Rapid quantification and isotopic analysis of dissolved sulfurspecies

Derek A. Smith1*,† , Alex L. Sessions1, Katherine S. Dawson1, Nathan Dalleska2 andVictoria J. Orphan1

1Division of Geological and Planetary Sciences, California Institute of Technology, 1200 E California Blvd, Pasadena, CA 91125,USA2Environmental Science and Engineering, California Institute of Technology, 1200 E California Blvd, Pasadena, CA 91125, USA

RATIONALE:Dissolved sulfur species are of significant interest, both as important substrates for microbial activities andas key intermediaries in biogeochemical cycles. Species of intermediate oxidation state such as sulfite, thiosulfate, andthiols are of particular interest but are notoriously difficult to analyze, because of low concentrations and rapid oxidationduring storage and analysis.METHODS: Dissolved sulfur species are reacted with monobromobimane which yields a fluorescent bimane derivativethat is stable to oxidation. Separation by Ultra-Performance Liquid Chromatography (UPLC) on a C18 column yieldsbaseline resolution of analytes in under 5min. Fluorescence detection (380 nm excitation, 480 nmemission) provides highlyselective and sensitive quantitation, and Time-of-Flight Mass Spectrometry (TOF-MS) is used to quantify isotopicabundance, providing the ability to detect stable isotope tracers (either 33S or 34S).RESULTS: Sulfite, thiosulfate, methanethiol, and bisulfide were quantified with on-column detection limits ofpicomoles (μM concentrations). Other sulfur species with unshared electrons are also amenable to analysis.TOF-MS detection of 34S enrichment was accurate and precise to within 0.6% (relative) when sample and standardhad similar isotope ratios, and was able to detect enrichments as small as 0.01 atom%. Accuracy was validated bycomparison to isotope-ratio mass spectrometry. Four example applications are provided to demonstrate the utilityof this method.CONCLUSIONS: Derivatization of aqueous sulfur species with bromobimane is easily accomplished in the field, andprotects analytes from oxidation during storage. UPLC separation with fluorescence detection provides low-μMdetection limits. Using high-resolution TOF-MS, accurate detection of as little as 0.01% 34S label incorporation intomultiple species is feasible. This provides a useful new analytical window into microbial sulfur cycling. Copyright© 2017 John Wiley & Sons, Ltd.

Sulfur exists in the natural environment in myriad forms,ranging in oxidation state from sulfate (SO4

2–) to sulfide(HS–) and including many species of intermediate oxidationstate, such as polysulfide (HSx

–), elemental sulfur (S0),thiosulfate, trithionate, sulfite, and others. Species such asthiosulfate, trithionate, and sulfite in particular are thoughtto be important intermediaries in microbial reactions thattransform (either oxidize, reduce, or disproportionate) sulfurand are frequently coupled to the biogeochemical cycles ofcarbon, oxygen, and iron.[1–4] ’Cryptic’ sulfur cycles drivenby microbial metabolism have been hypothesized to occur in

oxygen minimum zones of the world’s oceans,[5,6] the lowsulfate ’methanogenic’ zone of marine sediments,[7,8] saltmarshes,[9] freshwater sediments,[10,11] and within symbioticmicrobial associations.[9,12] In all cases, microbial sulfurcycling is thought to play key roles in the remineralization oforganic matter.

Sulfur speciation also dramatically affects the solubility andbioavailability of trace metals,[1–3] abiotic oxidation andreduction of manganese and iron,[1,4] and sulfurization oforganic matter.[13,14] Studies of all these processes, and others,would benefit greatly from improvedmethods for quantifyingdissolved sulfur species, particularly the most transient,intermediate compounds. Moreover, an ability to distinguishstable isotope labels in such compounds would enable awide variety of stable isotope probing experiments thatcould help illuminate the cryptic microbial sulfur cycle. Avariety of techniques to measure dissolved sulfur specieshave previously been described. One of the most widelyutilized techniques, the Cline assay, employs absorptionspectrophotometry of a N,N-dimethyl-p-phenylene diamine,chloride, and sulfide complex (methylene blue) to quantifyhydrogen sulfide.[15]

* Correspondence to: D. A. Smith, Department of Earth andPlanetary Sciences, Washington University in Saint Louis,Campus Box 1169, Rudolph Hall Rm 110, 1 BrookingsDrive, St. Louis, MO 63130, USA.E-mail: derek.smith@wustl.edu

† Current address: Department of Earth and PlanetarySciences, Washington University in Saint Louis, CampusBox 1169, Rudolph Hall Rm 110, 1 Brookings Drive, St.Louis, MO 63130, USA.

Copyright © 2017 John Wiley & Sons, Ltd.Rapid Commun. Mass Spectrom. 2017, 31, 791–803

Research Article

Received: 19 September 2016 Revised: 6 January 2017 Accepted: 23 February 2017 Published online in Wiley Online Library

Rapid Commun. Mass Spectrom. 2017, 31, 791–803(wileyonlinelibrary.com) DOI: 10.1002/rcm.7846

791

Ion chromatography (IC) has been employed to separatemultiple anions of sulfur (i.e., [16,17]). However, IC eluentchemistry can alter sulfur speciation; some sulfur species,e.g. polysulfides, are too reactive for accurate quantificationwith IC; further, IC provides no isotopic information,although it has been combined with multi-collectorinductively coupled plasma mass spectrometry (MC-ICP-MS) to achieve this.[18,19] Derivatization of sulfur species withmethyl trifluoromethanesulfonic acid (methyl triflate) iscapable of stabilizing polysulfides with quantificationdown to approximately 25 μM by UV absorptionspectrophotometry coupled to high-performance liquidchromatography (HPLC); however, with extraction andevaporation post-extraction, lower limits are reported.[20]

Drawbacks of the methyl triflate method include the toxicityof the reagent, multistep extraction procedure, addition ofsulfur to the derivatized product, and applicability only toreduced sulfur species. Cyclic voltammetry and in situ probeshave also been used to measure and quantify many of thereactive sulfur species in environmental samples at relativelyhigh spatial resolutions (mms).[9,21,22] The limitations of cyclicvoltammetry are that it cannot distinguish between theisotopes of sulfur.Here, we adapt a previously described method for

preserving aqueous nucleophilic sulfur species as bimanederivatives, and update their separation and detection tomodern instrumentation. Key features of the derivatizationare that (1) reaction with bromobimane (mBBr) isquantitative thus causing no isotopic fractionation; (2) theresulting bimane derivatives are very stable towardssubsequent oxidation, providing a convenient means topreserve samples in the field; and (3) bimane is stronglyfluorescent, enabling non-destructive fluorescencedetection.[23–26] The quantitative mBBr reaction proceeds asa nucleophilic substitution of sulfur compounds containingan unshared pair of electrons (Fig. 1), which includes sulfite(SO3

2–), thiosulfate (S2O32–), organic thiols (R-SH), bisulfide

(HS–), and polysulfides (Sn2–) but not elemental sulfur (S0)

or sulfate (SO42–). mBBr is a neutral molecule capable of

crossing the cellular membrane, enabling reactions with bothextra- and intracellular sulfur pools. Bimane derivatives

have been previously used to quantify sulfur species froma range of diverse samples including hydrothermal ventfluids, sulfur-oxidizing bacterial cultures (SO3

2–, S2O32–),

HS– in human blood, and cysteine and HS– inmethanogens.[27–31]

To improve on the separation of derivatized sulfur speciesby HPLC, which takes ~60 min per sample,[24,25] we employUltra-Performance Liquid Chromatography (UPLC) for rapid(<5 min) separation of derivatized sulfur species. Fordetection and quantitation, we evaluated both a conventionalfluorescence detector and a state-of-the-art Time-of-Flight(TOF) mass spectrometer (TOF-MS). Fluorescence detectionis less expensive and more sensitive than MS, andnondestructive so it can be coupled with preparative fractioncollection of individual sulfur species, e.g. for subsequenthigh-precision analysis of natural abundance isotope ratios.However, fluorescence itself provides no isotopic information,which is the main impetus for using MS. The high-resolutionTOF-MS employed here allows us to distinguish isobaricinterferences in the molecular ions, and accurately quantify34S enrichment down to levels of ~0.6%. The samemethodological approach could presumably be used for 33Senrichments. We note that while TOF-MS was used in thisstudy, quadrupole or other low-resolution MS could be analternative, with the caveat that isotopic discrimination mightsuffer.

