Research

Factors Affecting the Presence ofDissolved Glutathione in EstuarineWatersD E G U I T A N G , M A R T I N M . S H A F E R , *D A W N A . K A R N E R , J O E L O V E R D I E R , A N DD A V I D E . A R M S T R O N G

Environmental Chemistry and Technology Program,University of WisconsinsMadison, 660 North Park Street,Madison, Wisconsin 53706-1484

We investigated factors influencing the presence of thethiol glutathione (GSH) in estuarine waters. Our studyaddressed thiol phase-association, the biological releasefrom algal cultures, and the role of copper in both thiol releaseand preservation. Our measurements in three diverseestuaries in the continental United States (San Diego Bay,Cape Fear Estuary, and Norfolk Estuary) show thatdissolved GSH, present at sub-nanomolar levels, ispreferentially partitioned into the ultra-filtrate fraction(<1 kDa) in comparison with dissolved organic carbon(DOC). Concentrations of GSH generally increased withincreases in total copper (Cu) levels, although large variabilitywas observed among estuaries. In 30-h exposureexperiments, release of dissolved GSH from the diatomThalassiosira weissflogii into organic ligand-free experimentalmedia was a strong function of added Cu concentration.The released GSH increased from about 0.02 to 0.27fmol/cell as Cu was increased from the background level(0.5 nM) to 310 nM in the modified Aquil media. However,excretion of GSH was lower (up to 0.13 fmol/cell) when cellswere grown in surface waters of San Diego Bay, despitemuch higher total Cu concentrations. Experimentsconducted in-situ in San Diego Bay water indicated thathigh concentrations of added Cu destabilized GSH, whileboth Mn(II) and natural colloids promoted GSH stability. Incontrast, laboratory experiments in synthetic mediaindicated that moderate levels of added Cu enhancedGSH stability.

IntroductionNewly developed methods have enabled researchers todocument the presence of compounds with sulfhydrylgroups [e.g., glutathione (GSH: γ-glutamylcysteinylglycine,(HO2CCH(NH2)CH2CH2CONHCH(CH2SH)CONHCH2-CO2H) and sulfide] in oxic surface waters (1). However, theirrole in trace element cycling and significance to the sulfurbiogeochemical cycle (2) remains elusive. Present at micro-molar levels in eukaryotic cells, GSH plays a critical role incombating oxidative stress generated from xenobiotics orintermediate metabolites because of its exceptional stabilityand strong nucleophilicity (3). Processes involved in themetabolic transformation of GSH in live cells (4) undoubtedly

have analogues in natural surface waters. These mechanismsmay include (i) scavenging of free radicals as in the oxidationof the free form to the disulfide or other species (5); (ii)addition (conjugation) of GSH to dissolved organic carbon(DOC) (6), a mechanism associated with the diagenesis oforganic matter (7); and (iii) decomposition of GSH conjugatesas indicated by the facilitated production of carbonyl sulfurusing dissolved organic matter (8).

Thiol-like compounds are known to be released fromdifferent algal species (9) and likely constitute most of thecopper-complexing ligands released from the marine alga,Emiliania huxleyi (10). Field measurements from the coastalwestern North Sea and English Channel (1) and estuarinewaters of Galveston Bay, TX (11), indicate that low nanomolarlevels of dissolved GSH are closely related to in-situ algalproduction rather than terrestrial inputs (1, 11). GSH likelyplays an important role in trace metal speciation of surfacewaters; however, factors regulating its release and persistenceremain obscure. It is generally accepted that trace metalsfacilitate autoxidation of thiols, one reason that phototrophicmicroorganisms utilize GSH and not cysteine as the majorsulfur species when adapting to the oxygen-rich atmosphere(12). In oxic surface waters, however, the typically very labilesulfhydryl group is stabilized, likely through the formationof metal-thiolate clusters as indicated, for example, by thepresence of metal-sulfide complexes in freshwaters (13, 14).GSH thus could potentially contribute significantly to thepool of uncharacterized metal-complexing ligands (15),especially for soft metals [e.g., Ag(I), Hg(II), Cd (II), andCu(I)]. Field studies suggest that GSH may be partiallycomplexed by Cd, Cu, and Pb in estuarine waters (16).

In this paper, we summarize the current state of knowledgeabout the presence of GSH in surface estuarine waters. Wealso demonstrate the release of GSH from a coastal diatom,Thalassiosira weissflogii, in response to short-term exposuresto Cu and present new data on GSH distribution in threemarine estuaries around the coastal United States. Finally,we discuss the role of trace metals, especially Cu, in affectingGSH stability in surface waters.

Materials and MethodsSampling Sites. Three geochemically contrasting marineestuarine systems were chosen for study (Figure 1) in orderto capture large gradients in metal speciation. Two systemswere located on the U.S. East Coast (Norfolk-HamptonRoads/James River/Elizabeth River system in Virginia andthe Cape Fear River Estuary in North Carolina) and one onthe West Coast (San Diego Bay, California). Selected samplingsites within two of the systems (Norfolk and San Diego) reflectsignificant impacts from anthropogenic Cu loading, prin-cipally from Cu-based anti-fouling agents used on naval andpleasure craft (17, 18). These estuaries differ markedly intheir hydraulic loading and residence time and in sourcesand types of DOC. In brief, San Diego Bay is characterizedby very low fluvial loading (high salinity and minimalterrestrial inputs) such that autochthonous DOC dominates.Hydraulic and chemical residence times are relatively longwith strong gradients from North to South Bay. These factorsalong with strong point and sediment sources of Cucontribute to high total Cu concentrations and large gradientsin free Cu levels. In contrast, the Cape Fear system is entirelyfluvial dominated with predominantly allochthonous-ter-restrial-sourced DOC. Residence times are very short, andanthropogenic Cu loading is quite low, resulting in low total

* Corresponding author phone: (608)262-0140; fax: (608)262-0454;e-mail: mmshafer@facstaff.wisc.edu.

