Modeling and Separation–Detection Methods to Evaluate the Speciation of Metals for Toxicity Assessment

Page 1

41

Journal of Toxicology and Environmental Health, Part B, 9:41–61, 2005Copyright© Taylor & Francis Inc.ISSN: 1093–7404 print / 1521–6950 onlineDOI: 10.1080/15287390500196172

MODELING AND SEPARATION–DETECTION METHODS TO EVALUATE THE SPECIATION OF METALS FOR TOXICITY ASSESSMENT

Joseph A. Caruso1, Rodolfo G. Wuilloud1, Jorgelina C. Altamirano2, Wesley R. Harris3

1University of Cincinnati, Cincinnati, Ohio, 2U.S. Food and Drug Administration, Cincinnati, Ohio, and 3University of Missouri–St. Louis, St. Louis, Missouri, USA

There is an increasing appreciation for the importance of speciation in the assessment of metal toxicity. In this review, twoapproaches to speciation are discussed, with an emphasis on their application to biological samples. One approach is thedirect separation and detection of metal species of toxicological interest. Various “hyphenated” techniques, consisting of achromatographic system coupled to inductively coupled plasma–mass spectrometry (ICP-MS), are discussed. The chromato-graphic strategies employed for separation emphasize liquid chromatography (LC), but the increasing use of gas chromatog-raphy (GC) and capillary electrophoresis (CE) in speciation analysis is discussed. The second approach to speciation is theuse of computer models to calculate the speciation of a metal ion within a complex mixture of ligands. This approach isapplicable to systems in which the metal cation exchanges ligands rapidly, so that the sample represents an equilibrium mix-ture of metal complexes. These computational models are based on the equilibrium constants for the metal complexes anda series of mass balance equations and give the distribution of metal complexes in the original sample. This approach is illus-trated using the speciation of Al(III) in serum as an example.

The assessment of metal toxicity is complicated because bioavailability, mobility, and, ulti-mately, toxicity are dependent on the specific chemical forms (species) of the element present inthe biological system. For example, chromium(VI) is a more hazardous carcinogen than chro-mium(III) (Barceloux, 1999). Metals in lipophilic organometallic species, such as tetraethyllead, canbe more neurotoxic than ionic complexes because of their ability to diffuse across the blood–brainbarrier. The oxoanion vanadate (VO4

3−) enters cells because it structurally resembles phosphate,while vanadyl (VO2+) behaves as a divalent cation (Chasteen, 1983). The impact of speciation ontoxicity is discussed in more detail in Yokel et al. (2005).

As a result of these variations in toxicity among different chemical species, total element con-centration may be uninformative or even misleading in risk assessment. It is important to identifythe specific chemical species in a biological or environmental sample. One approach is the directseparation, identification, and quantification of individual species. This requires selective and sensi-tive analytical techniques. This review discusses several “hyphenated” techniques, which consist ofone technique for separation, such as liquid chromatography or capillary electrophoresis, combinedwith inductively coupled plasma–mass spectrometry (ICP-MS) as a sensitive, element-specificdetector. The scope and limitations of these techniques are discussed. Particular attention is givento sample preparation, sensitivity, and selectivity, since these influence the precision and accuracyof the analytical results. Numerous examples from the literature of the application of these methodsto both environmental and biological samples are cited, but the focus of this review is primarily onthe experimental methods.

Many divalent and trivalent metal ions are labile, meaning that the ligands in the first coordi-nation sphere of the metal ion exchange with free ligands in the solution (including watermolecules) within seconds or minutes. Under these circumstances, a complex mixture of metalions and ligands will quickly adopt an equilibrium composition of metal complexes. Attempts toseparate the metal complexes from each other and from free ligands will perturb this equilibriumdistribution, so that any subsequent analysis of various fractions will give an incorrect picture of

This article is based on a workshop entitled “Metal Speciation in Toxicology: Determination and Importance for Risk Assessment,”presented at the 42nd annual meeting of the Society of Toxicology, March 2003, Salt Lake City, UT.

Address correspondence to Wesley R. Harris, Department of Chemistry and Biochemistry, University of Missouri–St. Louis,St. Louis, MO 63121, USA. E-mail: [email protected].

Page 2

42 J. A. CARUSO ET AL.

the original species distribution. As an alternative approach, one can incorporate the stabilityconstants of the metal–ligand complexes into a computational model and calculate the concen-trations of the metal complexes in the original complex mixture. This review describes themethods used to construct such computational models. The focus is on the use of these modelsfor biological, rather than environmental samples, and a representative speciation model forAl(III) in human serum is presented.

WHY ICP-MS FOR ELEMENTAL SPECIATION?

The main reasons for the increasing use of ICP-MS are (a) low limits of detection, (b) multiele-ment detection, (c) wide linear calibration range, (d) high throughput, (e) ability to access both qual-itative and quantitative information, and (f) isotope ratio capability (Gray, 1989). However, ICP-MS(a) has limited mass resolution with a quadrupole, allowing some spectral interferences, (b) is prima-rily limited to liquid samples, and (c) produces total atomization of the various species, eliminatingthe possibility for qualitative structural information.

Some of the limitations just listed have been overcome by the introduction of new devices suchas a reaction/collision cell (Tanner et al., 2002). Additionally, the coupling of alternative sampleintroduction systems or different chromatographic separation techniques has extended the applica-bility of ICP-MS for accurate and sensitive elemental analysis in a complicated matrix.

ICP-MS provides element specific detection in the milligrams to submicrograms per liter rangefor many elements. It is a technique of choice for coupling with chromatographic methods such asgas chromatography (GC), liquid chromatography, (LC), supercritical fluid chromatography, andcapillary electrophoresis (CE) (Day et al., 2000; Heitkemper et al., 1998; Sutton et al., 1997; Uden,1995). Generally, the separation and the interface need to be optimized before ICP-MS detection.The ICP is an ion source, which operates at 5000–10,000 K and atmospheric pressure. The argonplasma is generated in a quartz torch under the conditions of a radiofrequency electromagneticfield (27–40 MHz, at a power of 600–1800 W) (O’Connor & Evans, 1999). Liquid samples areintroduced as an aerosol through the center tube of the torch into the plasma by means of a nebu-lizer connected to a spray chamber. Larger droplets are removed in the spray chamber. The nebu-lizer gas flow transports the aerosol to the plasma, where it is desolvated, vaporized, atomized,excited, and ionized (B’Hymer & Caruso, 2000b; Montaser et al., 1998c). Singly charged positiveions, which are very efficiently produced in the plasma, enter into the mass spectrometer throughsampler and skimmer cones and then are focused into the mass analyzer, where they are separatedbased on their mass to charge ratio and then detected (Houk, 1986). The response for mostelements is linear over 4–11 orders of magnitude with typical precision near the signal baseline of0.2–3% relative standard deviation (Montaser, 1998). Commonly, the spectral interferences are lesspronounced in ICP-MS than in other techniques such as ICP-optical emission spectrometry(Montaser, 1998).

Different approaches are utilized to overcome these interferences, including the use of high-resolution mass spectrometers (Moens et al., 1994) and cryogenic desolvation (Alves et al., 1992).More recently, the addition of the shield torch and collision/reaction cell to ICP-MS instruments hasreduced the problems associated with polyatomic species (Thomas, 2002). Extensive discussions ofplasma theory, structures, and applications are given by Montaser et al. (1998a, 1998b).

LIQUID CHROMATOGRAPHY COUPLED TO ICP-MS

Prior separation of different elemental species is required before element detection by ICP-MS.The coupling of high-performance liquid chromatography (HPLC) to ICP-MS to achieve this hasreceived special attention. The availability of different chromatographic modes (reversed-phase,reversed-phase ion pairing, ion-exchange, and size-exclusion chromatography) extends the applica-tion of HPLC–ICP-MS to many analytical situations.

Depending on the specific element and compounds in the sample, ICP-MS may improve detectionlimits for LC by 1000-fold compared to non-element-specific detectors such as ultraviolet (UV).

Page 3

EVALUATING METAL SPECIATION IN TOXICOLOGY 43

Coupling HPLC with ICP-MS is simple, as the separation flow rate used in LC (0.1–1 ml/min, depend-ing on the column) is in the working range of most nebulizers utilized for sample introduction into ICPs.Regular or microbore columns may be used. The latter require a micronebulizer due to the lowermobile phase flow rate. A typical LC–ICP-MS system for elemental speciation is shown in Figure 1.

The selection of the mobile phase has special consequences when plasma-based detectors areemployed. The organic solvent concentration should be less than 20%. During desolvation theargon plasma may pyrolize the organic solvent and leave deposits on the sample injector tube ofthe plasma containment torch and the sample introduction orifice of the mass spectrometer inter-face. With extended chromatographic run times, these narrow orifices may become clogged, lower-ing overall sensitivity. Adding oxygen to the nebulizer gas flow (∼10%, v/v) minimizes the problem,because it reacts with the solvents. Cooling the spray chamber may reduce organic solvent volatilityand hence diminish its introduction into the plasma (Boorn & Browner, 1982). Membrane desolva-tors can remove up to 100% of the organic solvent (Cairns et al., 1996).

Mobile phases containing salt concentrations >0.2% may degrade sensitivity because of saltbuildup on the interface (Sutton et al., 1997). The use of chromatographic columns with ID bores<4.6 mm reduces the amount of organic solvents or salts reaching the interface.

Chromatographic methods produce transient analytical signals requiring detectors that canacquire the signal within the time frame of the chromatographic elution. Most commercial ICP-MSinstruments have this capability.

Size Exclusion ChromatographySize exclusion chromatography (SEC) or gel permeation chromatography is an entropy-

controlled technique in which separation is based on the hydrodynamic molecular volume or sizeof the analyte (Blanco Gonzalez & Sanz-Medel, 2000). SEC is the premier technique for determin-ing molecular weight ranges of macromolecules such as proteins and peptides (Caruso & Montes-Bayon, 2003; Szpunar, 2000; Szpunar et al., 2003).

Molecular separation by SEC uses organic or inorganic stationary-phase packing materials withpores of a particular dimension. Molecules too large to enter the pores elute at the void volume of

FIGURE 1. Schematic diagram of typical LC–, GC–, and CE–ICP-MS systems (not shown to scale). Further detail found in various refer-ences cited in the text.

