UV Resonance Raman Detection and Quantitation of Domoic Acid in

Feb 29, 2000 - Cultures of the phytoplankton diatom, Pseudonitzschia multiseries, have been harvested under controlled growth conditions ranging from ...
0 downloads 0 Views 69KB Size
Anal. Chem. 2000, 72, 1666-1671

UV Resonance Raman Detection and Quantitation of Domoic Acid in Phytoplankton Q. Wu,† W. H. Nelson,*,† J. M. Treubig, Jr.,† P. R. Brown,† P. Hargraves,‡ M. Kirs,‡ M. Feld,§ R. Desari,§ R. Manoharan,§ and E. B. Hanlon§

Department of Chemistry and School of Oceanography, University of Rhode Island, Kingston, Rhode Island 02881, and George Russell Harrison Spectroscopy Laboratory, Massachusetts Institue of Technology, Cambridge, Massachusetts 02138

Cultures of the phytoplankton diatom, Pseudonitzschia multiseries, have been harvested under controlled growth conditions ranging from late logarithmic to late stationary phase (17-58 days). The amount of domoic acid (DA) present in the growth media and in the homogenized cells has been determined by HPLC. Defined samples of media, homogenized cells, whole cells, and whole cells in media have been laser excited at 251 nm for the purpose of selectively exciting intense UV resonance Raman spectra from DA in the samples. Neither media nor cell component spectra from algae seriously interfere with DA spectra. The spectral cross sections for the dominant 1652-cm-1 mode of DA have been determined for 242-, 251-, and 257-nm excitation. Maximum sensitivities are achieved with 251-nm excitation because cross sections for DA are a maximum, and interference from other algal components becomes very small. DA concentrations that have been determined with 251-nm excitation by resonance Raman methods correlate closely with values determined independently with HPLC, especially at higher DA concentrations. The UV resonance Raman analysis of DA in phytoplankton algae is shown to be very sensitive and quantitative as well as rapid and nonintrusive. Amnesic shellfish poisoning (ASP), a new malady,1,2 entered the public health lexicon in 1987. Through succeeding years3-6 * To whom correspondence should be addressed: (phone) 401-874-2498; (fax) 401-874-5072; (e-mail) [email protected]. † Department of Chemistry, University of Rhode Island. ‡ School of Oceanography, University of Rhode Island. § MIT. (1) Bates, S. S.; Bird, C. J.; de Freitas, A. S.; Foxall, R.; Gilgan, M.; Hanic, L. A.; Johnson, G. R.; McCulloch, A. W.; Odense, P.; Pocklington, R.; Quilliam, M. A.; Sim, P. G.; Smith, J. C.; Subba Rao, D. V.; Todd, E. C. D.; Walter, J. A.; Wright, J. L. C. Can. J. Fish. Aquat. Sci. 1989, 46, 1203-1215. (2) Wright, J. L. C.; Boyd, R. K.; de Freitas, A. S. W.; Falk, M.; Foxall, R. A.; Jamieson, W. D.; Laycock, M. V.; McCulloch, A. W.; McInnes, A. G.; Odense, P.; Pathak, V. P.; Quilliam, M. A.; Ragan, M. A.; Sim, P. G.; Thibault, P.; Walter, J. A. Can. J. Chem. 1989, 67, 481-490. (3) Noguchi, T.; Arakawa, O. Adv. Exp. Med. Biol. 1996, 391, 521-526. (4) Wekell, J. C.; Gauglitz, E. J. Jr.; Barnett, H. J.; Hatfield, C. L.; Eklund, M. J. Shellfish Res. 1994, 13 (2), 587-593. (5) Lundholm, N.; Skov, L.; Pocklington, R.; Moestrup, O. Phycologia 1994, 33 (6), 475-478. (6) Altwein, D. M.; Foster, K.; Doose, G.; Newton, R. T. J. Shellfish Res. 1995, 14 (1), 217-222.