10.1021/es030669g CCC: $27.50 2004 American Chemical Society VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4247Published on Web 07/14/2004

and free Cu levels. Salinity gradients are large (31-8 psu)and both spatially and seasonally variable. The Cape FearRiver drains the Piedmont wetlands, and organic carbon levelsin the upper estuary can be extremely high (>1000 µM). TheNorfolk system is very diverse geochemically and, in severalrespects, falls on a gradient between the Cape Fear and SanDiego systems. Salinity levels at our principal sampling sitesare moderate and relatively stable (22-18 psu). Thoughterrestrial inputs are large, both autochthonous and alloch-thonous DOM can be important, dependent upon the specificsampling location and season. A large range of total Cu levelsis observed across the system, reflecting the interplay ofcomplex hydrology and strong localized Cu loading. Samplingsites within each of the three study systems were selected tofurther exploit differences in the sources and nature of metal-binding ligands and dissolved-phase geochemistry.

Field Collections. Modern trace metal clean techniqueswere implemented throughout sampling and analyticalprocessing (19). All sample bottles (Teflon, LDPE, FLPE),sampling equipment (Teflon), and filters (PP) were exhaus-tively cleaned in either hot nitric (Teflon) or sequential roomtemperature acid leaches (FLPE, LDPE, PP). Samplingsupplies and equipment were double- or triple-bagged inclean plastic, and field personnel wore full-Tyvek gowns.Critical sample processing was performed under HEPA hoodsin the field laboratories. Surface waters (∼1.5 m depth) werecollected from small (7 m) fiberglass boats using an all-Teflonsampling train supported on a polyethylene boom. Filteredsamples were obtained, in-line, using acid-leached, allpolypropylene, 0.4 µm capsule filters (Meissner). Trace metalsamples (unfiltered and filtered) for total concentrationmeasurements were acidified in the field with ultrapure nitricacid (Ultrex). Large (25 L) composites for metal speciationwere sampled into acid-clean FLPE carboys. Time from field

collection to initiation of laboratory separations (e.g., ul-trafiltration) was typically in the range of 3-6 h. Samples forDOC measurement were collected in 20-mL acid-washedand ashed borosilicate vials with Teflon-lined septa caps andimmediately frozen. Samples for particulate thiol andchlorophyll a analysis were collected on combusted GF/Ffilters and immediately frozen. Samples for dissolved thiolanalysis were acidified immediately using methanesulfonicacid (MSA) at a ratio of 200 µL of MSA (50%):100 mL of sample(final pH around 1.7) and stored frozen until analysis. Amultiparameter sonde (Hydrolab) was used to obtain tem-perature, salinity, pH, dissolved oxygen, and depth measure-ments.

Laboratory Processing. Our trace metal clean ultrafil-tration protocols are described elsewhere in detail (20, 21).A brief summary is presented here. A tangential flowultrafiltration (UF) system, with all wetted components madefrom Teflon (except UF membrane), was used for separation.A dual-stage Teflon diaphragm pump generated the cross-flow (typically 2 L min-1) through the UF cartridge of 2.5 ft2

surface area (Millipore TFF-PrepScale Regenerated Cellulosewith 1 kDa nominal molecular mass cutoff). During typicaloperation little additional back pressure, beyond that gener-ated by the pump (6-10 psi), was applied, and resultingpermeate flow rates were in the range of 20-30 mL min-1.Concentration factors were held close to 8, a compromisebetween the need to minimize concentration polarizationbut still achieve the desired separation and concentration ofcolloids. Extensive cleaning and quality control enabled adetailed mass balance of carbon and trace elements to beconstructed for each field sample. Subsamples of feed,permeate, and retentate collected in Teflon bottles for tracemetals were acidified and analyzed later using ICP-MS (19).Throughout the paper, we define colloidal as the fractionfalling between 1 kDa (ultrafiltration) and 0.4 µm (filtration).Organic carbon-based phases dominate the colloidal mate-rial, and it is unlikely that non-OC phases contribute morethan 15% to the colloidal mass based upon Fe and selectedAl measurements.

