Article pubs.acs.org/ac
Rapid and in Vivo Quantification of Cellular Lipids in Chlorella vulgaris Using Near-Infrared Raman Spectrometry Tsung-Hua Lee,† Jo-Shu Chang,†,‡,§ and Hsiang-Yu Wang*,†,‡,§ †
Department of Chemical Engineering, National Cheng Kung University, Tainan, Taiwan Research Center for Energy Technology and Strategy, National Cheng Kung University, Tainan, Taiwan § University Center for Bioscience and Biotechnology, National Cheng Kung University, Tainan, Taiwan ‡
ABSTRACT: A rapid and noninvasive quantification method for cellular lipids in Chlorella vulgaris is demonstrated in this study. This method applied near-infrared Raman spectroscopy to monitor the change of signal intensities at 1440 cm−1 and 2845−3107 cm−1 along the nitrogen depletion period, and calibration curves relating signal intensity and cellular lipid abundance were established. The calibration curves show that signal intensity at 2845−3107 cm−1 and cellular lipid abundance were highly correlated. When the calibration curve was applied on the lipid quantification of two unknown samples, the differences between lipid abundances estimated by the calibration curve and measured by gas chromatography were less than 2 wt %. Carotenoids produced a strong and broad peak near 1440 cm−1, and it weakened the correlation between signal intensity and lipid abundance. The consistency of detection and effects of cellular contents and water on the Raman spectrogram of Chlorella vulgaris were also addressed. The sample pretreatment only involved centrifugation, and the time required for lipid quantification was shortened to less than 1.5 h. The rapid detection has great potential in high-throughput screening of microalgae and also provides valuable information for monitoring the quality of microalgae culture and determining parameters for the mass production of biodiesel from microalgae.
T
spectrometry. Raman spectrometry is built based on light scattering reported by Chandrasekhara Venkata Raman.13 It detects the emitted lights from light-excited molecules, and each wavenumber shift represents a distinct mode of vibration from a specific molecular structure. The quantitative analysis by Raman spectrometry was first reported in the early 1980s,14−16 and it has been combined with many statistic models to quantitatively analyze complex material17−20 and bioproducts such as saffron and fucoidan.21,22 Compared with infrared spectroscopy, water does not severely compromise signals in Raman spectrometry. Therefore, Raman spectrometry is suitable for rapid analysis of biological samples because drying is not necessary. The iodine value of microalgae cellular lipids has been successfully interpreted using Raman spectrometry on single-cell basis.23,24 However, the process requires the holding of single microalgal cell in a trap during the measurement and collecting statistically meaningful data is labor intensive and time-consuming. Cell ensemble measurements, on the other hand, provide straightforward and representative information for a microalgae culture. In this study, NIR Raman spectrometry was applied to obtain the relationship between signal intensity and average lipid abundance of a microalgae culture. Two forms of samples were
he shortage of fossil fuels and their negative impacts on the environment have driven scientists to find alternative energy sources that are renewable and environmentally-friendly. Among many options, biofuels are considered promising prospects because they are not only renewable but also low in carbon intensity. Microalgae have short growth/harvest cycles, high abundance of cellular lipids, low requirement in land area, and possible abilities in carbon fixation and hydrogen generation;1−4 therefore, they have been favorable choices in the biofuels development with the focal fuel in such processes being biodiesel.5,6 As the microalgae-based lipids are the raw materials for biodiesel synthesis, it is crucial to be able to detect and quantify the lipid content in the microalgal biomass. Conventional quantification methods for microalgae cellular lipid are typically invasive, time-consuming, demanding in instrument, or requiring additional reagents.7−10 Most methods are not able to provide real-time information for improving the cultivating parameters that significantly affect the lipid production3,11,12 or for determining the harvest timing during the large-scale cultivation. In most current studies, the accurate quantity of microalgae lipids can only be obtained after the crop is harvested; hence, the adjustment of cultivation parameters depends solely on experience and large amounts of experiments. The aim of this study was to develop a rapid, noninvasive, straightforward, and label-free quantification method for microalgae cellular lipids using near-infrared (NIR) Raman © 2013 American Chemical Society
Received: September 27, 2012 Accepted: January 18, 2013 Published: January 18, 2013 2155
dx.doi.org/10.1021/ac3028118 | Anal. Chem. 2013, 85, 2155−2160
Analytical Chemistry
Article
used: lyophilized microalgae powder and microalgae paste. The characteristic Raman shifts of microalgae lipids were identified using lyophilized microalgae powder, and selected shifts from microalgae paste were monitored along the nitrogen depletion period. Using microalgae paste as the sample shortens pretreatment and preserves cell viability. The signal intensities at 1440 cm−1 and 2845−3107 cm−1 increased with the total cellular lipid abundance. Consistency of the detection was also examined using multiplex aliquots from a sample as well as samples from three batches of microalgae. The calibration curve built with signals at 2845−3107 cm−1 was applied to predict the total lipid abundance of two microalgae samples, and the predicted values had differences less than 2 wt % from the values measured by the conventional gas chromatography method. The proposed method is undemanding in pretreatment and shortens the time for quantification to less than 1.5 h. It has great potential in high-throughput screening of microalgae and timely quality control of microalgae culture and biodiesel production.
