Dissolved organic metabolite extraction from high-salt media

We describe considerations and strategies for developing a nuclear magnetic resonance (NMR) sample preparation method to extract low molecular weight metabolites from high-salt spent media in a model coculture system of phytoplankton and marine bacteria. Phytoplankton perform half the carbon fixation and oxygen genera-tion on Earth. A substantial fraction of fixed carbon becomes part of a metabolite pool of small molecules known as dissolved organic matter (DOM), which are taken up by marine bacteria proximate to phytoplankton. There is an urgent need to eluci-date these metabolic exchanges due to widespread anthropogenic transformations on the chemical, phenotypic, and species composition of seawater. These changes are increasing water temperature and the amount of CO 2 absorbed by the ocean at energetic costs to marine microorganisms. Little is known about the metabolite-medi-ated, structured interactions occurring between phytoplankton and associated marine bacteria, in part because of challenges in studying high-salt solutions on various analytical platforms. NMR analysis is problematic due to the high-salt content of both natural seawater and culture media for marine microbes. High-salt concentration degrades the performance of the radio frequency coil, reduces the efficiency of some pulse sequences, limits signal-to-noise, and prolongs experimental time. The method described herein can reproducibly extract low molecular weight DOM from small-volume, high-salt cultures. It is a promising tool for elucidating metabolic flux between marine microorganisms and facilitates genetic screens of mutant microorganisms.

