- freely available
Mar. Drugs 2013, 11(11), 4158-4175; doi:10.3390/md11114158
Published: 29 October 2013
Abstract: Phytoplankton exudates play an important role in pelagic ecology and biogeochemical cycles of elements. Exuded compounds fuel the microbial food web and often encompass bioactive secondary metabolites like sex pheromones, allelochemicals, antibiotics, or feeding attractants that mediate biological interactions. Despite this importance, little is known about the bioactive compounds present in phytoplankton exudates. We report a stable-isotope metabolic footprinting method to characterise exudates from aquatic autotrophs. Exudates from 13C-enriched alga were concentrated by solid phase extraction and analysed by high-resolution Fourier transform ion cyclotron resonance mass spectrometry. We used the harmful algal bloom forming dinoflagellate Alexandrium tamarense to prove the method. An algorithm was developed to automatically pinpoint just those metabolites with highly 13C-enriched isotope signatures, allowing us to discover algal exudates from the complex seawater background. The stable-isotope pattern (SIP) of the detected metabolites then allowed for more accurate assignment to an empirical formula, a critical first step in their identification. This automated workflow provides an effective way to explore the chemical nature of the solutes exuded from phytoplankton cells and will facilitate the discovery of novel dissolved bioactive compounds.
Phytoplankton exude large amounts of organic compounds to the surrounding water . The exuded compounds, or exometabolome, contain primary metabolites such as lipids , carbohydrates, amino acids and secondary metabolites. Exudates mediate a variety of biological interactions , for example, sexual pheromones that assist mate finding , allelopathic compounds that deter or intoxicate grazers and suppress competing species , and metabolites that guide microbes and grazers to algal cells and patches of elevated resources [6,7]. Exudates are also an important carbon source for heterotrophic bacteria . All of these small scale molecular processes and biological interactions between species ultimately drive important, larger scale ecological mechanisms. In addition, phytoplankton contribute ca. 50% of the global carbon fixation . Despite these facts there is surprisingly little known about the chemical nature of bioactive compounds produced and released by phytoplankton.
At low cell densities, phytoplankton exudates form a solute cloud around the cell, limited by the release rate and molecular diffusion of these exudates, with high concentration only occurring close to the cell surface. If cell density increases the background concentrations of exudates build up and may have far reaching effects . Unknown toxic exudates from a Chrysochromulina poylepis bloom in the Kattegatt and Skagerrak Sea, for example, resulted in mass mortality of fish and benthic invertebrates [11,12]. Allelochemicals from Alexandrium tamarense structure pelagic communities . Chemical compounds from algal blooms or biofilms also guide other organisms, such as nematodes, fish and seabirds, to productive hotspots in the sea [13,14]. These types of chemically mediated effects may often be as, or more important than, direct trophic interactions in structuring the pelagic community . Identifying and characterizing the phytoplankton “footprint” from the complex background of dissolved organic matter in seawater is, however, a considerable analytical challenge and many bioactive compounds of phytoplankton origin still remain unknown.
Mass spectrometry (MS)-based metabolomics has advanced rapidly in the past decade and is increasingly routinely used for the identification and quantification of small molecules within complex mixtures, such as cells, tissues and whole organisms [16,17]. A variety of conceptual targeted and untargeted MS approaches have been developed, in particular metabolic fingerprinting, an untargeted method that combines analytical and computational analyses to reveal intracellular metabolism . Conversely, metabolic footprinting is an approach to simultaneously study metabolites consumed from, and/or released into, the cell’s local environment . Metabolic footprinting is a relatively rapid and information-rich method for studying multiple released compounds under different environmental conditions and therefore offers a potential short-cut in relation to traditional bioassay guided fractionation, in which samples are partitioned and biotested sequentially in the search for active compounds. Nonetheless, characterising low concentration algal exudates in seawater against the highly complex dissolved organic background is extremely challenging, requiring an analytical technique with high resolution, high mass accuracy and the ability to span a wide dynamic range. Fourier transform ion cyclotron resonance (FT-ICR) MS has the highest mass accuracy and mass resolution of all commercially available mass spectrometers and has proven to be powerful platform across a large variety of metabolomics studies [20,21,22]. Nevertheless, there remain considerable challenges in annotating the m/z (mass-to-charge) measurements in mass spectrometry-based metabolomics experiments .
