free Cu(II) in a simple metal ion buffer system results in an 85% decrease in fluorescence intensity (Figure 1) and a commensurate decrease in lifetime from 3.2 to 0.3 ns (results not shown). We can fit these intensity data to a simple binding isotherm and obtain a conditional stability constant (KD ) 0.09 ( 0.01 pM) rather close to that obtained by other means (0.1 pM).6 As expected, Ni(II) and Co(II) substantially quench the fluorescence as well, exhibiting KD’s of 11.6 ( 0.7 and 105 ( 11 nM, respectively. These values are also close to those measured previously (15 ( 7 nM for Ni(II) and 150 ( 70 nM for Co(II)) by a spectrophotometric technique. For all these data fitting to two binding sites gave only a small improvement in χ2. These concentrations of free Co(II) and Ni(II) are well above those usually observed in seawater, and thus, we do not anticipate these ions will interfere. This is also true of Cd(II), which exhibits a KD with the wild-type protein of ∼2.3 nM6 but which does not affect the intensity or lifetime of the Oregon Green. Zn(II) binds tightly (KD ) 4 pM)7 as mentioned above, and therefore, it is a potential interference as discussed below even though it, too, responds minimally. Small changes (