1666 Analytical Chemistry, Vol. 72, No. 7, April 1, 2000

many hundreds of people were made ill and a number of deaths were attributed to ASP. Survivors frequently suffered neurological damage, including memory loss. The causative agent was shown to be domoic acid (DA), a neuroexcitatory amino acid. The DA is produced by plankton diatoms7-14 in the genus Pseudonitzschia. Humans eating shellfish that have fed upon Pseudonitzschia are victim15 to ASP when DA in the shellfish approaches or exceeds 20 ppm. Concentrations as high as 100 ppm10 have been observed in shellfish, and 20 pg/cell in the diatoms cells16 themselves. Also, a number of ecologically and commercially valuable animals are affected adversely by DA as it is vectored16-18 through the food web. In recent years, the occurrence of DA in Pseudonitzschia has been demonstrated1,5,7-14 in many U.S. coastal locations and worldwide. There are a number of ways to detect DA in algae and in food. The low levels, which must be detectable, mandate prior separation of DA from many interfering substances in most currently used chemical methods.2,19-25 Such procedures are time-consum(7) Hasle, G. R.; Lange, C. B.; Syvertsen, E. E. Helgol. Meeresunters. 1996, 50, 131-175. (8) Fryxell, G. A.; Reap, M. E.; Valencic, D. L. Nova Hedwigia, Beiheft 1990, 100, 171-188. (9) Hillebrand, H.; Sommer, U. J. Plankton Res. 1996, 18 (2), 295-301. (10) Buck, K. R.; Uttal-Cooke, L.; Pilskaln, C. H.; Roelke, D. L.; Villac, M. C.; Fryxell, G. A.; Cifuentes, L.; Chavez, F. P. Mar. Ecol. Prog. Ser. 1992, 84, 293-302. (11) Garrison, D. L.; Conrad, S. M.; Eilers, P. P.; Waldron, E. M. J. Phycol. 1992, 28, 604-607. (12) Villac, M. C.; Roelke, D. L.; Chavez, F. P.; Cifuentes, L. A.; Fryxell, G. A. J. Shellfish Res. 1993, 12 (2), 457-465. (13) Martin, J. L.; Haya, K.; Burridge, L. E.; Wildish, D. J. Mar. Ecol. Prog. Ser. 1990, 67, 177-182. (14) Worms, J.; Bates, S. S.; Smith, J. C.; Cormier, P.; Legar, C.; Pauley, K. Proceedings of the Symposium on Marine Biotoxins; 1991; pp 35-42. (15) Douglas, D. J.; Kenchington, E. R.; Bird, C. J.; Pocklington, R.; Bradford, B.; Silvert, W. Can. J. Fish. Aquay. Sci., 1997, 54, 907-913. (16) Wright, J. L. C.; Quilliam, M. A Intergovernmental Oceanographic Commission of UNESCO Manuals and Guides; 1995; Vol. 33, pp 113-133. (17) Fritz, L.; Quilliam, M. A.; Wright, J. L. C.; Beale, A. M.; Work, T. M. J. Phycol. 1992, 28, 439-442. (18) Work, T. M.; Barr, B.; Beale, A. M.; Fritz, L.; Quilliam, M. A.; Wright, J. L. C. J. Zoo Wildl. Med. 1993, 24, 54-62. (19) Quilliam, M. A.; Xie, M.; Hardstaff, W. R. J. AOAC Int. 1995, 78, 543-554. (20) Subba Rao, D. V.; Quilliam, M. A.; Pocklington, R. Can. J. Fish. Aquat. Sci. 1988, 45, 2076-2079. (21) Hummert, C.; Reichelt, M.; Luckas, B. Chromatographia 1997, 45, 284288. (22) Lawrence, J. F.; Charbonneau, C. F.; Menard, C.; Quillium, M. A.; Sim, P. G. J. Chromatogr. 1989, 462, 349-356. 10.1021/ac991052d CCC: $19.00