GC

Interface directly

to torch

LC

Interface to torch

via a nebulizer

CE

Interface to torch

via a nebulizer

GC

Interface directly

to torch

LC

Interface to torch

via a nebulizer

CE

Interface to torch

via a nebulizer

Page 4

44 J. A. CARUSO ET AL.

the column. Smaller molecules migrate into the stationary phase and sequentially elute according tohydrodynamic size. SEC mobile phases employed are conventional buffered solutions (Szpunaret al., 2003). SEC allows correlation between elution volume and molecular mass, providing essen-tial qualitative information on the type of association(s) between elements and compounds in thesample. Mobile phases often contain high salt content to diminish polar interactions between the col-umn and the analyte molecules. Likewise, organic solvents such as methanol or acetonitrile can beused to reduce hydrophobic interactions that can alter the elution volumes. One of the main limita-tions of SEC arises from the reduced number of theoretical plates (imaginary separated layers of thecolumn in where is reached equilibration between the mobile and the stationary phases), whichreduces/compromises resolution, and multidimensional chromatography (when separation is basedon two or more different chromatographic principles) is often used (Szpunar & Lobinski, 2002).

Ion-Exchange ChromatographyIon-exchange chromatography (IEC) is based on competition between an analyte ion and ions

in the mobile phase for oppositely charged functional-group ions on the stationary phase. Ions thatinteract more strongly with the stationary phase move more slowly through the column. Separationsare highly controlled by the mobile phase pH, because it affects the dissociation of weakly acidic orbasic compounds, and by addition of agents that chelate metal analytes (de Leon et al., 2002).

Packing materials for IEC consist of beads of cross-linked styrene and divinylbenzene. Func-tional groups, such as sulfonic or carboxylic acids for cation exchange, and quaternary or primaryamines for anion exchange, are covalently attached to the beads (Fritz, 2000; Muraviev et al.,2000). The mobile phase is usually an aqueous salt/buffer solution, which can be mixed withrequired amounts of organic solvents such as methanol or acetonitrile.

Among the elements determined using IEC-ICP-MS, arsenic has been the most studied by anionexchange. Ionic species studied include inorganic arsenic, organoarsenicals, and arseno-sugars(Heitkemper et al., 2001; Pergantis et al., 2000). In samples that contain the element chlorine, theformation of the 40Ar35Cl+ ion (with m/z = 75) can interfere with the detection of 75As by ICP-MS.This interference can be diminished by separating Cl prior to its introduction into the argon plasma.High-resolution MS can separate 40Ar35Cl+ from 75As+ (Sheppard et al., 1990). Recently the reac-tion/collision cell has been implemented in ICP-MS, in which the 40Ar35Cl+ interference is elimi-nated by gas-phase reactions, energy discrimination, and/or collision with the ions formed in theplasma. Hydrogen, helium, and other gases are normally used for this purpose. 40Ar35Cl+interfer-ence has been effectively reduced with a hexapole or octapole reaction cell (O’Brien et al., 2003;Xie et al., 2002). Table 1, which is not intended to be a comprehensive review, highlights a few ele-ments, applications, and experimental conditions for ion exchange ICP-MS.

Chelating IEC adds the possibility of preconcentration and is an alternative to simple ionexchange. Its greater selectivity means separations for divalent and trivalent ions may be markedlyaffected relative to monovalent ions. Chelating IEC allows separation of transition and alkali metals(Sutton & Caruso, 1999).

Reversed-Phase ChromatographyReversed-phase (RP) separation is one of the most used strategies in LC. RP columns have a

nonpolar stationary phase [commonly octadecyl (C18) or octyl (C8) chains] bonded to a solid support(generally microparticulate silica gel, nonpolar). The mobile phases are polar, and the analytes par-tition between the mobile and stationary phases. Aqueous mobile phases containing organic modi-fiers (e.g. methanol, ethanol, acetonitrile, or tetrahydrofuran) to improve the selectivity of thedifferent species are normally used. The mobile-phase pH affects the dissociation of the analytesand hence, the separation. The main advantage of RP for separation of elemental species prior toICP-MS detection is its simplicity. Mobile-phase organic modifiers may cause difficulties with ele-ment measurement by ICP-MS, but methanol or ethanol concentrations up to 5–15% (v/v) do notseriously compromise the technique. Micronebulizers allow higher organic modifier concentrationsand the use of microcolumns for species separations at run times lower or similar to those obtainedwith conventional columns (Ackley et al., 2000; Sun et al., 2003). RP-HPLC–ICP-MS has been used

Page 5

45

TABL

E 1.

Sum

mar

y of

App

licat

ion

of L

iqui

d C

hrom

atog

raph

y C

oupl

ed to

ICP-

MS

for S

peci

atio

n of

Som

e Im

port

ant E

lem

ents

Chr

omat

ogra

phic

sep

arat

ion

Sam

ple

Anal

yzed

spe

cies

Sepa

ratio

n ty

peC

olum

nM

obile

pha

seD

etec

tion

limits

Refe

renc

e

As Free

ze-d

ried

carr

ots

As(

III),

As(V

), m

onom

ethy

lars

onic

ac

id, d

imet

hyla

rsin

ic

acid

, and

ar

seno

beta

ine

Ani

on e

xcha

nge

Wat

ers

IC-P

ak A

nion

H

R, (7

5 ×

4.6

mm

)10

mM

am

mon

ium

ca

rbon

ate,

pH

10.

00.

15, 0

.11,

0.1

3, 0

.24,

an

d 0.

14 n

g m

l−1,

resp

ectiv

ely

(Vel

a et

al.,

200

1)

Urin

eA

s(III

), As

(V),

mon

omet

hyla

rson

ic

acid

, dim

ethy

lars

inic

ac

id a

nd

arse

nobe

tain

e

Ani

on e

xcha

nge

Inte

ract

ion

chro

mat

ogra

phy

ION

120

(125

×

3m

m)

5 m

M a

mm

oniu

m

carb

onat

e, p

H

10.3

, gra

dien

t

0.4,

0.4

, 0.3

, 0.4

, and

0.

3 μg

ml−1

, re

spec

tivel

y

(Rits

ema

et a

l., 1

998)

Wat

erA

s(III

), As

(V),

mon

omet

hyla

rson

ic

acid

, and

di

met

hyla

rsin

ic a

cid

Ani

on e

xcha

nge

Pack

ed w

ith

Hyp

erca

rb 5

μm

(1

00 ×

4.6

mm

)

Form

ic a

cid

grad

ient

(0

.0–1

mol

l−1)

40, 7

0, 2

0, a

nd

10ng

L−1

, re

spec

tivel

y

(Maz

an e

t al.,

200

2)

Alga

Foc

us s

erra

tus

Four

ars

enos

ugar

s,

dim

ethy

lars

inat

e,

and

arse

nate

Ani

on e

xcha

nge

PRP-

X100

, 10

μm (2

50

× 4.

1 m

m).

T:40

°C20

mM

NH

4H2P

O4,

pH

5.6

Non

spec

ified

(Mad

sen

et a

l., 2

000)

Cat

ion

exch

ange

Zorb

ax 3

00 S

CX,

5 μ

m

(150

× 4

.6 m

m).

T:

30°C

20 m

M p

yrid

ine,

pH

2.

2

Cd

Rabb

it liv

erC

d-bi

ndin

g m

etal

-lo

thio

nein

-1 (M

T-1)

su

b-iso

form

s

Reve

rse-

phas

eV

ydac

C8

259

VHP

5415

, (15

0 ×

4.6)

A:10

mM

Tris

-HC

l, pH

7.4

; B: 1

0 m

M

Tris-

HC

l, pH

7.4

in

50%

MeO

H,

grad

ient

10 a

nd 2

1 ng

Cd

for

two

maj

or M

T-1

sub-

isofo

rms

(Fer

rare

llo e

t al.,

20

02)

Page 6

46

TABL

E 1.

Sum

mar

y of

App

licat

ion

of L

iqui

d C

hrom

atog

raph

y C

oupl

ed to

ICP-

MS

for S

peci

atio

n of

Som

e Im

port

ant E

lem

ents

(con

tinue

d)

Chr

omat

ogra

phic

sep

arat

ion

Sam

ple

Anal

yzed

spe

cies

Sepa

ratio

n ty

peC

olum

nM

obile

pha

seD

etec

tion

limits

Refe

renc

e

Hum

an u

rine

Cd-

bind

ing

met

al-

loth

ione

in-1

and

-2

sub-

isofo

rms

Ani

on e

xcha

nge

Prot

ein-

Pak

DEA

E-5

PW, 1

0 μm

(7.5

×

75 m

m)

A: 2

mM

TRI

S-H

Cl,

pH 7

.4; B

: 200

mM

Tr

is-H

Cl,

pH 7

.4,

grad

ient

45, 3

5.3,

and

60

pg C

d fo

r thr

ee M

T-1

sub-

isofo

rms

(Infa

nte

et a

l., 1

999)

Coc

oaC

d 2+

, Cd-

wat

er-

inso

lubl

e pr

otei

ns,

Cd-

wat

er-in

solu

ble

poly

sacc

harid

e, a

nd

Cd-

bioa

vaila

ble

com

plex

es

Size

exc

lusio

nSu

perd

ex-7

5 H

R 10

/30

, 13

μm (1

0 ×

300

mm

)

30 m

M T

ris-H

Cl,

pH

7.2

Not

spe

cifie

d(M

ouni

cou

et a

l.,

2002

)

Fish

cyt

osol

sC

d-bi

ndin

g m

etal

loth

ione

in-1

an

d -2

sub

-isof

orm

s

Ves

icle

med

iate

C18

Sph

eriso

rb O

DS

2 m

odifi

ed w

ith

dido

decy

ldim

ethy

lam

mon

ium

bro

mid

e (D

DA

B), 5

μm

(250

×

4.6

mm

)

A: 2

mM

Tris

-HC

l, pH

7.

4, a

nd 1

mM

D

DA

B; B

: 200

mM

Tr

is-H

Cl,

pH 7

.4,

grad

ient

25.5

, 40.

1, a

nd 4

6 pg

C

d fo

r the

thre

e M

T-1

sub-

isofo

rms;

50

and

58 p

g C

d fo

r the

th

ree

MT-

2 su

b-iso

form

s.