Stable isotope labelling is a well-established method in biochemistry and is routinely used to study biochemical pathways in plants [23,24], cells [25,26] and organisms [27,28]. More recently, the method has been used to assist metabolic profiling in pelagic cyanobacteria [29,30]. Stable isotopes are readily distinguishable from their more abundant natural isotopes by high resolution MS, even within a complex mixture. This approach provides an opportunity to focus selectively on biologically synthesised metabolites. By increasing the proportion of 13C in the inorganic carbon sources utilized solely by photoautotrophic organisms, metabolites produced by these autotrophs will obtain a unique isotope signature, hereafter referred to as a stable isotope pattern (SIP). Using high resolution MS-based approaches the 13C-labelled released organic compounds of algal origin can be discriminated from the highly complex dissolved organic compounds and other contaminants. Using a combination of unlabelled, 13C labelled, and 15N labelled alga, Baran and colleagues were able to eliminate approximately 90% of the features detected in Synechococcus cell extracts and growth media (i.e., compounds of non-biogenic origin). They were only able to detect a limited number of exudates in the cell media due to their low concentrations but concluded that the method is well suited for metabolic profiling of pelagic autotrophs .
Here we report the development of a methodological framework to selectively analyze and annotate exudates from aquatic autotrophs using both high resolution direct infusion (DI) FT-ICR MS and stable isotope labelling experiments (Figure 1). Alexandrium tamarense cultures were 13C-enriched by addition of 13C-bicarbonate to the seawater. Then the compounds exuded from these algae, as well as controls with natural isotope composition, were measured using DI FT-ICR MS. A novel computational workflow was developed to subsequently annotate, in an automated manner, only those released compounds that were selectively 13C-enriched in the mass spectral footprints. Building on the method described by Baran et al. (2010) , we describe a method that can also handle scenarios where near 100% labelling is not achievable, and methods to sample and desalt larger quantities of exudates.
2. Results and Discussion
2.1. High-Resolution Mass Spectrometry-Based Metabolic Footprinting
Four groups of samples (i.e., derived from unlabelled and 13C-labelled seawater without cells, and from unlabelled and 13C-labelled cultures of Alexandrium tamarense) were studied so as to provide a robust experimental design with appropriate controls. A. tamarense and the seawater-only controls were cultured in parallel under identical conditions and differed only in the carbon source that was added to the system, either unlabelled or 13C-labelled bicarbonate. An optimized DI FT-ICR MS selective ion monitoring (SIM) stitching method was used to maximize the detection and putative annotation of the A. tamarense exometabolome. This approach combines multiple, narrow mass spectra into a single wide-scan spectrum to maximize sensitivity and mass accuracy. The mass spectra of the unlabelled and 13C-labelled seawater controls were highly similar (Figure 2a,b), which indicates the expected lack of incorporation of 13C into dissolved organic matter in the seawater. However, several prominent SIPs (i.e., multiple stable-isotope patterns) were observed in the mass spectra of 13C-labelled cultures that were absent in the unlabelled cultures (Figure 2c,d), confirming successful incorporation of the stable isotope into A. tamarense’s biochemical pathways and subsequent transfer to the culture media. Collectively these findings demonstrate that we can visually discriminate biochemically synthesised exudates from the milieu of chemicals in seawater. Cultures were not axenic, and it is possible that algal metabolites were modified by associated bacteria. Mass spectra of the four sample groups were further investigated using principal component analysis (PCA) to visualize any differences in the overall metabolic footprints (Figure 3). The seawater control samples cluster together regardless of stable isotope labelling and are separated along the PC1 axis of the scores plot from both the unlabelled and 13C-labelled cultures. Samples from 13C-labelled and unlabelled cultures are clearly separated along the PC2 axis (Figure 3), with this metabolic difference caused by the presence of SIPs that were only present in 13C-labelled cultures (Figure 2d). A single outlier can be observed in the 13C-labelled algal samples, closer to the control samples, which we deduced was caused by a lower concentration of exudates in that particular sample, potentially due to a problem during sample preparation. Quality control (QC) samples are relatively well grouped which demonstrates the consistency of the mass spectrometric analyses (Supplementary Figure S1). These results confirm the successful uptake of 13C bicarbonate into the algal metabolism that shifts the overall metabolic footprints, or exudates, of those samples.
2.2. Locating Stable Isotope Patterns
SIPs were initially located in the 13C-labelled A. tamarense spectra by searching for groups of four peaks with specific signal intensity patterns (i.e., the first peak in the 4-peak pattern should have a lower intensity than the second peak, and the third peak in the pattern should have a higher intensity than the fourth peak) and 12C–13C peak differences, hereafter referred to as a “template”, which represents the highest intensity peaks of a stable isotope pattern (Figure 4, shown in green). Overall ca. 300 templates were observed in the FT-ICR MS dataset. To locate the complete SIP, additional 13C-labelled peaks were added on both sides of the template as described in the Methods. Finally, each SIP was matched to its corresponding all-12C peak (or if available a naturally occurring 12C–13C isotope peak pair) in the unlabelled A. tamarense mass spectra (Figure 4). This final step defines the original unlabelled peak in each SIP, which subsequently is used for putative annotation of the exuded metabolite (see Section 2.4). Despite the high sensitivity and mass accuracy of FT-ICR MS, several non-ideal stable-isotope patterns were found (Supplementary Figure S2); these arose because (1) the signal intensity of partially or fully-labelled compound dropped below the detection level of the FT-ICR MS, which resulted in missing peaks across the SIP; (2) a single m/z peak resulted from two or more metabolites of similar mass, which resulted in an altered isotope intensity profile; (3) some m/z features were falsely assigned to a particular SIP in part because of the finite mass accuracy of the FT-ICR; or (4) no all-12C peak or 12C–13C isotope peak pair could be located. Nonetheless, over 100 SIPs of the ca. 200 SIPs that were selected for putative annotation did not suffer from these problems.