Dissolved GSH was measured using HPLC fluorometricdetection coupled with solid-phase extraction of mBBr-derivatized thiols (22). Cellular thiols were determined byHPLC analysis using a monobromobimane (mBBr) deriva-tization technique (23). A GSH detection limit (3σ of a 0.1 nMGSH spike) of 0.03 nM was determined, and overall precisionof the measurements are in the range of 5-6% for GSHconcentrations between 0.5 and 20 nM, increasing to 10%at the 0.1 nM level (22). Pigments were extracted in 90:10acetone:water (24) and speciated by HPLC. DOC in both thedissolved and ultra-filtrate fractions was determined usinga high-temperature combustion analyzer (Shimadzu TOC VCSH/CSN). At the 100 µM DOC level, the coefficient ofvariation averaged 2.5%. Trace metals in aqueous phaseswere determined using automated solid-phase chelationICP-MS, after acidification and UV oxidation. Trace metalsin particles and algal cells were solubilized in Teflon bombs,using an automated, ultra-clean, microwave-assisted aciddigestion protocol (Milestone Ethos+). Digests were analyzedfor trace metals using ICP-MS (Thermo Elemental PQ ExCell).Copper method blanks for seawater and Aquil processingaveraged 0.36 nM (standard deviation ) 0.18 nM). For thealgal digests, Cu method blanks averaged 0.55 (( 0.2) nM,which equates to ∼0.01 fmol of Cu/cell.

Field and Laboratory Experiments. Experiments de-signed to address the stability of dissolved thiols wereconducted under two conditions. (i) In-situ degradation:filtered natural water samples (<0.4 µm) from station a inSan Diego Bay in February 2002 (Figure 1) were amendedwith three different Cu levels (+100, +200, +400 nM) inseparate, clear polycarbonate bottles with no headspace. At

FIGURE 1. Sampling stations in San Diego Bay (small lettersdesignate stations sampled in February 2002 and capital lettersdesignate those sampled in May 2002), Cape Fear Estuary in April2002, and Norfolk Estuary in November 2002.

4248 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004

the 200 nM Cu addition level, Mn (1.33 µM, ) 400× ambient)and natural colloids (which increased the DOC concentrationin the sample from 72 to 135 µM) from a sampling site inSouth San Diego Bay were also added to different bottles.The bottles were then immersed in the bay water 1 m belowthe surface on a polycarbonate incubation platform andexposed to the natural day-night illumination for 2 d. Lightflux was monitored with PAR sensor. Dissolved thiol con-centrations from days 1 and 2 were sampled from separatebottles. (ii) In lab degradation: a coastal diatom, T. weissflogii(CCMP 1336), was grown for 48 h in modified Aquil (25)media amended with Cu at several levels to induce thiolexcretion. The filtrate from this treatment was collected anddistributed into various polycarbonate flasks (125 mL each).Flasks were then placed into a 20 °C incubator for photo-degradation evaluation over 16 d under a 16:8 h of light:darkcycle (light intensity of 120 µeinstein m-2 s-1).

A bioassay experiment was performed to determine theeffects of short-term copper exposure on thiol release fromT. weissflogii. The alga was maintained in standard Aquil(25). Seven days prior to the bioassay experiment, cells weretransferred to a modified Aquil solution that lacked Cu, Zn,and EDTA. Exponentially growing cells from the deficientAquil were inoculated at a rate of 4000 cells/ml, into bothcollected natural waters and a series of flasks of modifiedAquil representing a gradient of 65Cu additions. All experi-ments were carried out in 500-mL acid-cleaned, preequili-brated, polycarbonate flasks. Cells were grown under con-tinuous illumination (120 µeinstein m-2 s-1) at 20 °C for 30h. Cells were then harvested under gentle vacuum filtrationfor analysis of cellular pigments and thiols on precombustedGF/F filters and trace metals on acid-cleaned Teflon mem-brane filters (25 mm, 1.0 µm porosity). The filtrate from theGF/F filters was collected for dissolved thiol analysis usingthe method mentioned previously (22). Cells were enumer-ated using a Beckman Coulter EPICS XL flow cytometeragainst standardized count beads (26).

Results and DiscussionIn general, dissolved GSH is present at sub- or low-nanomolarconcentrations in oceanic and coastal/estuarine waters(Table 1). Concentrations of GSH found in estuarine watersof San Diego, Cape Fear, and Norfolk, were lower and lessvariable than reported earlier for Galveston Bay (11). Amongour three sites, the lowest GSH concentrations were foundin the Cape Fear Estuary, where inputs of fluvial watersdominate the hydrology and primary production and au-tochthonous carbon are low. In San Diego Bay, dissolvedGSH concentrations were higher in May than February 2002,suggesting seasonal variability of GSH in estuarine waters.This trend is consistent with measured phytoplanktonpigment levels in the water column. GSH concentrations areexpected to be related to in-situ biological activity, assuggested by the vertical profile and horizontal transect datain Coastal England area (1) and North Atlantic (27). Similarly,elevated GSH concentrations observed in estuarine watersin Galveston Bay were associated with higher chl a values(11).

As shown in Figure 2, most of the GSH was associatedwith the ultra-filtrate (<1 kDa) fraction of the conventionallydefined “dissolved” pool (<0.4 µm), similar to the patternfound in Galveston Bay (11). GSH was a minor componentof DOC (refer to Table 1) and was preferentially distributedinto the ultra-filtrate fraction of the DOC (Table 1). Thepresence of a minor amount of GSH (MW ) 307) in the >1kDa fraction (colloidal fraction) may be due to associationwith DOC and/or due to its permeation/retention behaviorduring ultrafiltration.

