Rapid Screening of Fatty Acids Using Nanostructure-Initiator Mass

Anal. Chem. 2010, 82, 3751–3755

Rapid Screening of Fatty Acids Using Nanostructure-Initiator Mass Spectrometry Wolfgang Reindl and Trent R. Northen* Department of Bioenergy/GTL & Structural Biology, Life Sciences Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720 We present the application of nanostructure-initiator mass spectrometry (NIMS) as a fast and simple method for the analysis of plant and microbial fatty acids. NIMS allowed the direct detection of a broad range of saturated and unsaturated fatty acids in negative mode showing linearity over several orders of magnitude with a limit of detection at the femtomole level. Additionally, the fatty acid compositions of olive and soybean oil and the algal species Chlamydomonas reinhardtii could be determined both qualitatively and quantitatively with only minimal sample volumes and preparation steps. The unique properties of the NIMS surface allowed for an in situ sample cleanup step leading to a more than 10-fold increase of the signalto-noise ratio. Our data provide the basis for rapid screening of plant and microbial oils and may aid in the development of biodiesel fuels. Over the last years, biodiesel fuels based on animal fats and plant oils have gained increasing attention due to their relative compatibility with existing vehicles.1 However, this approach is limited by the relatively slow growth, the low yield of plant oil, and direct impact on food production.2 Therefore, there is growing interest in microbial production of biodiesel.2,3 Two major approaches are (1) algae-based approaches utilizing wastewater and nonarable land4,5 and (2) microbial engineering approaches to yield cellulosic biofuels from plant waste products.6,7 Both are being developed to produce triacylglycerols (TAGs), fatty acid methyl esters (FAMEs), or fatty acid ethyl esters (FAEEs), etc.4-7 As the resulting fuel performance is highly dependent on the properties and the composition of fatty acids,8 it is critical to screen for clones producing the optimal fatty acids5,7,9,10 with a through* To whom correspondence should be addressed. E-mail: trnorthen@ lbl.gov. Phone: +1 (510) 468-5240. Fax: +1 (510) 468-4545. (1) Vasudevan, P. T.; Briggs, M. J. Ind. Microbiol. Biotechnol. 2008, 35, 421– 430. (2) Rottig, A.; Wenning, L.; Broker, D.; Steinbuchel, A. Appl. Microbiol. Biotechnol. 2010, 85, 1713–1733. (3) Li, Q.; Du, W.; Liu, D. Appl. Microbiol. Biotechnol. 2008, 80, 749–756. (4) Chisti, Y. Trends Biotechnol. 2008, 26, 126–131. (5) Gouveia, L.; Oliveira, A. C. J. Ind. Microbiol. Biotechnol. 2009, 36, 269– 274. (6) Kalscheuer, R.; Stolting, T.; Steinbuchel, A. Microbiology 2006, 152, 2529– 2536. (7) Steen, E. J.; Kang, Y.; Bokinsky, G.; Hu, Z.; Schirmer, A.; McClure, A.; Del Cardayre, S. B.; Keasling, J. D. Nature 2010, 463, 559–562. (8) Knothe, G. Fuel Process. Technol. 2005, 86, 1059–1070. (9) Li, Y.; Han, D.; Hu, G.; Dauvillee, D.; Sommerfeld, M.; Ball, S.; Hu, Q. Metab. Eng. 2010, in press. (10) Xin, L.; Hong-Ying, H.; Jia, Y. New Biotechnol. 2010, 27, 59–63. 10.1021/ac100159y  2010 American Chemical Society Published on Web 03/31/2010

