Carboxythiazole is a key microbial nutrient currency and critical component of thiamin biosynthesis

Almost all cells require thiamin, vitamin B1 (B1), which is synthesized via the coupling of thiazole and pyrimidine precursors. Here we demonstrate that 5-(2-hydroxyethyl)-4-methyl-1,3-thiazole-2-carboxylic acid (cHET) is a useful in vivo B1 precursor for representatives of ubiquitous marine picoeukaryotic phytoplankton and Escherichia coli – drawing attention to cHET as a valuable exogenous micronutrient for microorganisms with ecological, industrial, and biomedical value. Comparative utilization experiments with the terrestrial plant Arabidopsis thaliana revealed that it can also use exogenous cHET, but notably, picoeukaryotic marine phytoplankton and E. coli were adapted to grow on low (picomolar) concentrations of exogenous cHET. Our results call for the modification of the conventional B1 biosynthesis model to incorporate cHET as a key precursor for B1 biosynthesis in two domains of life, and for consideration of cHET as a microbial micronutrient currency modulating marine primary productivity and community interactions in human gut-hosted microbiomes.

to use the newly detected precursor(s) found in seawater and produced by de novo B1-synthesizing planktonstrongly suggesting that the compound(s) is thiazole-related 18 .
Intrigued that these marine picoeukaryotic phytoplankton potentially use a novel thiazole precursor, we noted with interest that carboxythiazole, 5-(2-hydroxyethyl)-4-methyl-1,3-thiazole-2-carboxylic acid (cHET), specifically phosphorylated cHET (cHET-P), is produced by bacterial thiazole synthase 19 , and that the thiazole synthase of plants and yeast similarly generates cHET-ADP (adenylation of the precursor is used rather than phosphorylation) 20,21 . Further, cHET-P is a functional substrate for bacterial thiamin monophosphate synthase (ThiE) 22 . This rigorous biochemical evidence points to cHET as a core component of de novo B1 biosynthesis; nonetheless, the vast majority of B1-related research and reviews to date make no mention of cHET (or phosphorylated or adenylated forms) and instead describe only synthesis and use of HET(−P) 5,7,9,[23][24][25] . Given the apparent importance of cHET in B1 biosynthesis, and the ThiM (thiazole kinase) requirement for marine picoeukaryotic phytoplankton to use the newly detected thiazole precursor 18 , we hypothesized that cHET is a useful exogenous thiazole B1 precursor for phytoplankton.

Results
Marine picophytoplankton use exogenous cHET to meet their B1 demands. In experiments with vitamin-B1 limited Ostreococcus tauri RCC745, a ThiM-possessing marine picoeukaryotic phytoplankton unable to use HET 18 , low additions of cHET (plus the pyrimidine precursor HMP) promoted growth (Fig. 1), confirming our hypothesis and revealing that O. tauri is adapted to use minute amounts of cHET dissolved in seawater. The cHET stock used in these experiments contained no detectable B1 cross-contamination, but did contain trace (0.5%) amounts of HET contamination based on selected reaction monitoring mass spectrometry (LC-SRM) (Supplementary Table S1).
Unlike wildtype O. tauri, a ∆thiM (lacking ThiM) mutant line did not grow on supplied cHET (Fig. 1B) indicating that cHET utilization requires ThiM, which agrees with prior results showing that O. tauri RCC745 requires ThiM in order to use the thiazole precursor produced by B1-synthesizing marine plankton 18 . Besides O. tauri, another cosmopolitan picoeukaryotic marine phytoplankton organism, Bathyococcus sp. RCC4222, also grew on supplied cHET (and HMP) under B1-limiting conditions ( Supplementary Fig. S1), showing that use of exogenous cHET is a more general phenomenon in marine picoeukaryotic phytoplankton, particularly the Prasinophyceae.
ThiM prevalence in human microbiomes and cHET use by Escherichia coli. Diverse organisms, including freshwater algae, enteric bacteria, human pathogens, and terrestrial plants also possess ThiM 11,14,15,18 and hence might similarly salvage exogenous cHET for use in B1 synthesis. Bioinformatic surveys revealed that metagenomes from the human microbiome contain ~10× higher relative abundance of ThiM sequences than marine and terrestrial metagenomes (Supplementary Table S2), prompting the hypothesis that ThiM-possessing human-associated bacteria use exogenous cHET, in the same way as picoeukaryotic marine phytoplankton (Fig. 1).
