Use and detection of a vitamin B1 degradation product yields new views of the marine B1 cycle and plankton metabolite exchange

ABSTRACT Vitamin B1 (thiamin) is a vital nutrient for most cells in nature, including marine plankton. Early and recent experiments show that B1 degradation products instead of B1 can support the growth of marine bacterioplankton and phytoplankton. However, the use and occurrence of some degradation products remains uninvestigated, namely N-formyl-4-amino-5-aminomethyl-2-methylpyrimidine (FAMP), which has been a focus of plant oxidative stress research. We investigated the relevance of FAMP in the ocean. Experiments and global ocean meta-omic data indicate that eukaryotic phytoplankton, including picoeukaryotes and harmful algal bloom species, use FAMP while bacterioplankton appear more likely to use deformylated FAMP, 4-amino-5-aminomethyl-2-methylpyrimidine. Measurements of FAMP in seawater and biomass revealed that it occurs at picomolar concentrations in the surface ocean, heterotrophic bacterial cultures produce FAMP in the dark—indicating non-photodegradation of B1 by cells, and B1-requiring (auxotrophic) picoeukaryotic phytoplankton produce intracellular FAMP. Our results require an expansion of thinking about vitamin degradation in the sea, but also the marine B1 cycle where it is now crucial to consider a new B1-related compound pool (FAMP), as well as generation (dark degradation—likely via oxidation), turnover (plankton uptake), and exchange of the compound within the networks of plankton. IMPORTANCE Results of this collaborative study newly show that a vitamin B1 degradation product, N-formyl-4-amino-5-aminomethyl-2-methylpyrimidine (FAMP), can be used by diverse marine microbes (bacteria and phytoplankton) to meet their vitamin B1 demands instead of B1 and that FAMP occurs in the surface ocean. FAMP has not yet been accounted for in the ocean and its use likely enables cells to avoid B1 growth deficiency. Additionally, we show FAMP is formed in and out of cells without solar irradiance—a commonly considered route of vitamin degradation in the sea and nature. Altogether, the results expand thinking about oceanic vitamin degradation, but also the marine B1 cycle where it is now crucial to consider a new B1-related compound pool (FAMP), as well as its generation (dark degradation—likely via oxidation), turnover (plankton uptake), and exchange within networks of plankton.