To refine and test this updatedmethodology, we focused onfour sulfur analytes: SO3

2–, S2O32–, CH3SH, and HS–. Other

sulfur species are also potentially accessible with thisapproach. These analytes were measured in standardsolutions (both a dilution series of decreasing concentration,and a series of varying isotopic composition), thencharacterized in four separate proof-of-principle applications:(1) measurement of sulfite and thiosulfate concentrations inmarine sediment pore-waters, where they are present at lowμM concentrations; (2) measuring 34S appearance in bisulfidein a pure culture of sulfate-reducing bacteria amended with34SO4

2– tracer; (3) measuring 34S in bisulfide in methane seepsediments incubated with 34SO4

2– tracer; and (4) identifyingpreviously unreported intermediate sulfur species in a cultureof sulfur-oxidizing bacteria.

Figure 1. The nucleophilic substitution reaction occurs where a nucleophilic sulfurmolecule replaces bromide from bromobimane. The reaction is buffered by HEPES ata pH of 8 and EDTA is used as a chelating agent. In the case of bisulfide, two moles ofbromobimane are reacted per one mole of bisulfide. However, most other nucleophilicsulfur molecules form a terminal group at the end of a single bimane.

D. A. Smith et al.

wileyonlinelibrary.com/journal/rcm Copyright © 2017 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2017, 31, 791–803

792

EXPERIMENTAL

Derivatization protocol

Monobromobimane (hereafter just bromobimane or mBBr)powder (≥97%; Sigma Aldrich, St. Louis, MO, USA) wasdissolved in acetonitrile (ACN) to produce a working stocksolution of 50 mM mBBr. This was diluted prior to use to aconcentration of ~2-fold stoichiometric excess relative tosulfur analytes. Some sulfur species have multiple reactionsites, e.g. HS– binds two bimanes, necessitating the excessreagent. On the other hand, too much mBBr creates an HPLCpeak that can coelute with bimane thiols and produce largeamounts of background fluorescence; thus we avoidedadding a large excess. A typical mBBr reaction mixture wasprepared by combining the 50 mM working stock solutionof mBBr 1:1 with 50 mM HEPES and 5 mM EDTA in MilliQwater (at pH 8). An aliquot of 300 μL from the reactionmixture was added to 100 μL of sample (e.g. a ratio of3:1 v/v). Samples were reacted with mBBr for 1 h in the darkat room temperature. The reaction was then quenched with600 μL 65 mMmethanesulfonic acid in MilliQ water to 400 μLof sample and bromobimane reaction mixture (e.g. ratio of3:2 v/v). Samples were stored in amber glass vials at –20°Cuntil analysis. Both the mBBr reaction mixture andmethanesulfonic acid were bubbled with N2 gas prior touse to remove O2, and the entire derivatization procedurewas carried out in an anaerobic chamber to minimizeoxidation of analytes. Similar precautions to minimizeoxidation during field sampling and derivatization arestrongly recommended.

Liquid chromatography

Separation of the bimane derivatives was conducted on anACQUITY I-Class UPLC® system (Waters, Milford, MA,USA) equipped with a flow-through needle autosampler.Analytes were separated on an ACQUITY UPLC® BEH C18column (2.1 mm × 50 mm, 1.7 μm, 130 Å) maintained at 45°Cusing an injection volume of 1 or 2 μL. The elution profilecontained a mobile phase comprised of 0.1% formic acid(LCMSgrade), 1%ACN (LCMSgrade), and 98.9%MilliQwater(solvent A) and 100%ACN (LCMS grade) (solvent B) (Table 1).Effluent from the UPLC system was directed sequentially

through an ACQUITY fluorescence (FLR) detector and a XevoG2-S TOF mass spectrometer (both Waters). The FLR detectorwas operated at an excitation wavelength of 380 nm and anemission wavelength of 480 nm.

Mass spectrometry

The TOFmass spectrometer employed electrospray ionization(ESI) in both negative and positive ion mode, optimized asfollows: source temperature 120°C, desolvation temperature500°C, desolvation gas flow 800 L/h, capillary voltage2.5 kV, cone voltage 50 V, cone gas flow 40 L/h, and flight tube9.0 kV. Collision energy was 6.0 eV. Acquisition mass rangewas 50–800 Da and the calibration mass range was 91.183–1993.753 Da. A lock mass of 556.2771 Da was used to verifymass accuracy of ±3 mDa. The scan time was 0.2 s with aninterscan time of 0.014 s from 0.0 to 4.5min. Datawas collectedin centroid format. Sulfur species were quantified using themolecular ion.

Quantification of sulfur species

Calibration standards for quantification were created byserial dilution of sodium sulfite, sodium thiosulfate, sodiumsulfide and sodium thiomethoxide (all from Sigma-Aldrich)in N2-bubbled MilliQ water. Concentrations ranged from3.335 nM to 6.667 mM. The integrated mass spectral peakarea and integrated fluorescence peak area were recordedfor each standard and sample using MassLynx® andQuanLynx® software, which produced linear regressions ofintegrated peak area vs. concentration with average R2

values ≥0.99.

Quantification of stable sulfur isotope enrichment

Several approaches for extracting isotopic enrichment datafromLC/MS spectra are possible.Herewe adopt the approachof measuring the molecular ion and ~7–10 isotope peaks(depending on level of 34S enrichment) for each analytespecies, in both the sample and one or more standards. Foreach analyte, the average molecular weight of the bimanederivative was calculated as the weighted mean of themolecular ion and isotopic peaks, using peak height as theweighting factor for abundance because peak area is notavailable for data collected in centroid mode:

M ¼ ∑10

n¼1MnHn=∑Hn

where Mn and Hn are the accurate mass and height,respectively, of a given mass spectral peak. M can then becompared between sample and standard to calculate theshift in mean molecular weight of the sample:

ΔM ¼ Msamp �Mstd

Although this analysis confounds contributions fromisotopes of S, C, H, and N, the latter three are entirely derivedfrom atoms in the bimane derivative, which is the same forboth sample and standard and so cancel in the calculation ofΔM. Isotopes of oxygen are not necessarily identical, e.g. inS2O3 or SO3, but with a difference of 26 mDa between16O18O and 34S, the TOF mass spectrometer employed hereeasily resolves these two isobars. ΔM can thus be interpreteddirectly as the change in average mass due to sulfur isotopesubstitution, and is linearly proportional to the fractionalincorporation of isotope label. The slope of the relationship(between ΔM and atom% label) is ~1 if 33S is the label and

Table 1. Optimized elution program for separation ofbimane derivatives of inorganic sulfur species

Time (min) Flow rate %A %B

1 Initial 0.40 90.00 10.002 0.05 0.40 90.00 10.003 3.25 0.40 75.00 25.004 3.75 0.40 20.00 80.005 3.95 0.40 20.00 80.006 4.00 0.40 90.00 10.007 4.50 0.40 90.00 10.00

Quantification and isotopic analysis of dissolved sulfur species

Rapid Commun. Mass Spectrom. 2017, 31, 791–803 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