We also investigated cellular levels of GSH and release ofGSH from cells as influenced by Cu levels and nature of the TA

BLE

1.Su

mm

ary

ofDi

ssol

ved

Glut

athi

one,

Diss

olve

dOr

gani

cCa

rbon

,Dis

solv

edCu

,Maj

orNu

trien

ts,a

ndCh

loro

phyl

laCo

ncen

tratio

nsin

Surfa

ceEs

tuar

ine

Wat

ersa

loca

tion

date

[Chl

a](µ

gL-

1 )[D

OC]

(mg

L-1 )

DO

C u/d

b

(%)

[GSH

](n

M)

GSH

u/db

(%)

[Cu]

(nM

)[C

u]u/

db

(%)

nitr

ate

(µM

)ph

osph

ate

(µM

)si

licat

e(µ

M)

ref

Co

asta

lEn

gla

nd

Feb

-Ju

l199

50.

4-4.

0-

c-

0.7-

3.6

--

--

--

1N

ort

hA

tlan

tic

Mar

1991

--

-1-

15-

--

--

-27

Gal

vest

on

Bay

(GB

0899

)A

ug

1999

1.6-

26(1

0.1)

1.7-

5.5

(4.8

)22

-65

(46)

0.23

-6.

1(2

.6)

10-

95(8

6)-

-0.

05-

0.76

(0.0

9)0.

35-

5.93

(3.8

5)11

.2-

130

(64.

1)11

San

Die

go

Bay

(SD

0202

)Fe

b20

021.

0-3.

1(1

.5)

0.86

-1.

6(1

.02)

64-

73(6

8)0.

17-

0.52

(0.2

4)46

-10

6(6

5)4.

0-79

(37)

45-

46(4

5)0.

35-

0.49

(0.4

2)0.

80-

1.05

(1.0

0)7.

57-

12.7

(10.

2)th

isw

ork

San

Die

go

Bay

(SD

0502

)M

ay20

021.

7-2.

7(2

.1)

0.80

-1.

7(1

.06)

-0.

23-

0.79

(0.5

0)-

3.4-

69(3

7)-

--

-th

isw

ork

Cap

eFe

arE

stu

ary

(CF0

402)

Ap

r20

021.

7-4.

6(3

.0)

2.6-

10.4

(6.4

0)54

-64

(60)

0.08

-0.

13(0

.09)

44-

102

(73)

7.6-

16(1

2)45

-49

(47)

4.72

-17

.2(1

6.9)

0.62

-1.

47(1

.18)

13.3

-27

.2(2

4.9)

this

wo

rk

No

rfo

lkE

stu

ary

(NF1

002)

Oct

2002

1.0-

5.4

(1.5

)2.

4-3.

5(3

.52)

29-

42(3

8)0.

37-

0.45

(0.4

1)75

-82

(80)

12-

53(2

9)42

-51

(45)

5.49

-21

.6(1

5.5)

1.46

-4.

39(3

.00)

2.06

-37

.4(4

.83)

this

wo

rk

aD

ata

ran

ge

isg

iven

wit

hit

sm

edia

nva

lue

inp

aren

thes

es.

bS

ub

scri

pt

(u/d

)in

dic

ates

the

rati

oo

fco

nce

ntr

atio

ns

inth

eu

ltra

-filt

rate

(u,<

1kD

ao

fn

om

inal

mo

lecu

lar

wei

gh

t)to

tho

sein

the

trad

itio

nal

lyd

efin

edd

isso

lved

ph

ase

(d,

<0.

45µm

).c

No

tav

aila

ble

.

VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4249

growth medium (modified Aquil versus San Diego Bay water).Results from a 30-h Cu exposure experiment on May 2002showed that the cellular GSH quotas in T. weissflogii wererelatively insensitive to Cu addition within each sample set(Figure 3a). However, GSH levels differed significantly amongsamples, likely due to the differences in aqueous chemistrybetween these sample locations (28). In contrast to GSH, wefound total cellular Cu to increase almost linearly with Cuconcentration in the media (Figure 3b). With the exceptionof a few points from site E, Cu uptake was similar amongsample sites for the Cu-amended San Diego waters but slightlyhigher in modified Aquil medium. The cytoplasmic fractionof the total cellular Cu is reported to be relatively constant(∼40%) in this diatom (29), suggesting that cytoplasmic Cuconcentration also followed the trend of total Cu uptake inthis experiment (Figure 3b). We conclude that cellular GSHlevels are not directly coupled to Cu concentration. Thisfinding suggests that GSH does not regulate cytoplasmic Cuactivity through intracellular Cu binding, although Cu and

GSH are present at strikingly similar concentrations. Intra-cellular activities of metals have been shown to be orches-trated by several families of proteins, each a chaperone forvery specific regulatory activity (30).

GSH is the major nonprotein thiol in both eukaryoticorganisms and in prokaryotic cyanobacteria and purplebacteria (12). Maintaining GSH homeostasis is essential tocell growth and development under different environmentalstresses, to balance redox conditions, to scavenge radicals,and to detoxify toxic metals (3, 4, 31, 32). Cellular GSHbiosynthesis is limited by γ-glutamylcysteine (γEC) synthesisand also requires an intermediate from photorespiration (forglycine supply). Therefore, GSH concentrations in plants havebeen observed to be light-dependent (33). Under differentmetal exposure conditions, phytoplankton cellular GSHconcentrations have been shown to vary only within arelatively narrow range (34) at millimolar levels on abiovolume basis (23, 35), when harvested at similar growthtime, although GSH quotas differ among algal species. Ourfindings detailed above and in Figure 3 are consistent withthe very limited published data.