B-vitamin (3,6,7), and the necessity for B-vitamins is tied to their roles as cofactors in enzymatically driven reactions, translation regulatory elements that bind to riboswitches, or reactive oxygen quenching antioxidants (8)(9)(10)(11). B-vitamins are greatly understudied in the ocean relative to carbon, nitrogen, phosphorus, and iron; how ever, interest in them is reemerging (12,13). Broadly, marine plankton (and cells in general) meet their B-vitamin demands by: (i) making required vitamin de novo (prototrophs) and/or (ii) acquiring extracellular vitamin (auxotrophs). The latter lifestyle requires co-existence with the prior to some degree and is surprisingly common among marine plankton (14)(15)(16). The prevalence of the auxotrophic lifestyle is likely due to the elemental or energetic cost advantages of bypassing de novo vitamin synthesis and raises questions about how plankton meet their vitamin demands and stay alive.
Exogenous vitamers, vitamin-related compounds such as precursors or degradation products, are an alternative to intact vitamin that auxotrophs can use to meet their vitamin requirements. In the case of vitamin B1 (thiamin; called B1 herein), biosynthesis is a multi-step process that generates pyrimidine and thiazole precursor compounds as well as phosphorylated B1-the enzyme cofactor form of B1 (17). These compounds, along with B1 degradation products, are known to sustain select B1 auxotrophic plankton in culture (3,4,(18)(19)(20)(21). Recent work shows more specifically that key bacterio plankton and phytoplankton lineages in the ocean use select vitamers-bringing added attention to these alternatives to B1 (22)(23)(24)(25). As an example, B1 precursor 4-amino-5hydroxymethyl-2-methylpyrimidine (HMP) is required instead of B1 itself by lineages within the marine SAR11 bacterioplankton clade that dominates the surface ocean (22). Evidence of B1 deficiency in marine ecosystems at multiple trophic levels, from bacteria and phytoplankton to fish and birds (14,(26)(27)(28)(29), has also increased interest in vitamersas they may help to sustain populations at the base of the marine food web under B1 scarcity. Furthermore, vitamers appear to be readily exchanged, especially pyrimidine precursors based on recent experiments (24,30), which indicates flux to and from cells is likely commonplace in the ocean.
While appreciation for vitamers is growing, it remains a significant challenge to identify the complete B1 vitamer pool used by cells in nature, and this is important to resolve because knowing the true vitamin availability, as well as turnover, will ultimately allow prediction of plankton biomass and productivity in the ocean. Cell bioassays and liquid chromatography mass spectrometry (LC/MS) methods indicate vitamers occur in seawater (24,(31)(32)(33)(34). Solid phase extraction (SPE) and LC/MS have quantified specific vitamer molecules, including phosphorylated B1, pyrimidines, and thiazoles for which there are commercial standards (32,33). However, non-marine experiments and theoretical chemistry detail the formation of other B1 vitamers through alkaline or oxidative B1 degradation (35)(36)(37)(38)(39), and these have not yet been investigated in an oceanographic context nor considered key metabolite currency for marine microbes (40).
There is a notable gap in knowledge regarding in situ B1 degradation, cycling, and remodeling of degradation products in the ocean. Shedding light on these processes could reveal new microbial interdependencies, as well as explanations for the success of key populations impacting oceanic productivity and biogeochemistry. Our current understanding of oceanic B1 degradation is limited to scant laboratory and mesocosmbased exposures of dissolved B1 in seawater that highlight solar irradiance as an important degradative factor, along with increasing temperature (25,34,41).
Intracellular degradation of B1 has been especially overlooked in marine plank ton yet could be an important source of vitamers to seawater and a process that impacts B1 per cell and trophic transfer (42,43). Some marine plankton salvage B1 from pyrimidine degradation products; however, this has been minimally examined (18,19,25). Primarily exogenous photooxidation of B1 in the dissolved phase of seawater has been considered (25,34), but this may also occur intracellularly as described in plants. Plants experiencing high oxidative stress exhibit an increased B1 biosynthesis response based on transcriptomic, proteomic, and enzymatic datasets (36,44). Oxidation of B1 in plants is proposed to yield diverse vitamers, including N-formyl-4-amino-5-aminomethyl-2-methylpyrimidine (FAMP) (36,37). The B1 salvage enzyme TenA_E in Arabidopsis exhibits high affinity for FAMP compared with B1 and other pyrimidine B1 vitamers in vitro (36). Thus, FAMP production and salvage of B1 from FAMP are expected within plant cells. In cells that cannot salvage B1 from FAMP, e.g., cells lacking TenA_E such as low-light Prochlorococcus or SAR11 clade affiliates (22,45), possibly the vitamer is released into the environment as "overflow" (46); however, this has not been demonstrated directly. An intermediate in the pathway of salvaging B1 from FAMP ( Fig. 1), 4-amino-5-aminomethyl-2-methylpyrimidine (AmMP), is useful for marine haptophytes (phytoplankton) to meet their B1 demands (25) and is expected to be used by diverse marine bacterioplankton, based on genomes and metagenomes containing genes coding for TenA_C-the isoform of TenA that converts AmMP to HMP (25,36,47). Additionally, B1 auxotrophic marine phytoplankton, including cosmopolitan chlorophytes Ostreococcus and Micromonas spp., possess TenA and thus may also use AmMP or FAMP (24,48). AmMP has also been detected in seawater (32), suggesting B1 salvage and likely degradation occurs in the ocean. No equivalent tests or data are available for FAMP; thus, a component of the marine B1 cycle and potentially useful vitamer pool may be unaccounted for in the ocean.
Intrigued by the potential importance of FAMP in the ocean, we hypothesized the following: (i) exogenous FAMP can be used by marine plankton with TenA_E to salvage B1 from FAMP and grow, (ii) FAMP and genes for FAMP use in plankton are present in the ocean, and (iii) marine plankton are sources of FAMP. These hypotheses were tested in experiments with cultures of marine phytoplankton and bacterioplankton, searches for TenA sequences within the TARA Oceans Expedition dataset, and LC/MS-based measurements of FAMP in seawater and cell biomass.

Isolate growth conditions
Axenic cultures of B1-auxotrophic Ostreococcus lucimarinus CCE9901 were used for growth experiments as it possesses TenA and grows on B1 vitamers in medium without B1 (23,24,48). CCE9901 was maintained on L1 medium (49) (56,57) at room temperature before harvesting in stationary phase through filtration on a 0.22-µm polycarbonate filter for vitamin/vitamer extraction. Spent media and media blanks samples were collected in amber vials and frozen at −20°C until extracted.