793

~2 if 34S is the label. For most applications then, astraightforward approach is simply to generate a linearcalibration curve for ΔM using multiple standards of knownisotopic abundance.To be able to compare data collected by LC/MS to isotope

ratios measured by isotope-ratio mass spectrometry (IRMS),we first convertedΔM into the absolute (fractional) abundanceof 34S using:

34F ¼ ΔMþMVCDT

1:98759

� �� 16:08997

where MVCDT is themean atomicweight of sulfur in the VCDT(Vienna Canyon Diablo Troilite) international referencematerial, and the constants 1.98759 and 16.08997 derive fromthe masses and abundances of sulfur isotopes in thatreference material. The isotope ratio was then calculatedfrom the following equation:

34R def34S� �32S½ � ¼

34F1�34F

and an enrichment factor (EnrF, i.e., fractionation factor) wascalculated as:

αA=B ¼34RA34RB

analogous to the calculation typically employed by natural-abundance IRMS measurements.Sequentially diluted sulfur standards with concentrations

and isotopic compositions that bracketed those of the sampleswere used to quantify S isotope enrichment. To determine theeffect that non-bracketed standards have on the determinedEnrF of a sample, we performed the isotopic analysisdescribed above with standards of bisulfide that were morethan three orders of magnitude less concentrated than sampleconcentrations. This exercise produced an additional 6.2%decrease in the recorded EnrF of the sample; however, asstandard concentrations approach and overlap sampleconcentrations, this additional error is greatly decreased. Peakheights and exact masses were recorded either withMassLynx® or using a partially automated analysis withMAVEN (Metabolomics Analysis and VisualizationEngine).[32,33] As a test of precision, a zero-enrichmentexperiment (i.e. comparison of two aliquots of the samereference standard) yielded enrichment factors of1.000 ± 0.006 for both sulfide and thiosulfate.

Isotope-ratio mass spectrometry (IRMS)

Sulfur isotope ratios and weight percentage for precipitatedZnS weighed into tin capsules (5 × 7 mm) were determinedon a Delta™ XLPlus isotope-ratio mass spectrometer (ThermoQuest) attached to a NC2500 elemental analyzer (EA; CEInstruments). All sulfur isotopic compositions are reportedas per mil variations relative to VCDT following theconventional delta notation:

δ34S ¼ 34S=32Sð ÞSample

34S=32Sð ÞVCDT� 1

!

Values of δ34S were calibrated relative to an internal standardof AgS2 (18.03‰); precision and accuracy for thesemeasurements are estimated as <0.2‰.

To achieve cross-calibration of the two methods, twoaliquots from seven sulfide samples with increasing atomicpercentages of 34S, ranging from 4.27 to 17.4%, weregenerated. In an anaerobic chamber, 9.8 mL of a 23 mM4.27% 34S sulfide solution was combined with 0.2 mL of a185mM 100% 34S sulfide solution to produce a 17.4% 34S stocksolution. That stock was serially mixed with aliquots of the4.27% 34S sulfide solution to produce solutions of intermediate34S abundance. Parallel aliquots of these standards werederivatized with mBBr and measured by UPLC/TOF-MS,and precipitated as zinc sulfide for measurement byEA/IRMS.

Experiments with sulfate-reducing and sulfur-oxidizingbacterial cultures

Stocks of Desulfococcus multivorans DSM 2059 and Sulfurovumlithotrophicum DSM 23290[34] were obtained from the GermanCollection of Microorganisms (DSMZ) and initiallypropagated in the lab using the recommended media.Triplicate anaerobic cultures of D. multivorans were grown in30 mL butyl rubber stoppered serum bottles at 28°C toexponential phase in a modified MJ medium[34] (g/L): NaCl30.0, K2HPO4 0.434, CaCl2•2 H2O 0.28, MgSO4•7 H2O 3.40,MgCl2•6 H2O 4.18, KCl 0.33, NH4Cl 0.35, NiCl2•6 H2O0.50 mg, Na2SeO3•5 H2O 0.50 mg, FeCl2 0.01, NaHCO3 1.5,Na2S2O3•5 H2O 1.5, NaNO3 2.0, with the addition of 10 mLDSM 141 Trace Element solution and 10 mL of DSM 141Vitamin solution. The medium was supplemented withpyruvate to a final concentration of 10 mM, and sulfatesolution with a δ34SVCDT value of +2000‰ was added to afinal concentration of 0.28 mM. The medium was keptanaerobic by saturation with 80:20 N2:CO2 gas, followed byadjustment of the pH to 7.0. During exponential phase(47.5 h after inoculation), 100 μL from each culture wascollected into a 1 mL syringe and transferred after filtrationthrough a 0.2 μm acrodisk filter into 300 μL of the mBBrreaction mixture.

Continuous culturingofS. lithotrophicumwas carried out in aBioFlo115 fermentor/bioreactor (New Brunswick Scientific).Oxidant availability, in this case the concentration of nitrate(NO3

–), was the growth-limiting factor. A working volume of1.75 L was maintained with an agitation rate of 150 rpm. Aconstant pressure of 0.5 bar from a 80:20 N2:CO2 gas mixturecontinuously saturated the growth vessel. Modified MJmedium as described above was used[34] (and DSMZMedium1011) with the addition of 1.5 mL 400 mMNa2S•9H2O and noaddition of isotopically labeled substrates. An excess of theautotrophic carbon substrate, CO2, was continually suppliedvia the80:20N2:CO2gasmixture.Temperaturewasmaintainedat 28°C. Influent anoxic sterile media was supplied from a 2 Lglass bottle that was connected to a 3 L mylar gas bag overpressurized with the 80:20 N2:CO2 gas mixture and effluentwaste was collected into a sterile anoxic 2 L glass bottle with anon-pressurized 3 L mylar gas bag. A low dilution rate of0.036 day–1 was maintained throughout the experiment topromote slow growth. That dilution rate was approximatelyhalf of the optimal growth rate of 0.069 day–1 determined inbatch culture. Once growth conditions of the bioreactor

D. A. Smith et al.

wileyonlinelibrary.com/journal/rcm Copyright © 2017 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2017, 31, 791–803

794

reached steady state, 25 mL aliquots were removed from thevessel; from these a 100 μL aliquot was collected into a 1 mLsyringe and filtered through a 0.2 μm acrodisk filter into300 μL of the bromobimane reactionmixture.

Sediment incubations

Sediment samples were collected on May 9, 2013, from theSanta Monica Basin, offshore California, as part of anexpedition led by the Monterey Bay Aquarium ResearchInstitute on the R/V Western Flyer. Using the ROV Doc Ricketts(Dive DR463), sediment push cores PC 61 and PC 55 werecollected in an extensive white microbial mat overlying anactive methane seep at a water depth of 860 m and in situtemperature of 4°C (lat. 35.473431 N, long. 118.4007W). Uponcollection shipboard, PC 55 was processed shipboard in 1 cmintervals according to Orphan et al.[35] Pore-water aliquotsfrom each depth horizon were collected using an argon-pressurized Reeburgh-type squeezer, preserved forgeochemical analyses, and stored at –20°C. PC 61 wassectioned into two 6 cm intervals and heat sealed under anoxicconditions in amylar bag flushedwith argon and stored at 4°Cuntil processing for microcosm incubation experiments(established 40 days after collection).Stable isotope probing (SIP) experiments were conducted

at 10°C in 5 mL butyl stoppered serum bottles with 1 mLof sediment slurry (0–6 cm PC 61) mixed with anoxicartificial seawater containing 5 mM unlabeled sulfide and28 mM 100% 34SO4 for 40 days. Residual unlabeled SO4

2–

in the sediment slurry diluted the tracer to a level of ~80%34S, based on initial sampling. The incubation was allowedto settle before time points of the supernatant were collected.The supernatant was centrifuged at 20,000 g for 5 min. A50 μL aliquot of the overlying liquid was added to the mBBrreaction mixture and allowed to react as described above.Before analysis this solution was thawed and diluted 1:10in 20:80 ACN:65 mM methanesulfonic acid.