■
EXPERIMENTAL SECTION Microalgae Culture. Chlorella vulgaris ESP-31 was cultured according to previously published strategies to increase the cellular lipid abundance.25 Briefly, the microalgae cells were cultured on Basal medium with limited amount of nitrate and aerated with 2% carbon dioxide at a flow rate of 0.2 vvm. After reaching the stationary phase, the shortage of nitrogen source rendered Chlorella vulgaris accumulating cellular lipids. In this study, the cellular lipid abundances increased from a lower value (14 wt %, based on dry weight of biomass) to a ratio as high as 65 wt %. Sample Pretreatment. Lyophilized microalgae powder was prepared by a previously reported procedure to avoid the oxidation of lipids.3 Briefly, microalgae suspension was centrifuged at 9000 rpm for 30 min to remove the cultivation medium. The remaining microalgae paste was then dried in a lyophilizer (FDU-2100, EYELA Co., Japan) at 1.33 Pa and −83 °C for 48 h. The microalgae paste was prepared with the following procedures. The amount of Chlorella vulgaris in the wet paste was calibrated using optical density at 688 nm, which was mainly contributed by chlorophyll and commonly used for the quantification of algae amounts.26,27 The microalgae suspension gathered from the culture tank was first adjusted to an optical density of around 4.5. Approximately 1 mL of the suspension was then centrifuged at 13500 rpm for 5 min, and the supernatant was removed. The collected microalgae cells were then rinsed three times with phosphate buffer by repeating the centrifugation step. For the convenience of sampling, 100 μL of phosphate buffer was used to resuspend the microalgae cells after rinsing. Quantitative Analysis of Microalgae Lipid Abundance. Lyophilized microalgae powder and microalgae paste were analyzed on glass slides coated with a thin Au film (∼200 Å, fabricated by the E-beam evaporator, ULVAC, VT1-10CE, Japan). The lyophilized microalgae powder was spread into a 2mm circle for the characterization of Raman spectrum of Chlorella vulgaris cells. A fixed amount (5 μL) of microalgae paste was dropped onto the Au surface for the lipid quantification (Figure 1). The droplet was left for drying for 60 min before Raman spectrometry. At least three aliquots were analyzed to obtain each data point. A 1064 nm laser (Bayspec, CA, USA) was used to excite the microalgae sample, and the
Figure 1. The schematics of microalgae cellular lipid quantification using NIR Raman spectrometry.
scattering signals were captured by a deep-cooled detector (Bayspec, CA, USA) and recorded by computer software (Spec 20/20, Bayspec, CA, USA). The power for the laser was 94 mW/cm2, and the integration time for signal collection was 10 s. The signal intensity of each shift was obtained by integrating the area under its corresponding peak. In addition to Raman spectrometry, the lipid abundances and compositions were also determined by gas chromatography.25,28 These values (Table 1) serve as the reference for Table 1. Composition of Extracted Lipids from Chlorella vulgaris in Figure 5 individual abundance (wt%) total lipid abundance (wt%, based on dry biomass)
C16:0
C18:0
C18:1
C18:2
C18:3
others
13.95 14.56 25.03 33.87 37.83 44.87 44.80
2.23 4.40 7.12 9.03 9.97 11.73 12.08
-a -a -a 1.49 2.01 1.90 1.88
-a 2.84 4.88 8.34 8.40 10.04 9.51
3.34 4.33 6.55 8.07 9.05 11.44 11.44
1.37 -a 1.59 1.33 2.45 2.81 2.88
7.01 2.99 4.89 5.61 5.95 6.95 7.01
a
- not detected.
