Method optimization of the simultaneous detection of B 12 congeners leading to the detection of a novel isomer of hydroxycobalamin in seawater

Rationale: More than half of surveyed microalgae and over 90% of harmful algae have an obligate requirement for vitamin B 12 , but methods for directly measuring dissolved B 12 in seawater are scarce due to low concentrations and rapid light-induced hydrolysis. Methods: We present a method to detect and measure the four main congeners of vitamin B 12 dissolved in seawater. The method includes solid-phase extraction, separation by ultrahigh-performance liquid chromatography and detection by triple-quadrupole tandem mass spectrometry utilizing an electrospray ion source. This method was applied to coastal field samples collected in the German Bay, Baltic Sea and the Danish Limfjord system. Results: The total dissolved B 12 pool ranged between 0.5 and 2.1 pM. Under ambient conditions methyl-B 12 and adenosyl-B 12 were nearly fully hydrolyzed to hydroxy-B 12 in less than 1 h. Hydroxy-B 12 and a novel, corresponding isomer were the main forms of B 12 found at all field sites. This isomer eluted well after the OH-B 12 peak and was also detected in commercially available OH-B 12 . Both compounds showed very high similarity in their collision-induced dissociation spectra. Conclusions: The high instability of the biologically active forms of Me-B 12 and Ado-B 12 towards hydrolysis was shown, highlighting the importance of reducing the duration of the extraction protocol. In addition, the vitamin B 12 pool in the study area was mostly comprised of a previously undescribed isomer of OH-B 12 . Further studies into the structure of this isomer and its bioavailability are needed.


| INTRODUCTION
Phytoplankton, including microalgae and cyanobacteria, and bacterioplankton are the base of the oceanic food web and are responsible for roughly 50% of global primary production. 1,2 Thus, they play a major role in the biogeochemical cycles of most major elements in the ocean, including carbon, nitrogen, phosphorus and silicon. 3 Even though animals have an obligate requirement (are auxotrophic) for vitamin B 12 and thus need to acquire it through their diet, it is only produced by few bacteria and archaea. 4 In contrast, higher plants and fungi neither synthesize nor require vitamin B 12 , but instead possess B 12 -independent pathways (e.g. for methionine synthesis). 5,6 Microalgae on the other hand are photosynthetic organisms with a high occurrence (>50%) of vitamin B 12 auxotrophy. 4,5,7,8 Interestingly, there seems to be no apparent evolutionary pattern to this requirement with groups loosing or gaining B 12 -dependent pathways throughout history. 4 In the last decades many phytoplankton species have been identified as auxotrophic for one or more B-vitamins (thiamin (B 1 ), biotin (B 7 ), B 12 ) confirming that B-vitamin requirement in algae is the norm rather than the exception. 7 More recently, vitamin enrichment studies have illustrated that vitamin B 12 concentrations can significantly influence growth rates of phytoplankton communities in coastal and open ocean environments by co-limiting phytoplankton biomass with other nutrients (e.g. iron, nitrogen). [9][10][11][12][13] Vitamin B 12 itself is an organometallic tetrapyrrolic cofactor, which is structurally and biosynthetically related to heme and chlorophyll. 14 Many essential metabolic processes are catalyzed by B 12 -dependent enzymes, such as intramolecular rearrangements, methylations and reduction of ribonucleotides to deoxyribonucleotides. 6,15,16 B 12 consists of a central cobalt ion that is coordinated to an α-ligand of 5,6-dimethylbenzimidazole (DMB) known as the nucleotide loop and a β-ligand, such as methyl, adenosyl, cyano or hydroxy group (Me, Ado, CN, OH, respectively). 15 While the biologically active forms are Me-B 12 and Ado-B 12 , CN-B 12 and OH-B 12 have also been isolated from natural sources such as dairy products. 17,18 It is assumed that even though CN-B 12 and OH-B 12 cannot be utilized for intracellular functions, they are actively transformed into the active cofactors Me-B 12 and Ado-B 12 after transportation into the cell. 19 This transportation via a complex capturing mechanism has been reported to not distinguish between the various B 12 derivatives in mammals and bacteria. [20][21][22] It is not yet clear however if this is also the case for phytoplankton and if the β-ligand influences the respective bioavailability. Consequently, it is beneficial to distinguish the four B 12 congeners in biological studies.
Historically, vitamin B 12 concentrations were quantified with a laborious and error-prone bioassay method using B 12 -requiring microalgae. 23,24 Today's methods directly measure vitamin B 12 concentrations in seawater through a series of steps including filtration and preconcentration followed by analytical detection.
Okbamichael et al developed a C18 solid-phase extraction protocol followed by high-performance liquid chromatography (HPLC) separation and detection of CN-B 12 using ultraviolet-visible spectroscopy. 25,26 More recently advances have been made to establish simultaneous detection of all four B 12 congeners via triplequadrupole mass spectrometry. 27,28 These new methods reduced sample volume and processing time, thereby making vitamin B 12 data collection more accessible for marine sciences. However, even though these studies have increased knowledge about the importance and detection of B 12 in marine environments, the cycling of B 12 in the oceans remains mostly unexamined as data of particular and dissolved  rosette and subsequently mixed 1:1 from 10 m (or maximum depth) and surface water. Seawater was filtered immediately over 0.2 μm filters (AcroPak™ 1500, Pall Corporation, Port Washington, USA) and collected in polyethylene bottles. During sample collection and subsequent processing, care was taken to minimize the exposure of samples to light by wrapping all bottles and hoses in tinfoil.
In the case of self-packed columns, about 2-3 mL of the resin was loaded onto the column and conditioned with 6 mL of HPLC-grade methanol (Merck, Darmstadt, Germany), followed by at least 10 mL of deionized water. Conditioning followed by washing with water was also performed for commercially available columns. The pH of filtered seawater samples was adjusted to 6.2-6.6 with HCl (1 M) and loaded onto the solid-phase-extraction (SPE) column at a flow rate of approximately 1 mL min À1 . SPE columns were kept wet during the entire process and were washed with 10 mL of deionized water before freezing them at À20 C until further processing in the laboratory. SPE columns were then thawed, additionally flushed with 2% methanol in deionized water, followed by sample elution with 10 mL of 100% methanol. Clogged columns had to be eluted with compressed air (2-5 bar) and additionally spin-filtered (Millipore Ultrafree 0.45 μm, Eschborn, Germany) to remove co-eluted column material. The samples were concentrated using a SpeedVac (SC210A-230, Thermo Fisher Scientific, Schwerte, Germany) at 25 C, re-dissolved in methanol (100 μL) and stored at À20 C in the dark until mass spectrometric analysis. An initial subset of field samples was analyzed in order to characterize the background noise. If a high nonpolar background was detected in this subset of samples, the remaining samples were further extracted with n-hexane (Merck) in a HPLC vial and stored again at À20 C until mass spectrometric analysis.