RESULTS AND DISCUSSION

Quantitation of dissolved sulfur analytes

The use of bimane derivatives for quantitation of dissolvedsulfur species has many benefits. Experimental testing of thisassay showed reactivity with a broad range of aqueous sulfurspecies, and provided rapid, quantitative measurement withthe addition of a ~2-fold molar excess of the bromobimane(mBBr) reagent, and resulted in no isotopic fractionation.Bromobimane is reactive towards most sulfur species withunpaired electrons, and so has the ability to target a widevariety of trace, labile species in addition to those studiedhere (Table 2). Sulfate and elemental sulfur are notableexceptions that do not react with mBBr. The reagent, whilemoderately expensive, is safe and easily deployed in the fieldto preserve samples immediately after collection.[24,25,27]

Although we did not quantitatively study the stability ofbimane derivatives, qualitative evidence suggests they arehighly stable when stored at –20°C. Mixtures of derivatizedstandards stored in this way were measured with no lossin signal for at least 3 months (longer times were not tested).Marine pore-water samples, derivatized at the time of

collection and analyzed 29 months later, yielded HS–

concentrations that were identical within error to thoseobtained shipboard by the Cline assay, and SO3

2– andS2O3

2– concentrations that were similar to previous reports(data not shown). Underivatized sulfide and sulfite do notlast nearly this long in the presence of O2, even when storedat –20°C.

The optimized UPLC elution gradient described above andin Table 1 was able to fully resolve the bimane derivatives ofSO3

2–, S2O32–, CH3SH, and HS– within 4.5 min (Fig. 2(A)),

making this a very rapid separation method. However, thecomposition of the eluent was found to influence theseparation and required modification depending on thespecies of interest. For example, LC peaks representing SO3

2–

bimane and S2O32– bimane were observed to co-elute under

conditions containing >90% of the polar eluent (0.1% formicacid, 1% ACN, and 98.9% MilliQ water). Similarly, >70%organic eluent (100% ACN) was required to separate CH3S

–

bimane from S2– dibimane by 3.75 min. Additionally, as thepeak of unreacted mBBr elutes close to sulfide dibimane,excessively large reagent concentrations can lead to a broadpeak area that can influence the sulfidemeasurement. Limitingthe mBBr reagent to a 2-fold excess largely avoided thisproblem, while still enabling quantitative preservation ofsulfur intermediates.

Performance characteristics for quantitation of analytes byeither FLR or TOF-MS are given in Table 3. Comparison ofthe two methods revealed a higher degree of sensitivity usingfluorescence detection over TOF-MS. The TOF-MS methodwas found to be sensitive down to the low micromolar range(sample concentration before dilution in the reactionmixture);based on an injection volume of 1 μL, equivalent to <1 pmolinjected on-column. Both the FLR and TOF-MS detectorsbecame saturated at ≥3 mmol sulfur on-column, yielding adynamic range of ~9 orders of magnitude. Both detectors

Table 2. List of representative sulfur compounds and theirpredicted reaction with bromobimane

NameMolecularformula

Reaction withmBBr

Bisulfide HS– +Sulfite SO3

2– +Thiosulfate S2O3

2– +Polysulfides HSSn

– +Cysteine C3H7NO2S +Glutathione C10H17N3O6S +Coenzyme A C21H36N7O16P3S +Coenzyme M C2H5O3S2

– +Organic Thiols R-SH +Polythionates SnO6

2– 0Elemental Sulfur S0 0Adenosine 5’-phosphosulfate (APS)

C10H14N5O10PS 0

Sulfonates (i.e., HEPES) R-SO3– 0

Sulfate SO42– 0

+ indicates positive reaction, 0 indicates reaction will not occurR indicates alkyl or organic substituentData sourced from Fahey and Newton[27] and Zopfi et al.[39]

Quantification and isotopic analysis of dissolved sulfur species

Rapid Commun. Mass Spectrom. 2017, 31, 791–803 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

795

Figure 2. Chromatography of sulfur species using bromobimane preservation and UPLC/TOF-MS analysis showing aseparation of a mixture of sulfur standards consisting of SO3

2–, S2O32–, CH3SH, and HS– each at 250 μM. (A) Extracted ion

chromatogram of 1: sulfite bimane 0.55 min 271.0389 Da, 2: thiosulfate bimane 0.96 min 303.0609 Da, 3: methanethiol bimane2.97 min 239.0854 Da, and 4: sulfide dibimane 3.25 min 415.1440 Da. (B–E) Mass spectra with the chemical structure, chemicalformula, and isotopic distribution for the negative ion of sulfite bimane (B), negative ion of thiosulfate bimane (C), positive ionof methanethiol bimane (D), and positive ion of sulfide dibimane (E). [Color figure can be viewed at wileyonlinelibrary.com]

Table 3. Performance characteristics for FLR and TOF-MS detection of bimane derivatives

Sulfite bimane Thiosulfate bimane Sulfide dibimane

FLR Sensitivitya (S/N) 916:1 1333:1 1833:1TOF-MS Sensitivitya (S/N) 165:1 369:1 371:1FLR Precision (relative %) 2.35 6.10 3.95TOF-MS Precision (relative %) 4.71 3.81 4.00FLR Accuracya (± μM) 0.63 0.86 2.32TOF-MS Accuracya (± μM) 3.31 3.64 4.04aQuantified at 33.35 μM.

D. A. Smith et al.

wileyonlinelibrary.com/journal/rcm Copyright © 2017 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2017, 31, 791–803

796

yielded linear calibration curves over this range. Precision andaccuracy were similar for the two detectors, typically a fewpercent relative to sample concentration. In summary, bothdetectors were shown to be capable of quantitation ofdissolved sulfur species. Quantitation of sulfur species byTOF-MS is more expensive and complex than the FLRdetection but has the advantage of enabling positiveidentification of unknown sulfur species, while thenondestructive FLR detector is ideally suited for preparativefraction collection.

Measurement of sulfur isotopic enrichment

Several different strategies were tested for quantifyingisotope abundance based on the relative intensity of isotopicpeaks of the molecular ion (Figs. 2(B)–2(E)). These includedcalculation of M + 2/M ion abundance ratios, Σ(M + 2/M + …M + 7/M) ion abundance ratios, and fitting thecomplete envelope of isotopic peaks with a forward model.In the end, the calculation of a weighted-mean mass shift(as described in the Experimental section) was both the mostaccurate and simplest approach. A minor drawback is thatthis analysis cannot distinguish between simultaneousenrichments of 33S and 34S (i.e., from separately labeledsubstrates), for which deconvolution of the full isotopologueenvelope would be required. A spectrometer with sufficientmass resolution to distinguish 18O from 34S isotopologues,such as the TOF analyzer employed here, is essential to thisapplication.An estimate of precision in isotopic measurements is

provided by the zero-enrichment experiment. The same(natural isotopic abundance) standard solution used forcalibration of the TOF-MS data was measured repeatedly asa sample, yielding enrichment factors for sulfide andthiosulfate of 1.000 ± 0.006, equivalent to 6‰ uncertainty inthe δ34S nomenclature with no systematic error. Relativeprecision is better at higher levels of isotopic enrichment.

Further validation of method accuracy is complicated bythe fact that the standard analytical technique, IRMS, is notdesigned to measure the relatively large isotopic enrichmentsbeing targeted by our LC/MS method. Certified internationalreference materials for calibration span a range of only a fewtens of permil,[36] whereas an artificial enrichment in 34S ofonly 1% (absolute abundance) constitutes a shift in δ34S ofnearly 250‰. These issues notwithstanding, we measuredthe same series of sulfide isotopic standards (seeExperimental section) using both UPLC/TOF-MS andEA/IRMS (Fig. 3).