calibrating the results obtained from Raman spectrometry. The lipid abundances and compositions were determined after the transesterification, i.e., they were estimated as fatty acid methyl esters (FAMEs). Raman Spectrum for Chlorella vulgaris Cellular Contents. In this test, insoluble starch, cellulose, pectin, olive oil, β-carotene, and mixtures of olive oil and β-carotene (olive oil:β-carotene = 1:2 or 5:1) were examined, and the same laser power and integration time as above-mentioned were used to obtain their Raman spectrum. Insoluble starch, cellulose, and pectin were in the form of powder and spread into a 2-mm circle for the spectroscopy. Besides polysaccharides, Chlorella vulgaris contain abundant carotenoids and lipids; therefore, 5 2156
dx.doi.org/10.1021/ac3028118 | Anal. Chem. 2013, 85, 2155−2160
Analytical Chemistry
Article
μL of β-carotene, olive oil, and their mixtures were also dropped onto the Au-coated slide respectively for examination.
■
microalgal lipids,31 while pectin and cellulose were used to examine the signal from the cell wall.32 Insoluble starch was used to verify signals from starch granules. Under the same excitation energy and integration time, olive oil and β-carotene produced more prominent signals than other substances (Figure 3), and these signals were consistent with those
Figure 2. Raman spectra of lyophilized powder and wet paste of Chlorella vulgaris containing 65 wt % cellular lipids. a: 1266 cm−1, b: 1302 cm−1, c: 1444 cm−1, d: 1660 cm−1, e: 1749 cm−1, f: 2845 cm−1, g: 3107 cm−1.
Figure 3. Raman spectrograms of cellular contents in Chlorella vulgaris. a: 1266 cm−1, b: 1302 cm−1, c: 1440 cm−1, d: 1660 cm−1, e: 1749 cm−1, f: 2845 cm−1, g: 3107 cm−1.
powder produced 7 relatively sharp peaks in the range of 1000−1800 cm−1 and a broad peak between 2845 and 3107 cm−1. These shifts were similar to those contributed by lipids (1266−1302 cm−1, 1444 cm−1, 1660 cm−1, 1749 cm−1, and 2845−3107 cm−1) and carotenoids (1005 cm−1, 1156 cm−1, and 1525 cm−1) as indicated in previous reports (Table 2).29,30
obtained from the lyophilized microalgae powder. Pectin produced weak signals at 1302 cm−1, 1440 cm−1, and 3000− 3100 cm−1 and a signal having comparable intensity with olive oil at 1750 cm−1. Therefore, 1750 cm−1 was excluded as a marker for lipid quantification. Cellulose and starch did not generate noticeable signals in the range between 1200 cm−1 and 3100 cm−1. The cellular composition of Chlorella vulgaris changes with cultivation conditions. General estimations of cellular components are given in previous reports.31−33 When the lipid abundance of Chlorella vulgaris was around 18%, the cells consisted of 50% proteins and 32% carbohydrates. When the lipid abundance was around 65%, the percentages of proteins and carbohydrates were around 16% and 19%.33 Although proteins are one of the main components, they did not contribute noticeable signals on the Raman spectra of microalgae.4,23,24,35,36 Figure 2 also shows the Raman spectrogram of Chlorella vulgaris in the form of wet paste. Using microalgae paste simplifies the pretreatment and expedites the quantification. Compared with lyophilized powder, signals at 1266 cm−1, 1302 cm−1, and 1749 cm−1 from the wet paste were unnoticeable even with 65 wt % lipid abundance. Signals at 1440 cm−1, 1660 cm−1, and 2845−3107 cm−1 were stronger and presented room for decline; however, their intensities were still much lower than those from lyophilized powder. This shows that although water does not cause changes in the characteristic Raman spectrum of Chlorella vulgaris, it reduces the strength of signals, possibly by absorbing the energy released from vibrating structures. According to the results shown in Figures 2 and 3, signals at 1440 cm−1 and 2845−3107 cm−1 were chosen as markers for quantifying total cellular lipids. Before the proposed method was applied to the quantitative analysis of cellular lipids, the detection consistency was examined by measuring multiple aliquots from the same sample (Figure 4). The
RESULTS AND DISCUSSION Lyophilized microalgae powder was used in the first set of experiments to identify the characteristic shifts of Chlorella vulgaris. As shown in Figure 2, the lyophilized microalgae
Table 2. Raman Shifts of Lipids and Carotenoids29,30 lipids
carotenoids
Raman shift (cm−1)
assignment
3008 2970 2940 2885 2850 1750 1660 1444 1300 1266 1524 1157 1009
υas(C−H) υas(C−H3) υas(C−H2) υs(C−H3) υs(C−H2) υ(CO) υ(CC) cis δ(C−H2) τip(C−H2) δip(C−H) cis υ (CC) υ (C−C) ρ (C−C)
The shifts caused by lipids represent different intermolecular vibration modes: cis C−H in plane deformation (1266 cm−1), CH2 twisting motion (1302 cm−1), CH2 scissoring deformation (1444 cm−1), cis CC stretching (1660 cm−1), CO stretching (1749 cm−1), and C−H stretching (2845− 3107 cm−1). To validate that these signals were contributed mainly by cellular lipids, several chemicals that represent main compositions of condensed contents in Chlorella vulgaris were also tested. Olive oil was used to simulate the signals from 2157
dx.doi.org/10.1021/ac3028118 | Anal. Chem. 2013, 85, 2155−2160
Analytical Chemistry
Article
Figure 4. The consistency of Raman detection for cellular lipids in Chlorella vulgaris. The picture shows multiple aliquots of microalgae wet paste.