(Infa

nte

et a

l., 2

000)

Cr

Wat

er, w

aste

wat

er

and

solid

was

te

extra

ct

Cr(I

II) a

nd C

r(VI)

Ion

exch

ange

afte

r a

chel

atio

n w

ith

sodi

um

ethy

lene

diam

ine

tetra

ceta

te

IonP

Ac A

S7, 1

0 μm

(2

50 ×

4.0

mm

)35

mM

(NH

4)2S

O4,

pH

9.2

40 p

g C

r(III)

and

100

pg

Cr(V

I)(B

yrdy

et a

l., 1

995)

Was

tew

ater

from

the

galv

anic

indu

stry

Cr(I

II) a

nd C

r(VI)

Reve

rse

phas

e af

ter a

ch

elat

ion

with

am

mon

ium

-py

rrol

idin

edith

ioca

rbam

ate

Pack

ed w

ith

LiC

hros

pher

60

RP-

sele

ct B

mat

eria

l; 5

μm (1

25 ×

5 m

m)

Ace

toni

trile

/wat

er

(67:

33%

)0.

2 μg

l−1 C

r (III

) and

0.

1 μg

l−1 C

r (VI

)(A

ndrle

et a

l., 1

997)

Mus

sel h

epat

o pa

ncre

aC

r-m

etal

loth

ione

inlik

e pr

otei

nsSi

ze e

xclu

sion

Prep

arat

ive

Frac

-100

Se

phad

ex G

-75

colu

mn

10 m

M T

ris-H

Cl,

pH

7.4;

5 m

M 2

-m

erca

ptoe

than

ol,

0.1

mM

ph

enyl

met

hylsu

lfon

yl fl

uorid

e; 2

5 m

M

NaC

l

Not

spe

cifie

d(F

erra

rello

et a

l.,

2000

)

Ani

on e

xcha

nge

Mon

o-Q

HR

5/5,

10

μm (5

0 ×

50m

m)

A: 4

mM

Tris

-HC

l, pH

7.

4; B

: 250

mM

am

mon

ium

ace

-ta

te, 1

0 m

M T

ris-

HC

l, pH

7.4

Page 7

47

TABL

E 1.

Sum

mar

y of

App

licat

ion

of L

iqui

d C

hrom

atog

raph

y C

oupl

ed to

ICP-

MS

for S

peci

atio

n of

Som

e Im

port

ant E

lem

ents

(con

tinue

d)

Chr

omat

ogra

phic

sep

arat

ion

Sam

ple

Anal

yzed

spe

cies

Sepa

ratio

n ty

peC

olum

nM

obile

pha

seD

etec

tion

limits

Refe

renc

e

Hum

an s

erum

Cr-

seru

m p

rote

ins

Fast

pro

tein

liqu

id

chro

mat

ogra

phy

(ani

on e

xcha

nge)

Mon

o-Q

HR

5/5,

10

μm

(50

× 50

mm

)A

mm

oniu

m a

ceta

te

(0–0

.25

M g

radi

ent)

in 0

.05

M T

ris-H

Cl,

pH 7

.4

Not

spe

cifie

d(B

ayon

et a

l., 1

999)

Pb Coc

oaPb

2+, P

b-w

ater

-in

solu

ble

prot

eins

, Pb

-wat

er-in

solu

ble

poly

sacc

harid

e an

d Pb

-bio

avai

labl

e co

mpl

exes

Size

exc

lusio

nSu

perd

ex-7

5 H

R 10

/30

, 13

μm (1

0 ×

300

mm

)

30 m

M T

ris-H

Cl,

pH

7.2

Not

spe

cifie

d(M

ouni

cou

et a

l.,

2002

)

Win

ePb

-pec

tic

poly

sacc

harid

e, P

b-rh

amno

gala

ctur

onan

II,

and

Pb-

biom

olec

ules

not

id

entif

ied

Size

exc

lusio

nSu

perd

ex-7

5 H

R 10

/30

, 13

μm (1

0 ×

300

mm

)

30 m

M a

mm

oniu

m

form

ate,

pH

5.2

Not

spe

cifie

d(S

zpun

ar e

t al.,

199

8)

Rain

wat

erTr

imet

hylle

ad a

nd

triet

hylle

adRe

vers

e ph

ase–

ion

pair

Hyp

ersil

OD

S; 5

μm

(2

50 ×

4.2

mm

)M

etha

nol (

60%

v/v

)/H

2O (4

0% v

/v):

0.1

M a

cetic

/ac

etat

e so

lutio

n an

d 4

mM

sod

ium

1-

pent

asul

foni

c ac

id, p

H 4

.6

3 ng

g−1

Pb

and

14 n

g g−1

Pb,

resp

ectiv

ely

(Ebd

on e

t al,.

199

8)

Se Hum

an s

erum

Sele

nium

pre

sent

as

plas

ma

glut

athi

one

pero

xida

se,

sele

nopr

otei

n an

d al

bum

in

Ani

on e

xcha

nge

Mon

o-Q

HR

5/5

FPLC

an

alyt

ical

col

umn;

10

μm

(50

× 50

mm

)

A: 0

.05

M T

ris-H

Cl,

pH 7

.4; B

: 0.0

5 M

Tr

is-H

Cl,

0.5

M

amm

oniu

m

acet

ate,

pH

7.4

, gr

adie

nt

Not

spe

cifie

d(R

eyes

et a

l., 2

003)

Affi

nity

Hitr

ap H

epat

in-

Seph

aros

e an

d H

itrap

Blu

e-Se

phar

ose

(1 m

l ea

ch)

A: 0

.02

M T

ris-H

Cl,

pH 7

.4; B

: 0.0

2 M

Tr

is-H

Cl,

1.4

M

amm

oniu

m a

ce-

tate

, pH

7.4

Page 8

48

TABL

E 1.

Sum

mar

y of

App

licat

ion

of L

iqui

d C

hrom

atog

raph

y C

oupl

ed to

ICP-

MS

for S

peci

atio

n of

Som

e Im

port

ant E

lem

ents

(con

tinue

d)

Chr

omat

ogra

phic

sep

arat

ion

Sam

ple

Anal

yzed

spe

cies

Sepa

ratio

n ty

peC

olum

nM

obile

pha

seD

etec

tion

limits

Refe

renc

e

Yeas

tSe

leno

met

hion

ine,

Se

(IV) a

nd S

e(VI

)A

nion

exc

hang

eH

amilt

on P

RP-X

100,

10

μm

(250

×

4.1

mm

)

5 m

M a

mm

oniu

m c

it-ra

te in

2%

met

ha-

nol p

HA

3.65

, pH

B 8.

0, g

radi

ent p

H 5

Not

spe

cifie

d(H

uert

a et

al.,

200

3)

Whe

at fl

our

Sele

nom

ethi

onin

e an

d tw

o un

know

n sp

ecie

sSe

dim

ents

Sele

nour

ea,

sele

noet

hion

ine,

se

leno

met

hion

ine,

Se

(IV),

Se(V

I),

dim

ethy

lsele

nide

, di

met

hylse

leni

de

Ani

on e

xcha

nge

Dio

nex

AS

11 (4

×

300

mm

)A:

1 m

M N

aOH

in

2% m

etha

nol;

B:0.

5%

tetra

met

hyla

mm

oni

um h

ydro

xide

; T:

30°C

Not

spe

cifie

d(O

chse

nkuh

n-Pe

tropo

ulou

et a

l.,

2003

)

Reve

rse

phas

eC

apce

ll-C

18 (4

.6 ×

35

mm

)A:

10

mM

Tris

-HC

l, pH

7.3

, 1%

m

etha

nol;

B: 1

0 m

M

Tris-

HC

l, pH

7.6

, 50

% m

etha

nol;

T:15

°CBr

azil

nuts

Sele

nom

ethi

onin

e,

sele

noet

hion

ine,

an

d se

leno

cyst

eine

Reve

rse

phas

eA

ltim

a C

8, 5

μm

(4.6

× 15

0 m

m)

Met

hano

l (90

%)/

5m

M c

itric

aci

d,

5m

M h

exas

ulfo

nic

acid

, pH

3.5

(10%

), T:

30°

C

Not

spe

cifie

d(V

onde

rhei

de e

t al.,

20

02)

Hum

an u

rine

Sele

nom

ethi

onin

e (S

eMet

), m

ethy

lsele

nom

ethi

oni

ne (M

eSeM

et),

sele

nocy

stei

ne,

sele

noga

mm

aam

ino

buty

ric a

cid,

tri

met

hylse

leno

nium

io

n (T

MSe

)

Ion-

pairi

ng re

vers

e ph

ase

Luna

C8,

3 μ

m (1

.0 ×

10

0 m

m)

Met

hano

l (20

%)/

wat

er, 1

0 m

M

hept

aflu

orob

utan

oic

acid

, T: 2

5°C

0.8

(MeS

eMet

), 1.

7 (S

eMet

), 1.

0 (T

MSe

) μg

Se

l−1

(Gam

mel

gaar

d et

al.,

20

02)

Met

hano

l (20

%)/

wat

er, 2

mM

trifl

u-or

oace

tic a

cid,

T:

25°C

Page 9

EVALUATING METAL SPECIATION IN TOXICOLOGY 49

to study metalloproteins such as metallothioneins (Ferrarello et al., 2002). The resolution is highenough to separate species that differ by a single amino acid.

In RP ion-pairing (IP) chromatography, ionic and nonionic compounds are separated by addingan ion-pairing reagent to the RP mobile phase. These reagents have a polar head group and a non-polar tail. Examples are tetraalkylammonium or trialkylammonium salts, or anions such as alkylsul-fonates, added at 0.001–0.005 mol/L. The ion pair that is formed between the solute ion and anappropriate counterion in the mobile phase is more hydrophobic than the original analyte speciesand shows a higher affinity for the stationary phase. The mobile phases used in RPIP-HPLC are sim-ilar to those used in RP-HPLC (water–methanol and water–acetonitrile). Table 1 gives a selection ofsome of the species that have been separated and quantified using RPIP-HPLC–ICP-MS. As anexample, a typical separation by RPIP-HPLC is shown in Figure 2 for selenium species using trifluo-roacetic acid (TFA) as the ion-pairing reagent.

CAPILLARY ELECTROPHORESIS COUPLED TO ICP-MS

Capillary electrophoresis is also known as high-performance CE since it has far greater separationefficiency than conventional slab-gel electrophoresis. The CE mode most widely used is capillary zoneelectrophoresis. Analyte separation depends on the solute’s mobility in an electric field rather than achemical interaction and partition between a stationary phase and a mobile phase (Olesik, 2000). Theadvantages of CE were highlighted by Jorgenson and Lukacs (1981, 1983). Performing electrophoreticseparations in capillaries allowed the use of automated analytical equipment, short analysis times, andonline detection of the separated peaks.