Seven samples ranging from 4.27 to 17.4 atom percent 34Swere measured. Data from both instruments were highlyreproducible and linear over the entire range of isotopiccompositions. However, both yielded apparent 34Sabundances significantly lower than those calculated forthe standards based on mixing volumes. The measuredvalues averaged 98.8% and 90.4% of the theoretical valuesfor the TOF-MS and IRMS data, respectively (Fig. 3(A)).Comparison of the two datasets with each other is alsohighly linear (Fig. 3(B)).

Both datasets (i.e., from TOF-MS and IRMS) were collectedby comparing unknown, 34S-enriched samples with a singleisotopic standard with a natural abundance of 34S.Presumably, the deviations of measured data from correctisotope ratios therefore reflect instrumental fractionations thatdo not scale uniformly with isotopic composition. Such effectsare well known in measurements of natural-abundanceisotopes where they are often referred to as "scalecompression/expansion". For this reason, two-pointcalibration of isotopic compositions is almost universallyrecommended.[36–38] However, in our case no 34S-enrichedreference materials exist, so we are unable to explicitly testaccuracy across the range of isotopic compositions to a levelthat approaches measurement precision. Nevertheless, thelinearity and reproducibility of the responses of bothinstruments in Fig. 3 strongly suggest that – with calibration

Figure 3. Cross-calibration of the same samples of HS– with increasing atomicpercentage of 34S derivatized with bromobimane and precipitated with zinc acetaterun using UPLC/TOF-MS and EA/IRMS, respectively: (A) Atomic percentage of 34Sin HS– plotted against enrichment factor (red circles) (EnrF = –0.1077 + 0.2613*%34S,R2 = 0.999), as determined by the UPLC/TOF-MS methodology, and δ34SVCDT (δ34S =�1027.3 + 244.0*%34S, R2 = 0.999), as measured by EA/IRMS (blue squares). (B) Asystematic and reproducible linear cross-calibration (gold circles) was establishedlinking our enrichment factors (EnrF) with conventional IRMS δ34SVCDT measurementsof sulfur (EnrF =0.9935 + 0.001071*δ34S, R2 = 0.999). The solid black line represents thetheoretical relationship between atom percentage of 34S, enrichment factors, andδ34SVCDT. [Color figure can be viewed at wileyonlinelibrary.com]

Quantification and isotopic analysis of dissolved sulfur species

Rapid Commun. Mass Spectrom. 2017, 31, 791–803 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

797

standards having suitable 34S-enriched compositions –accuracy for isotopic abundance should approachmeasurement precision, i.e., 0.6% relative.Specific reasons for the deviation of TOF-MS data at high 34S

enrichments are not currently known. Fractionations duringionization, transmission, and detection are all possible. Ionshading, in which the earlier-arriving 32S ions induce a slightelectric field at the detector that repels the later-arriving 34Sions, is a possibility specific to the TOF thatwould clearly scalenonlinearly with isotope ratio.

Application 1: Quantitation of pore-water SO32– and S2O3

2–

in methane seep sediments

As an initial test of themethod’s ability to quantify trace sulfurspecies in complex natural samples, UPLC separation withFLR and TOF-MS detection were used to compareenvironmental pore-water concentrations of SO3

2– andS2O3

2– in a depth profile of sediments from the Santa MonicaBasin. The sediments were collected within an activemethane seep, where microbial sulfur metabolism associatedwith anaerobic oxidation of methane and chemosyntheticsulfide oxidation is enhanced. Both the FLR and TOFdetectors were used to quantitate SO3

2– and S2O32–; we

present the FLR data for SO32– and the TOF S2O3

2–

concentrations highlighting the abilities of both instrumentsfor accurate quantitation (Fig. 4(A)). The FLR and TOFdetectors produced S2O3

2– and SO32– values that were

inconsistent with previous reports. SO32– and S2O3

2– reachedmaximum concentrations of 12.5 and 42.5 μM at 1.5 and3.5 cm depth, respectively. SO3

2– concentrations were belowdetection limits by 4.5 cm depth. S2O3

2– concentrationsvaried across much of the 18.5 cm core. Conventionalanalysis of HS– by the Cline assay and SO4

2– by ionchromatography show typical profiles of HS– building withdepth to a maximum of 12.5 mM by 10.5 cm and SO4

2–

decreasing from 25 mM at the surface down to 2 mM at

depth (Fig. 4(B)). We did not observe an increase in SO32–

or S2O32– concentrations at the HS– and SO4

2– transition asmight be expected. The concentrations of SO3

2– and S2O32–

were reproducible and fall within the range of previouslyreported environmental marine sediments (see Zopfiet al.[39] for review).

Application 2: Tracing 34S-labeled sulfate in asulfate-reducing bacterial culture

The second experimental demonstration of this methodwas totracktheconversionof 34S-labeledsulfate into 34S-sulfide,usingbatch cultures of the sulfate-reducing bacterium (SRB)Desulfococcus multivorans. Cultures were amended with280 μM labeled sulfate (34FSO4 = 12.4%) and unlabeledthiosulfate (6.44 ± 0.15 mM; 34FS2O3 = 4.2%). Here wedemonstrate the utility of the UPLC/TOF-MS analyticalmethodology for tracking the production of isotopicallyenriched sulfur species (Fig. 5(A)), and through the additionof 34S-enriched sulfate, we examine the potential for sulfurdisproportionation and preferred sulfur substrate usage byD.multivorans. Previous reports suggested the potential forthiosulfate reduction in addition to the canonical sulfatereductionbyD.multivorans, but the specificpreferenceof sulfursubstrate by this organism has not been directly tested.[40]

As the D. multivorans cultures reached the exponentialphase of growth, HS– accumulated to concentrations of1.66 ± 0.49 mM. Abiotic control bottles for this experimenthad no quantifiable HS– or SO3

2– and an average thiosulfateconcentration of 6.44 ± 0.15 mM, indistinguishable from theamount added. The triplicate experimental bottles had finalthiosulfate concentrations of 5.42 ± 1.35 mM. Significantlymore HS– was produced than the 280 μM SO4

2– provided,which implies a contribution of both SO4

2– and S2O32–

reduction to the HS– produced. The mass balance of sulfurwas conserved, within error; however, we cannot rule outthe possibility of a minor amount of non-enriched HS–

Figure 4. Pore-water profile from methane seep sediments underlying sulfide-oxidizing microbial mat (push core PC55) recovered from the Santa MonicaBasin. PC55 was sectioned at 1 cm intervals for the upper 6 cm, then at 3 cmintervals down to 18.5 cm: (A) Concentrations, with depth, of bromobimane-preserved intermediate sulfur species (sulfite and thiosulfate) measured byFLR and TOF detectors, respectively. (B) Colorimetric analysis (Cline assay) ofsulfide concentrations and IC measurements of sulfate and dissolved inorganiccarbon (DIC) measured from aliquots of the same pore-water samples. [Colorfigure can be viewed at wileyonlinelibrary.com]

D. A. Smith et al.

wileyonlinelibrary.com/journal/rcm Copyright © 2017 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2017, 31, 791–803

798

transferred with the initial inoculum. Assuming that all of the34SO4

2– was converted into HS–, more than 1.3 mM of HS–

had to come from unlabeled S2O32–. There were also μM

levels of SO32– detected in the experimental bottles, which

might have been produced via a number of differentpathways. Notably, there was no detectable enrichment of34S in SO3

2– indicating it did not derive from the reductionof the 34SO4

2– substrate. At exponential phase, 34FHS was5.4 ± 0.4% and 34FS2O3 was 4.5 ± 0.5%. Approximately one-third of the 34S-enriched HS– was created by D. multivoransreducing labeled sulfate, but the mechanism resulting inenrichment of the S2O3

2– pool remains unclear. The minorincorporation of the 34S label into thiosulfate could suggest

that SO42– is being incompletely reduced to S2O3

2– duringsulfate reduction, or possibly that thiosulfate is being slowlyformed from HS–.