| Light-induced hydrolysis of B 12 congeners
Triplicates of two concentrations (10 and 100 nM) of each B 12 congener were exposed to ambient natural light (20-200 μmol m À2 s À1 ) in both transparent glass and opaque vials.
The experiment was conducted at 20 C in deionized water adjusted to pH 6.5 with 1 M HCl.

| OH-B 12 isomer
A standard solution of OH-B 12 in deionized water and in deionized water adjusted to pH 8.2 according to the pH in seawater was prepared and OH-B 12 concentrations were subsequently analyzed via mass spectrometry on a regular basis ( Figure A2). In addition, greater amounts of the isomer of OH-B 12 (Iso-OH-B 12 ) were prepared by lowering the pH and adding the Lewis acid FeCl 3 to a standard solution of OH-B 12 , followed by incubation for 1 h. equivalents. In addition, in order to evaluate the replicability of the SPE preconcentration, four samples were collected at two stations in the Baltic Sea (stations 32 and 33, Figure 1).

| Data correction
Since Me-B 12 and Ado-B 12 are very sensitive towards hydrolysis, it was not possible to obtain standard calibration curves without substantial amounts of OH-B 12 . Thus, the obtained peak integral of a standard measurement had to be corrected by the detected amount of OH-B 12 . This amount of OH-B 12 was quantified using the OH-B 12 standard and Ado-B 12 and Me-B 12 concentrations were corrected accordingly. This data correction was not necessary for CN-B 12 due to its higher stability.
In addition, the transition used for quantification of OH-B 12 consistently had an impurity that could not be removed. This impurity fluctuated between 1 and 2 nM depending on the sampling day; however, it was consistent within each sample run (SD < 5%

| Limit of detection (LoD) and limit of quantification (LoQ)
LoDs and LoQs were defined as signal-to-noise ratio (S/N) = 3 and 10, respectively. The S/N ratio was determined as the mean of one high-and one low-noise region using MassLynx V4.2 SCN982 software (Waters).