|Table 1. Pearson’s correlation coefficient (r) and associated p-value for each comparison between an experimental and simulated stable isotope pattern; specifically to determine which simulated SIP (across five different colour coded labelling efficiencies) is most similar to each experimental SIP in Figure 4, this analysis is used to estimate the overall 13C-isotope labelling efficiency in the experiment.|
|m/z (all-12C containing peak as [M − H]− ion)||Empirical formula||Labelling Efficiency (%)|
2.3. Estimation of the Stable Isotope Labelling Efficiency
As we have shown, SIPs allow the discrimination between m/z signals from released metabolites and the background milieu of chemicals in seawater. Furthermore, SIPs represent a discrete probability distribution of the number of 12C atoms that have been replaced with 13C atoms as a result of the stable isotope labelling. Therefore this distribution includes information regarding the total number of C atoms in a particular metabolite, which can help to annotate unknown metabolites (see Section 2.4). However, such calculations are dependent upon knowing the 13C-isotope labelling efficiency in the experiment (Equation 2). This labelling efficiency has a direct effect on the observed isotope distribution, becoming right-shifted as the percentage of labelling efficiency increases (Supplementary Figure S3). Consequently we can accurately estimate the efficiency of 13C-labelling from the isotope distributions in the experimental data, but this should only be calculated from isotope distributions for which we know the empirical formula unambiguously (termed “well-behaved” SIPs, below). Therefore, first several SIPs were investigated manually, and based on the number of m/z measurements and the profile of the isotope distribution a crude labelling efficiency of approximately 60% was estimated (Supplementary Figure S3, spanning a possible labelling efficiency from 1.1% to 100%). Subsequently, based on this estimate, a more accurate labelling efficiency was calculated by modelling a series of binomial distributions (Equation 2) and then comparing these to several well-behaved experimental SIPs using Pearson's correlation analysis (Figure 4 and Table 1). The criteria for the well-behaved SIPs included: only one empirical formula could be assigned to that SIP, and all possible 12C- and 13C-containing peaks were detected for that formula. For example, the three SIPs and corresponding empirical formulae listed in Table 1 were well-behaved, and were putatively annotated by examination of the KEGG database as [cyclohex-1-enecarboxylic acid or toluene-cis-dihydrodiol], [6-hydroxyhexanoic acid, d-2-hydroxyisocaproate, ethyl (R)-3-hydroxybutanoate, or paraldehyde] and [myristic acid] (a common saturated fatty acid), respectively. For still greater confidence in this annotation, the highest intensity SIP in Figure 4 was further investigated using solid phase extraction and tandem MS, revealing an MS/MS mass spectra consistent with a myristic acid standard (Supplementary Figure S4). Using just these well-behaved SIPs, the average labelling efficiency (i.e., that corresponds to the highest and most significant Pearson’s correlation) was estimated to be 58%.
2.4. Putative Annotation of Exuded Metabolites
Having discovered ca. 300 SIPs that arise from biologically synthesised metabolites and that have subsequently been transferred to the surrounding media, and having determined the average 13C-labelling efficiency in the experiment, we then putatively annotated this compound list. One of the greatest challenges in metabolomics is the annotation of peaks in mass spectra, in part because many of the mass features in a complex biological mass spectrum can be assigned to multiple empirical formulae. Using a mass spectrometer with high mass accuracy is important for this annotation step to reduce the number of possible assignments. The stable isotope labelling approach described here can further improve the confidence in metabolite annotation by eliminating incorrect putative empirical formula(e) assignments (in terms of the number of carbon atoms in the formula). The annotation workflow was applied to all SIPs that matched to an unlabelled 12C–13C isotope peak pair, specifically to ca. 200 SIPs. In this section, however, we focus only on a subset of these as shown in Figure 5 to highlight the metabolite annotation workflow in more detail. First, all possible empirical formulae were calculated for each of the three all-12C containing peaks in the SIPs (Table 2; mass accuracy of 1.5 ppm). As can be seen, high mass accuracy alone is insufficient to determine a single empirical formula for any of these metabolites, and for the highest molecular weight metabolite there are four possible formulae. The experimental SIPs were used to estimate the number of C atoms per metabolite by comparing them to simulated isotope distributions for a range of C atoms in the elemental composition of interest (Figure 5 and Table 2). The putative empirical formulae assignments were then ranked based on the degree of correlation between each simulated and experimental SIP; e.g., the all-12C containing peaks in the two SIPs shown in Figure 5a,b were both putatively annotated with two empirical formulae. However, based on the calculated correlations each of the two SIPs was clearly reduced to a single empirical formula, specifically C10H18O3 and C12H22O2, respectively. Both empirical formulae were assigned multiple putative metabolite names. The third experimental SIP was highly correlated with the theoretical SIPs of three empirical formulae, while a fourth empirical formula was discounted due to a poor correlation. None of the three empirical formulae was annotated as a KEGG compound. However, the highest correlated empirical formulae (i.e., C22H38O3, not present KEGG database) may represent a furan fatty acid which is known to exhibit radical-scavenging ability and anti-inflammatory properties in several organisms, including algae .