BLASTp interrogation of reference genomes and TARA Oceans data
The presence/absence of B1-related proteins in specific isolate genomes, e.g., CCE9901, DSS-3, E-37, etc., was determined using online Uniprot BLASTp (59) searches with default parameters and restricting searches to a specific strain of interest. Local BLASTp searches were conducted to search for B1-related transporters in downloaded reference genomes for the same isolate strains. TenA sequences were aligned using MUSCLE (60) and the default settings; some sequences included in the alignment were anno tated as TenA fragments, although similar in length to full sequences (Uniprot ID's provided): A0A7S4E370 Pelagomonas calceolata, F0YQK2 Aureococcus anophagefferens. A phylogenetic tree of TenA amino acid sequences was generated using MEGA11 (61). The evolutionary history was inferred by using the Maximum Likelihood method and Whelan and Goldman model (62). The percentage of trees in which the associated taxa clustered together is shown next to the branches, and 500 replicated trees were generated. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Joining and BioNJ algorithms to a matrix of pairwise distances estimated using the Jones-Thornton-Taylor (JTT) model and then selecting the topology with superior log likelihood value. A discrete gamma distribution was used to model evolutionary rate differences among sites [five categories (+G, parameter = 2.5862)]. The rate variation model allowed for some sites to be evolutionarily invariable [(+I), 0.40% sites].
The Ocean Gene Atlas (OGA) v2.0 online resource (63) was used to search with BLASTp for B1-related protein sequences within metagenomic, metatranscriptomic, metagenome assembled genome (MAG), and single-cell amplified genome (SAG) TARA Oceans data. The prokaryotic databases searched were: OM-RGCv2+G, OM-RGCv2+T, and BAC_ARC_MAGs, while the eukaryotic databases searched were MATOUv1+G, MATOUv1+T, and EUK_SMAGs. Initially Rubrobacter xylanophilus TenA_C and TenA_E reference sequences (36) were used for BLASTp searches. TenA_C and TenA_E sequen ces from Thalassospira spp. were evident in the results; thus, Thalassospira reference sequences from Uniprot were retrieved and used for BLASTp searches for prokaryotic TenA sequences. A more stringent E-value of −17 was used instead of the default −10 as fewer non-specific hits were returned based on manual inspection using BLASTp searches against the NCBI NR database. OGA BLASTp searches for eukaryotic TenA_C and TenA_E used Emiliania huxleyi strain PLY M219 sequences retrieved from the Marine Microbial Eukaryotic Transcriptome Sequencing Project (MMETSP) database. In some cases, a sizeable percentage of recovered hits from the OGA BLASTp searches could not be taxonomically assigned. Manual inspection of these sequences confirmed relatively low percent sequence identity to prokaryotic or eukaryotic TenA sequences within the NCBI NR database (27 July 2022).

Seawater sampling
Near surface water seawater samples were collected from different regions of the North Atlantic Ocean and adjacent estuarine waters. Seawater samples from the Scotian Shelf and Slope (SSS) were obtained from 5 m depth during 9 October 2020 and 14 September 2021 in collaboration with the Atlantic Zone Monitoring Program, at station HL02 (on shelf ) (Lat: 44.2663; Lon: −63.3159) and station HL012 (off shelf ) (Lat: 41.4100; Lon: −60.6774). One liter of water was collected from the CTD rosette bottles in amber bottles rinsed with sample water, then gently vacuum filtered through 0.2 µm pore-size nylon filters. Samples were protected from light during filtration. Dissolved samples were frozen at −20°C in acid-washed, MilliQ water-rinsed, and sample-rinsed amber High-den sity polyethylene (HDPE) bottles until processing.
Neuse River Estuary (NRE) water was collected from ~0.5 m at stations NRE0 and NRE180 in collaboration with the University of North Carolina at Chapel Hill Institute of Marine Sciences (UNC-IMS) Neuse River Estuary Modeling and Monitoring Project (ModMon) program (64) on 11 November 2021. Prior to water collection, opaque amber sampling bottles were cleaned with 0.1N HCl, then rinsed with High performance liquid chromatography (HPLC) grade MillliQ water, HPLC grade methanol, and rinsed again with MillliQ water. Collected water was stored at room temperature until filtration the next day. Filtration units and collection bottles for filtrate were rinsed with methanol and MilliQ prior to use. NRE water was serially filtered through 90 µm Nitex mesh, 3 µm pore-sized polycarbonate filters (Isopore, Millipore), and 0.2 µm pore-sized nylon filters. In total, six bottles of 200 mL 0.2 µm filtrate each were prepared per station and stored at −20°C in amber HDPE bottles.