Application 3: Tracing 34S-labeled sulfate in methane seepsediment incubations

To test the utility of the sulfur SIP method in environmentalsamples, we conducted 40-day incubations with labeledsulfate (28mM, FSO4 = 1, i.e., 100% 34S label) added tomethaneseep sediments. UPLC chromatograms of the mBBr-derivatized supernatant contained many more peaks relativeto the experiments with pure cultures of D. multivorans,

Figure 5. (A) Representative negative ionization chromatogram from triplicate culturesof Desulfococcus multivorans, a sulfate-reducing bacterium, incubated with 34S-labeledsulfate. Peaks shown in the chromatogram are labeled with the compound, elutiontime, and most abundant ion, and include 1: HEPES 0.31 min 237.095 Da, 2: sulfitebimane 0.52 min 271.042 Da, 3: thiosulfate bimane 0.86 min 303.015 Da, 6: brominecluster 2.89 min 375.607 Da, 7: sulfide dibimane 3.11 min 413.130 Da. (B) Positiveionization chromatogram from a methane seep sediment incubation with 34SO4

2–; asthis is a complex natural sample, more compounds were detected relative to purecultures or standards. Peaks are 1: HEPES 0.31 min 239.111 Da, 3: thiosulfate bimane0.89 min 305.030 Da, 4: bimane 1.51 min 193.102 Da, 5: 34S sulfide bimane 2.67 min227.063 Da, 6: bromobimane 2.89 min 271.012 Da, 7: 34S sulfide dibimane 3.11 min415.146 Da, 8: disulfide dibimane 3.46 min 469.098 Da. (C) Mass spectrum of thepositive ion of sulfide dibimane from methane seep sediment incubations (peak 7 in(B)). The peak heights for 32S (415.1455 Da) and 34S (417.1416 Da) were approximatelyequal due to the extensive conversion of 34SO4

2– into sulfide (34SH–) presumablyassociated with sulfate-coupled anaerobic oxidation of methane. [Color figure can beviewed at wileyonlinelibrary.com]

Quantification and isotopic analysis of dissolved sulfur species

Rapid Commun. Mass Spectrom. 2017, 31, 791–803 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

799

including several peaks that were clearly not SO32–, S2O3

2–,CH3S

–, or HS–. Each peak in the environmental samples wassubsequently identified through comparisons with standardsor de novo determination from the mass spectral data (Fig. 5(B)).Two separate bimane derivatives of HS– were observed:sulfide monobimane and sulfide dibimane. We infer that theamount of mBBr added to these samples (at the time ofcollection) was insufficient to fully react with bisulfide dueto other reactive sulfur species. For natural samples preciseconcentrations of total reactive sulfur species are likely to beunknown; however, if an estimation can be provided, then atwo-fold stoichiometric excess of mBBr can still be supplied.This caused considerable problems for quantitation becausegeneration of a sulfide monobimane calibration curve is notstraightforward. Nevertheless, based on sulfide dibimane weobserved the build-up of 4.35 mM sulfide with FSO4 = 0.498(i.e., 50% 34S label) (Fig. 5(C)). Sulfate reduction was clearlyprevalent in these sediments, likely coupled to the anaerobic

oxidation of methane (AOM). In addition to large quantitiesof HS–, the sediment incubations also contained 1.65 mMS2O3

2– with no detectable 34S enrichment (FS2O3 = 0.042).Disulfide dibimane, corresponding to the polysulfide HS2

–,was also detected in the incubations. The zero valent (HS2

–)has been hypothesized as the chemical intermediate in theAOM microbial syntrophy between a sulfate-reducingmethanotrophic ANME archaea and a disulfide-disproportionating deltaproteobacteria.[12] Milucka et al.[12]

used the methyl triflate technique to measure polysulfidesfrom 3-x S chain length; however, this method was incapableof directly measuring the primary proposed intermediate,HS2

–. Here we show that the TOF-MS method for measuringbromobimane derivatives of sulfur is capable of HS2

–

detection; however, to develop this into a quantitative assay,further work and reliable standards for HS2

– are needed aswe are currently unable to reliably create and derivatizeindividual chain-length polysulfides with bimane.

Figure 6. (A) Negative ionization chromatogram from slow growth and oxidant limitedcontinuous culture of Sulfurovum lithotrophicum, a sulfur-oxidizingepsilonproteobacterium. The three resolvable peaks are 1: thiosulfate bimane 1.38 min303.0114 Da, 2: disulfanemonosulfonate (–S-S-SO3H) bimane 2.06 min 334.9837 Da, 3:trisulfanemonosulfonate (–S-S-SO3H) bimane 2.94 min 366.9554 Da. (B) Isotope modelof the negative ion of trisulfanemonosulfonate (–S-S-SO3H) bimane with theoreticalmass and isotopic distribution. (C) Mass spectrum of the negative ion oftrisulfanemonosulfonate (–S-S-SO3H) bimane from the oxidant-limited chemostatexperiment (peak 3 in (A)). [Color figure can be viewed at wileyonlinelibrary.com]

D. A. Smith et al.

wileyonlinelibrary.com/journal/rcm Copyright © 2017 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2017, 31, 791–803

800

Application 4: Identification of novel sulfur species in acontinuous culture of the sulfur-oxidizing bacteriumSulfurovum lithotrophicum

Coupling bimane derivatization to mass spectral analysisshould enable the identification of potentially novel andimportant intermediate redox species of sulfur. In thisapplication, we studied the sulfur-oxidizing bacteriumSulfurovum lithotrophicum in oxidant-limited continuousculture, with thiosulfate as the supplied sulfur source. Wehypothesized that partial oxidation of thiosulfate could resultin the accumulation of unusual and potentially importantsulfur species of intermediate redox state. Chromatogramsof mBBr-derivatized samples from these cultures revealedtwo unidentified peaks (Fig. 6(A)) in addition to thiosulfate.Using de novo analysis of the mass spectral data, these twounknown bimane-derivatized sulfur compounds weretentatively identified as disulfanemonosulfonate (–S-S-SO3H)bimane (mass 334.9837 Da) and trisulfanemonosulfonate(–S-S-S-SO3H) bimane (mass 366.9553 Da) (Figs. 6(A), 6(B)and 6(C)). These compounds appeared to be producedduring S2O3

2– oxidation by S. lithotrophicum and were notobserved in the standard solutions or in any of the othersamples we analyzed. Quantitation and identification, vialiquid chromatography and FLR detection, of thesecompounds was not possible as authentic standards arenot available to construct calibration curves and retentiontimes.The chemistry of sulfanemonosulfonic acids has been

previously described and the formation of these acids isproposed to be a result of the nucleophilic degradation ofsulfur chains.[41] The proposed reaction mechanisminvolves either SO3

2– or S2O32– interacting with elemental

sulfur or polysulfide chains. Through microscopicexamination of S. lithotrophicum cells, structures consistentwith intracellular sulfur globules were visible (data notshown).[42,43] Sulfanemonosulfonic acids were previouslydetected in the cell pellet of Allochromatium vinosum, apurple photosynthetic sulfur oxidizer, but these sulfurcompounds were not detected in the supernatant.[28] Dueto the analytical limitations of XANES spectroscopy, thatstudy was unable to determine the chain length of thesemolecules. The experiments with S. lithotrophicum heredemonstrate the added advantage of using TOF-MS, withits high mass resolving capabilities, for untargetedidentification and resolution of unknown sulfurcompounds.The detection of sulfanemonosulfonic acids in two distantly

related sulfur-oxidizing bacteria points to the potentialimportance of these compounds. One possibility is thatintracellular sulfanemonosulfonic acids may represent apathway by which microorganisms first begin to utilize theirstored reserves of sulfur globules (elementalsulfur/polysulfides). Indeed, previous experiments with A.vinosum revealed a correlation between intracellularsulfur and the relative concentration/presence ofsulfanemonosulfonic acids.[28] The reaction mechanismoriginally proposed by Meyer et al.[41] would be consistentwith the ability of dissimilatory sulfite reductase (Dsr)subunits to produce SO3

2– which could then interact withthe intracellular elemental sulfur/polysulfides creatingpotentially important intracellular sulfanesulfonic acids.