From the ca. 200 SIPs that were located, only the SIPs that were matched at least one simulated SIP with an r-value ≥0.7 were considered reliable, which was a total of 126 (Supplementary Table S1 and Figure S5). Many common primary metabolites from marine algae were observed, such as docosohexanoic acid and other polyunsaturated fatty acids. In addition we detected putative structures corresponding to more exotic compounds such as isophorone, known to attract sea lice parasites to salmonid fish , grayanotoxin, originally isolated from rhododendrons, and hormones such as hydroxycorticosterone and 11-β,21-dihydroxy-5-β-pregnane-3,20-dione. There is a bias towards compounds retained on the reversed phase SPE columns used here, and the extraction could be made more complete by using additional resins.
|Table 2. Stable isotope patterns located and annotated in FT-ICR mass spectra collected from 13C-labeled cultures of Alexandrium tamarense.|
|m/z||Colour (in Figure 5)||Empirical Formula (peak)||Ion Form||Theoretical Mass (Da)||Mass Error (ppm)||KEGG Compound||r-value||p-value|
|185.11855||blue||C10H18O3||[M − H]−||185.11832||1.25||[(3R)-6-Hydroxy-3-isopropenyl-heptanoate, (3S)-6-Hydroxy-3-isopropenyl-heptanoate, (5R)-6-Hydroxy-5-isopropenyl-2-methylhexanoate, (5S)-6-Hydroxy-5-isopropenyl-2-methylhexanoate, 10-Oxodecanoate, 2-Oxodecanoic acid, Epomediol]||0.95||0.0115|
|green||C8H14O||[M + Acetate]−||185.11832||1.25||[Sulcatone]||−0.09||0.9149|
|197.15491||blue||C12H22O2||[M − H]−||197.15470||1.05||[(−)-Menthyl acetate, Citronellyl acetate, Decanoyl acetaldehyde, Neomenthyl acetate]||1.00||0.0000|
|green||C10H18||[M + Acetate]−||197.15470||1.05||0.58||0.1735|
|409.29594||red||C22H38O3||[M + Acetate]−||409.29595||−0.02||0.99||0.0000|
|blue||C21H45N2O2P||[M + Acetate]−||409.29653||−1.45||0.91||0.0000|
|orange||C24H42O5||[M − H]−||409.29595||−0.02||0.91||0.0000|
|green||C19H39N8P||[M − H]−||409.29625||−0.77||0.55||0.0506|
3. Experimental Section
3.1. Cell Culture and Stable Isotope Labelling
The dinoflagellate Alexandrium tamarense strain 975 was obtained from the Scandinavian Culture Collection of Algae and Protozoa . Alexandrium spp. are known producers of paralytic shellfish toxin, a group of neurotoxic alkaloids , and in addition, produce allelopathic compounds that lyse competing species and ciliates causing changes in the pelagic community structure at relatively low cell concentrations . Alexandrium cells are ca. 30 µm in diameter. Culture media was prepared from filtered seawater (0.22 µm) that had been pre-purified through two 300 mg Isolute ENV+ (Biotage) solid phase extraction (90 µm, 800 Å hydroxylated polystyrene-divinylbenzene co-polymer; SPE) columns to reduce background levels of dissolved organic compounds. The purified water was enriched with K-min medium , and spiked with either 13C bicarbonate or 12C bicarbonate (as a control) to a concentration of 4 mM. An Alexandrium culture was diluted with either 13C or 12C bicarbonate-containing media and divided between six 500 mL glass flasks for each treatment, approximately 400 mL in each. The percentage of 13C of total carbon in the alga (Ct) will asymptotically approach that of the surrounding media as the cells divide according to the formula:
3.2. Metabolite Extraction from Media
At the end of the two-day incubation of the twenty-four culture beakers (six cell free controls of the 12C and 13C media, six 12C Alexandrium cultures, and six 13C Alexandrium cultures), we removed the algal cells by gentle suction filtration onto 45 mm GF/F filters (Whatman). This procedure may lyse some cells or cause the release of compounds from cells stressed by the filtration, and therefore one should be aware that the sample analysed by the mass spectrometer may contain, in addition to algal exudates, some intracellular algal compounds. The cell free filtrate was then pumped through a 100 mg ENV+ SPE column after washing with 1 mL MeOH and equilibrating with 1 mL water, using a peristaltic pump attached to the outlet luer of the column, to trap the exudates. The ENV+ is a polymeric (polystyrene) resin functionalized with phenol groups. Functionalized polymers have previously been shown to be superior to, for example, C18 packings to extract DOC from seawater samples . Columns were desalted with 1 mL MQ grade water before elution with 1 mL methanol. The eluate was evaporated in a centrifugal concentrator, and resolved in 80:20 methanol:water containing 20 mM ammonium acetate, vortexed and then centrifuged prior to DI FT-ICR MS analyses. A set of QC samples were prepared by pooling an aliquot of each of these samples. The cell free media controls were produced exactly as described above, providing both 13C-labelled and unlabelled seawater controls (each n = 6).