Vitamin/vitamer capture and dissolved phase quantification
B1 and vitamers were captured using C18 solid phase extraction columns similar to previously published methods (31, 65) but with modifications as described below. Bottles with filtrate were thawed overnight at 4°C and then pH adjusted to 6.5 with molecular grade 1M HCl. The pH of SSS samples was not adjusted. Select samples had vitamers added for calculation of percent recovery using targeted LC/MS (Table S1) as described below (Table S2). Stocks of vitamer were made from the same primary stocks listed above for isolate experiments. Dried vitamer extracts were stored at −20°C until analysis when they were resuspended with buffer A (see below).
NRE environmental samples were spiked with 75 pM final concentration of 13Cthiamin (thiamine-(4-methyl-13C-thiazol-5-yl-13C3) hydrochloride (Sigma-Aldrich). SPE columns (Waters, WAT043345) were conditioned by soaking overnight in HPLC grade methanol and washed with 25 mL of HPLC plus grade water prior to use for capture of dissolved organics in seawater. Seawater (0.2 µm filtered) was passed through SPE columns in a dark 4°C room at a flow rate of ~1 mL min −1 . After pumping was complete, the SPE columns were washed with 100 mL HPLC plus grade water, purged of water, and stored sealed at −80°C until further processing. Columns were thawed at room temperature for 30 min, placed in a vacuum manifold (Waters), and then washed with another 100 mL HPLC grade water before gently purging residual water and eluting with 35 mL methanol. All vacuum manifold steps were performed with less than 5 in. Hg vacuum applied, resulting in a flow rate or approximately 5 mL min −1 . Solvent was removed using a roto evaporator (Centrivap, Labconco) for 12 to 24 h.
For SSS environmental samples (20 mL) and Pibocella spent media samples (~14 mL), vitamers were extracted using 500 mg and 100 mg HyperSep C18 SPE columns (Thermo Scientific, 03-251-258), respectively. Columns were preconditioned by passing two times 0.85 mL methanol then 0.85 mL MilliQ water before sample was loaded onto the column at 1 mL min −1 in a dark room. Columns were then washed with 0.850 mL MilliQ and eluted with 1 mL methanol. Solvent was removed using a roto evaporator (Vacufuge, Eppendorf, Mississauga, ON, Canada).
Ostreococcus spent media samples were thawed at room temperature for 3 h and spiked with 80 pM final concentration of 13C-thiamine hydrochloride (4,5,4-methyl-13C3, 97%; Cambridge Isotope Laboratories). One gram C18 SPE columns (Waters, WAT043345) were conditioned with 5 mL methanol followed by 5 mL HPLC grade water (Optima, ThermoFisher). Spent media was then loaded in dim light at a flow rate of ~1 mL min −1 . Columns were washed with 50 mL HPLC grade water and eluted with 5 mL methanol. Solvent was removed using a roto evaporator (Vacufuge, Eppendorf, Mississauga, ON, Canada).
All reported concentrations (Table 3 to 5) are corrected for percent recoveries, which are described in Table S2. To determine these recoveries in NRE samples, select representative samples were spiked with vitamers: FAMP (150 pM), cHET (150 pM), HMP (150 pM), and HET (40 pM). Percent recovery of vitamers in SSS samples was determined by spiking HMP (400 pM), FAMP (400 pM), and HET (80 pM) into HL2 and HL12 samples, calculating the amount recovered after subtracting the concentration of endogenous analytes in unspiked sample (Table S2). Percent recovery for FAMP in all environmental samples regardless of extraction method was consistent and roughly 50% (see Table S2), which is expected based on recoveries of similar compounds using similar methods (65). Percent recovery of SSS samples was applied to dissolved Pibocella samples because they were subject to the same SPE method. Ostreococcus samples were assumed to have a 50% recovery rate for FAMP based on the consistency of this recovery rate across SPE methods (see Table S2).