Human-associated enteric bacterium Escherichia coli K-12 as well as >400 other E. coli strains possess ThiM 15 (Supplementary Table S3), making Escherichia coli a suitable model for testing our hypothesis. Experiments with an E. coli mutant lacking ThiG (∆thiG), the enzyme that synthesizes the thiazole precursor of B1 in de novo biosynthesis, showed the bacterium is also adapted to use low concentrations of exogenous cHET, specifically down to subpicomolar concentrations (Fig. 2). In contrast, ~1 million times more (>100 nM) HET was necessary to support comparable growth (Fig. 2), which confirms our hypothesis and is also consistent with prior experiments indicating that micromolar concentrations of HET are needed to sustain E. coli growth 15  but only at relatively high extracellular concentrations of HET, which presumably enters the cell via low-affinity transporters and/or diffusion.
cHET use by the terrestrial plant Arabidopsis. Broadly contextualizing microbial use of cHET, we also tested whether ThiM-possessing terrestrial plants such as Arabidopsis 14 can use exogenous cHET. Growth experiments with Arabidopsis, using the wild-type and a mutant (tz-1) unable to synthesize thiazole precursor, confirmed that Arabidopsis can use cHET because the mutant grew well when given high concentrations of cHET ( Fig. 3). In contrast to picoeukaryotic phytoplankton and E. coli (Figs 1, 2), equivalent concentrations of HET also sustained growth of the Arabidopsis mutant ( Fig. 3). This result highlights a key difference between plants and (aquatic) microbes, in that the latter are equipped to salvage B1 from very low concentrations of exogenous precursors (Figs 1, 2), likely as a result of adaption within an environment where precursor(s) are a community currency circulated between producers and consumers.
Our results alter this paradigm as cHET is clearly useful for prokaryotic and eukaryotic microorganisms, and moreover is accessible at extremely low concentrations (Figs 1, 2). Specifically, acquisition of exogenous cHET sustains key primary producers in the ocean as well as important enterobacteria, presumably enabling them to bypass the energetic and/or elemental costs of de novo biosynthesis of cHET(−P). The ability of these cells, which are endemic to marine euphotic waters and the human body, to use very low exogenous cHET concentrations strongly suggests that cHET is bioavailable in nature and integral to interactions between B1 auxotrophs and B1-synthesizing microorganisms or hosts. However, direct evidence of this is lacking. Our observation that diverse organisms grow on cHET (Figs 1-3, Supplementary Fig. S1), alongside prior biochemical evidence of cHET-P generation by bacteria and plants 19,21 , solidifies cHET-P as a fundamental and widely overlooked 5,7,9,14,[23][24][25] component of de novo B1 biosynthesis (Fig. 4). Establishing that cHET is central to de novo B1 biosynthesis pinpoints B1-prototrophic organisms as sources of the thiazole to co-occurring populations, i.e. key microorganisms that influence marine primary productivity (Fig. 1) or human health (e.g. Shiga-toxin producing E. coli (STEC) 26 possessing ThiM, e.g. E. coli STEC_O31; Supplementary Table S3). Our findings also improve understanding of B1 biosynthesis in general -a vital process for life on Earth, and target for industrial and biomedical applications with human impact, e.g. efforts to increase crop nutritive value or resilience 27,28 and to develop drugs targeting pathogenic microbes 29 . For clarity, we propose that HET be regarded as a degradation-derived precursor as it comes from B1 degradation 1,30 , whereas cHET is generated via the de novo biosynthesis pathway [19][20][21] (Fig. 4). Comparable descriptors also apply to pyrimidine B1 precursors 4-amino-5-aminomethyl-2-methylpyrimidine (AmMP; degradation product) and HMP (product of de novo biosynthesis) 1,10,30 .