3.3. FT-ICR MS Analysis, Spectral Processing and Unsupervised Multivariate Analyses
Negative ion mode MS analyses were conducted using a hybrid 7-Tesla linear ion trap FT-ICR mass spectrometer (LTQ FT Ultra, Thermo Fisher Scientific, Germany) equipped with a Triversa chip-based nanoelectrospray ion source (Advion Biosciences) using conditions as described previously [37,38]. Each of the 24 samples was analysed in triplicate from m/z 70 to 590 using the selected ion monitoring (SIM)-stitching approach, which was developed previously to maximise the dynamic range and mass accuracy of metabolomics analyses [37,38]. Mass spectra were processed using a three-stage signal filtering method as described by Payne et al. . The resulting intensity matrices, representing the different metabolic profile for each sample, were further processed by the imputation of missing values , probabilistic quotient normalization and generalized-log transformed [41,42]. Finally, principal components analyses were conducted using the Matlab PLS-Toolbox (Eigenvector Research) to assess the metabolic similarities and differences between samples.
3.4. Locating and Putative Identification of SIPs
A novel computational algorithm was developed to locate stable isotope patterns (SIPs) in the partially processed (untransformed and not normalised) mass spectral data. First, three additional peak lists were created by subtracting the exact masses of 13C (13.00335 m/z), 13C2 (26.00670 m/z) and 13C3 (39.01005 m/z) from the one original peak list. These four lists were concatenated and then specific groupings of 4 closely spaced peaks were extracted if they met the following two criteria: (1) all 4 peaks should have the same m/z value, within an error range of 1.5 ppm (this is used to find 13C isotopic spacings); (2) the first peak in the pattern (ordered from low to high m/z) should have a lower intensity than the second peak, and the third peak in the pattern should have a higher intensity than the fourth peak (this is used to select the centre of a stable isotope pattern for compounds with a minimum of 4 carbon atoms). Each grouping of 4 peaks (see Figure 5a,b), hereafter referred to as a “template”, was further extended by including all the observed m/z values on both sides of the template with n × 13C–12C m/z difference(s) away from the first or last peak of the pattern. Once all the peaks (arising from 13C isotopes) that surround the template are located, termed a SIP, the original measured m/z values and intensities across each of the four sample groups (13C-labelled and unlabelled exudates and corresponding seawater-only controls were determined). This allows the examination of overlapping isotopic peaks between 13C-labelled and unlabelled mass spectra. Using MI-Pack metabolite annotation software , the all-12C containing peaks (observed in the unlabelled mass spectra and directly related to the SIP observed in the labelled spectra) were assigned an empirical formula(e) based upon their accurate mass measurement (i.e., elements and ion forms were restricted as follows: 12C0–34H0–72N0–15O0–19P0–7S0–8 and [M − H]−, [M + 35Cl]− and [M + Acetate]−). Only those empirical formulae within a mass tolerance of 1.5 ppm were recorded. Next, all empirical formulae were filtered using the heuristic rules reported previously . In many cases, each empirical formula was putatively assigned a metabolite name using the “single-peak search” approach in MI-Pack as well as the KEGG (Kyoto Encyclopedia of Genes and Genomes) database [43,45]. Next, for each observed SIP, a set of theoretical SIPs were calculated using a binomial density function (Equation 2). In short, the probability of finding a 13C atom in a particular elemental composition follows a binomial distribution, where k is the number of 13C atoms, n is the total number of carbon atoms in the empirical formula of interest, and p is the probability that any single carbon atom is 13C (i.e., the labelling efficiency, ~58% in this study, derived in Results).