We describe considerations and strategies for developing a nuclear magnetic resonance (NMR) sample preparation method to extract low molecular weight metabolites from high-salt spent media in a model coculture system of phytoplankton and marine bacteria. Phytoplankton perform half the carbon fixation and oxygen generation on Earth. A substantial fraction of fixed carbon becomes part of a metabolite pool of small molecules known as dissolved organic matter (DOM), which are taken up by marine bacteria proximate to phytoplankton. There is an urgent need to elucidate these metabolic exchanges due to widespread anthropogenic transformations on the chemical, phenotypic, and species composition of seawater. These changes are increasing water temperature and the amount of CO 2 absorbed by the ocean at energetic costs to marine microorganisms. Little is known about the metabolite-mediated, structured interactions occurring between phytoplankton and associated marine bacteria, in part because of challenges in studying high-salt solutions on various analytical platforms. NMR analysis is problematic due to the high-salt content of both natural seawater and culture media for marine microbes. High-salt concentration degrades the performance of the radio frequency coil, reduces the efficiency of some pulse sequences, limits signal-to-noise, and prolongs experimental time. The method described herein can reproducibly extract low molecular weight DOM from small-volume, high-salt cultures. It is a promising tool for elucidating metabolic flux between marine microorganisms and facilitates genetic screens of mutant microorganisms. perform half of all carbon fixation ( Figure 1A). 1 A major fraction goes on to become part of the largest pool of active carbon on Earth known as dissolved organic matter (DOM), which can include carbohydrates, lipids, nitrogenous compounds, organic acids, and growth factors. 2,3 This collection of up to hundreds of thousands of organic metabolites plays vital roles in global carbon, nitrogen, sulfur, and phosphorus cycles; nutrient and xenobiotic catabolism; and ecological diversity. [4][5][6] Marine bacteria in the surface ocean scavenge, transform, and decompose DOM as part of the microbial loop ( Figure 1A). 7,8 Although ocean microorganisms would be physically distant from each other if evenly spaced in seawater (with a density of 10 4 to 10 6 cells per ml), phytoplankton and marine bacteria form concentrated spatial networks that maximize their interactions ( Figure 1A). [9][10][11] These unique exchanges between phytoplankton and bacteria influence the physiology and ecology of one another. The variety of DOM released by phytoplankton informs the community structure of bacteria, while the taxonomic identity and abundance of bacteria inform the concentration and variety of DOM released by phytoplankton. 8,11 Eukaryotic diatoms represent a particularly vital taxon of phytoplankton, generating 40% of all photosynthetic output from the ocean. 4,9 Diatoms and marine bacteria share specific associations with interactions tracing back on an evolutionary timescale greater than 200 million years.
There is strong evidence that hundreds of bacterial genes were incorporated into diatoms over that time. 12 Further substantiation of ecological relationships is the bacterial production of essential vitamins required by diatoms. 13 Much research has centered on cobalamin (vitamin B 12 ), which is required by organisms lacking the cobalamin-independent methionine synthase gene. 9,14 One such example is the provision of B 12 to the diatom Thalassiosira pseudonana by the marine bacterium Ruegeria pomeroyi, representatives of two microbial taxa commonly found in close ecological association ( Figure 1B). Vitamin B 12 is an essential exogenous nutrient for T. pseudonana, which exhibits poor growth rates as an axenic culture in B 12 -null media. 4 Addition of R. pomeroyi to the culture causes upregulation of the cobalamin acquisition protein (CBA1) gene that allows T. pseudonana to take up and utilize the form of B 12 produced by R. pomeroyi. As a result, growth rates recover to that of axenic phytoplankton grown in B 12 -enriched media. 4 Concurrently, in coculture with T. pseudonana, R. pomeroyi upregulates several genes linked to the transport and catabolism of the DOM substrate 2,3-dihydroxypropane 1-sulfonate (DHPS) (structure shown in Figure 1B). Thalassiosira pseudonana exudes large amounts of DHPS, which R. pomeroyi is able to use as a sole carbon source. Three of the upregulated genes (hpsKLM) were experimentally annotated as components of a tripartite ATP-independent periplasmic (TRAP) transporter for the import of DHPS from the extracellular environment into the cytoplasm of R. pomeroyi ( Figure 1B). 4,15 F I G U R E 1 Phytoplankton and marine bacteria interact specifically, playing a pivotal role in global oxygen production and carbon flux. (A) Atmospheric CO 2 is absorbed by the ocean and taken up by phytoplankton, which performs photosynthesis to yield O 2 and fixed carbon. Approximately half of the fixed carbon is released as DOM, which is scavenged by marine bacteria within the microbial loop. (B) In a model coculture, Thalassiosira pseudonana and Ruegeria pomeroyi upregulate transcription of CBA1 and hpsKLM genes, respectively. T. pseudonana (teal) upregulates the CBA1 gene to enhance uptake of R. pomeroyi-produced vitamin B 12 . Simultaneously, R. pomeroyi (pink, upper right) upregulates hpsKLM genes, which encode for a transporter to uptake DHPS into the cytoplasm of the cell. DHPS, 2,3-dihydroxypropane 1-sulfonate; DOM, dissolved organic matter We obtained bulk quantities of 13 C-labeled spent medium from axenic T. pseudonana and cocultures of T. pseudonana grown with either wildtype R. pomeroyi or a transporter knockout mutant (ΔhpsKLM). We developed and utilized a method of exometabolome footprint analysis to extract low molecular weight (LMW) metabolites contained within the high-salt spent media of marine microorganisms. Exudate is advantageous because it provides a snapshot of metabolites potentially participating in flux. 13 C-DHPS was used as an indicator of extraction efficacy because it is a known metabolite that we could monitor via nuclear magnetic resonance (NMR) spectroscopy.
We used the method to confirm the presence of 13 C-DHPS in 13 C-labeled spent media containing exudate from T. pseudonana cocultured with the transporter knockout mutant ΔhpsKLM. The metabolite footprint of the mutant coculture was compared with T. pseudonana cocultured with wild-type R. pomeroyi with the rationale that 13 C-DHPS should be present in the transporter knockout mutant due to an inability of the bacteria to uptake 13 C-DHPS. Conversely, absence of 13 C-DHPS was predicted in the wild-type cocultures, which possessed a functional transporter for the molecule.
We then applied this method to further investigate the exometabolome of spent media from cocultures of T. pseudonana and representative marine bacteria from several taxa to explore differential uptake of LMW DOM by members of several bacterial clades. These data are available in Ferrer-González et al. 17 To our knowledge, this is the first method for extracting LMW metabolites from high-salt spent media for NMR analysis. In the results of this paper, we will describe our optimization process for these analyses, as well as the strengths and limitations of the method for other applications.
For clarity, the material is organized into three sections based on spent media preparation strategies: D 2 O reconstitution, methanol (MeOH) extraction, and deuterated dimethyl sulfoxide (DMSO-d 6 ) extraction. We conclude by describing our optimized method of lyophilization, mechanical homogenization, and extraction in DMSO-d 6 .