Particulate vitamin extractions
Pibocella and Ostreococcus biomass samples were extracted following Heal et al. (15), except that solvent was removed by a roto evaporator (Eppendorf, Mississauga, ON, Canda) instead of an N 2 gas evaporator. The entire procedure was conducted in a dark room with red LED lights; samples were kept on ice whenever possible. The Ostreococ cus samples were spiked with 2 pmol 13C-thiamine hydrochloride (4,5,4-methyl-13C3, 97%; Cambridge Isotope Laboratories) prior to extraction. Percent recoveries were not assessed for particulate analyses.
Samples were resuspended in 100, 200, or 400 µL HPLC buffer A (20 mM ammonium formate, 0.1% formic acid) and diluted as required. Samples were grouped by sample type (SS, NRE, bacterial culture particulate and bacterial culture media, and Ostreococcus culture particulate and media), and Quality Control (QC) pools were created for each matrix grouping by combining equal portions of each sample within that sample type. Vitamins/vitamers were quantified in each sample using the standard addition method. Calibration curves were prepared with authentic metabolite standards for each matrix grouping (sample set) using the corresponding QC as a matrix. Duplicate injections were performed with 0, 25, 50, and 250 fmol of B1, HMP, cHET, FAMP, and AmMP and 0, 5, 10, 50 for HET added.
Data analysis methods were adapted from Heal et al. (15) and Boysen et al. (66) Briefly, raw files generated with Xcalibur software (ThermoFisher) were uploaded into Skyline Daily (University of Washington), and the transitions with the best signal to noise and lowest interference were selected for quantification purposes. Summed peak areas were exported and processed in Excel or R. Some thiamin (B1) peaks were normalized to the heavy internal standard peak, thereby reducing instrument and sample preparation variability. Normalization of other compounds to heavy B1 did not reduce variability. Vitamins/vitamers were quantified from these peak areas using the standard curves generated from their respective QC pools. Limits of detection and quantification were determined according to the Guidelines for Data Acquisition and Data Quality Evaluation in Environmental Chemistry (67) and are provided in Table S2. Additionally, samples with concentrations that fell between Limit of detection (LOD) and limit of quantification (LOQ) were further visually inspected and analyzed in a batch per batch method based on the following criteria modified from Boysen et al. (66). Concentrations were reported in samples that fell below the calculated LOQ if, (i) the peak has the same retention time (±0.2 min) as the authentic standard, (ii) two daughter fragments were present with co-occurring peaks, (iii) daughter fragments were present in same order of intensity as authentic standard, and (iv) the integrated peak area was at least two times greater than the average peak found in the blanks in the appropriate retention time window.

Statistical analyses
Testing of significant differences in maximum yield of O. lucimarinus CCE9901 cultures provided B1 or vitamers at different concentrations was done using log-transformed Chl-a fluorescence data and two-way ANOVA with Tukey-Kramer post hoc testing in Prism (GraphPad). Significant differences in the maximum yield of Roseobacteria (DSS-3, E-37) cultures provided vitamin/vitamers, versus no add negative controls was assessed using paired two-tailed t-tests in Prism. Culture experiments were pragmatically small in scale with triplicate cultures per treatment; as a result, data normality was assumed for parametric tests.
Isolates representing marine Roseobacteria, R. pomeroyi DSS-3 and S. stellata E-37, were used in comparable experiments to test their ability to use FAMP and AmMP. These strains were ideal for comparative experiments as they have publicly available complete genome sequences, grow on defined medium, and are B1 auxotrophs that vary in their possession of TenA_C and TenA_E with DSS-3 possessing only TenA_C and  Table 1). Considering their gene repertoires, DSS-3 was expected to grow on AmMP but not FAMP while E-37 would grow on both. Growth tests partly confirmed our hypotheses, as DSS-3 and E-37 grew on exogenous AmMP as expected but unexpectedly neither strain grew on exogenous FAMP (Fig. 4). Notably, these Roseobacteria, as well as O. lucimarinus and other marine phytoplankton and bacteria do not clearly possess YlmB, the best studied enzyme known to convert FAMP to AmMP in Bacillus (38) ( Table 1).