| LC/MS/MS conditions
Similar to a previous study, an aqueous eluent consisting of ammonium formate (10 mM) and 0.02% formic acid at pH 4 was found to be optimal. Increasing the buffer and/or the concentration of formic acid did not significantly increase the ionization efficiency, so lower concentrations were used to prolong column lifetime and to prevent contamination of the MS instrument. 27 In contrast to Heal et al and Suffridge et al, using solely methanol or acetonitrile as organic eluents resulted in poor peak separation in this study. 27,28 However, a mixture of acetonitrile and methanol (4:1) reduced peak tailing and additionally resulted in a slightly higher elution strength in comparison to solely acetonitrile, thereby also reducing sampling time.

| Ionization parameters
The ionization parameters were optimized to favor formation of (ideally) one specific molecular ion for each vitamin B 12 27,28 This transition is thus not specific to OH-B 12 , but rather pertains to cobalamins in general since removal of the β-ligand results in a molecular framework shared by all cobalamins. 16 This dissociation of the β-ligand can also be observed for all other B 12 congeners, however only to a negligible extent.
The highest ionization efficiency was obtained when the desolvation temperature and all interconnected gas flow parameters were set to the highest device-specific settings ( Figure A1). These were maintained for the optimization of all other parameters, thereby additionally contributing to a reduction in contamination of the ion source. However, the capillary and cone voltages had major influences on the ionization efficiency of the highest abundant vitamin B 12 ions.
A clear trend to higher ionization efficiencies at low capillary voltages was found for all B 12 congeners (Figure 2).

| Light-induced hydrolysis of B 12 congeners
Both Me-B 12 and Ado-B 12 , at concentrations of 10 and 100 nM, were fully hydrolyzed (>99%) to OH-B 12 after 1 h when exposed to the mild acidic conditions (pH 6.2-6.6) of the SPE preconcentration and ambient light irradiation (Figure 4). In contrast, CN-B 12 was only partially hydrolyzed after 1 h (2%; Figure 4). Hydrolysis for all congeners was reduced in opaque compared to transparent vials, resulting in 20%, 10% and 0% conversion for Me-B 12 , Ado-B 12 and CN-B 12 after 1 h, respectively.
Nevertheless, hydrolysis of Me-B 12 and Ado-B 12 was also observed even when the samples were mostly protected from light irradiation by utilizing opaque vials.

| Replicability of SPE preconcentration step and recovery rates
The preconcentration step through SPE has previously been described as the highest source of error in the analysis of vitamin  accounted for more than 99% of total isomer abundance and was more stable towards isomerization. Here, we only present data and spectra of OH-B 12 and the main observed isomerization product Iso-OH-B 12 , as these were the only variants detected in field samples.
This isomerization was observable even under mild conditions such as using deionized water, where it occurred faster than in deionized water adjusted with NH 3 to pH 8.2 (pH of seawater; Figure A2). After three months, 23% of Iso-OH-B 12 could be observed in deionized water compared to only 15% in NH 3 . However, this conversion was rapidly enhanced by lowering the pH and adding Lewis acids, such as in this study compared to previous studies. 27,28 Another factor reducing the LoD is the use of methanol as solvent of field samples and standard solutions, which increased the ionization efficiency and thus sensitivity of cobalamin detection.
Furthermore, the use of methanol as sample solvent is more convenient in sample handling as it does not freeze when stored at À20 C and additionally reduces the hydrolysis rate of Me-B 12 and Ado-B 12 . Additional drying of the methanol via a molecular sieve or with chemical drying agents may be useful to further reduce the hydrolysis rate. However, additional care must be taken to reduce solvent evaporation and to fully dissolve standards while preparing high-concentration stock solutions as solubility of Ado-B 12 is lower in methanol compared to water.

| Light-induced hydrolysis of B 12 congeners
Previously, photosensitivity of vitamin B 12 congeners was only

| Replicability of SPE preconcentration step
The preconcentration step through SPE has been described as the highest source of error by other studies. 27,28 Consistent with this, the error margin of the SPE preconcentration step in this study was 50% or higher ( Figure 5) as can be seen by the standard deviation of four identical field samples.

| OH-B 12 isomer
The discovery of a novel OH-B 12 isomer in both the commercially available standard and field samples from all analyzed geographical locations was surprising, as it has not previously been reported in other publications dealing with cobalamins in the marine environment.
This is especially noteworthy since the extraction protocol used here was identical to previously published methods. 23

DATA AVAILABILITY STATEMENT
Data available on request from the authors.