| MeOH extraction
Coculture extract was prepared by rotary evaporating 50 ml of spent media and dissolving solutes into 10 ml of MeOH. This suspension was centrifuged for 15 min at 22 C at 5000 x g. The supernatant was rotary evaporated again and resuspended in 1 ml of D 2 O with 5 μM deuterated sodium trimethylsilylpropanesulfonate (DSS-d 6 , Cambridge Isotope Laboratories, Inc.). This suspension was placed into a 1.7 ml Eppendorf tube and centrifuged for 15 min at room temperature (RT) at 5000 x g. The supernatant (600 μl) was transferred to a 5-mm NMR tube.

| DMSO-d 6 extraction
25 and 15 ml starting material Coculture extracts were prepared by lyophilizing 25 or 15 ml of spent media in 50 ml conical tubes. The resulting solid salt and DOM mixture was then manually homogenized with a metal spatula. The dry material was extracted with 1.5 or 1 ml of DMSO-d 6 (Cambridge Isotope Laboratories, Inc.), respectively, containing 50 μM DSS-d 6 . Tubes were centrifuged for 10 min at 22 C at 5000 x g. The supernatant was extracted into 1.7 ml Eppendorf tubes and centrifuged for 15 min at RT at 5000 x g. The final supernatant (550 μl) was transferred to 5-mm NMR tubes.

ml starting material
Coculture extracts were prepared by lyophilizing 5 ml of spent media in 15 ml conical tubes. The resulting solid salt and DOM mixture was then homogenized (MP Biomedicals FastPrep-96) to a fine powder through three 30 s cycles at 1800 rpm using 5 x 3.5-mm glass beads. The dry material was extracted with 200 μl of DMSO-d 6 containing 50 μM DSS-d 6 by further homogenizing with three 30 s cycles at 1800 rpm. Tubes were centrifuged for 10 min at 22 C at 5000 x g. The supernatant was extracted into 1.7 ml Eppendorf tubes and centrifuged for 15 min at RT at 5000 x g. The final supernatant (40 μl) was transferred to 1.7-mm NMR tubes.

| 800-MHz Bruker NEO sample storage
Samples were kept in a refrigerated sample changer (SampleJet) followed by temperature equilibration at 300 K for 5 min within the NMR probehead immediately prior to analysis.

| Experimental parameters
The utilized NMR experiments are summarized in Table 1. 2.3.3 | Spectral processing 1 H-13 C-HSQC spectra were processed in MNOVA by applying a squared cosine apodization, 4K zero-filling, and Fourier transformation.
Transformed spectra were auto-phased, baseline-corrected, and referenced along f1 and f2 to DSS-d 6 .

| DHPS concentration
Spent media from a 5 ml culture of axenic T. pseudonana were extracted with DMSO-d 6

| RESULTS
Here, we discuss the challenges, strategies undertaken (Scheme 1), and outcomes of each attempt to develop a method for extracting LMW metabolites from high-salt spent media.
The large salt concentration of spent media required for successful culturing of marine microorganisms presents numerous technical challenges to NMR. It is well established that high-salt solutions affect the tuning and response of the probe, leading to poor signal response and prolonged experimental acquisition times. 2,16 It also limits the use of more demanding NMR experiments that rely on multiple radio frequency (RF) pulses or high-power decoupling.

| Limitations of existing extraction methods
Most sample preparation for the chemical analysis of marine cultures has utilized solid phase extraction (SPE), diafiltration, and/or C18 silica columns to desalt spent media and extract DOM. [18][19][20][21] These techniques typically have an approximate 1-kDa cutoff that excludes most LMW DOM. 22 A principal challenge for the elucidation of dissolved metabolites is retaining small, polar, ionic DOM while separating it from the more than 20 inorganic salts present in the medium at a total concentration of 34 grams per kilogram! Additionally, the concentration of individual components of DOM in the culture medium and ocean is in the picomolar to micromolar range 2 and is often scavenged within minutes by bacteria. 23 Therefore, samples must be concentrated to reach detectable levels for 1 H NMR; this unfortunately increases the salt concentration as well.