© 2000 American Chemical Society Published on Web 02/29/2000

ing. Bioassays26-32 require either the sacrifice of animals or the use of expensive reagents. Procedures are costly in technician time and often are not available in a manner timely enough (given the short shelf life of live shellfish) to remove the potential threat to public health. Normal pollution indicators, coliform bacteria for example, tell us nothing about the presence of toxic DA as a biomarker indicating contaminated shellfish. A rapid and inexpensive method to ensure the absence of DA in shellfish and in coastal waters is badly needed to forestall outbreaks of ASP. As marine harvesting and mariculture operations proliferate, and environmental stressors alter biodiversity, the use of biomarkers becomes imperative; not only in areas of public health but as warnings of potential disruption of normal food web dynamics. Attempts33-35 have been made to identify algae from resonance Raman spectra and to determine whether toxic species can be detected in this way. Different classes of algae do show distinctly different resonance-enhanced carotenoid spectra. Principal component analysis score plots show a distinct clustering of spectral data according to algal class. But there seems to be no corresponding clustering of spectral data for Pseudonitzschia based on toxicity. The Raman carotenoid spectra35 cannot distinguish between toxic and nontoxic cells. To identify the toxic cells by Raman methods, it appears necessary to excite selectively the DA itself. It is known36 that the detection limits for DA in water approximate 1 ppm with 242- and 257-nm excited Raman spectra. However, the Raman cross sections for DA have not been calculated for those excitation wavelengths even though the necessary data had been published36 earlier. In this study, the Raman spectral cross sections are calculated for 242- and 257-nm excitation, but of greater significance, cross sections have been determined with 251-nm excitation and found to be larger. At the optimum wavelength of excitation, 251 nm, spectra have been obtained from pure algae cultures that contain known numbers of plankton cells. Enumeration of the cultures of known volume allows calculation to be made of the amount on the average of DA in each cell. The pure cultures have been divided into duplicate (23) Pocklington, R.; Milley, J. E.; Bates, S. S.; Bird, G. J.; DeFreitas, A. S. W.; Quillium, M. A. Int. J. Environ. Anal. Chem. 1990, 38, 351-368. (24) Wang, R.; Maranda, L.; Hargraves, P. E.; Shimizu, Y. Toxic Phytoplankton Blooms in the Sea. Proc. Int. Conf. Toxic Mar. Phytoplankton, 1993; pp 637641. (25) Zhao, J.-Y.; Thibault, P.; Quilliam, M. A. Electrophoresis 1997, 18, 268276. (26) Osada, M.; Mark, L. J.; Stewart, J. E. Bull. Environ. Contam. Toxicol. 1995, 54, 797-804. (27) Smith, D. S.; Kitts, D. D. J. Agric. Food Chem. 1995, 43, 367-371. (28) Lawrence, J. F.; Cleroux, C.; Truelove, J. F. J. Chromatogr., A 1994, 662, 173-177. (29) Bates, S. S.; Leger, C.; Keafer, B. A.; Anderson, D. M. Mar. Ecol. Prog. Ser. 1993, 100, 185-195. (30) Van Dolah, F. M.; Leighfield, T. A.; Haynes, B. L.; Hampson, D. R.; Ramsdell, J. S. Anal. Biochem. 1997, 245, 102-105. (31) Iverson, F.; Truelove, J.; Nera, E.; Tryphonas, L.; Campbell, J.; Lok, E. Food Chem. Toxicol. 1989, 27 (6), 377-384. (32) Truelove, J.; Mueller, R.; Pulido, O.; Iverson, F. Food Chem. Toxicol. 1996, 525-529 (33) Brahma, S. K.; Hargraves, P. E.; Howard, W. F. Jr.; Nelson, W. H. Appl. Spectrosc. 1983, 37, 55-58. (34) Dalterio, R. A.; Nelson, W. H.; Britt, D.; Sperry, J.; Purcell, F. J. Appl. Spectrosc. 1986, 40, 271-272. (35) Wu, Q.; Nelson, W. H.; Hargraves, P.; Zhang, J.; Brown, C. W.; Seelenbinder, J. A. Anal. Chem. 1998, 70, 1782-1787. (36) Yao, Y.; Nelson, W. H.; Hargraves, P.; Zhang, J. Appl. Spectrosc. 1997, 51 (6), 785-791.