Analyte mobility depends on the charge and size of the analyte as well as the electrical fieldapplied across a capillary filled with a suitable electrolyte (Olesik, 2000). Electrophoretic mobility isgoverned by coulombic attraction between the ions and the oppositely charged electrode. The elec-troosmotic flow (EOF) comes from the migration of the double layer, formed with hydrated cations toneutralize the inner, negatively charged silanol groups at the surface of the silica-capillary wall, towardthe more negative potential end of the capillary. Controlling the EOF and the voltage gradient enablesimproved separation of cationic, anionic, and neutral analytes (Kannamkumarath et al., 2002).

FIGURE 2. Separation of selenium compounds by RP-HPLC–ICP-MS obtained by B’Hymer & Caruso, 2000a). Reproduced with permission.

Page 10

50 J. A. CARUSO ET AL.

Although CE is an efficient separation technique and attractive for element speciation, the smallcapillary diameter (20–100 cm long and 25–100 μm ID) compromises the detection limits (Olesik,2000; Timerbaev & Buchberger, 1999), making a sensitive and specific detection system highlydesirable. Lu et al. (1995) and Olesik et al. (1995) first described coupling CE to ICP-MS for rapidelement speciation. Detection limits are now in the low milligrams per liter range. Although ICP-MSis a suitable detector for CE, the low flow rate (microliters per minute) and the small volume sample(high nanoliters) of CE are scarcely compatible with the typical nebulization conditions for ICP-MS(milliters per minute) (Tomlinson et al., 1995). Another issue is maintaining electrical conductivity tothe nebulizer tip. Figure 1 shows schematically CE–ICP-MS coupling. Its high resolution is shown bythe excellent peak shapes in Figure 3. Differences in migration times were less than 3%, and detec-tion limits were ∼1 mg/L. Applications of CE–ICP-MS to samples of biological, environmental,nutritional, and toxicological interest were recently reviewed (Kannamkumarath et al., 2002). WhileCE–ICP-MS is attractive when very small sample sizes are necessitated (e.g., radioactive samples), itcannot yet compete with HPLC–ICP-MS for concentration detection limits, ease of operation andreproducibility.

GAS CHROMATOGRAPHY COUPLED TO ICP-MS

Over the last two decades, species-selective analysis for volatile organometallic compounds hasattracted increasing attention in the toxicological, environmental, and nutritional fields (Lobinski &Adams, 1997). One of the most successful “hyphenated” techniques for separation and determina-tion of volatile elemental species has been GC–ICP-MS (Figure 1).

FIGURE 3. Simultaneous separation of 12 species of four elements by CE-ICP-MS. Concentration of the elements: As, Sb, Te 100 μg L−1

each, Se 1000 μg L−1. From Prange and Schaumloffel (1999), with permission.

Page 11

EVALUATING METAL SPECIATION IN TOXICOLOGY 51

Mercury species are of great concern since they can accumulate in food, especially seafood,and volatile Hg species can be naturally generated by environment processes. They represent a riskfor humans due to their tumorogenic and teratogenic properties and their negative effect on thecentral nervous system (Clarkson, 2002). Selenium has a major nutritional role and is thought topossess cancer chemoprevention properties, and speciation studies to define its biological roles arechallenging (Uden, 2002).

Capillary GC is a high-resolution separation technique and is the preferred separation methodto couple with ICP-MS. GC has a sample-introduction efficiency into the ICP of about 100%. Theuse of solid-phase microextraction improves the detection limits of GC–ICP-MS by preconcentrat-ing volatile species (Wuilloud et al., 2003).

The use of an element-specific detector coupled to GC has been described (Lobinski & Adams,1997). The most common element-specific detectors coupled to a gas chromatograph are a plasmasource using excitation by microwave-induced plasma (MIP-AES) or ICP-MS. ICP-MS can accuratelydetermine elemental isotope ratios. Its sensitivity is unrivaled (Montaser, 1998; Vanhaecke &Moens, 1999). As suggested in Figure 1, the analyte transfer line is directly inserted into the centralchannel of the torch. The addition of oxygen to the plasma gas is advised to prevent carbon deposi-tion and metal entrapment and to reduce the solvent peak, which also may be avoided by solid-phase extraction and selective vaporization (Caruso & Montes-Bayon, 2003).

Three types of columns are used: packed, capillary, and multicapillary (Chong & Houk, 1987;Van Loon et al., 1986). Method sensitivity may be depressed by dispersing the analyte on the col-umn. Several methodologies involving hydride generation purge and trap techniques allow analysisof highly volatile species at temperatures below 100�C (Amouroux et al., 1998; Feldmann, 1999).Capillary GC allows better resolution than packed columns, although in older instruments coolingthe oven can extend analysis time (Rodriguez et al., 1999). Multicapillary GC consists of a pack of900–2000 capillaries of 20–40 mm ID (Lobinski et al., 1999; Rodriguez et al., 1999). This designeliminates deficiencies associated with the use of capillary and packed columns but preserves theadvantages of both. Multicapillary GC is suitable for high flow rates, minimizing the dispersion andthereby facilitating the transport of the analytes to the plasma.

COMPUTATIONAL MODELING OF MIXTURES OF LABILE COMPLEXES

Ligand Exchange RatesMetals ions can be categorized as inert or labile based on the exchange rate of the inner sphere

water molecules of the metal aquo ion (Helm & Merbach, 1999). Metal ions with an exchange rateconstant >1/s are considered labile. Exchange rates vary widely, from 109/s for metal ions such asCu2+ and Gd3+ to 10−6/s for the inert Cr3+ ion. Most metal ions have exchange rates ≥100/s andequilibrate rapidly with a mixture of ligands.

The rate of water exchange is often controlled by the rate at which one of the coordinatedwater molecules dissociates from the metal aquo complex (Huheey, 1993). Filling this coordinationvacancy with a new water molecule is very rapid. Ligand exchange rates tend to slow down withincreasing ionic charge, and for a given ionic charge the rates slow down with decreasing ionicradius. This is illustrated by comparing the exchange rates for the d10 ions Al3+, Ga3+, and In3+. Therates change from about 1/s for the smallest ion, Al3+ (ionic radius = 0.535 Å) to about 250/s forGa3+ (ionic radius = 0.63 Å) to over 106/s for In3+ (ionic radius = 0.800 Å) (ionic radii taken fromShannon, 1976).

A labile metal ion in the presence of a mixture of ligands rapidly adopts an equilibrium distribu-tion of metal complexes. The composition of this mixture can be calculated from known stabilityconstants. Certain configurations of metal d-electrons energetically disfavor the transient complexesformed during ligand exchange, which slows down ligand exchange (Huheey, 1993). These ligandfield effects account for almost all inert metal ions. Rates tend to be particularly slow for d3 ions(Cr3+) and low-spin d6 ions (Co3+, Ru2+). Such metals are generally inappropriate choices for com-puter speciation models.

Page 12

52 J. A. CARUSO ET AL.

Definition of Stability ConstantsThe generic equilibrium for the formation of a metal complex is

where M represents the metal aquo ion, L represents the ligand, and H+ is a hydrogen ion (Martell &Motekaitis, 1988). Charges on the metal and ligand vary and are omitted for simplicity. Formal equi-librium constants are expressed in terms of activities, not concentrations. However, we will adopt astandard definition of βijk as

where the square brackets denote molar concentrations. Such constants are “effective” stabilityconstants in that they apply to a solution at a specific temperature and ionic strength. This issue isaddressed laters in the discussion of the selection of the appropriate stability constants for a model.

Most ligands bind protons as well as metal ions, and it is critical to understand that [L] in Eq. (2)always refers to the fully deprotonated form of the ligand. The presence of the H+ term in Eq. (2)accounts for the fact that with multidentate ligands, the metal may bind to a subset of the possibledonor groups while one or more donor groups remain protonated.

One should be aware that other conventions for writing stability constants are used. As anexample, consider a system in which a metal and ligand combine at low pH to form a mixture ofML and a protonated chelate MHL. The stability constants for such species are often reported as

A comparison of Eqs. (2) and (3) shows that KML = β110. However, there can be no βijk constantthat corresponds to KMHL. The formulation of KMHL contains a complex species (ML) in the denomi-nator, whereas the equilibrium quotients for βijk must contain only elementary components, that is,[M], [L], and [H]. In order to include the MHL species in the calculation, it is necessary to use theconstants in Eqs. (3) and (4) to calculate the appropriate βijk value as

KMHLKML = β111

Mass Balance EquationsSpeciation calculations are based on a series of simultaneous mass balance equations. For a sys-

tem containing one metal ion (M2+) and one ligand (HL), which combine to form ML+, ML2, andML3

−, the mass balance equation for the total metal ion concentration ([M]tot) is

[M]tot = [M2+] + [ML+] + [ML2] + [ML3�]

i j kijk

i j kM L H M L H+ + ⎯ →⎯⎯← ⎯⎯⎯+ β (1)

βijki j k

i j k= +

[ ]

[ ] [ ] [ ]

M L H

M L H(2)

KMLMLM L

= [ ][ ][ ]

(3)

KMHLMHLML H

= [ ][ ][ ] (4)

(5)

(6)

Page 13

EVALUATING METAL SPECIATION IN TOXICOLOGY 53

Equation (6) can be rewritten as

[M]tot = [M2+] + β110[M2+][L�] + β120[M

2+][L�]2 + β130[M2+][L�]3

A similar mass balance equation can be constructed for the total ligand concentration. Once [H+] isspecified, the two mass balance equations can be solved to determine [M2+] and [L−].

The metal, the ligand, and the hydrogen ion are the components of the system, and [M2+], [L−],and [H+] are the free component concentrations. Any chemical entity formed by the combinationof two or more free components is a species. There is a mass balance equation for each compo-nent. Thus each model consists of n mass balance equations with n unknown free componentconcentrations.

Several programs for speciation calculations are available. ECCLES is a Fortran program devel-oped by David Williams and co-workers (May et al., 1977). The program can be adjusted toaccommodate essentially any number of components and species. This program has been usedvery extensively for modeling the distribution of metal complexes in biological fluids such as serumand gasterointestinal fluid (Harris, 1992; Brumas et al., 1993; Jarvis et al, 1995, Whitburn et al.,1999). ECCLES gives a very detailed output report as a text file. It lacks any graphical user interfaceand does not allow for solid phases.

HySS is a Windows-based program (Alderighi et al., 1999). It can be downloaded from http://www.chem.leeds.ac.uk/People/Peter_Gans/hq2000.htm. species is also a Windows-based programthat is bundled with a searchable database of stability constants (Pettit & Powell, 1997). It canaccommodate solid phases, but is rather restricted as to the maximum number of componentsand species. The program and the database can be purchased from Academic Software(www.acadsoft.co.uk). HySS and species report results either as tables or as plots of species as afunction of either pH or a component concentration.