| Approach to method development
The key goals were as follows: • To develop a practical and reproducible high-throughput method to analyze and compare a large quantity of samples.
• To extract as many small, polar compounds as possible while minimizing both salt concentration and sample modification.
S C H E M E 1 Strategies to isolate DOM from high-salt spent media. Several methods of dehydration, separation, homogenization, and solvent selection were employed. NMR experiment choice, probe type, and tube diameter were also evaluated.
The results are organized into three subsections that describe the combination of dehydration, homogenization, and NMR techniques employed with each of the following extraction solvents: • 3.3 | NMR analysis: Nuclei observed, probe type, and tube size To develop a practical and sensitive NMR method that would enable analyses of many samples-for differing biological conditions as well as replicates-we considered the advantages of both 13 C detection and small-volume cryoprobes. The 13 C nucleus is less sensitive than 1 H by several orders of magnitude due to possessing a smaller gyromagnetic ratio and a natural abundance of only 1.1%. 24 However, isotopic labeling with 13 C increases the percentage of 13 C nuclei for detection by enhancing the effective 13 C concentration of the metabolites being actively synthesized. 24 Axenic T. pseudonana and cocultures with R. pomeroyi were thus cultured using media labeled with NaH 13 CO 3 . We capitalized on the unique assets of several NMR magnet strengths, probes, and experiments to analyze the extracted spent media. These include 600-, 800-, and 900-MHz spectrometers, specialty cryoprobes, 13 C observe direct-detection and two-dimensional (2D) 1 H-13 C-HSQC. We describe these and other key strategies of sample drying, homogenization, and solvent choice below.

| D 2 O
In an effort to modify the sample as little as possible, our investigation began by utilizing a simple drying method combined with D 2 O reconstitution.

| SpeedVac drying and D 2 O reconstitution
The process of sample drying can exert a powerful influence over the metabolite profiles determined by NMR. 25 SpeedVac drying involves the concentration of a sample by using a centrifugal vacuum evaporator. 25 This is a high throughput method of sample drying that can be performed with temperature control in some instruments. Salt is concentrated to the point of saturation, so the sample is unsuitable for later analysis in some NMR probes. Initially, to maintain as much material as possible, we simply added D 2 O as an extraction solvent for its strong polarity and ability to solubilize a range of molecules. Although this only slightly diluted the salt concentration, we exploited the 13 C enrichment in our cultures and attempted to detect 13 C directly with a high-sensitivity carbon cryoprobe.

| Direct-detection 13 C-NMR and 1 H-13 C-HSQC
We utilized a Bruker NEO 600-MHz NMR spectrometer equipped with a 5-mm DCH 13 C observe cryoprobe to perform direct 13 C detection on spent media concentrated to saturation with a SpeedVac and reconstituted in D 2 O. Cryoprobes have greater sensitivity than RT probes, and the DCH probe is optimized for the detection of 13 C nuclei. 26 Because of the lower frequency, this probe is less susceptible to interference from high-salt concentrations compared with 1 H-optimized cryoprobes at the same field strength, which can suffer dramatic signal-to-noise (S/N) losses as sample conductivity increases. 27 After 16 h of acquisition, this experiment allowed us to discern several differences between axenic T. pseudonana and the coculture of the diatom with R. pomeroyi in the region containing DHPS carbons ( Figure 2). Multiple peaks appeared to be drawn down by R. pomeroyi, but signal overlap and complicated multiplet structures made it difficult to identify the carbons of DHPS. Although a carbon detected 2D heteronuclear correlation (HETCOR) experiment would help with signal assignments, this was not attempted because of the low S/N and long acquisition time. Low throughput, challenging peak identification, and inadequate S/N then led us to reconsider proton detected 2D 1 H-13 C-HSQC experiments to resolve both 1 H and 13 C nuclei with a 5-mm CN{H} TXO cryoprobe on a 900-MHz Bruker NEO spectrometer.
We analyzed samples of T. pseudonana cocultured with the R. pomeroyi transporter knockout mutant ΔhpsKLM. As a baseline, we first analyzed unmodified exudate to determine the feasibility of using 1 H-13 C-HSQC experiments in moderate salt concentrations. As expected, the metabolites in this spectrum were too dilute to yield any usable peaks ( Figure 3A). Next, we concentrated the exudate with rotary evaporation and reconstituted in D 2 O, resulting in a spectrum in which all three peaks of DHPS were present ( Figure 3B). However, the overall spectrum had few peaks ( Figure 3D), and the starting exudate volume was 50 ml, which was a challenge-both upstream and downstream-in preparing large numbers of cultures. The salt content was now very high, reflected by the long 1 H and 13 C pulse widths (26.2 and 52.42 μs, respectively), while the S/N was relatively low (Figure 4). We elected to continue refining the method to further reduce starting sample volumes, experimental times, and salt.