In contrast to the influence on intracellular GSH, theextracellular release of GSH from T. weissflogii was enhancedat the higher Cu exposure (Figure 4; slopes of lines are allsignificantly different than zero), whether expressed in termsof (i) the thiol concentration (nM) in solution or (ii) the cellnumber normalized concentration (fmol/cell). The variationbetween field sites in the concentration of dissolved GSH(Figure 4a) was related in part to intracellular quotas (Figure3a) but also to total cell counts in the cultures. GSH releaseprofiles for sites A and C are significantly different from eachother and from those of sites B, D, and E. On a per cell basis(Figure 4b), the release of GSH increased from 0.02 fmol/cellat background levels of Cu to as high as 0.27 fmol/cell at the

FIGURE 2. Dissolved GSH concentrations in the <0.4 µm fractionfor all sampling stations and in the <1 kDa fraction for stationswhere ultrafiltration was performed.

FIGURE 3. Cellular GSH and Cu concentrations in the diatom, T.weissflogii, exposed to different levels of Cu in a 30-h growthexperiment. Uncertainties were typically within the bounds of thesymbol shape with overall precision of the cellular GSH measure-ments in the range of 0.03-0.06 fmol/cell.

FIGURE 4. Extracellular GSH release from the diatom, T. weissflogii,exposed to different levels of Cu in a 30-h growth experiment. Datafrom Aquil (EDTA-depleted) experiments are shown with filledsymbols, and San Diego field site (SD0502 A-E) data are shownwith open symbols. Uncertainties were typically within the boundsof the symbol shape with overall precision of the measurementsin the range of 5-6% for GSH concentrations between 0.5 and 20nM, increasing to 10% at the 0.1 nM level.

4250 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004

highest Cu addition (310 nM) in the modified Aquil medium.However, when the diatom was grown in various surfacewaters of San Diego Bay, the amounts of GSH released weresubstantially lower. This is expected because complexingligands were not present in the modified Aquil medium (noEDTA added). Thus, Cu activities are lower and have lessimpact on algal cells in the natural water where metal-complexing ligands are present. However, even when Cuadditions were increased to levels sufficient to saturatecomplexing ligands (M.M.S., unpublished data) and producefree-Cu levels comparable to those in Cu-spiked Aquil, GSHrelease in the natural water cultures was less than measuredin the Aquil medium. These medium-dependent differencesmay reflect the role of (i) nutrients in intracellular thiolinduction and/or (ii) microbial activities and solutionchemistry in transformation of released thiols (5, 8).

It is noteworthy that GSH was released even underbackground Cu conditions (0.5 nM in Aquil; 4-40 nM in SanDiego waters) and that release increased with Cu exposurewhether expressed as thiol excretion in solution (Figure 4a)or on a per cell basis (Figure 4b). We know that in Aquilmedia the diatom begins to show evidence of stress at Cuconcentrations higher than 100-150 nM; at much higherconcentrations, the cells are dying (as evidenced by reduc-tions in growth rates and increases in dead/live cell ratios)but not yet dead after the 30 h exposure (within exponentialgrowth phase). Depending upon the concentrations of naturalligands, much higher Cu exposure levels were required innatural samples to produce similar effects. If release of GSHwere due to passive leakage from cells at a constant rate (36),irrespective of Cu exposure, one would expect [given thatGSH cell quota remains relatively constant (Figure 3a)] thatthe dissolved GSH concentrations to be higher in fast-growingcultures due to higher cell numbers. The passive leakagehypothesis cannot exclusively explain the observed resultsin that released GSH concentrations were lower at lower Cuadditions, corresponding to fast growing cultures (Figure 4a).We believe the enhanced release of GSH upon Cu additionis closely related to Cu-induced cell membrane damage,especially under high Cu exposures. Excretion of GSH reflectsphysiological conditions during algal growth, and its releaseis likely not an enzymatic response (i.e., positive feedback)of the algae to control the trace metal speciation in the media(D.T., submitted for publication).

The physical-chemical forms of GSH released from thediatom are uncertain. The relative levels of cellular GSH andCu shown in Figure 3 suggest, but do not prove, that GSH-Cu complexes were released. The observed low molecularweight nature of detected GSH (Figure 2 and Table 1) suggeststhat GSH was not present as metal-thiol clusters of mediumto high molecular weights, such as the metal-sulfidecomplexes found in some locations (13, 14). The free thiolform of GSH can be easily oxidized by common oxidants[e.g., the reversible reaction with hydrogen peroxide to formthe disulfide (GSSG) at neutral pH (5)] or irreversiblytransformed to other species [e.g., the facilitated productionof carbonyl sulfide through reaction with DOC (8)]. As shownin the Supporting Information (Figure S1), the oxidation ofdissolved free GSH to GSSG is a relatively fast process (apseudo-first-order loss constant of 0.2 h-1 was determined).The generated disulfide (GSSG), on the other hand, isrelatively stable and readily detectable as part of the totalGSH (22) reported in Table 1. Although free GSH is thoughtto be the major fraction detected in voltammetric measure-ment (1, 27, 37), the presence of the disulfide form cannotbe ruled out because reduction potentials for the free anddisulfide forms are the same (38).