sets, which allow HPLC methods to be used to determine independently the total concentration of DA in each aliquot studied by Raman methods. EXPERIMENTAL SECTION Instrumentation and Spectral Measurement. Raman spectra were obtained with a Lexel model 95 CW argon ion laser equipped with an extended cavity. Intercavity frequency doubling with a BBO crystal allowed excitation at 251 nm. A flow system was used to circulate the sample at a rate of 15 mL/min by means of a Masterflex pump in a closed loop consisting of the pump and a flat quartz sample cell with a path length of 5-8 mm. The laser beam impinged on the flat quartz cell surface at an angle of 30° to the optic axis, defined by the spectrometer, to avoid collecting reflected laser light. The power of the light at the cell surface varied between 3 and 8 mW. The Raman signal was collected and depolarized by means of a quartz lens and prism, respectively, located along the direction perpendicular to the sample surface. A 150-µm exit slit was used throughout which corresponded to a spectral slit of 8 cm-1. Collection times were 20 or 30 s. Duplicates were taken of all spectra. A SPEX 1000M 1-m, single-grating spectrometer was equipped with a 2400 grids/mm grating. This produced a dispersion of 0.4 nm/mm across a 1100B, Princeton Instruments, Inc. CCD camera which was cooled by liquid nitrogen. A solid edge filter (Barr Associates, Inc., Westford, MA) rid the spectra of most Rayleigh scattering in the region observed and provided high throughput (90-95%) for Raman lines. The incident laser beam was estimated to give a spot size of 10 × 20 to 50 × 50 µm at the sample cell. The Raman shift axis was calibrated by ethanol. Enumeration of the algae cells was accomplished by means of a Sedgwick-Rafter counting chamber. Selected algae samples were centrifuged for 2 min and resuspended in 5 mL of D2O to avoid all interference from the water peak. All the raw data were collected digitally and imported into GRAMS 386 or GRAMS 32 software (Galactic Industries, Inc., Salem, NH) for processing and display. Peak heights, areas, and baselines were determined with the aid of the GRAMS 386 or GRAMS 32 software for all spectra. Differences in concentrations calculated using peak areas differed from concentrations calculated using peak heights by ∼20%. Since the spectral bandwidths were not entirely dominated by the instrument band-pass, spectral peak areas were used in calculations of DA concentrations. Background subtraction was used each time to remove solvent Raman bands as well as spectral contributions due to the quartz cell. Background spectra were obtained, always keeping the cell in a position identical to that of the sample. Raman Cross-Section Calculations. Raman spectral cross sections were determined by direct comparison with that of a sulfate internal standard. Basically, the cross section of an analyte peak is taken as proportional to the integrated intensity of the sulfate 981-cm-1 peak. The relation between the Raman cross section and the integrated Raman peak intensities of the analyte and internal standard peaks are defined by the equation

σN ) σS(IN/IS)(CS/CN){(νO - νS)/(νO - νN)}4

(1)

The Raman cross section (σN) of a Raman band of the analyte at frequency νN is determined by comparison of its peak intensity Analytical Chemistry, Vol. 72, No. 7, April 1, 2000