Several speciation programs have been developed for the primary purpose of modeling the spe-cies distribution of metal ions in environmental samples. A partial list of such programs includesMINTEQL (Allison et al., 1991), PHREEQC (Parkhurst, 1995), CHESS (Santore & Driscoll, 1995),and EQ3/6 (Wolery, 1992). Full versions of these programs often include a database of formal sta-bility constant corrected to 25�C and zero ionic strength. Free versions of the software with a morelimited database are available for MINTEQA2 (http://www.epa.gov/ceampubl/mmedia/minteq/index.htm) and PHREEQC (http://wwwbrr.cr.usgs.gov/projects/GWC_coupled/phreeqc).

When the software database consists of stability constants at zero ionic strength (e.g.,MINTEQA2), the user specifies an ionic strength for each calculation, and the program applies cor-rections to the constants to match the specified conditions. This approach works better for systemssuch as freshwater samples, for which the ionic strength corrections are relatively small. For rela-tively high-ionic-strength biological samples such as serum (∼0.16 M ionic strength), the correctionsare less accurate. For this reason, we have emphasized the programs like ECCLES that allow theuser to select stability constants that have been measured at the desired ionic strength.

DEVELOPMENT OF A SPECIATION MODEL

The first step in the development of a speciation model is to assign the total concentration ofeach component in the chosen biological fluid. These values are typically selected from standardcompilations, such as the Geigy Scientific Tables (Lenther, 1981). The model must include allpotentially important chelating agents. This typically includes the common amino acids as well asother low-molecular-mass (LMM) ligands such as citrate, phosphate, oxalate, and so on. Authorssometimes include proteins such as albumin and transferrin if the appropriate metal-protein bindingconstants are known.

The second step is to specify the species to be included. The art of performing speciation calcu-lations lies largely in the identification of all the important chemical species for the model. Themodel must include the important metal-ligand complexes, including possible 2:1 and 3:1 com-plexes. However, other species need to be included in a complete model. Some divalent and most

(7)

Page 14

54 J. A. CARUSO ET AL.

trivalent metal ions are extensively hydrolyzed at neutral pH. Hydrolysis constants can be writtenusing the βijk formalism as

where the stoichiometric coefficient ( j ) for the ligand is zero, and the stoichiometric coefficient forthe hydrogen ion is negative to indicate that the [H+] term appears in the numerator, rather thanthe denominator, of the equilibrium quotient. These constants are used to include the metal–hydroxo complexes in the mass balance equation. For some trivalent metal ions such as Ga(III) andAl(III), the omission of the appropriate hydrolysis terms can lead to errors in the free componentconcentrations of several orders of magnitude (Harris et al., 1994).

The interaction of metal ions with hydroxide can also be represented as the binding of OH− as aligand,

The expressions for hydrolysis shown in Eqs. (8) and (9) are related to one another by Kw, theautoionization constant for water. At zero ionic strength and 25�C,

Kw = [H+][OH�] = 10�14

A value of KOH can be converted to β11–1 as

β10–1 = KOHKw

Speciation programs usually require that hydrolysis constants be entered in the β10–1 format.Neutral metal–hydroxo complexes are typically very insoluble. For a dihydroxo species such as

M(OH)2 (solid), the solubility product is

Ksp ≥ [M2+][OH�]2

If any amount of the solid phase is present, the ion product on the right-hand side of Eq. (12) mustequal Ksp. When a speciation program does not allow for solid phases, the user should check thefinal values of the free components against known solubility products to ensure no solid phaseswould be expected to form under the conditions modeled.

The set of species should also include protonated forms of the ligands. Omitting protonatedligands can lead to serious errors. For example, at pH 7.4, only 0.005% of phosphate is present asthe fully deprotonated free component PO4

3−. Failure to include the phosphate pKas will overesti-mate the free phosphate concentration by four orders of magnitude.

The speciation of trace metals can also be affected by competition from metal ions such asCa(II) and Mg(II). These are present at such high concentrations that a significant fraction of the totalligand may be present as the Ca(II) or Mg(II) complex. The effective binding affinity of the ligandtoward the target metal ion will be reduced in direct proportion to the fraction of the total ligandthat is bound to Ca(II) and Mg(II).

Speciation models for metal ions in biological fluids seldom attempt to include redox equilibria,and programs such as ECCLES do not allow metal ions to change oxidation state. It is more com-mon to include redox equilibria in geochemical models, so programs such as MINTEQ, CHESS, andPHREEQC do include the solution reduction potential as a variable in the model.

β10-kM OH H

M=

+[ ( ) ][ ][ ]

kk

(8)

KOHMOHM OH

= [ ][ ][ ]

(9)

(10)

(11)

(12)

Page 15

EVALUATING METAL SPECIATION IN TOXICOLOGY 55

The final step is the selection of the appropriate equilibrium constant for each species. As notedearlier, the effective equilibrium constants vary with temperature and solution ionic strength. Thusin selecting the most appropriate value for a particular equilibrium constant, one should search forvalues that match as closely as possible the temperature and ionic strength of the solution that oneis modeling. Even for the same experimental conditions, reported stability constants often vary con-siderably, forcing one to make subjective judgments as to the most appropriate value. If several ofthe reported log β values cluster within a reasonably narrow range, outliers can be identified andavoided. Other factors to consider include the experimental method used to determine the con-stant and the reputation of the laboratory reporting the constants.

To conduct a complete assessment of binding constants from the primary literature would beenormously time-consuming, so one usually relies on stability constant databases. For many years thebest-known database was the six-volume set of stability constants compiled by Martell and Smith(Martell & Smith, 1974, 1977, 1982; Smith & Martell, 1975, 1976, 1989). These authors conducted a“critical” assessment of the primary literature and reported only the “best” value for each stability con-stant. The National Institute of Standards and Technology now maintains an updated version of thisdatabase in a searchable, electronic format (http://www.nist.gov/srd/nist46.htm).

Another electronic database of stability constants is available from Academic Software in theUnited Kingdom (Pettit & Powell, 1997). It lists essentially all the values (with references) that havebeen reported for a given equilibrium constant, so the user must choose the most appropriatevalue. There is one additional source specifically for hydrolysis constants (Baes & Mesmer, 1976).Although quite old, it is a very useful, comprehensive treatment of hydrolysis reactions.

SPECIATION OF ALUMINUM IN SERUM

Aluminum (Al) speciation has been of some interest, primarily in relation to its neurotoxicity(Yokel, 2000). Unfortunately, the agreement among the published speciation models for Al inserum is very poor (Harris et al., 1996). The variation among the models reflects the disarray in theprimary literature for aluminum binding constants for citrate and phosphate. This system isdescribed here because it illustrates the difficulties that can be encountered in the construction of acomputer model.

The 1:1 Al–citrate complex (Al(cta)) undergoes one or two sequential deprotonation reactionsto form species designated as Al(H−1cta)− and Al(H−1cta)(OH)2− (Harris et al., 1996). The presumedstructures for these complexes are shown in Figure 4. For most metal ions it would be straightfor-ward to determine the stability constants for these complexes by potentiometric titration. However,at the millimolar concentrations of Al required for potentiometric titrations, the Al(H−1cta)− com-plex trimerizes to form a very stable Al3(H−1cta)3(OH)4− complex (Feng et al., 1989). This trimer is sodominant that it is very difficult to determine accurate stability constants of the Al(H−1cta)− andAl(H−1cta)(OH)2− monomers (Harris et al., 2003).

The reported values for the fraction of low-molecular-mass (LMM) Al in serum vary from 57 to80% (Harris et al., 1996). Two studies report that Al(H−1cta)− is the most important Al–citrate species

FIGURE 4. Proposed structures for mononuclear aluminum–citrate complexes.

Al3+ OH

2

O OH2

O

O

OH2

O

O

OH

O

Al3+ OH

2

O OH2

O

O

OH2

O

O

O

O

Al3+ OH

2

O OH-

O

O

OH2

O

O

O

O

Al(cta) Al(H-1cta)- Al(H-1cta)(OH)

2-

Page 16

56 J. A. CARUSO ET AL.

(Harris, 1992; Jackson, 1990), one reports that Al(H−1cta)(OH)2− is the most important citrate spe-cies (Duffield et al., 1991), and one study omits both these species and reports no binding to citrate(Dayde et al., 1990).

There are also serious difficulties in the determination of Al-phosphate binding constants. Theneutral Al(PO4) precipitates at pH ∼3.5 in potentiometric studies (Atkári et al., 1996), which makesit very difficult to measure the binding constants for Al(PO4) and the Al(PO4)(OH)− complex thatforms at higher pH. In the absence of experimental binding constants, one can estimate bindingconstants using linear free energy relationships (LFER). For a hard metal ion such as Al(III), there isoften a positive linear correlation between the affinity of the donor group for H+ and its affinity forAl(III) (Harris, 1992; Atkári et al., 1996). The pKa for the HPO4

2− anion has been used to predict abinding constant of 106.13 for Al(HPO4)

+ (Atkári et al., 1996). Harris (1992) has used similar LFER toestimate binding constants for both Al(PO4) and Al(PO4)(OH)−.

At the low micromolar Al concentrations found in serum, the citrate trimer is much less stable,and monomeric complexes of citrate and phosphate are the dominant species. Harris et al. (2003)used difference ultraviolet (UV) spectroscopy to measure effective binding constants for Al–phosphate and Al–citrate at a total Al concentration of only 12 μM. These constants have beenincorporated into a new speciation model for Al with a total of 75 species, including 27 Al com-plexes. In this new model (model 1 in Table 2) transferrin binds 97% of serum Al, slightly higherthan the experimental value of ∼90% (Sanz-Medel et al., 2002). For a total Al concentration of3 μM, the total concentration of LMM Al complexes is 78 nM, consisting almost entirely of 4 species:Al(H−1cta)− (48%), Al(OH)3 (22%), Al(PO4)(OH)− (14%), and Al(PO4)2

3− (5%).Model 1 has been modified to illustrate the importance of some of the variables involved

in constructing a valid equilibrium model. To illustrate the importance of including Ca(II) andMg(II) as competitors to Al(III), both these metal ions were deleted as components in model 2.This increases the total LMM Al from 78 to 289 nM, primarily due to a fivefold increase in the[Al(H−1cta)] (Table 2).