coefficients of variation for signals at 1440 cm−1 and 2845− 3107 cm−1 were around 0.08, showing that the signal intensities were consistent among aliquots. The signal of 1660 cm−1 is proven useful in determining the degree of saturation of microalgal lipids,4,23,24 and its coefficient of variation is around 0.24. Figure 5a shows the Raman spectra of microalgae wet paste along the nitrogen depletion period, in which the lipid abundance of Chlorella vulgaris increased from 14 wt % to 45 wt %. The signal intensities at 1440 cm−1 and 2845−3107 cm−1 increased as the cellular lipids accumulated. Figure 5b shows that signal intensity at 2845−3107 cm−1 was highly correlated with total lipid abundance. The Raman shifts between 2845 cm−1 and 3107 cm−1 have been used in quantifying cellular lipids in human tissue for medical applications.34,35 The results indicate that these shifts are equally effective in quantifying cellular lipids in microalgae. To the best of our knowledge, this is the first study that successfully quantifies the total cellular lipids in microalgae using Raman spectrometry. Previous reports using Raman spectrometry for microalgae analysis are based on single-cell analysis and limited to the prediction of degree of saturation for cellular lipids 24,35,36 and the examination of cell nutrient status.37 In this study, the signal intensities at 1660 cm−1 also increased as the amount of unsaturated lipids increased (Figure 5a). However, the correlation between the ratio of signal intensities at 1660 cm−1 and 1440 cm−1 (I1660/I1440, I is the signal intensity) and the ratio of unsaturated bonds (NCC/NC−H2, N is the number of bonds) was weak (R2 = 0.22, data not shown). The weak correlation was possibly a result of the oxidation of unsaturated bonds during extraction and esterification as well as the limitation of the current setup of gas chromatography. The correlation between signal intensity at 1440 cm−1 and total lipid abundance was not as strong as that from C−H stretching (2845−3107 cm−1). This can be attributed to the interference of carotenoids and chlorophyll. Although signals from β-carotene were well separated from those of lipids in a lower concentration, its high abundance in a sample resulted in the peak broadening. This leads to the overlap of peaks centered at 1440 cm−1 and 1524 cm−1 (Figure 6) and brings difficulties in straightforward and accurate interpretation of individual peak area (signal intensity). Microalgae cells can produce many pigments having similar structures with βcarotene; therefore, it will be challenging to quantify total lipids
Figure 5. (a) The Raman spectrum of Chlorella vulgaris wet paste having different lipid abundances. a: 1440 cm−1, b: 1660 cm−1, c: 2845 cm−1, d: 3107 cm−1. (b) The relationship between total lipid abundances and intensities of Raman shifts at 1440 cm−1 and 2845−3107 cm−1 from the wet paste of Chlorella vulgaris.
Figure 6. Raman spectrograms of olive oil and β-carotene mixtures. a: 1440 cm−1, b: 1524 cm−1.
using signals at 1440 cm−1 without sophisticated statistical models. Chlorophyll also contributes a Raman shift37 around 1438 cm−1, and the quality of its spectrum can be enhanced by water when excited with light sources having wavelengths ranging from 406 to 647 nm.38 Similar effects may elevate the 2158
dx.doi.org/10.1021/ac3028118 | Anal. Chem. 2013, 85, 2155−2160
Analytical Chemistry
Article
interference of chlorophyll on the spectrum of Chlorella vulgaris wet paste. The signal intensities at 2845−3107 cm−1 for three batches of Chlorella vulgaris were monitored along nitrogen depletion period to construct the calibration curve shown in Figure 7.