1667

Table 1. Names and Origin of the Algae Clones Studied clone

species

location

MK-3 V11A CLN-1

Cosdinodiscus Thalassiosira sp. P. multiseries

Narragansett Bay, RI Vero Beach, FL New London Bay, PEI, Canada

(IN) with the peak intensity (IS) of the internal standard of known absolute Raman scattering cross section (σS) at frequency, νS. CN and CS are the molar concentrations of the analyte and standard, respectively. The laser frequency is νO. The absolute Raman cross section of the sulfate 981-cm-1 symmetrical stretching vibration has been determined previously37,38 as a function of the UV excitation wavelength. Algae Culture. Three algae clones listed in Table 1 were grown at 22 °C in a 12:12 light/dark cycle with “K + Si” medium in 500-mL volumetric flasks. Four sample types (A-D) were grown for CLN-1 clones under the following conditions. Samples A-C were initiated at the same time with the same number of cells in the same volume of nutrient medium and were grown for different times: 16, 39, and 58 days, respectively. Sample D was grown independently of samples A-C and grown for 49 days. Reagents. Domoic acid, sodium sulfate and D2O were purchased from Sigma Chemical Co. and used as received. Sodium sulfate was used in all experiments designed to measure analyte Raman cross sections. This was possible since the 981-cm-1 peak of sulfate did not overlap with any analyte peaks. HPLC Instrumentation. The HPLC system consisted of a Waters 600E system controller with a Waters 600 multisolvent delivery system. Injections were automated with a Perkin-Elmer ISS 200 autosampler. An Applied Biosystems 783 programmable multiwavelength detector was used for detection. Data acquisition was accomplished by means of a Spectra Physics SP-4270 integration unit. The column used for separation was a reversedphase Ultrasphere octadecylsilica (ODS) Beckman, 5 µm, with dimensions of 4.6 mm i.d. and 250 mm length. The separation of the algae samples was performed using an isocratic mobile phase that had a composition of 90% water and 10% acetonitrile, adjusted to a pH of 2.5 with phosphoric acid. The flow rate was adjusted to 0.8 mL/min. The detection wavelength was programmed to 242 nm. Injection volumes of 25 µL were made, unless otherwise indicated. Sample Preparation of Algae Cells for UVRR and HPLC Analysis. All algae cell samples for UVRR and HPLC analysis were prepared as follows. Original algae cultures (∼200 mL growing in a 500-mL volumetric flask) were preconcentrated by centrifugation for 10 min at 9000 rpm to 15 mL (RC-5 superspeed refrigerated centrifuge (Du Pont Instruments, Sorvall, Wilmington, DE). For UVRR direct detection, a 5-mL sample was saved. The remaining 10-mL samples were centrifuged for 10 min at 4000 rpm (IEC HN-SII centrifuge, International Equipment Co.). The 10-mL supernatant (“culture medium supernatant”) was decanted and split into two 5-mL samples, one for a UVRR test and one for an HPLC test. The remaining cells were lysed in 1-2 mL of D2O (37) Fodor, S. P. A., Copeland, R. A., Grygon, C. A., Spiro, T. G. J. Am. Chem. Soc. 1989, 111 (15), 5509-5518. (38) Dudik, J. M., Johnson, C. R., Asher, S. A. J. Chem. Phys. 1985, 82 (4), 1732-1740.

1668 Analytical Chemistry, Vol. 72, No. 7, April 1, 2000

Table 2. Original Volumes and Cell Concentrations of CLN-1 Samples of Types A-D sample no.

culture age (days)

original vol (mL)

original concn (×105 cells/mL)

A B C D

16 39 58 49

221 211 202 245

0.95 1.26 1.64 0.88

Figure 1. Structure of DA.

Figure 2. UV absorption spectrum of DA (8.03 × 10-6 M).

and then homogenized for 5 min using a Glas-Col glass homogenizer (Terre Haute, IN). After the cells were homogenized and lysed, they were brought to 10 mL (with D2O) and mixed using a vortex mixer. After mixing, the homogenized cells were centrifuged for 15 min at 4000 rpm to remove solids. This resultant suspension is called the “homogenized cell supernate”. This supernate (10 mL) was split into two 5-mL parts, one for the HPLC test and one for the UVRR test. Table 2 lists the algae culture volumes and the cell concentrations for the original cultures prior to cell concentration. RESULTS Figure 1 shows the structure of DA with its single conjugated double bond. The UV absorption spectrum of DA appears in Figure 2. The absorption maximum occurs at 242 nm. The comparisons of the spectra of toxic and nontoxic algae are shown in Figure 3 with the spectra of toxic CLN-1 placed above the others. Spectra of serially diluted DA solutions with 0.1 M Na2SO4 as the internal standard are shown in Figure 4. The Raman spectral cross sections for aqueous DA determined at 251 nm, and calculated from spectra excited earlier at 242 and 257 nm, are listed in Table 3.