TenA_C and TenA_E in the global ocean
Recent studies have noted select marine bacterioplankton and haptophytes possess TenA (14,22,25); however, newly available expansive in situ genetic datasets can provide insight into TenA in marine plankton globally-and accordingly their potential use of FAMP and AmMP. We searched TARA Oceans metagenomic, metatranscriptomic, MAG, and SAG data for TenA_C and TenA_E sequences (Table S3; see Tables S4-S18 at https:// doi.org/10.5061/dryad.4b8gththk), revealing that diverse prokaryotes [mostly eubacte ria, few archaea (<1%)], and eukaryotes possess at least one TenA isoform (Fig. 5). tenA_C was approximately eight times more common than tenA_E in prokaryotic metagenomic (OM-RGCv2) data (Table 2), pointing to a putatively larger investment in use of AmMP than FAMP in B1 salvage. Exogenous use of AmMP over FAMP by marine eubacteria may be even larger, given that some possessing TenA_E cannot use exogenous FAMP but can use AmMP (Fig. 4). Most TenA_C sequences taxonomically affiliated with Proteobacteria (Alphaproteobacteria, Gammaproteobacteria), Firmicutes, and Cyanobacteria hits with low E-value scores revealed nebulous annotations for the sequences that were related to M20 dimer proteins (e.g., ArgE and DapE), not YlmB. "X" = no hit was returned using BLASTp.
Research Article mBio (Prochlorococcus). Several tenA_C sequences with coarse taxonomic assignment were noted ( Fig. 5; marked "other bacteria 1 and 2"). Searches of these sequences using BLASTp against the NCBI NR database recovered a mixture of matches with highest sequence identity (<90%) to largely Chloroflexi, SAR202, and Halomonas sequences, but some exhibited low sequence identity match to anything in the NCBI NR database (see Tables S4-S18 at https://doi.org/10.5061/dryad.4b8gththk). Rhodobacteraceae and Candidatus Pelagibacter sp. IMCC9063 tenA_C sequences were notably abundant (jointly 52%)-the latter in particular fits with the prior observation that Pelagibacter sp. IMCC9063 is the only Pelagibacter isolate genome possessing TenA (22). The taxa distribution of sequences retrieved from metatranscriptomes was very similar, demonstrating active gene transcription in generally similar taxa proportions as in metagenomes (Fig. 6). Furthermore, many prokaryotic MAGs (n = 159) possessed tenA_C and represent 40 different eubacterial orders, including several that were not well resolved from bulk metagenome searches with the OGA tool, e.g., SAR202, Chloroflexi, Cyanobacteria (Trichodesmium) as well as Proteobacteria (Fig. 7). Not all prevalent marine eubacteria and archaea possess TenA_C (or any TenA) based on these search resultse.g., SAR11 affiliates-but this was expected based on prior isolate genome analyses (14,22,45) and overall points to an advantage gained by select populations possessing TenA. Furthermore, representatives of biogeochemically impactful lineages with widely different lifestyles possess TenA_C (Fig. 5 to 7); for example, photoautotrophic high-light Prochlorococcus and Trichodesmium spp. (Cyanobacteria) that occupy the surface ocean (the latter being a significant global contributor to nitrogen fixation), as well as hetero trophic SAR202 representatives (Chloroflexi) that thrive in aphotic waters of the deep ocean and influence global sulfur cycling (68).
Recovered tenA_E sequences from the TARA Oceans data were largely affiliated with Proteobacteria (Alphaproteobacteria, Gammaproteobacteria) and Candidatus for triplicate cultures. Unique letters above columns denote significant differences between treatment groups based on two-tailed paired t-tests.

Research Article mBio
here as possessing TenA were not previously recognized as such, outside of select Chlorophytes (e.g., Ostreococcus) and haptophytes (25,48). Select eukaryotic and prokaryotic populations possess TenA_C and TenA_E (Fig. 2) (25,36). Furthermore, some populations possess multiple copies of tenA_C or tenA_E (e.g., Thalassospira sp.; P. calceolata) (Fig. 2); however, it appears most isolate genomes do not possess multiple tenA copies ( Fig. 2; Table 1) (14,22,25). The advantages of these different genotypes (possessing tenA_C and tenA_E, as well as multiple copies of either) are unknown. Speculatively, possession of multiple copies of tenA_C or tenA_E may lead to more protein copies per cell, an increased rate of B1 salvage from pyrimidine vitamers, and a competitive advantage over cells with only a single copy of tenA_C or tenA_E.

Detection of FAMP in seawater and marine plankton cultures
Particle free near-surface water samples from estuarine and marine monitoring stations contained FAMP at mean concentrations, adjusted for percent recovery, ranging from 13.5 to 36.1 pM. FAMP was present at all sampled stations and was comparable in concentration with B1 pyrimidine precursor HMP, as well as other measured B1 vitamers and B1 (Table 3). Albeit a limited dataset, higher FAMP concentrations occurred in waters beyond the continental shelf in the western North Atlantic (HL12) and the marine "end-member" station within the Neuse River Estuary (NRE180) ( Table 3; Table S2).  Tables S7, S11, S14, and S18 at https://doi.org/ 10.5061/dryad.4b8gththk.