CONCLUSIONS

We present a method for preservation, separation, andquantitative analysis of dissolved sulfur-containing analytes.It is based on derivatization with monobromobimane (mBBr),which is rapid, quantitative, and easily adapted to fieldconditions. Qualitative evidence suggests that mBBr-derivatized samples are stable relative to unreacted samples,with minimal oxidation over the course of months to years.Baseline separation of four key analytes, sulfite, thiosulfate,methanethiol, and bisulfide, was achieved by UPLC in just5 min, making this a high-throughput analytical approach.Detection by either fluorescence or time-of-flight massspectrometry provided highly sensitive, precise, and accuratequantitation. Using TOF-MS, we demonstrate the ability toquantify 34S isotopic enrichments with ~0.6% relativeprecision, over a dynamic range from natural abundance(4.2%) to at least 50%. In the proof-of-principle applicationsdescribed here, we (1) demonstrated thiosulfate utilizationduring sulfate reduction by pure cultures of the sulfatereducing bacterium D. multivorans; (2) detected zero valentdisulfide (HS2

–) in methane seep sediments, a polysulfidespecies that has been historically challenging to measure withother methods; and (3) resolved the identity of two unknownsulfur metabolites as disulfanemonosulfonate andtrisulfonemonosulfonate produced by S. lithotrophicum duringthiosulfate oxidation under oxidant limited conditions.

The methodology demonstrated here has numerouspotential applications. The ability to quantify reactive, low-abundance species in complex natural samples will beextremely useful in studies of biogeochemical sulfur cycling,especially in environments supporting rapid sulfur cyclingsuch as oxygen minimum zones,[5,6] below the sulfatereduction zone in marine sediments,[7,8] sulfur-metabolizingmicrobial symbioses,[9,42] and freshwater sediments.[10,11]

Moreover, coupling UPLC with nondestructive FLRdetection to preparative fraction collection provides ananalytical route to high-precision IRMS-based isotopicanalysis of these compounds at natural isotopic abundance.The ability to detect a large dynamic range of isotopicenrichments is also extremely useful for stable-isotopeprobing (SIP)-type experiments, either in environmentalsample incubations or in culture. This is likely to haveapplications in microbial ecology, biochemistry, and evenbiomedicine.[45]

AcknowledgementsWe thank chief scientist Peter Brewer (MBARI) and theshipboard, scientific crew and pilots of the R/V Western Flyerand the ROV Doc Ricketts for allowing us to participate in theexpedition. We also thank S. Scheller and H. Yu forproviding and maintaining the methane seep sedimentincubations, F. Wu and S. Connon for technical assistance,and C. Neubaur, M. R. Raven and M. S. Sim for helpfuldiscussions. The work was funded by the Gordon and BettyMoore Foundation through Grant No. GBMF3306 to VJOand ALS, and the NASA Astrobiology Institute LifeUnderground (Grant Award # NNA13AA92A to VJO). Thisis NAI-Life Underground Publication Number 111. DAS wassupported as an Agouron Institute Geobiology Fellow.

Quantification and isotopic analysis of dissolved sulfur species

Rapid Commun. Mass Spectrom. 2017, 31, 791–803 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

801

REFERENCES

[1] J. Zopfi, M. E. Bottcher, B. B. Jorgensen. Biogeochemistry ofsulfur and iron in Thioploca-colonized surface sediments inthe upwelling area off central Chile. Geochim. Cosmochim.Acta 2008, 72, 827.

[2] M. A. Huerta-Diaz, A. Tessier, R. Carignan. Geochemistry oftrace metals associated with reduced sulfur in freshwatersediments. Appl. Geochem. 1998, 13, 213.

[3] J. W. Morse, G. W. Luther III. Chemical influences on tracemetal-sulfide interactions in anoxic sediments. Geochim.Cosmochim. Acta 1999, 63, 3373.

[4] A. Schippers, in Sulfur Biogeochemistry – Past and Present,(Eds: J. P. Amend, K. J. Edwards, T. W. Lyons). GeologicalSociety of America, Colorado, 2004, pp. 49–62.

[5] D. E. Canfield, F. J. Stewart, B. Thamdrup, L. De Bradandere,T. Dalsgaard, E. F. Delong, N. P. Revsbech, O. Ulloa. Acryptic sulfur cycle in oxygen-minimum-zone waters offthe Chilean coast. Science 2010, 330, 1375.

[6] D. T. Johnston, B. C. Gill, A. Masterson, E. Beirne,K. L. Casciotti, A. N. Knapp, W. Berelson. Placing an upperlimit on cryptic marine sulphur cycling. Nature 2014,513, 530.

[7] L. Holmkvist, T. G. Ferdelman, B. B. Jorgensen. A crypticsulfur cycle driven by iron in the methane zone of marinesediment (Aarhus Bay, Denmark). Geochim. Cosmochim. Acta2011, 75, 3581.

[8] M. Aoki, R. Kakiuchi, T. Yamaguchi, K. Takai, F. Inagaka,H. Imachi. Phylogenetic diversity of aprA genes insubseafloor sediments on the Northwestern Pacific Marginoff Japan. Microbes Environ. 2015, 30, 276.

[9] E. G. Wilbanks, U. Jaekel, V. Salman, P. T. Humphrey,J. A. Eisen, M. T. Facciotti, D. H. Buckley, S. H. Zinder,G. K. Druschel, D. A. Fike, V. J. Orphan. Microscale sulfurcycling in the phototrophic pink berry consortia of theSippewissett Salt Marsh. Environ. Microbiol. 2014, 16, 3398.

[10] C. M. Hansel, C. J. Lentini, Y. Tang, D. T. Johnston,S. D. Wankel, P. M. Jardine. Dominance of sulfur-fueled ironoxide reduction in low-sulfate freshwater sediments. TheISME J. 2015, 1.

[11] K. á. Norđi, B. Thamdrup, C. J. Schubert. Anaerobicoxidation of methane in an iron-rich Danish freshwater lakesediment. Limnol. Oceanogr. 2013, 58, 546.

[12] J. Milucka, T. G. Ferdelman, L. Polerecky, D. Franzke,G. Wegener, M. Schmid, I. Lieberwirth, M. Wagner,F. Widdel, M. M. M. Kuypers. Zero-valent sulphur is a keyintermediate in marine methane oxidation. Nature 2012,491, 541.

[13] M. R. Raven, A. L. Sessions, W. W. Fischer, J. F. Adkins.Sedimentary pyrite δ34S differs from porewater sulfide inSanta Barbara Basin: Proposed role of organic sulfur.Geochim. Cosmochim. Acta 2016, 186, 120.

[14] M. R. Raven, A. L. Sessions, J. F. Adkins, R. C. Thunell. Rapidorganic matter sulfurization in sinking particles from theCariaco Basin water column. Geochim. Cosmochim. Acta2016, 190, 175.

[15] J. D. Cline. Spectrophotometric determination of hydrogensulfide in natural waters. Limnol. Oceanogr. 1969, 14, 454.