In conclusion, prevailing views of thiazole precursor biosynthesis, use, and exchange require reassessment, as it is now evident that the role of B1 requirements in host-associated or aquatic microbiomes cannot be fully understood without consideration of the widely overlooked thiazole B1 precursor cHET(−P). Exchange of cHET and its influence upon interactions between taxa 31 , especially producers and consumers of cHET, deserves particular attention given that exogenous cHET acquisition is integral to the survival of microorganisms with environmental, industrial, and/or biomedical impacts, e.g. key marine microbial primary producers, E. coli, and other ThiM-possessing taxa 7 (Figs 1, 2; Supplementary Table S2). Since model organisms readily utilize exogenous cHET, it should be possible for future research to decipher the intricacies of cHET flux and its influence upon cell interactions.  18,32 . Antibiotics (penicillin G 50 μg mL −1 , streptomycin 200 μg ml −1 ) were added to prevent the growth of Marinobacter bacteria associated with O. tauri RCC745 cultures 18 . RCC745 cell growth and the absence of Marinobacter were determined on an Accuri C6 flow cytometer (Becton Dickinson) using SYBR Green Il staining. To start cHET bioassays, algal cells inoculated at 0.5 to 1 × 10 6 cells mL −1 were first grown for 7 days in B1-deplete ASW medium (ASW-B1). These B1-deprived cultures were used to inoculate ASW culture medium containing HMP (1 nM) and various concentrations of cHET at 0.5 to 1 × 10 6 cells mL −1 in triplicate. Similar experiments were performed to Escherichia coli growth experiments. E. coli JW5549 ΔthiG761::kan (Keio Collection) cells were cultured in LB medium. Cells were harvested via centrifugation (9 × 1000 g; 3 min), washed, and resuspended over three iterations in M63 minimal growth medium 33 lacking B1 and only 1 mM glucose as a carbon source to minimize the potential for vitamin contamination from glucose stock. Washed/resuspended cells were added to triplicate sterile 4.5 mL polystyrene tubes containing M63 B1-free test medium (at a ratio of 0.5 µL cells: 1.5 mL medium) with various concentrations of HET or cHET, B1 (as a positive control), or without amendment (as a negative control). Static culture tubes were incubated at room temperature, in the dark. Cultures were thoroughly vortexed before sampling after one and two weeks of incubation. Optical density (590 nm) of cultures was measured using a FLUOstar Optima Plate Reader (Bmg-Labtech) and clear 96-well plates. Cell abundances were determined from fixed (2% formaldehyde) and frozen (−80 °C) culture samples via SYBR green I (Molecular Probes, Eugene, OR, USA) staining and flow cytometry 34 using a FACS CANTO II (Beckton Dickenson, Heidelberg, Germany).

Plant growth experiments.
Arabidopsis thaliana thiazole-auxotrophic mutant (tz-1; ABRC stock number CS3375) 35   LC-SRM: Working stocks of all four compounds were prepared in 5 mM ammonium formate, 0.1% formic acid and 10% methanol. One microliter injections onto a 150 × 0.3 mm ID column (Acclaim PepMap RSLC, C-18, 2 µm, 100 Å), held at 50 degrees C and subject to an HPLC gradient of 2-6% B over 4 min, then 6 to 10% B over 1.5 min (A, 0.1% formic acid; B, 80% acetonitrile, 0.08% formic acid) at 7 µl per min. This was coupled to a Thermo Quantiva triple quadrupole mass spectrometer in selected reaction monitoring (SRM) mode, operating under the following conditions: Q1 and Q3 resolution 0.4 (FWHM), 50 msec dwell time, spray voltage 3500 (positive ion mode), sheath gas 6, aux gas 2, ion transfer tube 235 C, vaporizer temp 70 C°. SRM parameters for each compound are given in Table S4. Limits of quantitation and limits of detection were calculated as described previosuly 36 using a standard curve created from repeat injections of 0, 10, 100, 200, 500 and 1000 fmol of each compound on column and are given in Table S3. Cross contamination was assessed and reported for triplicate injections of 1000 fmol of each compound on column.
Metagenomic queries. The E. coli ThiM amino acid sequence (Uniprot ID P76423) was searched against diverse metagenomic libraries using BLASTP (and default BLOSUM62 settings) via GenomeNet (http://www. genome.jp/tools/blast/; Kyoto University Bioinformatics Center). Queries of soil metagenomic libraries were performed using BLASTP searches via the Joint Genome Institute (JGI) Integrated Microbial Genome (IMG) portal 37 .