In model 3 the pKas for phosphate were eliminated, and the total concentration of LMM Alincreased from 78 nM to 457 nM. The impact of eliminating the phosphate pKas, while clearly sig-nificant, is moderated by increased Ca(II) and Mg(II) binding to phosphate. Otherwise the effect ofremoving the phosphate pKas on the aluminum distribution would be much greater.

There is growing experimental evidence for the formation of mixed-ligand complexes with cit-rate and phosphate such as Al(PO4)(cta)3− (Bantan et al., 1999; Lakatos et al., 2001). A new modelbased on the binding constants from Lakatos et al. (2001) predicts that 31% of Al would be presentas LMM Al, well above the 10% experimental value. It may be that mixed ligand complexes areresponsible for increasing the fraction of LMM Al from the 3% calculated in model 1 to the 10%observed experimentally. However, additional experimental work is needed before these mixed-ligand species can be included in computer models.

TABLE 2. Calculated Speciation of Al in Serum

Model 1, basic model Model 2, omit Ca, Mg Model 3, omit phosphate, pKa

pAla 14.3 14.3 14.1Percent Al bound to transferrin 97.4 90.4 84.8Conc. of LMM Al 78 nM 289 nM 457 nMPercentage of low-molecular-mass Al

Al(H−1cta)− 47.6 66.3 13.8Al(H−1cta)(cta)4− — 18.1 —Al(OH)3 21.8 7.6 8.1Al(PO4)(OH)− 13.5 4.3 23.0Al(PO4)2

3− 5.4 2.1 53.8

apAl = −log [Al3+].

Page 17

EVALUATING METAL SPECIATION IN TOXICOLOGY 57

SPECIATION IN TOXICITY MODELS

In the free ion activity model (FIAM), it is assumed that the toxicity of a metal ion more directlyrelates to the free, rather than total, metal ion concentration (Brown & Markich, 2000; Morel,1983). The free metal ion concentration can be calculated using the computational methodsdescribed above. The FIAM is used primarily in aquatic toxicology, where dissolved organic matter(DOM) is one of the metal complexing agents. This is a significant complication because DOM con-sists of large, complex, heterogeneous molecules. Programs such as WHAM and NICA-Donnanhave been developed that are especially suited to model the interactions of metal ions with DOM(Tipping, 1994; Milne et al., 2003).

Pagenkopf (1983) refined the FIAM by attributing metal toxicity specifically to metal binding tofish gills, creating the gill surface interaction model (GSIM). The GSIM explicitly included equilib-rium constants for the binding of the metal ions to cell receptor sites. The concept of explicitlyincluding the cell-surface ligands in the computational speciation model has been generalized asthe biotic ligand model (BLM) (Bell et al., 2002; Di Toro et al., 2001). The BLM can account for theprotective effect of nonbiological ligands, which reduce the free ion concentration, as well as that ofmetal ions such as Ca(II) and Mg(II), which compete with the target metal ion for binding to thebiotic ligand.

The BLM focuses on the role of a specific receptor for the free metal ion in mediating cellularuptake. The BLM tends to fail when a metal ion can enter the cell by a different pathway thatdoes not involve the biotic ligand. The metal may form a lipophilic complex that can passivelydiffuse across the cell membrane. Another possibility is that the metal binds to a LMM ligand thathas its own transport pathway. For example, methylmercury binds to the thiol group of cysteine toform a complex that is taken into cells by the neutral amino acid transporter (Aschner & Clarkson,1989), and it has been suggested that the 1:1 complex of Cd2+ with citrate is taken into cellsvia the citrate transport system (Errécalde & Campbell, 2000). It has also been shown that thecomplex between Ag+ and thiosulfate is taken up by algae via the sulfate transport system (Fortin &Campbell, 2001).

CONCLUSIONS

“Hyphenated” techniques are used for elemental speciation by coupling GC, LC, or CE in theirvarious modes with ICP-MS for detection of the separated analytes. Providing speciation analysessupports the growing need to know the chemical forms of metals in toxicologically important sam-ples. The total amount of an element or metal in a sample is important, but information about thedifferent elemental species is also needed.

For mixtures of labile metal complexes, computer models can be used to calculate the equilib-rium distribution of species in the sample. The appropriate metal chelate stability constants must beknown to construct these models. In addition to the complexes of the target metal ion, it is impor-tant to include other species such as hydroxo species, protonated ligands, and complexes withcompetitive metal ions such as Ca2+ and Mg2+.

REFERENCESAckley, K. L., Sutton, K. L., and Caruso, J. A. 2000. A comparison of nebulizers for microbore LC-ICP-MS with mobile phases containing

methanol. J. Anal. Atom. Spectrom. 15:1069–1073.Alderighi, L., Gans, P., Ienco, A., Peters, D., Sabatini, A., and Vacca, A. 1999. Hyperquad simulation and speciation (HySS): A utility pro-

gram for the investigation of equilibria involving soluble and partially soluble species. Coord. Chem. Rev. 184:311–318.Allison, J. D., Brown, D. S., and Novo-Gradac. 1991. MINEQA2/PRODEFA2, a geochemical assessment model for environmental systems:

Version 3.0 user’s manual. Report, EPA/600/3-91/021.Alves, L. C., Wiederin, D. R., and Houk, R. S. 1992. Reduction of polyatomic ion interferences in inductively coupled plasma mass-

spectrometry by cryogenic desolvation. Anal. Chem. 64:1164–1169.Amouroux, D., Tessier, E., Pecheyran, C., and Donard, O. F. X. 1998. Sampling and probing volatile metal(loid) species in natural waters

by in-situ purge and cryogenic trapping followed by gas chromatography and inductively coupled plasma mass spectrometry (P-CT-GC-ICP/MS). Anal. Chim. Acta. 377:241–254.

Page 18

58 J. A. CARUSO ET AL.

Andrle, C. M., Jakubowski, N., and Broekaert, J. A. C. 1997. Speciation of chromium using reversed phase-high performance liquid chro-matography coupled to different spectrometric detection methods. Spectrochim. Acta B 52:189–200.

Aschner, M., and Clarkson, T. W. 1989. Methyl mercury uptake across bovine brain capillary endothelial cells in vitro: the role of aminoacids. Pharmacol. Toxicol. 64:293–297.

Atkári, K., Kiss, T., Bertani, R., and Martin, R. B. 1996. Interactions of aluminum(III) with phosphates. Inorg. Chem. 35:7089–7094.Baes, C. F., and Mesmer, R. E. 1976. The hydrolysis of cations. New York: Wiley.Bantan, T., Milacic, R., Mitrovic, B., and Pihlar, B. 1999. Investigation of low molecular weight Al complexes in human serum by fast pro-

tein liquid chromatography (FPLC)-ETAAS and electrospray (ES)-MS-MS techniques. J. Anal. Atom. Spectrom. 14:1743–1748.Barceloux, D G. 1999. Chromium. J. Toxicol. Clin. Toxicol. 37:173–194.Bayon, M. M., Cabezuelo, A. B. S., Gonzalez, E. B., Alonso, J. I. G., and Sanz-Medel, A. 1999. Capabilities of fast protein liquid chroma-

tography coupled to a double focusing inductively coupled plasma mass spectrometer for trace metal speciation in human serum.J. Anal. Atom. Spectrom. 14:947–951.

Bell, R. A., Ogden, N., and Kramer, J. R. 2002. The biotic ligand model and a cellular approach to class B metal aquatic toxicity. Comp.Biochem. Physiol. C 133:175–188.

B’Hymer, C., and Caruso, J. A. 2000a. Evaluation of yeast-based selenium food supplements using high-performance liquid chromatog-raphy and inductively coupled plasma mass spectrometry. J. Anal. Atom. Spectrom. 15:1531–1539.

B’Hymer, C., and Caruso, J. A. 2000b. Nebulizer sample introduction for elemental speciation. In Elemental speciation: New approachesfor trace element analysis, ed. K. L. Ackley, Vol XXXIII, pp. 213–226. New York: Elsevier.

Blanco Gonzalez, E., and Sanz-Medel, A. 2000. Liquid chromatographic techniques for trace element speciation analysis. In Elementalspeciation: New approaches for trace element analysis, ed. K. L. Ackley Vol. 33, pp. 81–121. Amsterdam: Elsevier Science.

Boorn, A. W., and Browner, R. F. 1982. Effects of organic solvents in inductively coupled plasma atomic emission spectrometry. Anal.Chem. 54:1402–1410.

Brown, P. L., and Markich, S. J. 2000. Evaluation of the free ion activity model of metal–organism interaction: Extension of the concep-tual approach. Aquat. Toxicol. 51:177–194.

Brumas, V., Alliey, N., and Berthon, G. 1993. A new investigation of copper(II)-serine, copper(II)-histidine-serine, copper(II)-asparagine,and copper(II)-histidine-asparagine equilibria under physiological conditions, and implications for simulation models relative toblood plasma. J. Inorg. Biochem. 52:287–296.

Byrdy, F. A., Olson, L. K., Vela, N. P., and Caruso, J. A. 1995. Chromium speciation by anion-exchange high-performance liquid chroma-tography with both inductively-coupled plasma-atomic emission spectroscopic and inductively-coupled plasma-mass spectrometricdetection. J. Chromatogr. A 712:311–320.

Cairns, W. R. L., Ebdon, L., and Hill, S. J. 1996. A high performance liquid chromatography–inductively coupled plasma-mass spectrometryinterface employing desolvation for speciation studies of platinum in chemotherapy drugs. Fresenius J. Anal. Chem. 355:202–208.

Caruso, J. A., and Montes-Bayon, M. 2003. Elemental speciation studies - new directions for trace metal analysis. Ecotoxicol. Environ.Safety. 56:148–163.

Chasteen, N. D. 1983. The biochemistry of vanadium. Struct. Bond. 53:107–138.Chong, N. S., and Houk, R. S. 1987. Inductively coupled plasma-mass spectrometry for elemental analysis and isotope ratio determina-

tions in individual organic-compounds separated by gas-chromatography. Appl. Spectrosc. 41:66–74.Clarkson, T. W. 2002. The three faces of mercury. Environ. Health Perspect. 110:11–23.Day, J. A., Caruso, J. A., Becker, J. S., and Dietze, H. J. 2000. Application of capillary electrophoresis interfaced to double focusing sector

field ICP-MS for nuclide abundance determination of lanthanides produced via spallation reactions in an irradiated tantalum target.J. Anal. Atom. Spectrom. 15:1343–1348.