Pearson’s correlation coefficients (and corresponding p-values) were calculated between the intensities of the observed SIP and each of the corresponding theoretical SIPs, with the highest correlation revealing the most likely number of carbons in the empirical formula. The intensities of all observed and theoretical SIPs were normalised to a maximum intensity of one to facilitate their comparison. This estimated number of carbons in each empirical formula was then used to guide the annotation of the exuded compounds. Custom scripts were developed using Python and the R-language and are available on request from the corresponding author.
We have developed a methodological framework to explore the chemical nature of exudates from phytoplankton cells. This work represents a stable-isotope mass spectrometry-based metabolic footprinting experiment to map the exudates from Alexandrium tamarense. The labelling procedure applied here, using 13C-enriched bicarbonate, is straightforward, cost efficient, and functional for all micro algae species. The algorithm developed automatically separates the biologically synthesised metabolites from the complex seawater background, and then derives the carbon isotope signatures of these compounds which subsequently enables more accurate putative annotation. Liquid chromatography, tandem or multi-stage mass spectrometry (including the use of authentic standards for structural confirmation), are valuable additions for this methodological framework. Ultimately this approach, in combination with a suitable bioassay, should help to facilitate the discovery of novel bioactive compound candidates secreted from micro algae species.
This work was supported by the NERC Biomolecular Analysis Facility at the University of Birmingham (R8-H10-61) and the European Union Marie Curie Project 237517-LIDPOP (Linking Inducible Chemical Defences and Phytoplankton Population Dynamics).
Conflicts of Interest
The authors declare no conflict of interest.
- Bjornsen, P.K. Phytoplankton exudation of organic-matter—Why do healthy cells do it. Limnol. Oceanogr. 1988, 33, 151–154, doi:10.4319/lo.1988.33.1.0151.
- Kind, T.; Meissen, J.K.; Yang, D.; Nocito, F.; Vaniya, A.; Cheng, Y.-S.; VanderGheynst, J.S.; Fiehn, O. Qualitative analysis of algal secretions with multiple mass spectrometric platforms. J. Chromatogr. A 2012, 1244, 139–147, doi:10.1016/j.chroma.2012.04.074.
- Van Donk, E. Chemical information transfer in freshwater plankton. Ecol. Inf. 2007, 2, 112–120, doi:10.1016/j.ecoinf.2007.03.002.
- Gillard, J.; Frenkel, J.; Devos, V.; Sabbe, K.; Paul, C.; Rempt, M.; Inze, D.; Pohnert, G.; Vuylsteke, M.; Vyverman, W. Metabolomics enables the structure elucidation of a diatom sex pheromone. Angew. Chem. Int. Ed. 2013, 52, 854–857, doi:10.1002/anie.201208175.
- Fistarol, G.O.; Legrand, C.; Selander, E.; Hummert, C.; Stolte, W.; Graneli, E. Allelopathy in Alexandrium spp.: Effect on a natural plankton community and on algal monocultures. Aquat. Microb. Ecol. 2004, 35, 45–56, doi:10.3354/ame035045.
- Stocker, R. Marine microbes see a sea of gradients. Science 2012, 338, 628–633, doi:10.1126/science.1208929.
- Kiorboe, T. A Mechanistic Aproach to Plankton Ecology; Princeton University Press: Princeton, NJ, USA, 2008; pp. 1–209.
- Azam, F.; Fenchel, T.; Field, J.G.; Gray, J.S.; Meyerreil, L.A.; Thingstad, F. The ecological role of water-column microbes in the sea. Mar. Ecol. Prog. Ser. 1983, 10, 257–263.
- Falkowski, P.G.; Barber, R.T.; Smetacek, V. Biogeochemical controls and feedbacks on ocean primary production. Science 1998, 281, 200–206, doi:10.1126/science.281.5374.200.
- Jonsson, P.R.; Pavia, H.; Toth, G. Formation of harmful algal blooms cannot be explained by allelopathic interactions. Proc. Natl. Acad. Sci. USA 2009, 106, 11177–11182, doi:10.1073/pnas.0900964106.
- Nielsen, T.G.; Kiorboe, T.; Bjornsen, P.K. Effects of a chrysochromulina-polylepis subsurface bloom on the planktonic community. Mar. Ecol. Prog. Ser. 1990, 62, 21–35, doi:10.3354/meps062021.
- Underdal, B.; Skulberg, O.M.; Dahl, E.; Aune, T. Disastrous bloom of Chrysochromulina polylepis (prymnesiophyceae) in Norwegian Coastal Waters 1988—Mortality in Marine Biota. AMBIO 1989, 18, 265–270.
- Nevitt, G.A. Olfactory foraging by Antarctic procellariiform seabirds: Life at high Reynolds numbers. Biol. Bull. 2000, 198, 245–253, doi:10.2307/1542527.