put comparable to current metabolic engineering approaches (i.e., multiwell plate formats).11 This requires rapid and high sensitivity analytical methods to characterize the fatty acid distribution. Current screening approaches typically rely on fluorescence assays, for example the Nile red dye which changes fluorescence in a lipid environment.12 While this is well suited for screening high lipid producing strains, it does not provide a detailed description of the lipid composition. Conventionally, the composition is determined using gas or liquid chromatographies, i.e., gas chromatography or high performance liquid chromatography (HPLC), coupled to mass spectrometry (GC/MS and LC/MS).13-15 These techniques increase the specificity, yet the throughput is limited by the chromatography step, though this time can be reduced by directly infusing the sample into the mass spectrometer at the cost of a sensitivity loss due to increased signal suppression. Surface-based mass spectrometry approaches provide information on lipid composition and do not include the time-consuming chromatographic or solution pumping steps and, therefore, allow for a higher throughput. Though, like for direct infusion methods, signal suppression is a major challenge. The most common surface-based technique is matrix assisted laser desorption ionization (MALDI) where the analyte is mixed with a large excess of matrix, a light absorbing compound that ionizes and volatilizes the analyte when irradiated. Since the matrix also ionizes, small molecule analysis using this approach has generally been complicated by the presence of abundant matrix background ions. However, this limitation has been overcome by a variety of different approaches, e.g., using new nonionizing matrix molecules,16 solvent-free MALDI mass spectrometry,17 or chargeremote fragmentation.18 Nanostructure-initiator mass spectrometry (NIMS) is a new matrix-free technique which combines soft ionization and high sensitivity and is ideally suited for rapid analysis with a measure(11) Beer, L. L.; Boyd, E. S.; Peters, J. W.; Posewitz, M. C. Curr. Opin. Biotechnol. 2009, 20, 264–271. (12) Chen, W.; Zhang, C.; Song, L.; Sommerfeld, M.; Hu, Q. J. Microbiol. Methods 2009, 77, 41–47. (13) Hejazi, L.; Ebrahimi, D.; Guilhaus, M.; Hibbert, D. B. Anal. Chem. 2009, 81, 1450–1458. (14) Sanchez-Avila, N.; Mata-Granados, J. M.; Ruiz-Jimenez, J.; Luque de Castro, M. D. J. Chromatogr., A 2009, 1216, 6864–6872. (15) Chen, S.-H.; Chuang, Y.-j. Anal. Chim. Acta 2002, 465, 145–155. (16) Shroff, R.; Svatos, A. Anal. Chem. 2009, 81, 7954–7959. (17) Saraiva, S. A.; Cabral, E. C.; Eberlin, M. N.; Catharino, R. R. J. Agric. Food Chem. 2009, 57, 4030–4034. (18) Trimpin, S.; Clemmer, D. E.; McEwen, C. N. J. Am. Soc. Mass Spectrom. 2007, 18, 1967–1972.

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3751

ment time of seconds per sample.19 Another advantage is that the NIMS surface can act like a reverse phase to selectively adsorb compounds of interest, which allows for greatly increased sensitivity by reducing signal suppression.20 NIMS analysis has exclusively been used in positive ion mode, and it was not clear if it could effectively form negative ions until it was recently shown that NIMS generates negatively charged nucleotides.21 Here, we demonstrate the application of NIMS for the detection of fatty acids in negative mode. Using NIMS, fatty acids were efficiently adsorbed from solution, desorbed, and ionized, allowing for a rapid fatty acid analysis of plant oils and crude algal extracts. An in situ sample cleanup step prior to NIMS analysis, taking advantage of the interactions between the hydrophobic carbon chain of fatty acids and the NIMS surface, was used to enhance the signal-to-noise ratio. EXPERIMENTAL SECTION Materials. Lauric acid (C12:0), myristic acid (C14:0), palmitic acid (C16:0), and stearic acid (C18:0) were all part of the “Fatty Acids, Saturated, Even Carbon Kit” from Supelco (Bellefonte, PA). Oleic acid (C18:1), linoleic acid (C18:2), and linolenic acid (C18: 3) were purchased from Sigma-Aldrich (St. Louis, MO). Olive oil and soybean oil were commercially available cooking oils. A Chlamydomonas reinhardtii 4A+ culture22 was kindly provided by Prof. Krishna Niyogi (Department of Plant and Microbial Biology, University of California, Berkeley, CA) and grown in TAP medium23 at 33 °C. Sample Preparation. For the analysis of the fatty acid standards, 10 mM stocks of all acids were prepared in methanol. The final samples were obtained by mixing and diluting the stocks to the desired concentrations in 25 mM ammonium hydroxide in 50% methanol (v/v). The plant oil samples were prepared by adding 1 µL of undiluted oil to 100 µL of 1 M ammonium hydroxide in 80% acetone (v/v) and vortexing for 1 min. Chlamydomonas cells were extracted by resuspending the pellet of a 40 mL culture in 1 mL of isopropanol followed by sonication, addition of ammonium hydroxide to a final concentration of 1 M, and incubation in an ultrasound bath at 50 °C for 30 s. The resulting supernatant of the subsequent centrifugation was dried down and resuspended in 100 µL of methanol. Ten µL of extract was added to 90 µL of an ammonium hydroxide/methanol solution with a final concentration of 25 mM ammonium hydroxide in 50% methanol (v/v). Fabrication of NIMS chips. NIMS has been described in extensive detail elsewhere.19,24 In brief, a 4 in. silicon wafer (singlesided polished P/Boron, orientation , resistivity 0.01-0.02 Ωcm, thickness 525 ± 25 µm) obtained from Silicon Quest (19) Northen, T. R.; Yanes, O.; Northen, M. T.; Marrinucci, D.; Uritboonthai, W.; Apon, J.; Golledge, S. L.; Nordstrom, A.; Siuzdak, G. Nature 2007, 449, 1033–1036. (20) Go, E. P.; Uritboonthai, W.; Apon, J. V.; Trauger, S. A.; Nordstrom, A.; O’Maille, G.; Brittain, S. M.; Peters, E. C.; Siuzdak, G. J. Proteome Res. 2007, 6, 1492–1499. (21) Amantonico, A.; Flamigni, L.; Glaus, R.; Zenobi, R. Metabolomics 2009, 5, 346–353. (22) Dent, R. M.; Haglund, C. M.; Chin, B. L.; Kobayashi, M. C.; Niyogi, K. K. Plant Physiol. 2005, 137, 545–556. (23) Gorman, D. S.; Levine, R. P. Proc. Natl. Acad. Sci. U.S.A. 1965, 54, 1665– 1669. (24) Woo, H. K.; Northen, T. R.; Yanes, O.; Siuzdak, G. Nat. Protoc. 2008, 3, 1341–1349.