construct calibration curve using samples from three batches of Chlorella vulgaris culture. The calibration curve was applied to predict the lipid abundance of two unknown samples, and the differences between predicted values and measured values (by gas chromatography) were less than 2 wt %.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Corresponding author address: Department of Chemical Engineering, National Cheng Kung University, No. 1 University Road, Tainan, Taiwan 70101. Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors would like to thank Dr. Ching-Lung Chen in the Department of Chemical Engineering of National Cheng Kung University at Taiwan for providing Chlorella vulgaris and assistance in the analysis of lipid compositions. The authors would also like to thank the National Science Council of Taiwan for funding this work via NSC 101-2311-B-006-003 and NSC 101-3113-E-006-015. This research also received funding from the Headquarters of University Advancement at the National Cheng Kung University, which is sponsored by the Ministry of Education, Taiwan, ROC.
Figure 7. The calibration curve of Raman signal intensity (2845−3107 cm−1) vs total lipid abundance in three batches of Chlorella vulgaris. The estimated value was obtained from the calibration curve, while the measured value was obtained by gas chromatography.
Using Raman spectrometry for quantitative study of contents in a sample was considered challenging because the detection is based on scattered lights. Therefore, it is more common to quantify the relative amount of unknown contents to the other known compound using the ratiometric method.24,39 In spite of changes in cellular contents during nitrogen depletion period, the relationship between total lipid amount and Raman signal intensity remained linear when the optical density of samples was kept constant. However, the quality of the gold film had considerable effects on the signal intensity (data not shown) and inconsistency of the gold film contributed to the deviation of signal intensity from the calibration curve. To ensure the accuracy of tests, a reference sample should be applied to calibrate the detection. Two samples from a Chlorella vulgaris culture undergoing lipid accumulation were subjected to the Raman spectrometry analysis, and the estimated lipid abundances by the calibration curve are noted as “estimated” in Figure 7. The lipid abundances of these two samples were also measured by the gas chromatography, and the obtained values are indicated as “measured” in Figure 7. The differences between these two values are less than 2 wt %, and this shows the great potential of the proposed method in rapid and noninvasive quantification of microalgae lipid abundance.
■
REFERENCES
(1) Chisti, Y. Biotechnol. Adv. 2007, 25, 294−306. (2) Singh, J.; Gu, S. Renewable Sustainable Energy Rev. 2010, 14, 2596−2610. (3) Widjaja, A.; Chien, C. C.; Ju, Y. H. J. Taiwan Inst. Chem. Eng. 2009, 40, 13−20. (4) Wu, H.; Volponi, J. V.; Singh, S. Biophys. J. 2010, 98, 744a−744a. (5) Chen, C. Y.; Yeh, K. L.; Aisyah, R.; Lee, D. J.; Chang, J. S. Bioresour. Technol. 2011, 102, 71−81. (6) Ho, S. H.; Chen, W. M.; Chang, J. S. Bioresour. Technol. 2010, 101, 8725−8730. (7) Chen, W.; Zhang, C.; Song, L.; Sommerfeld, M.; Hu, Q. J. Microbiol. Methods 2009, 77, 41−47. (8) Blokker, P.; Pel, R.; Akoto, L.; Brinkman, U. A. T.; Vreuls, R. J. J. J. Chromatogr., A 2002, 959, 191−201. (9) Lacaze, J.; Stobo, L. A.; Turrell, E. A.; Quilliam, M. A. J. Chromatogr., A 2007, 1145, 51−57. (10) Wawrik, B.; Harriman, B. H. J. Microbiol. Methods 2010, 80, 262−266. (11) Liang, Y. N.; Sarkany, N.; Cui, Y. Biotechnol. Lett. 2009, 31, 1043−1049. (12) Yeh, K. L.; Chang, J. S.; Chen, W. M. Eng Life Sci. 2010, 10, 201−208. (13) Raman, C.; Krishnan, K. Nature 1928, 121, 501−502. (14) Kauppinen, J. K.; Moffatt, D. J.; Mantsch, H. H.; Cameron, D. G. Appl. Spectrosc. 1981, 35, 271−276. (15) Thomas, G. J., Jr; Agard, D. A. Biophys. J. 1984, 46, 763−768. (16) Laane, J. Appl. Spectrosc. 1983, 37, 474−475. (17) Tsuchihashi, H.; Katagi, M.; Nishikawa, M.; Tatsuno, M.; Nishioka, H.; Nara, A.; Nishio, E.; Petty, C. Appl. Spectrosc. 1997, 51, 1796−1799. (18) Ryder, A. G.; O’Connor, G.; Glynn, T. J. J. Forensic Sci. 1999, 44, 1013−1019.