| MeOH extraction
To reduce the salt concentration in the final NMR samples, we switched to extracting rotary evaporated samples with MeOH, a solvent that is suitable for both polar and nonpolar molecules.

| Rotary evaporation and MeOH extraction
The sample was dehydrated using rotary evaporation and subjected to MeOH extraction. This extract was rotary evaporated again and redissolved in D 2 O. Rotary evaporation has several drawbacks. Heating samples impacts metabolite integrity 28 and is low throughput for multiple samples and replicates. Only one sample can be dried at a time, requiring continuous observation. There is also inevitable sample loss from solutes adhering to the walls of the relatively large volume flasks. Accordingly, variability with rotary evaporation is typically higher than other methods.

| MeOH extraction and 1 H-13 C-HSQC
The MeOH extract was rotary evaporated again and reconstituted in D 2 O in a 3-mm NMR tube and analyzed on the CN{H} TXO cryoprobe ( Figure 3C). In addition to the DHPS peaks, many other signals were faintly observed, indicating that the extraction was effective ( Figure 3E). The combination of lower salt and the smaller volume tube reduced the 1 H and 13 C pulse widths to 17 and 40.2 μs, respectively, and S/N was comparable, even although the starting culture volume was half as much ( Figures 3B and 4). We aimed to further improve the method to increase S/N, decrease salt concentration, and decrease the volume of the coculture medium utilized.

| Other method considerations
Two important aspects of optimizing the method were the refinement of our drying technique and homogenization practices.

| Lyophilization
To achieve our goal of high-throughput sample drying, lyophilization was used instead of rotary evaporation. Lyophilization is the high-throughput process of removing liquid and volatile components from samples through freeze-drying with a vacuum 6-30 times the pressure of the SpeedVac. 25 In this process, the solvent sublimates from the sample, going directly from the solid to the gas phase. 25 Lyophilization has been shown to provide consistent results with strong reproducibility and minimal degradation of metabolites. 29 We transitioned to lyophilization in the middle of the DMSO-d 6 stage (in 3.7) of method development.

| Homogenization of solid residue
Homogenization is a short but crucial step in sample preparation, affecting the surface area of the dried sample available to the solvent for extraction. Initially, dry solids in the rotary evaporated and lyophilized samples were manually ground using a metal spatula. The three main drawbacks to this method were sample loss, particle size heterogeneity, and low throughput. A significant portion of solid sample was unable to be scraped To quantitatively and reproducibly extract the DOM, dried material must be ground to a fine powder. Our optimized method employed the use of a mechanical homogenizer to bead-beat the samples in two rounds: first dry, then again after the addition of deuterated solvent. Beadbeating is high throughput and produces a finely powdered sample compared with manual homogenization. As such, the reconstitution solvent has greater access to the entirety of the sample, thus extracting more efficiently. The primary disadvantage of bead-beating is that the bead friction can introduce heat, potentially affecting temperature-sensitive metabolites. 30