Recognizing the susceptibility of GSH to chemical andbiological transformations, we investigated the stability ofGSH in San Diego Bay waters. We explored the influence of

Cu and two other factors (Mn and DOC levels). We focusedon Cu because it is the only trace metal likely to have asignificant influence on dissolved GSH or sulfide speciation(1). The Cu exposure levels used in this series of experimentsspan the range of concentrations used in our laboratory-based bioassays. The lower end concentrations are nontoxicto T. weissflogi, and at the higher levels, varying degrees oftoxicity are expressed depending upon natural ligand levels.Our in-situ degradation experiment using San Diego Baywater collected at station a in February 2002 produced twoseemingly contradictory results regarding the influence ofCu addition (Figure 5). Cu additions resulted in increaseddissolved GSH concentrations (nearly doubled) comparedwith the control ([Cu] ) 4.0 nM) in the first 24 h. It is not clearwhat caused this increase. Planktonic algae had beenremoved by filtration (0.4 µm), but the increased GSH mayhave been derived from remaining microbes (39). It is alsoconceivable that Cu(II) may have promoted release of GSHfrom its DOC conjugates, due to thiol oxidation by Cu (II)and its complexation with Cu(I). However, after 48 h, thehigher Cu additions caused increasing losses of GSH ascompared to the controls. Among treatments with addedCu, GSH values were statistically identical in all treatmentsafter 24 h, except for the slightly higher value in the sampleamended with colloids. GSH concentrations in all samplesdecreased after 48 h, with the Cu-alone treatments showingthe most dramatic change (Figure 5).

High Cu concentrations appear to facilitate the loss ofGSH during the degradation experiments. The 120 nMaddition lost (on an absolute basis) similar amounts of GSHas the control (6.5 nM); however, the trend in GSH lossbetween 24 and 48 h with increasing Cu additions is clearwith 6.5 nM (12%) lost at 120 nM Cu, 24.5 nM (52%) lost at220 nM Cu, and 41 nM (88%) lost at 400 nM. Organic carbonenrichment (natural colloids from site b) had little influenceon GSH release at 48 h when compared to the Cu-aloneaddition at a similar level (200 nM). However, a high levelof Mn(II) (1.33 µM) decreased the loss of GSH. The mech-anisms accounting for the influence of Mn (II) on GSHbiogeochemistry are uncertain. We note that the kineticallyslow oxidation reaction of Mn(II) has been shown to beenhanced via a photochemical mechanism upon the intro-duction of reactive oxygen species (mainly superoxide radical)produced from humic substances (40). Also, bacterial-mediated Mn(II) oxidation was shown to require multicopperoxidase and thus was enhanced in the presence of Cu (41).Because the oxidation of Mn(II) and GSH likely compete forelectron acceptors, the presence of Mn(II) may reduce theoxidation rate of GSH. Other trace metals may also stabilizeGSH. In Galveston Bay (16), both Cd(II) and Pb(II) co-varied

FIGURE 5. Dissolved GSH concentration changes during in-situincubation of filtered natural estuarine water of San Diego Baycollected at station a in February 2002. The horizontal arrow indicatesthe original GSH concentration (0.17 nM) in the samples.

VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4251

with GSH, suggesting that GSH may be present and stabilizedas metal complexes in surface waters.

We also investigated the stability of dissolved thiolsreleased by T. weisssflogii cultured in modified Aquil mediain a 16-d laboratory-based degradation experiment. Wecompared the release of GSH with that of γEC, a precursorin GSH biosynthesis. Significant differences were foundbetween GSH and γEC (Figure 6). In contrast to γEC, GSHreleased from this diatom showed a dose-response pat-tern upon Cu addition (similar to results shown in Fig-ure 4), and GSH concentration changes were proportionallymuch larger over the 16-d experiment. After an initial in-crease in GSH concentrations from day 0 to day 4, GSHconcentrations tended to decrease in the absence of Cubut remain unchanged in the presence of added Cu. Thesefindings suggest that Cu acts to stabilize released GSH,and that released GSH species are readily detectable after a16-d incubation experiment. Therefore, it is not surprisingthat GSH can be detected in surface estuarine waters,although the exact chemical form of the GSH present is stillunclear.

Seemingly contradictory results on the role of Cu in thiolstabilization are suggested by the lab and field incubationexperiments (Figures 5 and 6), in that Cu appeared to stabilizeGSH in the lab experiment while facilitating GSH loss in thein-situ experiment. However, the experimental conditionsin these two experiments, in terms of exposure time, addedCu levels, nature and amount of DOC, microbial activities,and available light spectrum (natural sun light vs fluorescentlight) were different. More Cu was added in the in-situexperiment (Figure 5) in order to saturate the ambient strongCu-complexing ligands (M.M.S., unpublished data). Muchlower DOC levels in the lab experiments could result inreduced superoxide production, and lower microbial activityin the lab experiment may have resulted in less degradationof GSH in the lab. Additionally, low concentrations of Cu

likely stabilize GSH, while high concentrations appear topromote degradation. This hypothesis is supported by thestability of GSH over several days in estuarine waters atambient Cu concentrations. Other metals (e.g., Mn) also helpto stabilize GSH, similar to the reported stabilization of sulfidein river waters (13).