Figure 3. Raman spectra of toxic and nontoxic algae excited by 251 nm. Clones in order, top to bottom: CLN-1, V11A, and MK-3.

Figure 5. Raman spectra of domoic acid (pg/cell) in Pseudonitzschia multiseries (CLN-1) from homogenized cells.

Figure 6. Raman spectrum of toxic algae cells (CLN-1) in the growth media excited by 251-nm light. Figure 4. Raman spectra of serial diluted standard domoic acid with 0.1 M sodium sulfate internal standard in D2O excited by 251 nm. Table 3. Raman Cross Section of the 1652-cm-1 Mode vs Excitation Wavelength λe(nm)

cross section (cm2/molecule sr) × 10-24

242 251 257

5.1 8.0 2.6

Figure 5 shows how spectra of CLN-1 suspensions in aqueous Na2SO4 grown from 2 to 8 weeks can vary with time. An unmodified spectrum of DA from CLN-1, taken from cells in growth media directly, with 251-nm excitation is shown in Figure 6. A typical chromatogram of CLN-1 observed by HPLC in Figure 7 shows the excellent separation after 20 min of DA from other algal components. Table 4 lists and allows a comparison of the DA concentrations calculated by means of the UVRR and the HPLC methods using identical samples from identical cultures. It allows as well the comparison of the Raman- and HPLC-derived

DA concentrations in those cases when Raman was used for determinations from suspensions of whole cells and the HPLC was used to determine the DA concentrations for corresponding sets of homogenized cells. Table 5 shows the results obtained from a separate set of cell homogenates prepared from cultures grown for the allotted times, but which show DA concentrations very different from those listed in Table 4. These data are from homogenized cells and reflect DA amounts in the cells alone. Still, even though the amounts of DA present in a sample are very different from those in Table 4, the values determined independently by Raman and HPLC agree fairly closely. Once again the concentrations of DA calculated from Raman data show systematically larger DA values. DISCUSSION An examination of Figure 1 shows the basis for the strong absorption illustrated in Figure 2. The absorption band is attributed to an allowed π f π* transition associated with the conjugated double bond. Presumably, the strong resonance Raman enhancement of the 1652-cm-1 vibrational band of DA is due to the coupling of the symmetric CdC mode, associated with the Analytical Chemistry, Vol. 72, No. 7, April 1, 2000

1669

Table 4. Comparison of Domoic Acid Concentrations Obtained from UVRR and HPLC Methods culture medium supernate (sea H2O) (µg/mL)

homogenized cell supernate (D2O) (µg/mL)

total (µg/mL)

UVRR

sample

culture age (days)

cells/mL × 106

UVRR

HPLC

UVRR

HPLC

UVRR

HPLC

cells in medium (µg/mL)

A B D C

16 39 49 58

1.40 1.77 1.44 2.21

2.2 9.0 12.7 5.0

2.5 9.2 10.9 6.2

1.9 5.9

0.56 3.4 1.58 15.4

4.1 14.9 15.0 24.7

3.6 12.6 12.5 21.6

19.8

Figure 7. Typical DA chromatogram. The large peak on the right belongs to DA. Table 5. Comparison of Domoic Acid Determinations Using HPLC and UV Resonance Raman domoic acid (pg/cell) from CLN-1 homogenized cells culture age (days)