| DMSO-d 6 extraction
Based on the increased number of extracted components seen in the MeOH extract, we tried using DMSO-d 6 as a possibly more effective solvent. DMSO is often considered a universal solvent and is ideal for small polar and ionic organic compounds. The drawbacks are that the high viscosity of DMSO-d 6 can cause line broadening, and it can also be an oxidant to some molecules. 31 If later recovery is desired, DMSO-d 6 can be challenging to remove completely from the sample, even with the dehydration methods discussed above. Finally, DMSO-d 6 has a strong affinity for water, which can cause broad water peaks in the spectra if the solvent is left exposed to air or in an unsealed vial, even for short periods of time. 31 This can also make standardizing chemical shifts difficult as water/DMSO-d 6 ratios change over time or between samples. Using a smaller amount of deuterated DMSO for extraction and as the NMR solvent provided a promising spectrum. The 1 H-13 C-HSQC spectra of various starting volumes of exudate, lyophilized and extracted with DMSO-d 6 , are shown in Figure 5. This method addressed all three main problems of analyzing spent media from marine microorganisms: high-salt concentrations, low concentrations of DOM, and difficulty in isolating small, polar metabolites. All three peaks of DHPS were visible with starting volumes of 25 ( Figure 5A) and 15 ml ( Figure 5B) exudate. DMSO-d 6 further reduced the 1 H and 13 C pulse widths (to ranges of 11-13 and 20-24 μs, respectively) and increased S/N by a factor of $2.5-3.5 ( Figure 4).
To develop a higher throughput method for our future experimental needs of chemically differentiating between multiple cocultures with many replicates and monitoring cocultures grown under varied experimental conditions, we continued to reduce the starting volume. We utilized a 1.7-mm cryoprobe at 800 MHz that allowed starting culture volumes of 5 ml and final NMR sample volumes of 35-40 μl. Small-volume probes are advantageous for increasing S/N in mass-limited exudate samples. 24 The smaller diameter of both the probe and the sample decreases the effects of salt and increases the amount of the sample volume observed by the RF coil due to an improved filling factor. 32 When combined with high-temperature super conducting material, small-volume probes can improve mass sensitivity by up to 14 times compared with typical 5-mm probes. 32 Our 800-MHz 1.7-mm cryoprobe provided high-quality, high-throughput spectra with just a small fraction of the starting volume of exudate  Figure S1). This allowed us to calculate the concentration of endogenous DHPS present in the axenic sample as 28.4 μM.
We used our optimized extraction method ( Figure 6) to compare three different culture conditions with suitable numbers of replicates: axenic T. pseudonana, T. pseudonana + wild-type R. pomeroyi, and T. pseudonana + ΔhpsKLM R. pomeroyi. These cocultures were a proof-of-principle as we expected 13 C-DHPS to be present in the axenic culture and the T. pseudonana coculture with ΔhpsKLM R. pomeroyi, which lacked the DHPS transporter. It was expected that DHPS would be absent in the T. pseudonana + wild-type R. pomeroyi coculture because of the bacteria possessing a functional DHPS transporter. Our extraction method detected the disruption of DHPS uptake in the mutant coculture and supported previous experimental annotation of the hpsKLM genes as DHPS transporter genes 15 (Figure 7). Dispensing the sample into 1.7-mm tubes for analysis in a small-volume cryoprobe allows characterization of LMW DOM peaks of low concentration via NMR spectroscopy. Ultimately, this method is a step forward in elucidating the metabolites present in DOM. It also provides additional benefits by requiring only small coculture volumes. Low culture sample volumes enhance the ability for high-throughput sample analysis in future experiments and conserve both 13 C label and deuterated solvents.

| DISCUSSION
In the future, we hope to further concentrate the extracts to achieve de novo metabolite identification and chemical analysis, but this method also has potential as a genetic screening tool for microorganisms cultured in high-salt media. Extracted exudate from mutant microorganisms can be compared with axenic and wild-type cocultures to support gene assignment, as was demonstrated with the R. pomeroyi ΔhpsKLM mutant ( Figure 7). Additionally, this method enables the discovery of metabolic interactions between ocean microbes across a range of bacterial taxa commonly found in ecological association with phytoplankton. This extraction method has already been utilized to support the hypothesis that metabolic interactions between phytoplankton and marine bacteria generate niche-specific footprints that differentiate cocultures (see Ferrer-González et al. 17 ).
Extraction methods of greater complexity may be developed and optimized to extract metabolites in higher concentrations. One such method we explored is a two-step process employing lyophilization, MeOH extraction, a second round of lyophilization, and DMSO-d 6 extraction. This method resulted in good spectra, but DHPS peaks of interest were in lower concentrations than our optimized direct DMSO extraction ( Figure S2). A drawback to multiday extraction methods is that they could add variation and greater potential for error.
The ocean absorbs approximately 26 million tons of CO 2 from anthropogenic sources each day. 34,35 As a result, oceans are acidifying and deoxygenating, causing a decrease in phytoplankton populations due to their sensitivity to environmental changes. 1,35 This has implications for species diversity, distribution, phytoplankton productivity, and marine food web structure, of which phytoplankton form the basis. 35,36 It is predicted that carbon fixation will decrease and selection will favor smaller over larger phytoplankton due to lower nutrient obtainability. 10 Our understanding of the chemical ecology of phytoplankton and marine bacteria is limited, with the bulk of seawater metabolites yet to be chemically characterized. 22 Doing so is essential for understanding metabolic turnover, reactivity, and chemical transformation to establish a baseline of carbon flux through the ocean and to monitor any changes caused by anthropogenic influences.