All GSH data from this investigation and a previous studyof Galveston Bay (11, 16) are plotted against the total dissolvedCu in Figure 7a and plotted against DOC in Figure 7b. Dataare normalized to chl a concentrations in an attempt tofactor out production influences and isolate preserva-tion effects in these waters. Considered in total, the datafrom the pooled sites do not show clear relationships be-tween normalized GSH concentration and either Cu or DOCconcentration. However, quite strong trends are apparent indata from individual sites (e.g., positive correlation of GSHto Cu except at Cape Fear; negative relation of GSH to DOCexcept at Norfolk). The narrow ranges in both DOC and Cuconcentrations at Galveston Bay confound trend analysis.The differences between sites or sampling times may berelated in part to the presence of different algal commu-nities. For example in San Diego Bay, normalized GSHconcentrations were higher in May than in February, as wastotal phytoplankton pigment levels, but pigment profilessuggest an even greater shift in species composition. Thecarotenoid pigment 19-hexanoyl fucoxanthin was detectedat all sampling sites in February and site A in May (data notshown), indicating the presence of E. huxleyi. The chla-normalized GSH concentrations are reported to be lowerin E. huxleyi than T. weissflogii and to decrease rapidly withCu exposure (34), indicating that E. huxleyi could beresponsible for the lower GSH in San Diego Bay in February.

FIGURE 6. Concentration changes of filterable GSH and γEC duringa 16 d incubation period after release into the growth media of T.weissflogii.

FIGURE 7. Chlorophyll a-normalized GSH concentrations as relatedto both dissolved Cu and DOC concentrations in estuarine watersaround the United States [data from this study and that reportedpreviously (11)].

4252 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 16, 2004

However, other phytoplankton species with different GSHcell quotas (26, 27) could also contribute to the observeddifferences in GSH levels in San Diego Bay in February andMay.

The lack of a strong relationship of GSH concentrationacross systems, sites, and season to a single variable is notsurprising in view of the expected influences of biota speciestype, other metals (e.g., Mn), and concurrent and possiblyopposing effects of Cu and DOC on GSH release andpreservation. Copper very likely promotes GSH release butcan enhance either preservation or degradation, dependingon the system, as demonstrated in our lab and fieldexperiments. Similarly, DOC may enhance either degradation[via indirect photolysis (8)] or preservation [via GSH con-jugation or sorption (6)].

In summary, we have shown that GSH, the major lowmolecular weight thiol in phytoplankton, is released fromphytoplankton in significant amounts and that release isenhanced by Cu. Given the observed lability of GSH insolution and our measurements documenting the presenceof significant levels of GSH in natural waters, it is plausiblethat the dissolved species of GSH detected is the more stabledisulfide form (GSSG) and/or complexed with metals. Theinfluences of Cu on GSH stability are complex. Our experi-ments indicated that added Cu promoted degradation ofGSH in a natural water but enhanced stability in a syntheticmedium. These differences may be due to contrasts in GSHconcentration, GSH/Cu ratios, and/or chemical compositionof the natural water and synthetic medium. Dissolved organicmatter may promote stability through conjugation or enhancedegradation through photochemical reactions. Although ourunderstanding of the biogeochemical transformation of GSHand other thiols in natural waters is improving, our currentlevel of knowledge does not enable us to make clearassessments as to the specific importance of thiols in thesulfur biogeochemical cycle, including the important issuesof trace metal bioavailability and the related mechanisms ofatmospheric carbon dioxide sequestration (42).

AcknowledgmentsThis work was supported in part by SERDP, a partnership ofU.S. Department of Defense, the Department of Energy, andthe Environmental Protection Agency. The authors are verygrateful for support from colleagues at SPAWAR-San Diego,Old Dominion University, and University of North Carolina-Wilmington in field sampling. Special thanks also go toJocelyn Hemming, Miel Barman, and others at the WisconsinState Laboratory of Hygiene for the help with the cultureexperiments.

Supporting Information AvailableFigure showing time course of relative absorbance changeof GSH-DTNB adducts. This material is available free ofcharge via the Internet at http://pubs.acs.org.

Literature Cited(1) Al-Farawati, R.; Van Den Berg, C. M. G. Environ. Sci. Technol.

2001, 35, 1902-1911.(2) Watts, S. F. Atmos. Environ. 2000, 34, 761-779.(3) Meister, A.; Anderson, M. E. Annu. Rev. Biochem. 1983, 22, 711-

760.(4) Noctor, G.; Arisi, A. C. M.; Jouanin, L.; Kunert, K. J.; Rennenberg,

H.; Foyer, C. H. J. Exp. Bot. 1998, 49, 623-647.(5) Deutsch, J. C.; Santhosh-Kumar, C. R.; Kolhouse, J. F. J.

Chromatogr. A 1999, 862, 161-168.(6) Ciglenecki, I.; Cosovic, B.; Vojvodic, V.; Plavsic, M.; Furic, K.;

Minacci, A.; Baldi, F. Mar. Chem. 2000, 71, 233-249.

(7) Urban, N. R.; Ernst, K.; Bernasconi, S. Geochim. Cosmochim.Acta 1999, 63, 837-853.

(8) Flock, O. R.; Andreae, M. O.; Drager, M. Mar. Chem. 1997, 59,71-85.

(9) Vasconcelos, M.; Leal, M. F. C.; van den Berg, C. M. G. Mar.Chem. 2002, 77, 187-210.