UVRR

HPLC

16 39 49 58

1.1 2.8 1.3 7.4

0.4 1.9 1.1 7.0

conjugated double bond, with the electronic transition residing in the same functional group. If DA is to be detected with high sensitivity from intact algae, it must be detected very selectively. This implies that the multitude of other biochemical components of the algal cell must not interfere strongly. Figure 3 compares the whole cell spectra of toxic algae with spectra of comparable nontoxic algae cells all taken from suspensions of ∼2 × 106 cells/mL. The upper spectrum belongs to CLN-1, the toxic species. The strong DA signal in the CLN-1 spectrum shows very clearly that the DA is much more sensitively excited than other cell components with 251-nm excitation. Figure 6 implies this as well. Sensitive quan1670 Analytical Chemistry, Vol. 72, No. 7, April 1, 2000

19.7

cells in D2O (µg/mL)

2.3

titation is possible because there is as interference only relatively weak nucleic acid signals from normal algae cell components. Aside from the 981-cm-1 peak from the sulfate internal standard, only weak nucleic acid peaks are shown in Figure 3 by the nontoxic species even at high concentration. To obtain the cross section of the intense DA Raman mode, a serial dilution of standard DA in D2O was prepared and excited at 251 nm. Cross sections of DA with 242- and 257-nm excitation calculated from spectra published earlier33 are listed in Table 3 along with the 251-nm value. Table 3 shows that 251-nm light produces the largest value of the cross section for DA, even though the UV maximum absorption is at 242 nm. Such an outcome is not unusual. It is likely that with 242-nm excitation the sample’s self-absorption of the Raman signal is higher than with 251-nm excitation. For reasons of both sensitivity and selectivity, 251-nm light was chosen to excite directly Raman spectra of all prepared algae cultures. Quantitation via Raman is accomplished through the calculation of Raman cross sections of specific vibrational modes. Once the cross section is known, the calculation of the concentration of the analyte can be accomplished from the measurement of peak intensities alone so long as an internal standard is used. Domoic acid concentrations have be obtained through the intensity measurement of the 1652-cm-1 peak in water or of the corresponding 1647-cm-1 mode in D2O excited selectively by 251-nm light. The use of sulfate of known concentration, typically 0.1 M, as internal standard permits the direct calculation of the DA concentration through the comparison of the DA and the 981-cm -1 sulfate peak areas. Because of the very small amounts of DA weighed in the experiments, the uncertainty in the DA spectral cross section approaches 5%. The average DA concentration within the cells is calculated in a two-step process. First, a DA concentration in picograms per milliliter is calculated from the DA peak intensity. Once that has been determined, the concentration per cell in picograms per cell is calculated directly by dividing the concentration of DA by the number of cells per unit volume. The cell numbers have been determined by direct microscopic observation. A typical time-dependent set of data shown in Figure 5 and belonging to suspensions of ∼2 × 106 cells/mL indicates that, at the early stage of the culture growth, the concentration of DA is relatively low. When the culture grows older, typically a larger amount of DA is present. In practice, the correlation between cultural age and DA amount is difficult to predict. The contents of Tables 4 and 5 reflect this. Normally, the DA is associated in increasing amounts with cells in the later stages of growth, but