Research Article mBio
Following confirmation of dissolved FAMP in seawater, we sought to determine whether plankton could be sources of FAMP. Thus far, purely abiotic processes have been considered the drivers of B1 degradation in the ocean, e.g., photooxidation of B1 dissolved in seawater (12,25,34). We hypothesized that planktonic cells are sources of FAMP, given that TenA_E (also TenA_C) occurs in B1 prototrophic plankton (Fig. 2), and FAMP generation is thought to occur in plants due to B1 oxidation (36,37). To test this hypothesis, we examined particulate and dissolved (0.22 µm prefiltered seawater) phases of axenic bacterioplankton and phytoplankton cultures.
Flavobacterium Pibocella sp. was used to test this hypothesis as it was co-isolated along with diatom cells from coastal North Atlantic water and used previously in broader metabolite exchange experiments between bacteria and diatoms (Bertrand et al. per. comm.). Spent medium as well as Pibocella biomass collected from cultures grown in the dark contained FAMP ( Table 4). The mineral growth medium (LNHM, see Materials and Methods) alone contained FAMP (~40 nM), likely due to abiotic degradation of B1 added when initially making the medium (~590 nM B1 final concentration). Autoclaving the medium or its storage at room temperature with ambient room light exposure may have facilitated degradation. Nonetheless, spent medium contained 18 ± 11 nM more FAMP than the starting medium (Table 4). Particulate samples collected on day 4, during stationary phase, were also rich in FAMP (~15 nM), pointing to intracellular production and/or import and accumulation (Table 4). AmMP also was detected in particulate samples (~15 nM) but markedly less so in the starting medium and spent medium-contrasting with FAMP (Table 4). These results suggest FAMP flux from cells and exchange between cells are likely in the ocean-e.g., Pibocella and diatoms or other plankton, such as Ostreococcus, capable of using exogenous FAMP (Fig. 3). FAMP (14 ± 4 HL12 were assessed, we do not present standard deviations. All measurements are corrected for percent recovery. X, measurements of the given compound were not attempted; nd, compound was present at levels lower than our limit of detection; and nq, where the compound was present at levels below our limit of quantification.  (Table 4), further highlighting marine bacterial cells as sources of B1 degradation products to their environment. We also investigated whether O. lucimarinus CCE9901, a B1 auxotroph that requires exogenous B1 or vitamers to survive, produced FAMP and AmMP when it is grown on B1 or other vitamers (HMP and cHET). Both FAMP and AmMP were detected in CCE9901 biomass in cultures supplied with only B1 or vitamers (HMP and cHET) ( Table  5). Additionally, though AmMP quantification is difficult (Table S2), based on compar isons between the Pibocella and CCE9901 results, there is some evidence that the FAMP:AmMP ratios vary between populations and/or growth phases, which deserves further investigation (Tables 4 and 5; Table S2).