3752

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

International (Santa Clara, CA) was cut into a 70 × 70 mm square, washed with methanol, and etched with 25% hydrofluoric acid in ethanol in a custom-made Teflon etching chamber under constant current of 2.4 A for 15 min. After etching, chips were coated with 400 µL of initiator (bis(heptadecafluoro-1,1,2,2-tetrahydrodecyl)tetramethyl-disiloxane). Regarding the production of the NIMS chips, we would like to note that methanol cleaned silicon wafers are better suited for measurements in negative mode than the standard Piranha (sulfuric acid/hydrogen peroxide) cleaned wafers24 presumably due to improved anion formation by less acidic surfaces. Nanostructure-Initiator Mass Spectrometry. In each case, 1 µL of a solution containing the fatty acid standard(s), plant oil, or Chlamydomonas extract was spotted onto the NIMS surface and air-dried or removed after a short incubation. Chips were loaded using a modified standard MALDI plate. NIMS was performed on a Voyager RP BioSpectrometry Workstation MALDITOF mass spectrometer from Applied Biosystems (Foster City, CA) in negative reflector mode. RESULTS AND DISCUSSION The general process of fatty acid detection using NIMS is depicted in Figure 1 and builds off of the work of Go et al.20 NIMS analysis is performed using a perfluorinated NIMS surface: the hydrophobic surface results in a large interfacial energy when in contact with polar extracts which drives amphipathic metabolites, such as fatty acids, to the surface. In case of a standard or pure sample, fatty acids “bind” to the NIMS surface and can be analyzed (Figure 1A) resulting in spectra with a high signal-to-noise ratio. However, the situation is different for a complex sample mixture. If a complex sample is spotted onto the NIMS surface and allowed to dry, obviously all sample components remain on the surface (Figure 1B, top), causing strong signal suppression from salts, etc. However, suppression can greatly be reduced by removing the sample before it dries (Figure 1B, bottom), leaving only the fatty acids on the surface. Initially, NIMS was applied for the detection of fatty acids using mixtures of various saturated (lauric, myristic, palmitic, stearic acid; Figure 2A left) and unsaturated (linolenic, linoleic, oleic acid; Figure 2A right) fatty acid standards. Indeed, signals for all used acids could be detected at the expected m/z values for the deprotonated acids (Figure 2A). Despite using equimolar fatty acid mixtures, the acquired signal intensities for each acid varied with chain length or degree of saturation. It seems that these fatty acid properties critically influence the molecules’ desorption/ionization. Therefore, like for other approaches, it is necessary to use internal standards for absolute quantification. Palmitic acid gave the most intense signal of the saturated fatty acids tested (Figure 2A). A pure palmitic acid standard was used to determine linearity and limit of detection for the used NIMS approach. The signal intensities for the palmitic acid standard showed good linearity over more than 3 orders of magnitude in the low picomole range with an R2 value of 0.979 (Figure 2B). The limit of detection was determined at around 100 fmol (Figure 2C). In order to validate our method with a biological sample, we used NIMS to analyze the fatty acid composition of plant oils. These are primarily triacylglycerols and must be converted to fatty acids before analysis. A rapid sample preparation approach was used to saponify plant oils for subsequent NIMS analysis. To this end, 1 µL of olive