(10) Leal, M. F. C.; Vasconcelos, M.; van den Berg, C. M. G. Limnol.Oceanogr. 1999, 44, 1750-1762.

(11) Tang, D. G.; Hung, C. C.; Warnken, K. W.; Santschi, P. H. Limnol.Oceanogr. 2000, 45, 1289-1297.

(12) Fahey, R. C.; Buschbacher, R. M.; Newton, G. L. J. Mol. Evol.1987, 25, 81-88.

(13) Rozan, T. F.; Lassman, M. E.; Ridge, D. P.; Luther, G. W. Nature2000, 406, 879-882.

(14) ] Rozan, T. F.; Benoit, G.; Luther, G. W. Environ. Sci. Technol.1999, 33, 3021-3026.

(15) Donat, J. R.; Bruland, K. W. In Trace Metals in Natural Waters;Salbu, B., Steinnes, E., Eds.; CRC Press: Boca Ration, FL, 1995;pp 247-281.

(16) Tang, D. G.; Warnken, K. W.; Santschi, P. H. Mar. Chem. 2002,78, 29-45.

(17) [17] Chadwick, D. B.; Zirino, A.; Rivera-Duarte, I.; Katz, C. N.;Blake, A. C. Limnol. Oceanogr. 2004, 49, 355-366.

(18) Johnson, H. D.; Grovhoug, J. G.; Valkirs, A. O. Copper Loadingto U.S. Navy Harbors, Norfolk, VA; Pearl Harbor, HI; and SanDiego, CA; SPAWAR Sytems Center: San Diego, CA, 1998; 79 pp.

(19) Shafer, M. M.; Overdier, J. T.; Hurley, J. P.; Armstrong, D.; Webb,D. Chem. Geol. 1997, 136, 71-97.

(20) Wen, L. S.; Stordal, M. C.; Tang, D. G.; Gill, G. A.; Santschi, P.H. Mar. Chem. 1996, 55, 129-152.

(21) Hoffmann, S. R.; Shafer, M. M.; Babiarz, C. L.; Armstrong, D. E.Environ. Sci. Technol. 2000, 34, 3420-3427.

(22) Tang, D.; Vang, K.; Karner, D. A.; Armstrong, D. E.; Shafer, M.M. J. Chromatogr. A 2003, 998, 31-40.

(23) Rijstenbil, J. W.; Wijnholds, J. A. Mar. Biol. 1996, 127, 45-54.(24) Hurley, J. P.; Garrison, P. J. Can. J. Fish. Aquat. Sci. 1993, 50,

2713-2722.(25) Price, N. M.; Harrison, G. I.; Hering, J. G.; Hudson, R. J.; Nirel,

P. M. V.; Palenik, B.; Morel, F. M. M. Biol. Oceanogr. 1988/1989,6, 443-461.

(26) Franklin, N. M.; Stauber, J. L.; Lim, R. P. Environ. Toxicol. Chem.2001, 20, 160-170.

(27) Le Gall, A. C. A.; Van Den Berg, C. M. G. Deep-Sea Res. PartI-Oceanogr. Res. Pap. 1998, 45, 1903-1918.

(28) Rijstenbil, J. W.; Dehairs, F.; Ehrlich, R.; Wijnholds, J. A. Aquat.Toxicol. 1998, 42, 187-209.

(29) Chang, S. I.; Reinfelder, J. R. Environ. Sci. Technol. 2000, 34,4931-4935.

(30) Finney, L. A.; O’Halloran, T. V. Science 2003, 300, 931-936.(31) Grill, E.; Loffler, S.; Winnacher, E.-L.; Zenk, M. H. Proc. Natl.

Acad. Sci. 1989, 86, 6838-6842.(32) May, M. J.; Vernoux, T.; Leaver, C.; Van Montagu, M.; Inze, D.

J. Exp. Bot. 1998, 49, 649-667.(33) Noctor, G.; Foyer, C. H. Annu. Rev. Plant Physiol. Plant Mol.

Biol. 1998, 49, 249-279.(34) Ahner, B. A.; Wei, L. P.; Oleson, J. R.; Ogura, N. Mar. Ecol. Prog.

Ser. 2002, 232, 93-103.(35) Satoh, M.; Hirachi, Y.; Yoshioka, A.; Kobayashi, M.; Oyama, Y.

J. Phycol. 2002, 38, 983-990.(36) Bjornsen, P. K. Limnol. Oceanogr. 1988, 33, 151-153.(37) Leal, M. F. C.; Van den Berg, C. M. G. Aquat. Geochem. 1998,

4, 49-75.(38) Forsman, U. J. Electroanal. Chem. 1981, 122, 215-231.(39) Lomans, B. P.; van der Drift, C.; Pol, A.; Op den Camp, H. J. M.

Cell. Mol. Life Sci. 2002, 59, 575-588.(40) Nico, P. S.; Anastasio, C.; Zasoski, R. J. Geochim. Cosmochim.

Acta 2002, 66, 4047-4056.(41) Francis, C. A.; Tebo, B. M. Appl. Environ. Microbiol. 2001, 67,

4272-4278.(42) Morel, F. M. M.; Price, N. M. Science 2003, 300, 944-947.

Received for review October 22, 2003. Revised manuscriptreceived May 19, 2004. Accepted May 20, 2004.

ES030669G

VOL. 38, NO. 16, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4253