■
CONCLUSION This study demonstrates a rapid and in vivo quantification method based on near-infrared Raman spectroscopy for the total abundance of cellular lipids in Chlorella vulgaris. Lipids and carotenoids contributed the main signals from Chlorella vulgaris, and signals of lipids increased as the total lipid abundance increased. Raman shifts at 1440 cm−1 and 2845− 3107 cm−1 of microalgae wet paste were observed during the nitrogen depletion period of the microalgae culture, and the relationship between Raman signal intensity and total lipid abundance was investigated. The correlation between the signal at 1440 cm−1 and lipid abundance was weaker than that at 2845−3107 cm−1 due to the interference of carotenoids. Therefore, the signal at 2845−3107 cm−1 was selected to 2159
dx.doi.org/10.1021/ac3028118 | Anal. Chem. 2013, 85, 2155−2160
Analytical Chemistry
Article
(19) Sadeghi-Jorabchi, H.; Wilson, R. H.; Belton, P. S.; EdwardsWebb, J. D.; Coxon, D. T. Spectrochim. Acta, Part A 1991, 47, 1449− 1458. (20) Muehlethaler, C.; Massonnet, G.; Esseiva, P. Forensic Sci. Int. 2011, 209, 173−182. (21) Anastasaki, E. G.; Kanakis, C. D.; Pappas, C.; Maggi, L.; Zalacain, A.; Carmona, M.; Alonso, G. L.; Polissiou, M. G. J. Agric. Food Chem. 2010, 58, 6011−6017. (22) Pielesz, A.; Biniaś, W.; Paluch, J. Carbohydr. Res. 2011, 346, 1937−1944. (23) Wu, H.; Volponi, J. V.; Oliver, A. E.; Parikh, A. N.; Simmons, B. A.; Singh, S. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 3809. (24) Samek, O.; Jona, A.; Pilat, Z.; Zemanek, P.; Nedbal, L.; Triska, J.; Kotas, P.; Trtilek, M. Sensors 2010, 10, 8635−8651. (25) Yeh, K. L.; Chang, J. S. Biotechnol. J. 2011, 6, 1358−1366. (26) Tang, J.; Hoagland, K. D.; Siegfried, B. D. Environ. Toxicol. Chem. 1998, 17, 1085−1090. (27) Dolan, D. M.; Bierman, V. J., Jr.; Dipert, M. H.; Geist, R. D. J. Great Lakes Res. 1978, 4, 75−83. (28) Yeh, K. L.; Chen, C. Y.; Chang, J. S. Biochem. Eng. J. 2012, 64, 1−7. (29) Schulz, H.; Baranska, M. Vib. Spectrosc. 2007, 43, 13−25. (30) Wood, B. R.; Heraud, P.; Stojkovic, S.; Morrison, D.; Beardall, J.; McNaughton, D. Anal. Chem. 2005, 77, 4955−4961. (31) Khasanova, V.; Gusakova, S.; Taubaev, T. Chem. Nat. Compd. 1978, 14, 37−40. (32) Macfie, S. Arch. Environ. Contam. Toxicol. 2000, 39, 413−419. (33) Spoehr, H. A.; Milner, H. W. Plant Physiol. 1949, 24, 120−149. (34) Lin, K.; Cheng, D. L. P.; Huang, Z. W. Biosens. Bioelectron. 2012, 35, 213−217. (35) Mo, J. H.; Zheng, W.; Low, J. J. H.; Ng, J.; Ilancheran, A.; Huang, Z. W. Anal. Chem. 2009, 81, 8908−8915. (36) Wu, H. W.; Volponi, J. V.; Oliver, A. E.; Parikh, A. N.; Simmons, B. A.; Singh, S. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 3809−3814. (37) Heraud, P.; Beardall, J.; McNaughton, D.; Wood, B. R. FEMS Microbiol. Lett. 2007, 275, 24−30. (38) Thomas, L. L.; Kim, J. H.; Cotton, T. M. J. Am. Chem. Soc. 1990, 112, 9378−9386. (39) Zhang, D. M.; Ansar, S. M. Anal. Chem. 2010, 82, 5910−5914.
2160
dx.doi.org/10.1021/ac3028118 | Anal. Chem. 2013, 85, 2155−2160