[16] B. B. Jørgensen. A thiosulfate shunt in the sulfur cycle ofmarine sediments. Science 1990, 249, 152.

[17] M. Yasuyuki, K. Fukasawa, T. Koh. Determination of sulfuranions at the ppb level by ion chromatography utilizing theircatalytic effects on the postcolumn reaction of iodine withazide. J. Chromatogr. A 1998, 804, 143.

[18] G. Paris, A. L. Sessions, A. V. Subhas, J. F. Adkins. MC-ICP-MS measurement of δ34S and Δ33S in small amounts ofdissolved sulfate. Chem. Geol. 2013, 45, 50.

[19] Y. Zakon, L. Halicz, F. Gelman. Isotope analysis of sulfur,bromine, and chlorine in individual anionic species by ionchromatography/multicollector-ICPMS. Anal. Chem. 2014,86, 6495.

[20] A. Kamyshny Jr, A. Goifman, J. Gun, D. Rizkov, O. Lev.Equilibrium distribution of polysulfide ions in aqueoussolutions at 25°C: A new approach for the study ofpolysulfides’ equilibria. Environ. Sci. Technol. 2004, 38, 6633.

[21] G. W. Luther III, B. T. Glazer, L. Hohmann, J. I. Popp,M. Taillefert, T. F. Rozan, P. J. Brendel, S. M. Theberge,D. B. Nuzzio. Sulfur speciation monitored in situ with solidstate gold amalgam voltammetric microelectrodes:polysulfides as a special case in sediments, microbial matsand hydrothermal vent waters. J. Environ. Monit. 2001, 3, 61.

[22] E. S. Boyd, G. K. Druschel. Involvement of intermediatesulfur species in biological reduction of elemental sulfurunder acidic, hydrothermal conditions. Appl. Environ.Microbiol. 2013, 79, 2061.

[23] G. L. Newton, R. Dorian, R. C. Fahey. Analysis of biologicalthiols: Derivatization with monobromobimane andseparation by reverse-phase high-performance liquidchromatography. Anal. Biochem. 1981, 114, 383.

[24] R. D. Vetter, P. A. Matrai, B. Javor, J. O’Brien, in BiogenicSulfur in the Environment, (Eds: E. S. Saltzman, W. J. Cooper).American Chemical Society, Washington, DC, 1989,pp. 243–261.

[25] J. Rethmeier, A. Rabenstein, M. Langer, U. Fischer. Detectionof traces of oxidized and reduced sulfur compounds in smallsamples by combination of different high-performanceliquid chromatography methods. J. Chromatogr. A 1997,760, 295.

[26] C. Arndt, F. Gaill, H. Felbeck. Anaerobic sulfur metabolismin thiotrophic symbioses. J. Exp. Biol. 2001, 204, 741.

[27] R. C. Fahey, G. L. Newton. Determination of low-molecular-weight thiols using monobromobimane fluorescent labelingand high-performance liquid chromatography. MethodsEnzymol. 1987, 143, 85.

[28] B. Franz, T. Gehrke, H. Lichtenberg, J. Hormes, C. Dahl,A. Prange. Unexpected extracellular and intracellular sulfurspecies during growth of Allochromatium vinosum withreduced sulfur compounds. Microbiology 2009, 155, 2766.

[29] E. A. Wintner, T. L. Deckwerth, W. Langston, A. Bengtsson,D. Leviten, P. Hill, M. A. Insko, R. Dumpit, E. VandenEkart,C. F. Toombs, C. Czabo. A monobromobimane-based assayto measure pharmacokinetic profile of reactive sulphidespecies in blood. Br. J. Pharmacol. 2010, 160, 941.

[30] Y. Liu, M. Sieprawska-Lupa, W. B. Whitman, R. H. White.Cysteine is not the sulfur source for iron-sulfur cluster andmethionine biosynthesis in the methanogenic ArchaeonMethanococcus maripaludis. J. Biol. Chem. 2010, 285, 31923.

[31] J. Rodriguez, J. Hiras, T. E. Hanson. Sulfite oxidation inChlorobaculum tepidum. Frontiers Microbiol. 2011, 2, 1.

[32] E. Melamud, L. Vastag, J. D. Rabinowitz. Metabolomicanalysis and visualization engine for LC-MS data. Anal.Chem. 2010, 82, 9818.

[33] M. F. Clasquin, E. Melamud, J. D. Rabinowitz. LC-MS dataprocessing with MAVEN: A metabolomics analysis andvisualization engine. Curr. Protocols Bioinformatics 2012,https://doi.org/10.1002/0471250953.bi1411s37.

[34] F. Inagaki, K. Takai, K. H. Nealson, K. Horikoshi. Sulfurovumlithotrophicum gen. nov., sp. nov., a novel sulfur-oxidizingchemolithoautotroph within the ε-Proteobacteria isolatedfrom Okinawa Trough hydrothermal sediments. Int. J.Systemat. Evolut. Microbiol. 2004, 54, 1477.

[35] V. J. Orphan, C. H. House, K. Hinrichs, K. D. McKeegan,E. F. DeLong. Methane-consuming Archaea revealed bydirectly coupled isotopic and phylogenetic analysis. Science2001, 293, 484.

D. A. Smith et al.

wileyonlinelibrary.com/journal/rcm Copyright © 2017 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2017, 31, 791–803

802

[36] W. A. Brand, T. B. Coplen, J. Vogl, M. Rosner, T. Prohaska.Assessment of international reference materials for isotope-ratio analysis (IUPAC Technical Report). Pure Appl. Chem.2014, 86, 425.

[37] R. A. Werner, W. A. Brand. Referencing strategies andtechniques in stable isotope ratio analysis. Rapid Commun.Mass Spectrom. 2001, 15, 501.

[38] T. B. Coplen, W. A. Brand, M. Gehre, M. Gröning,H. A. J. Meijer, B. Toman, R.M. Verkouteren. NewGuidelinesfor δ13C Measurements. Anal. Chem. 2006, 78, 2439.

[39] J. Zopfi, T. G. Ferdelman, H. Fossing, in SulfurBiogeochemistry – Past and Present, (Eds: J. P. Amend, K. J.Edwards, T. W. Lyons). Geological Society of America,Colorado, 2004, pp. 97–116.

[40] K. S. Habicht, D. E. Canfied, J. Rethmeier. Sulfur isotopefractionation during bacterial reduction anddisproportionation of thiosulfate and sulfite. Geochim.Cosmochim. Acta 1998, 62, 2585.

[41] B. Meyer, L. Peter, K. Spitzer. Trends in the chargedistribution of sulfanes, sulfanesulfonic acids,sulfanedisulfonic acids, and sulfurous acid. Inorg. Chem.1977, 16, 27.

[42] A. S. Engel,M. L. Porter, L. A. Stern, S. Quinlan, P. C. Bennett.Bacterial diversity and ecosystem function of filamentousmicrobial mats from aphotic (cave) sulfidicsprings dominated by chemolithoautotrophic"EpsilonproteobacteriaI". FEMS Microbiol. Ecol. 2004, 51, 31.

[43] T. L. Hamilton, D. S. Jones, I. Schaperdoth, J. L. Macalady.Metagenomic insights into S(0) precipitation in a terrestrialsubsurface lithoautotrophic ecosystem. Frontiers Microbiol.2015, 5, 1.

[44] K. S. Dawson, S. Scheller, J. G. Dillon, V. J. Orphan. Distinctmicrobial ecophysiologies and sulfur-cycling in theenvironment. Frontiers Microbiol. 2016, 7, 1.

[45] J. W. Taraska. Mapping membrane protein structure withfluorescence. Curr. Opin. Struct. Biol. 2012, 22, 507.

Quantification and isotopic analysis of dissolved sulfur species

Rapid Commun. Mass Spectrom. 2017, 31, 791–803 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/rcm

803