Figure 1. Scheme depicting the process of fatty acid detection using nanostructure-initiator mass spectrometry. (A) Analysis of a relatively pure sample. Sample is spotted and dried. (B) Analysis of a complex mixture. Top: Sample is spotted and dried. When dried, cellular components (salts, etc) interfere with desorption/ionization and cause low signal sensitivities. Bottom: After spotting, fatty acids interact with the NIMS surface but other sample components are removed using an in situ sample cleanup step improving NIMS sensitivity.

Figure 2. NIMS analysis of mixtures of fatty acid standards in negative detection mode. (A) NIMS spectrum of a mixture of the saturated fatty acids (lauric acid (C12:0), myristic acid (C14:0), palmitic acid (C16:0), and stearic acid (C18:0) (left)), and the unsaturated fatty acids (linolenic acid (C18:3), linoleic acid (C18:2), and oleic acid (C18:1) (right)); 1 nmol of each acid was used. (B) Signal-to-noise ratio for different amounts of a pure palmitic acid standard. The plot in the inset shows the linearity of the signal over more than 3 orders of magnitude in the low picomole range. (C) Spectrum showing the limit of detection for a pure palmitic acid standard at 100 fmol (S/N ∼3).

or soybean oil was dissolved in 100 µL of a 1 M ammonium hydroxide/acetone solution to convert the esters into fatty acids. As plant oils are relatively pure samples (Figure 1A), 1 µL of the solution was spotted onto the NIMS surface and allowed to dry. NIMS analysis detected deprotonated palmitic, linoleic, and oleic acid for both olive oil (Figure 3A) and soybean oil (Figure 3B) and additional signals

for stearic and linolenic acid in soybean oil only (Figure 3B). In both cases, the recorded spectra were in high agreement with published data for these oils,25 both in fatty acid composition and abundance (Figure 3, insets). In the next step, we tested NIMS with a more complex biological sample. Unlike plant oils that are relatively pure Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3753

Figure 3. Spectra for different plant oils. Insets show the most abundant acids (each more than at least 4% of total fatty acid content) of the published fatty acid (FA) composition.25 (A) Spectrum for olive oil with signals for palmitic acid (C16:0), linoleic acid (C18:2), and oleic acid (C18:1). (B) Spectrum for soybean oil with signals for palmitic acid (C16:0), linolenic acid (C18:3), linoleic acid (C18:2), oleic acid (C18:1), and stearic acid (C18:0).

Figure 4. Spectra for a Chlamydomonas reinhardtii 4A+ extract. The inset shows the reported quantity of fatty acids for C. reinhardtii 137c.26 Signals could be detected for hexadecatetraenoic acid (C16:4), hexadecatrienoic acid (C16:3), palmitoleic acid (C16:1), palmitic acid (C16:0), linolenic acid (C18:3), linoleic acid (C18:2), oleic acid (C18:1), and stearic acid (C18:0). Upper trace: sample was spotted using in situ cleanup. Lower trace: sample was spotted and allowed to dry.

solutions (Figure 1A), algal extracts contain a diversity of metabolites that can interfere with analysis (Figure 1B). Therefore, we tested our NIMS approach using an extract from Chlamydomonas reinhardtii 4A+ cells to further validate the application of NIMS for complex samples. A Chlamydomonas sample was simply extracted with a 1 M ammonium hydroxide/isopropanol solution. One microliter drops were spotted onto the NIMS surface and allowed to dry (Figure 1B, top). In the following analysis, strong signal suppression was observed resulting in only weak signal intensities with a signal-to-noise ratio of 10 for the linolenic acid signal (Figure 4, lower trace). In situ sample cleanup (Figure 3754