Dayde, S., Filella, M., and Berthon, G. 1990. Aluminum speciation studies in biological fluids. Part 3. Quantitative investigation of aluminum-phosphate complexes and assessment of their potential significance in vivo. J. Inorg. Biochem. 38:241–259.

de Leon, C. A. P., Montes-Bayon, M., and Caruso, J. A. 2002. Elemental speciation by chromatographic separation with inductively cou-pled plasma mass spectrometry detection. J. Chromatogr. A. 974:1–21.

Di Toro, D. M., Allen, H. E., Bergman, H. L., Meyer, J. S., Paquin, P. R., and Santore, R. C. 2001. Biotic ligand model of the acute toxic-ity of metals. I. Technical basis. Environ. Toxicol. Chem. 20:2383–2396.

Duffield, J. R., Edwards, K., Evans, D. A., Morrish, D. M., Vobe, R. A., and Williams, D. R. 1991. Low molecular mass aluminum com-plexes speciation in biofluids. J. Coord. Chem. 23:277–290.

Ebdon, L., Hill, S. J., and Rivas, C. 1998. Lead speciation in rainwater by isotope dilution-high performance liquid chromatographyinductively coupled plasma mass spectrometry. Spectrochim. Acta B 53:289–297.

Errécalde, O., and Campbell, P. G. C. 2000. Cadmium and zinc bioavailability to Selenastrum capricornutum (chlorophyceae): Acciden-tal metal uptake and toxicity in the presence of citrate. J. Phycol. 36:473–483.

Feldmann, J. 1999. Determination of Ni(CO)4, Fe(CO)5, Mo(CO)6, and W(CO)6 in sewage gas by using cryotrapping gas chromatographyinductively coupled plasma mass spectrometry. J. Environ. Monitor. 1:33–37.

Feng, T. L., Gurian, P. L., Healy, M. D., and Barron, A. R. 1989. Aluminum citrate: Isolation and structural characterization of a stable tri-nuclear complex. Inorg. Chem. 29:408–411.

Ferrarello, C. N., Bayon, M. M., de la Campa, R. F., and Sanz-Medel, A. 2000. Multi-elemental speciation studies of trace elements asso-ciated with metallothionein-like proteins in mussels by liquid chromatography with inductively coupled plasma time-of-flight massspectrometric detection. J. Anal. Atom. Spectrom. 15:1558–1563.

Ferrarello, C. N., Encinar, J. R., Centineo, G., Alonso, J. I. G., de la Campa, M. R. F., and Sanz-Medel, A. 2002. Comparison of three dif-ferent ICP-MS instruments in the study of cadmium speciation in rabbit liver metallothionein-1 using reversed-phase HPLC andpost-column isotope dilution analysis. J. Anal. Atom. Spectrom. 17:1024–1029.

Page 19

EVALUATING METAL SPECIATION IN TOXICOLOGY 59

Fortin, C., and Campbell, P. G. C. 2001. Thiosulfate enhances silver uptake by a green alga: Role of anion transporter in metal uptake.Environ. Sci. Technol. 35:2214–2218.

Fritz, J. S. 2000. Ion chromatography. New York: Wiley-VCH.Gammelgaard, B., Bendahl, L., Sidenius, U., and Jons, O. 2002. Selenium speciation in urine by ion-pairing chromatography with perflu-

orinated carboxylic acids and ICP-MS detection. J. Anal. Atom. Spectrom. 17:570–575.Gray, A. L. 1989. The origins, realization and performance of ICP-MS systems. In Applications of inductively coupled plasma mass spec-

trometry, ed. A. L. Gray, pp. 1–42. London: Blackie.Harris, W. R. 1992. Equilibrium model for speciation of aluminum in serum. Clin. Chem. 38:1809–1818.Harris, W. R., Chen, Y., and Wein, K. 1994. Equilibrium constants for the binding of indium(III) to human serum transferrin. Inorg. Chem.

33:4991–4998.Harris, W. R., Berthon, G., Day, J. P., Exley, C., Flaten, T. P., Forbes, W. F., Kiss, T., Orvig, C., and Zatta, P. 1996. Speciation of alumi-

num in biological systems. J. Toxicol. Environ. Health. 48:543–568.Harris, W. R., Wang, Z., and Hamada, Y. Z. 2003. Competition between transferrin and the serum ligands citrate and phosphate for the

binding of aluminum. Inorg. Chem. 42:3262–3273.Heitkemper, D., Zimmer, B. S., Wolnik, K. A., and Caruso, J. A. 1998. Chromatography/plasma mass spectrometry. In Encyclopedia of

environmental analysis and remediation, ed. R. A. Meyers, New York: Wiley.Heitkemper, D. T., Vela, N. P., Stewart, K. R., and Westphal, C. S. 2001. Determination of total and speciated arsenic in rice by ion

chromatography and inductively coupled plasma mass spectrometry. J Anal. Atom. Spectrom. 16:299–306.Helm, L., and Merbach, A. E. 1999. Water exchange on metal ions: Experiments and simulations. Coord. Chem. Rev. 187:151–181.Houk, R. S. 1986. Mass spectrometry of inductively coupled plasmas. Anal. Chem. 58:97A–105A.Huerta, V. D., Reyes, L. H., Marchante-Gayon, J. M., Sanchez, M. L. F., and Sanz-Medel, A. 2003. Total determination and quantita-

tive speciation analysis of selenium in yeast and wheat flour by isotope dilution analysis ICP-MS. J. Anal. Atom. Spectrom.18:1243–1247.

Huheey, J. E. 1993. Inorganic chemistry, 4th ed. New York: Harper and Row.Infante, H. G., Sanchez, M. L. F., and Sanz-Medel, A. 1999. Cadmium-bound species in human urine using high-performance liquid

chromatography–vesicular hydride generation–inductively coupled plasma mass spectrometry. J. Anal. Atom. Spectrom.14:1343–1348.

Infante, H. G., Sanchez, M. L. F., and Sanz-Medel, A. 2000. Vesicle-mediated high performance liquid chromatography coupled tohydride generation inductively coupled plasma mass spectrometry for cadmium speciation in fish cytosols. J. Anal. Atom. Spectrom.15:519–524.

Jackson, G. E. 1990. Aluminum, gallium and indium in biological fluids – a computer model of blood plasma. Polyhedron 9:163–170.Jarvis, N. V., Wagener, J. M., and Jackson, G. E. 1995. Metal-ion speciation in blood plasma as a tool for elucidating the in vivo behavior

of radiopharmaceuticals containing 153Sm and 166Ho. J. Chem. Soc. Dalton Trans. 1411–1415.Jorgenson, J. W., and Lukacs, K. D. 1981. Zone electrophoresis in open-tubular glass capillaries. Anal. Chem. 53:1298–1302.Jorgenson, J. W., and Lukacs, K. D. 1983. Capillary zone electrophoresis. Science 222:266–272.Kannamkumarath, S. S., Wrobel, K., B’Hymer, C., and Caruso, J. A. 2002. Capillary electrophoresis-inductively coupled plasma-mass

spectrometry: An attractive complementary technique for elemental speciation analysis. J. Chromatogr. A 975:245–266.Lakatos, A., Evanics, F., Dombi, G., Bertani, R., and Kiss, T. 2001. Speciation of Al(III) in blood serum—The Al(III)-citrate-phosphate ter-

nary system. Eur. J. Inorg. Chem. 3079–3086.Lenther, C., ed. 1981. Geigy Scientific Tables , 8th ed. West Caldwell, NJ: CIBA-Geigy.Lobinski, R., and Adams, F. C. 1997. Speciation analysis by gas chromatography with plasma source spectrometric detection. Spectro-

chim. Acta B 52:1865–1903.Lobinski, R., Sidelnikov, V., Patrushev, Y., Rodriguez, I., and Wasik, A. 1999. Multicapillary column gas chromatography with element-

selective detection. Trends Anal. Chem. 18:449–460.Lu, G. H., Bird, S. M., and Barnes, R. M. 1995. Interface for capillary electrophoresis and inductively-coupled plasma-mass spectrometry.

Anal. Chem. 67:2949–2956.Madsen, A. D., Goessler, W., Pedersen, S. N., and Francesconi, K. A. 2000. Characterization of an algal extract by HPLC-ICP-MS and

LC-electrospray MS for use in arsenosugar speciation studies. J. Anal. Atom. Spectrom. 15:657–662.Martell, A. E., and Smith, R. M. 1974. Critical stability constants, vol. 1. Amino acids. New York: Plenum Press.Martell, A. E., and Smith, R. M. 1977. Critical stability constants, vol. 3. Other organic ligands. New York: Plenum Press.Martell, A. E., and Smith, R. M. 1982. Critical stability constants, vol. 5. First supplement. New York: Plenum Press.Martell, A. E., and Motekaitis, R. J. 1988. Determination and use of stability constants, pp. 9–11. New York: VCH.May, P. M., Linder, P. W., and Williams, D. R. 1977. Computer simulation of metal-ion equilibria in biofluids: Models for the low-

molecular-weight complex distribution of calcium(II), magnesium(II), manganese(II), iron(III), copper(II), zinc(II), and lead(II) ions inhuman blood plasma. J. Chem. Soc. Dalton. 588–595.

Mazan, S., Cretier, G., Gilon, N., Mermet, J. M., and Rocca, J. L. 2002. Porous graphitic carbon as stationary phase for LC-ICPMS sepa-ration of arsenic compounds in water. Anal. Chem. 74:1281–1287.

Milne, C. J., Kinniburgh, D. G., Van Riemsdijk, W. H., and Tipping, E. 2003. Generic NICA-Donnan model parameters for metal-ionbinding by humic substances. Environ. Sci. Technol. 37:958–971.

Moens, L., Verrept, P., Dams, R., Greb, U., Jung, G., and Laser, B. 1994. New high-resolution inductively-coupled plasma-massspectrometry technology applied for the determination of V, Fe, Cu, Zn and Ag in human serum. J. Anal. Atom. Spectrom.9:1075–1078.

Montaser, A. 1998. Inductively coupled plasma mass spectrometry. New York: Wiley-VCH.

Page 20

60 J. A. CARUSO ET AL.

Montaser, A., McLean, J. A., Liu, H., and Mermet, J.-M. 1998a. An introduction to ICP spectrometries for elemental analysis. In Induc-tively coupled plasma mass spectrometry, ed. A. Montaser, pp. 1–31. New York: Wiley-VCH.