- Hockelmann, C.; Moens, T.; Juttner, F. Odor compounds from cyanobacterial biofilms acting as attractants and repellents for free-living nematodes. Limnol. Oceanogr. 2004, 49, 1809–1819, doi:10.4319/lo.2004.49.5.1809.
- Hay, M.E. Marine chemical ecology: Chemical signals and cues structure marine populations, communities, and ecosystems. Ann. Rev. Mar. Sci. 2009, 1, 193–212, doi:10.1146/annurev.marine.010908.163708.
- Viant, M.; Sommer, U. Mass spectrometry based environmental metabolomics: A primer and review. Metabolomics 2013, 9, 144–158, doi:10.1007/s11306-012-0412-x.
- Dunn, W.; Erban, A.; Weber, R.M.; Creek, D.; Brown, M.; Breitling, R.; Hankemeier, T.; Goodacre, R.; Neumann, S.; Kopka, J.; et al. Mass appeal: Metabolite identification in mass spectrometry-focused untargeted metabolomics. Metabolomics 2013, 9, 44–66.
- Draper, J.; Lloyd, A.; Goodacre, R.; Beckmann, M. Flow infusion electrospray ionisation mass spectrometry for high throughput, non-targeted metabolite fingerprinting: A review. Metabolomics 2013, 9, 4–29, doi:10.1007/s11306-012-0449-x.
- Kell, D.B.; Brown, M.; Davey, H.M.; Dunn, W.B.; Spasic, I.; Oliver, S.G. Metabolic footprinting and systems biology: The medium is the message. Nat. Rev. Microbiol. 2005, 3, 557–565.
- Taylor, N.; Weber, R.M.; Southam, A.; Payne, T.; Hrydziuszko, O.; Arvanitis, T.; Viant, M. A new approach to toxicity testing in Daphnia magna: Application of high throughput FT-ICR mass spectrometry metabolomics. Metabolomics 2009, 5, 44–58.
- Oikawa, A.; Nakamura, Y.; Ogura, T.; Kimura, A.; Suzuki, H.; Sakurai, N.; Shinbo, Y.; Shibata, D.; Kanaya, S.; Ohta, D. Clarification of pathway-specific inhibition by Fourier transform ion cyclotron resonance/mass spectrometry-based metabolic phenotyping studies. Plant Physiol. 2006, 142, 398–413, doi:10.1104/pp.106.080317.
- Brown, S.C.; Kruppa, G.; Dasseux, J.L. Metabolomics applications of FT-ICR mass spectrometry. Mass Spectrom. Rev. 2005, 24, 223–231, doi:10.1002/mas.20011.
- Giavalisco, P.; Hummel, J.; Lisec, J.; Inostroza, A.C.; Catchpole, G.; Willmitzer, L. High-resolution direct infusion-based mass spectrometry in combination with whole 13C metabolome isotope labeling allows unambiguous assignment of chemical sum formulas. Anal. Chem. 2008, 80, 9417–9425, doi:10.1021/ac8014627.
- Giavalisco, P.; Kohl, K.; Hummel, J.; Seiwert, B.; Willmitzer, L. 13C isotope-labeled metabolomes allowing for improved compound annotation and relative quantification in liquid chromatography-mass spectrometry-based metabolomic research. Anal. Chem. 2009, 81, 6546–6551, doi:10.1021/ac900979e.
- Mueller, D.; Heinz, E. Stable isotope-assisted metabolomics to detect metabolic flux changes in mammalian cell cultures. Curr. Opin. Biotechnol. 2013, 24, 54–59, doi:10.1016/j.copbio.2012.10.015.
- Rodgers, R.; Blumer, E.; Hendrickson, C.; Marshall, A. Stable isotope incorporation triples the upper mass limit for determination of elemental composition by accurate mass measurement. J. Am. Soc. Spectrom. 2000, 11, 835–840, doi:10.1016/S1044-0305(00)00158-6.
- Wu, L.; Mashego, M.R.; van Dam, J.C.; Proell, A.M.; Vinke, J.L.; Ras, C.; van Winden, W.A.; van Gulik, W.M.; Heijnen, J.J. Quantitative analysis of the microbial metabolome by isotope dilution mass spectrometry using uniformly 13C-labeled cell extracts as internal standards. Anal. Biochem. 2005, 336, 164–171, doi:10.1016/j.ab.2004.09.001.
- Birkemeyer, C.; Luedemann, A.; Wagner, C.; Erban, A.; Kopka, J. Metabolome analysis: The potential of in vivo labeling with stable isotopes for metabolite profiling. Trends Biotechnol. 2005, 23, 28–33, doi:10.1016/j.tibtech.2004.12.001.