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

1B, bottom) resulted in a >10-fold higher signal-to-noise ratio of 115 for the linolenic acid signal (Figure 4, upper trace). Analysis of the algae extract showed clear signals for the deprotonated anions of hexadecatetraenoic, hexadecatrienoic, palmitoleic, palmitic, linolenic, linoleic, oleic, and stearic acid (Figure 4, upper trace). Comparison with the published fatty acid composition of C. reinhardtii 137c showed that the detected signals match with the most abundant fatty acids and the ratios of the signal intensities closely resemble the published data26 (Figure 4, upper trace and inset). We would like to note that 1 µL of the resuspended and diluted extract spotted onto the NIMS surface

equals 40 µL of the original liquid culture showing the high sensitivity of the NIMS approach. Regarding the throughput of the presented method, pure sample spectra with a signal-to-noise of >100 could be acquired with less than 25 shots, whereas a signal-to-noise of >50 was obtained with 100 shots for more complex biological samples. The Voyager MALDI-TOF used has a repetition rate of 3 Hz resulting in an acquisition time of ∼10 s/sample for pure and ∼30 s/sample for complex samples, though instruments with more than 10 times higher repetition rate are common. Hence, utilization of a faster instrument and automation of the liquid handling/spotting process has the potential to further increase the throughput. Sample preparation (saponification and spotting) for the plant oils and algae extracts required ∼2 min. Chlamydomonas extracts were obtained using a standard lab protocol and required an additional 25 min, including spinning down the initial Chlamydomonas culture, ultrasonically lysing the cells, solvent extracting the cells, centrifuging, and filtering the extract. Sample preparation time can likely be reduced using automating liquid handling systems and 96 well plate format. The fabrication of one NIMS chip required about 1 hour; however, depending on the spotting volume, one chip can fit over 1000 samples. In some cases, researchers may also be interested in the percentage of free fatty acids. Since the saponification converts all TAGs, FAMEs, FAEEs, etc into fatty acids, it would be necessary to separate the fatty acids from neutral lipids and analyze them separately. There are many protocols described for accomplishing this.27 For example, extracting the cell lysate with hexane, individually mixing the hexane and aqueous fractions with the ammonium hydroxide/isopropanol solution, and determining the fatty acid content using NIMS. CONCLUSIONS In summary, we present NIMS as a new soft matrix-free desorption/ionization method for direct screening of fatty acids and its application for the rapid analysis of the fatty acid content of plant and microbial oils. Using NIMS, we were able to identify

both the composition and the relative quantities of fatty acids from plant oils and algae extracts. The sample preparation involved only a minimal number of steps mainly for saponification of TAGs, FAMEs, FAEEs, etc and, if necessary, sample extraction. Our approach takes advantage of the unique properties of the NIMS surface that selectively adsorbs hydrophobic compounds, like lipids or fatty acids, while nonhydrophobic or polar compounds can simply be removed from the surface. Therefore, NIMS combines the fast data acquisition of a surface-based approach with “precleaning” to remove salts, etc. Additionally, due to the high sensitivity of NIMS, only very low amounts of sample are required for analysis. The presented method could be coupled to an automated liquid handling and sample spotting system, so that together with its rapid sample analysis, it has the potential to be used for high throughput screening of microalgae and most likely other microorganisms. Altogether, NIMS presents a new technique to assist in biofuel development and complements already existing techniques for biofuel analysis.28 It is expected that a major application will be for screening large libraries of metabolically engineered fatty acid producing strains of microbes. ACKNOWLEDGMENT We gratefully acknowledge support from the US Department of Energy [DE-AC02-05CH11231] and Prof. Krishna Niyogi (Department of Plant and Microbial Biology, University of California, Berkeley, CA).

Received for review January 19, 2010. Accepted March 19, 2010. AC100159Y (25) Paul, A. A.; Southgate, D. A. T. McCance and Widdowson’s The Composition of Foods, 4th ed.; Elsevier/North-Holland Biomedical Press: New York, 1978. (26) Giroud, C.; Gerber, A.; Eichenberger, W. Plant Cell Physiol. 1988, 29, 587– 595. (27) Kim, H. Y.; Salem, N., Jr. J. Lipid Res. 1990, 31, 2285–2289. (28) Abdelnur, P. V.; Eberlin, L. S.; de Sa, G. F.; de Souza, V.; Eberlin, M. N. Anal. Chem. 2008, 80, 7882–7886.

Analytical Chemistry, Vol. 82, No. 9, May 1, 2010

3755