DISCUSSION
Here we identify an abundant, previously unrecognized, B1 vitamer pool (FAMP) in the ocean and new components of the marine B1 cycle (FAMP generation, use) that require future consideration with respect to plankton interactivity and ecology. Experimental evidence (O. lucimarinus CCE9901) and culture-independent genetic evidence ( Fig. 3 to 7) point to extensive use of FAMP by marine eukaryotic phytoplankton. In contrast, more of the prevalent bacterioplankton surveyed have the genetic potential to use AmMP than FAMP. This is based on fewer TARA Oceans tenA_E sequences, which code for the protein that uses FAMP as its primary substrate (Table 2), as well as the results of our Roseobacter growth experiments (Fig. 4). Collectively this points to a newly realized interconnectivity between abundant eukaryotic phytoplankton and other cells, especially bacterioplankton-like Flavobacteria (Table 4)-that produce and release FAMP and AmMP. On the other hand, abundant bacterioplankton capable of salvaging B1 from pyrimidines (those possessing TenA) seem to have evolved to more commonly use AmMP (Fig. 5; Table 2), a product of FAMP deformylation by co-occurring plankton ( Fig. 1) or possibly a yet-to-be identified abiotic process.
Why TenA_E is not more common in marine bacterioplankton and why there is mixed use of FAMP and AmMP among marine plankton overall ( Fig. 2 and 5) is not clear. Potentially there are significant costs associated with producing and maintaining TenA_C and/or TenA_E, otherwise it would seem advantageous for all cells to possess TenA_E and be able to use FAMP and AmMP. Transport costs may also factor in, but this remains unclear-e.g., what transporters are crucial for FAMP and/or AmMP uptake (17,70)? At a basic level, the ability to use exogenous FAMP or AmMP offers a competitive advantage when B1 or HMP is unavailable, which may occur in regions of the ocean, given that a wide diversity of plankton are expected to use exogenous B1 and HMP (5,12,14,22,23,48). Alternatively, use of FAMP or AmMP in addition to use of available B1 and other vitamers may simply boost growth rates or metabolism over populations that cannot use FAMP or AmMP. Presently, too few measurements of FAMP, AmMP, and B1, along with growth rates, are available to address this directly (Table 3) (32).  Table S2). b Values are in picomolar (pM). Abbreviations: 'Exp' = exponential growth phase; 'Stat' =944 stationary growth phase. c nd = not detected (i.e., below the LOD).
Some TenA-possessing cells may salvage B1 only from autochthonous B1 degrada tion, i.e., these cells regenerate B1 from degradation happening within the cell and do not import exogenous FAMP or AmMP. E-37 appears to fit this phenotype, as it possesses TenA_E but did not use exogenous FAMP in growth experiments (Table 1; Fig. 4). A working hypothesis is that E-37 lacks an FAMP transporter to import and use FAMP. To this point, the only B1-related transporter found in E-37 was ThiB (Table 1), which is best linked to B1 transport (71). We speculate high-light adapted Prochlorococcus spp. are phenotypically similar and do not use exogenous AmMP since they lack any known B1 transporter but possess TenA_C for intracellular salvage of degraded B1 (Fig.  5) (45). Clarifying the genotypes that can or cannot use exogenous FAMP and AmMP, as well as those that export high versus low amounts of these vitamers, is important to elucidate in the future and likely the identification of FAMP and AmMP transporters will be illuminating on this front.
The detection of FAMP in seawater, as well as bacterial and phytoplankton cultures (Tables 3 to 5), is strong evidence that B1 degradation is commonplace in the ocean. Moreover, plankton must be considered as important sources of B1 degradation (Tables 3 to 5) in addition to the purely abiotic reactions in seawater that have been deemed important thus far (12,25,34,41). Dark generation of FAMP and AmMP by Pibocella cells (Table 4) illustrates the importance of aphotic B1 degradation within the marine B1 cycle and expands thinking beyond just photodegradation in the surface ocean (12,25,34). In agreement, characteristic taxa of the deep ocean, e.g., SAR202 and Chloroflexi representatives, possess TenA proteins for use of FAMP and AmMP (Fig. 5 to 7). Thus, B1 degradation and production of FAMP, AmMP is expected-be it within cells or extracellularly-in the vast and dark ocean interior. Altogether, B1 degradation in the ocean is likely more widespread than previously considered and elevates the need to better quantify it and its impact on global ocean productivity and plankton community composition.
We hypothesize that reactive oxygen species (ROS) production and subsequent B1 oxidation widely occur in marine plankton. Both processes and their interplay have been best studied in terrestrial plants (36,37,44,72) but not considered in marine plankton or the ocean. ROS production is considered widespread and continuous in the ocean due to aerobic respiration and metabolisms, photosynthetic activity, as well as photooxidation of organics (73)(74)(75)(76). As a result, the potential for B1 oxidation by ROS is high in the ocean, especially within cells where concentrations of B1 and ROS will be locally high. Extracellular B1 degradation by ROS in seawater is also possible; however, reaction kinetics should be evaluated as ROS and vitamin concentrations are lower in the dissolved phase.
Thinking about Earth's broad biogeochemical evolution, cyanobacterial oxygenation and subsequent promotion of aerobic metabolism likely increased B-vitamin chemical complexity and vitamin interdependencies among microbes, including cycling of B1 oxidation products FAMP and AmMP. Looking to the future, linkages between ROS-form ing cell stressors (73,77), climate change-related factors, e.g., regional to global shifts in temperature, irradiance, salinity (78,79), and rates of FAMP and AmMP production (also B1 degradation) should be investigated. These factors would not only alter the availability of B1 and vitamers within seawater and plankton networks but also impact the amount of B1 per cell (23,25,80) and thus trophic transfer of B1 in marine food webs (43).
In conclusion, speculation on vitamers in the ocean is long running (3,4,12,18,19), and here we add to a broader effort to identify precursors and their use to plankton (16,22,23,25,48,58) by demonstrating the use of FAMP by key marine plankton and its occurrence in the ocean. Previously, the compound was given little consideration as a useful exogenous nutrient or as an exchanged metabolite between marine plankton (4,5,12,40). Many details of FAMP formation, use, and exchange are unknown and deserve future exploration as they likely will help explain niches of important micro bial populations, cell interactions, altered cell quotas of vitamin, and fundamentals of B-vitamin degradation. We anticipate other B-vitamins and organic nutrients undergo oxidative degradation similar to B1 (37). Confirming the resulting compounds is a significant challenge but important to address in order to fully explain marine plankton metabolite exchange (40,81,82) and the co-existence of diverse populations (83) and identified genotypes (12,14,22). We thank Anna Gleason for assistance with culturing. Jeremy Braddy, Amy Barten felder, Hans Paerl, and the MODMON program at UNC-IMS assisted with seawater collection and/or provided access to seawater from the Neuse River Estuary.
The authors have no financial or non-financial competing interests to report.

DATA AVAILABILITY
Data used in this study that are not directly presented in figures or tables are in the supplemental material.

ADDITIONAL FILES
The following material is available online.