- Baran, R.; Bowen, B.P.; Bouskill, N.J.; Brodie, E.L.; Yannone, S.M.; Northen, T.R. Metabolite identification in Synechococcus sp. PCC 7002 using untargeted stable isotope assisted metabolite profiling. Anal. Chem. 2010, 82, 9034–9042, doi:10.1021/ac1020112.
- Baran, R.; Bowen, B.P.; Northen, T.R. Untargeted metabolic footprinting reveals a surprising breadth of metabolite uptake and release by Synechococcus sp. PCC 7002. Mol. Biosyst. 2011, 7, 3200–3206, doi:10.1039/c1mb05196b.
- Wakimoto, T.; Kondo, H.; Nii, H.; Kimura, K.; Egami, Y.; Oka, Y.; Yoshida, M.; Kida, E.; Ye, Y.; Akahoshi, S.; et al. Furan fatty acid as an anti-inflammatory component from the green-lipped mussel Perna canaliculus. Proc. Natl. Acad. Sci. USA 2011, 108, 17533–17537, doi:10.1073/pnas.1110577108.
- Ingvarsdottir, A.; Birkett, M.A.; Duce, I.; Genna, R.L.; Mordue, W.; Pickett, J.A.; Wadhams, L.J.; Mordue, A.J. Semiochemical strategies for sea louse control: Host location cues. Pest Manag. Sci. 2002, 58, 537–545, doi:10.1002/ps.510.
- Scandinavian Culture Collection of Algae and Protozoa. Available online: http://www.sccap.dk/ (accessed on 18 October 2013).
- Schantz, E.J.; Lynch, J.M.; Vayvada, G.; Matsumoto, K.; Rapoport, H. The purification and characterization of the poison produced Gonyaulax catenella in axenic cultures. Biochemistry 1966, 5, 1191–1195, doi:10.1021/bi00868a011.
- Keller, M.D.; Selvin, R.C.; Claus, W.; Guillard, R.R.L. Media for the culture of oceanic ultraphytoplankton. J. Phycol. 1987, 23, 633–638.
- Dittmar, T.; Koch, B.; Hertkorn, N.; Kattner, G. A simple and efficient method for the solid-phase extraction of dissolved organic matter (SPE-DOM) from seawater. Limnol. Oceanogr. Meth. 2008, 6, 230–235, doi:10.4319/lom.2008.6.230.
- Weber, R.J.; Southam, A.D.; Sommer, U.; Viant, M.R. Characterization of isotopic abundance measurements in high resolution FT-ICR and Orbitrap mass spectra for improved confidence of metabolite identification. Anal. Chem. 2011, 83, 3737–3743, doi:10.1021/ac2001803.
- Southam, A.D.; Payne, T.G.; Cooper, H.J.; Arvanitis, T.N.; Viant, M.R. Dynamic range and mass accuracy of wide-scan direct infusion nanoelectrospray fourier transform ion cyclotron resonance mass spectrometry-based metabolomics increased by the spectral stitching method. Anal. Chem. 2007, 79, 4595–4602.
- Payne, T.G.; Southam, A.D.; Arvanitis, T.N.; Viant, M.R. A signal filtering method for improved quantification and noise discrimination in fourier transform ion cyclotron resonance mass spectrometry-based metabolomics data. J. Am. Soc. Mass Spectr. 2009, 20, 1087–1095, doi:10.1016/j.jasms.2009.02.001.
- Hrydziuszko, O.; Viant, M. Missing values in mass spectrometry based metabolomics: An undervalued step in the data processing pipeline. Metabolomics 2012, 8, 161–174, doi:10.1007/s11306-011-0366-4.
- Parsons, H.M.; Ludwig, C.; Gunther, U.L.; Viant, M.R. Improved classification accuracy in 1- and 2-dimensional NMR metabolomics data using the variance stabilising generalised logarithm transformation. BMC Bioinf. 2007, 8, 234, doi:10.1186/1471-2105-8-234.
- Dieterle, F.; Ross, A.; Schlotterbeck, G.; Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in 1H NMR metabonomics. Anal. Chem. 2006, 78, 4281–4290, doi:10.1021/ac051632c.
- Weber, R.J.M.; Viant, M.R. MI-Pack: Increased confidence of metabolite identification in mass spectra by integrating accurate masses and metabolic pathways. Chemom. Intell. Lab. Syst. 2010, 104, 75–82, doi:10.1016/j.chemolab.2010.04.010.
- Kind, T.; Fiehn, O. Seven Golden Rules for heuristic filtering of molecular formulas obtained by accurate mass spectrometry. BMC Bioinf. 2007, 8, 105, doi:10.1186/1471-2105-8-105.
- Kanehisa, M.; Goto, S.; Sato, Y.; Furumichi, M.; Tanabe, M. KEGG for integration and interpretation of large-scale molecular data sets. Nucleic Acids Res. 2012, 40, D109–D114, doi:10.1093/nar/gkr988.
© 2013 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).