the amounts of DA produced can vary considerably from culture to culture. A comparison of the measurements of DA concentration determined by HPLC and by UVRR independently, listed in Tables 4 and 5, shows that the results of the two methods agree fairly well. However, there are substantial and systematic differences in the results of the two methods that need to be noted, and an attempt needs to be made to understand these differences. It is not surprising that there is some variance in the results of the HPLC and the Raman methods, especially at low DA concentrations. From whole cells, whole cell homogenate, and the supernate of whole cell homogenate, the Raman method produced systematically larger values for DA. These differences are not large. It is suggested that the lower HPLC values may reflect actual losses of DA in the HPLC experiment or, more likely, in sample preparation and transfer. The difference in values for DA determined by Raman and HPLC from culture medium supernate may in part be the result of the attenuation by internal absorption of the Raman intensities due to the medium, but more likely reflects experimental uncertainty. At the higher DA concentration values in algae, the agreement of the results of the two methods is remarkably good. The observation that the differences between HPLC and Raman values are not simply proportional to DA concentration indicates that the differences are not due to error in the calculation of the original DA cross section values alone. The larger percentage differences observed for the lower concentrations of DA are consistent with the suggestion that there are losses of DA occurring in the HPLC method associated with the multiple purification steps needed prior to separation and analysis. For example, a significant amount of the DA may remain with the cellular debris and never reach the column. Part of the difference could be explained as due to a measured cross section for the DA peak, which is a little too large. In this study, the permeability of cell walls to DA has to be noted. Experiments have been complicated because of the permeability of the cell walls to DA, which may cause loss of DA upon washing and resuspension of samples. In this work, sample transfer was done as quickly as possible to minimize the effects of cell permeability. Little is known of the Raman spectra of isomers. The HPLC data exemplified by Figure 7 indicate that concentrations of isomers, if present in our samples, are probably negligibly low. Data have been interpreted with the assumption that spectral contributions from isomers can be ignored. CONCLUSION Domoic acid is very sensitively excited by 251-nm laser light. In D2O solution, concentrations of domoic acid as low as 40 ppb have been determined directly. The sensitivity of the method is due to the very large resonance enhancement of a single symmetric A1 Raman mode of DA. Domoic acid per cell on the average can be measured very sensitively in phytoplankton if the number of algal cells is known. Algal cells not belonging to (39) Boustany, N. N. Ph.D. Dissertation, Biochemical Characterization of Mucosal Dysplasia with Ultraviolet Resonance Raman Spectroscopy. HarvardsMIT Division of Health Science and Technology, 1997.

Pseudonitzschia and not expected to produce DA have been studied as well. These have weak spectra which do not exhibit a DA peak or other strong peaks interfering with DA measurements if excitation occurs at 251 nm. All phytoplankton studied do show nucleic acid peaks that do interfere slightly, but only at DA concentrations well below those domoic acid levels known to be toxic, i.e., 0.1 pg/cell. At very low DA concentrations, the cellular nucleic acid contributions do appear and may need to be subtracted as part of the background spectra. No attempt at such a correction was attempted here. Data show that, from media, cell supernate, homogenized cells, and homogenized cell supernate, the Raman and HPLC results agree within experimental uncertainty. There is little reason to believe that whole cell spectra should be different from homogenized cell spectra. Indeed, the HPLC method assumes that the DA content must be the same. It is expected that the UV light will penetrate the cells and not be attenuated, as shown39 in earlier work with human cells. Data for samples C and D in Table 4 indicate that DA can be quantitated by UVRR not only from separated purified whole cells but also from cells in culture media. This is a remarkable finding. The potential for practical application must be recognized. That the DA can be measured directly from the cells without prior extraction or separation steps is a large advantage. The Raman method like HPLC potentially is subject to automation, and will be relatively inexpensive to apply, since potentially little labor will be involved in testing. Instrumentation is expected to be modest in cost. There appears to be little or no interference from nontoxic cells. Potential for use in remote sensing exists not only because sensitivity is high, and there is little interference from normal cellular material with 251-nm excitation, but because in the environment of blooms a significant fraction of the DA is excreted into the surrounding water. Thus, it is likely that both DA in the cells and in the water column can be detected together. Normally, in environmental samples there is little need for DA per cell values in routine measurements. Total DA per milliliter, the quantity of greatest interest, can be measured from complex systems in a single step. Results of this study show only that it is possible to detect quantitatively by resonance Raman relatively small amounts of domoic acid from algae samples that are isolated by centrifugation or present in cultures. Development of effective environmental sampling methods remains to be done. ACKNOWLEDGMENT Support for this work at URI has come from Sea Grant, the U.S. Department of Energy, and NSF through Grant NSF/DEB9400669. Use of the facilities of NIH-supported Laser Biomedical Research Center (George Russell Harrison Spectroscopy Laboratory) MIT, Grant P41-RR 02594, and NSF-supported Laser Research Facility, Grant CHE-9708265, is gratefully acknowledged.

Received for review September 10, 1999. Accepted January 3, 2000. AC991052D

Analytical Chemistry, Vol. 72, No. 7, April 1, 2000

1671