Montaser, A., Minnich, M. G., Liu, H., Gustavsson, A. G. T., and Browner, R. F. 1998b. Fundamental aspects of sample introduction inICP spectrometry. In Inductively coupled plasma mass spectrometry, ed. A. Montaser, pp. 335–420. New York: Wiley-VCH.

Montaser, A., Minnich, M. G., McLean, J. A., Liu, H., Caruso, J. A., and McLeod, C. W. 1998c. Sample introduction to ICPMS. In Induc-tively coupled plasma mass spectrometry, ed. A. Montaser, pp. 83–264. New York: Wiley-VCH.

Morel, F. M. M. 1983. Principles of aquatic chemistry, pp. 300–309. New York: Wiley.Mounicou, S., Szpunar, J., Lobinski, R., Andrey, D., and Blake, C. J. 2002. Bioavailability of cadmium and lead in cocoa: Comparison of

extraction procedures prior to size-exclusion fast-flow liquid chromatography with inductively coupled plasma mass spectrometricdetection (SEC-ICP-MS). J Anal. Atom. Spectrom. 17:880–886.

Muraviev, D., Gorshkov, V., and Warschawsky, A. 2000. Ion exchange. New York: Marcel Dekker.O’Brien, S. E., Acon, B. W., Boulyga, S. F., Becker, J. S., Dietze, H. J., and Montaser, A. 2003. Reduction of molecular ion interferences

with hexapole collision cell in direct injection nebulization–inductively coupled plasma mass spectrometry. J. Anal. Atom. Spec-trom. 18:230–238.

Ochsenkuhn-Petropoulou, M., Michalke, B., Kavouras, D., and Schramel, P. 2003. Selenium speciation analysis in a sediment usingstrong anion exchange and reversed phase chromatography coupled with inductively coupled plasma-mass spectrometry. Anal.Chim. Acta. 478:219–227.

O’Connor, G., and Evans, E. H. 1999. Fundamental aspects of ICP-MS. In Inductively coupled plasma spectrometry and its applications,ed. S. J. Hill, pp. Sheffield: Sheffield Academic Press.

Olesik, J. 2000. Capillary electrophoresis for elemental speciation studies. Comp. Anal. Chem. 33:151–211.Olesik, J. W., Kinzer, J. A., and Olesik, S. V. 1995. Capillary electrophoresis inductively-coupled plasma spectrometry for rapid elemental

speciation. Anal. Chem. 67:1–12.Pagenkopf, G. K. 1983. Gill surface interaction model for trace-metal toxicity to fishes: Role of complexation, pH, and water hardness.

Environ. Sci. Technol. 17:342–347.Parkhurst, D. L. 1995. User’s guide to PHREEQC: A computer program for speciation, reaction-path, advective-transport, and inverse

geochemical calculations. Report USGS/WRIR-95-4227.Pergantis, S. A., Wangkarn, S., Francesconi, K. A., and Thomas-Oates, J. E. 2000. Identification of arsenosugars at the picogram level

using nanoelectrospray quadrupole time-of-flight mass spectrometry. Anal. Chem. 72:357–366.Pettit, L. D., and Powell, K. J. 1997. IUPAC stability constant data-base for windows, 2nd ed. Trimble, Otley, UK: Academic Software.Prange, A., and Schaumloffel, D. 1999. Determination of element species at trace levels using capillary electrophoresis-inductively cou-

pled plasma sector field mass spectrometry. J. Anal. At. Spectrom. 14:1329–1332.Reyes, L. H., Marchante-Gayon, J. M., Alonso, J. I. G., and Sanz-Medel, A. 2003. Quantitative speciation of selenium in human serum by

affinity chromatography coupled to post-column isotope dilution analysis ICP-MS. J. Anal. Atom. Spectrom. 18:1210–1216.Ritsema, R., Dukan, L., Navarro, T. R. I., van Leeuwen, W., Oliveira, N., Wolfs, P., and Lebret, E. 1998. Speciation of arsenic com-

pounds in urine by LC-ICP MS. Appl. Organomet. Chem. 12:591–599.Rodriguez, I., Mounicou, S., Lobinski, R., Sidelnikov, V., Patrushev, Y., and Yamanaka, M. 1999. Species selective analysis by micro-

column multicapillary gas chromatography with inductively coupled plasma mass spectrometric detection. Anal. Chem.71:4534–4543.

Santore, R. C., and Driscoll, C. T. 1995. The CHESS model for calculating chemical equilibria in soils and solutions. SSSA Special Publica-tion 42:357–375.

Sanz-Medel, A., Soldado Cabezuelo, A. B., Milacic, R., and Bantan Polak, T. 2002. The chemical speciation of aluminum in humanserum. Coord. Chem. Rev. 228:373–383.

Shannon, R. D. 1976. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. ActaCryst A32:751–767.

Sheppard, B. S., Shen, W. L., Caruso, J. A., Heitkemper, D. T., and Fricke, F. L. 1990. Elimination of the argon chloride interference onarsenic speciation in inductively coupled plasma mass-spectrometry using ion chromatography. J. Anal. Atom. Spectrom. 5:431–435.

Smith, R. M., and Martell, A. E. 1975. Critical stability constants, Vol. 2. Amines. New York: Plenum Press.Smith, R. M., and Martell, A. E. 1976. Critical stability constants, Vol. 4. Inorganic complexes. New York: Plenum Press.Smith, R. M., and Martell, A. E. 1989. Critical stability constants, Vol. 6. Second supplement. New York: Plenum Press.Sun, Y. C., Lee, Y. S., Shiah, T. L., Lee, P. L., Tseng, W. C., and Yang, M. H. 2003. Comparative study on conventional and low-flow

nebulizers for arsenic speciation by means of microbore liquid chromatography with inductively coupled plasma mass spectrometry.J. Chromatogr. A 1005:207–213.

Sutton, K. L., and Caruso, J. A. 1999. Liquid chromatography-inductively coupled plasma mass spectrometry. J. Chromatogr. A856:243–258.

Sutton, K., Sutton, R. M. C., and Caruso, J. A. 1997. Inductively coupled plasma mass spectrometric detection for chromatography andcapillary electrophoresis. J. Chromatogr. A. 789:85–126.

Szpunar, J. 2000. Bio-inorganic speciation analysis by hyphenated techniques. Analyst 125:963–988.Szpunar, J., and Lobinski, R. 2002. Multidimensional approaches in biochemical speciation analysis. Anal. Bioanal. Chem. 373:404–411.Szpunar, J., Pellerin, P., Makarov, A., Doco, T., Williams, P., Medina, B., and Lobinski, R. 1998. Speciation analysis for biomolecular

complexes of lead in wine by size-exclusion high-performance liquid chromatography inductively coupled plasma mass spectrome-try. J. Anal. Atom. Spectrom. 13:749–754.

Szpunar, J., Lobinski, R., and Prange, A. 2003. Hyphenated techniques for elemental speciation in biological systems. Appl. Spectrosc.57:102A–112A.

Page 21

EVALUATING METAL SPECIATION IN TOXICOLOGY 61

Tanner, S. D., Baranov, V. I., and Bandura, D. R. 2002. Reaction cells and collision cells for ICP-MS: A tutorial review. Spectrochim. Acta B57:1361–1452.

Thomas, R. 2002. A beginner’s guide to ICP-MS, Part IX—Mass analyzers: Collision/reaction cell technology. Spectroscopy 17:42–48.Timerbaev, A. R., and Buchberger, W. 1999. Prospects for detection and sensitivity enhancement of inorganic ions in capillary electro-

phoresis. J. Chromatogr. A 834:117–132.Tipping, E. 1994. WHAM—A chemical equilibrium model and computer code for waters, sediments, and soils incorporating a discrete

site/electrostatic model of ion-binding by humic substances. Comp. Geosci. 20:973–1023.Tomlinson, M. J., Lin, L., and Caruso, J. A. 1995. Plasma-mass spectrometry as a detector for chemical speciation studies. Analyst

120:583–589.Uden, P. C. 1995. Element-specific chromatographic detection by atomic absorption, plasma-atomic emission and plasma-mass spec-

trometry. J. Chromatogr. A 703:393–416.Uden, P. C. 2002. Modern trends in the speciation of selenium by hyphenated techniques. Anal. Bioanal. Chem. 373:422–431.Vanhaecke, F., and Moens, L. 1999. Recent trends in trace element determination and speciation using inductively coupled plasma mass

spectrometry. Fresenius J. Anal. Chem. 364:440–451.Van Loon, J. C., Alcock, L. R., Pinchin, W. H., and French, J. B. 1986. Inductively coupled plasma-source mass spectrometry: New ele-

ment/isotope-specific mass spectrometry detector for chromatography. Spectrosc. Lett. 19:1125–1135.Vela, N. P., Heitkemper, D. T., and Stewart, K. R. 2001. Arsenic extraction and speciation in carrots using accelerated solvent extraction,

liquid chromatography and plasma mass spectrometry. Analyst 126:1011–1017.Vonderheide, A. P., Wrobel, K., Kannamkumarath, S. S., B’Hymer, C., Montes-Bayon, M., De Leon, C. P., and Caruso, J. A. 2002. Char-

acterization of selenium species in Brazil nuts by HPLC-ICP-MS and ES-MS. J. Agric. Food Chem. 50:5722–5728.Whitburn, J. S., Wilkinson, S. D., and Williams, D. R. 1999. Chemical speciation of ethylenediamine-N, N′-disuccinic acid (EDDS) and

its metal complexes in solution. Chem. Spec. Bioavail. 11:85–93.Wolery, T. J. 1992. EQ3/6, a software package for geochemical modeling of aqueous systems: Package overview and installation guide.

Report UCLR-MA-1100662-Pt. 1.Wuilloud, R. G., de Wuilloud, J. C. A., Vonderheide, A. P., and Caruso, J. A. 2003. Determination of iodinated phenol species at parts-

per-trillion concentration levels in different water samples by solid-phase microextraction/offline GC-ICP-MS. J. Anal. Atom. Spec-trom. 18:1119–1124.

Xie, Q. L., Kerrich, R., Irving, E., Liber, K., and Abou-Shakra, F. 2002. Determination of five arsenic species in aqueous samples by HPLCcoupled with a hexapole collision cell ICP-MS. J. Anal. Atom. Spectrom. 17:1037–1041.

Yokel, R. A. 2000. The toxicology of aluminum in the brain: A review. Neurotoxicology 21:813–828.Yokel, R. A., Lasley, S. M., and Dorman, D. 2005. Metal speciation influences metal toxicokinetics, metal toxicodynamics and human

health assessment. J. Toxicol. Environ. Health B.

Page 22