Anal Bioanal Chem (2015) 407:7037–7046

DOI 10.1007/s00216-015-8842-8

RESEARCH PAPER

An improved method for retrospective quantification of sulfur

mustard exposure by detection of its albumin adduct
using ultra-high pressure liquid chromatography-tandem
mass spectrometry
ChangCai Liu1,2 & LongHui Liang1 & Yu Xiang1 & HuiLan Yu1,2 & ShiKun Zhou1,2
2 1,2 1
& HaiLing Xi & ShiLei Liu & JingQuan Liu

Received: 1 May 2015 / Revised: 5 June 2015 / Accepted: 9 June 2015 / Published online: 12 July 2015
# Springer-Verlag Berlin Heidelberg 2015

Abstract Sulfur mustard (HD) adduct to human serum albu- 1.00 and 250 ng•mL–1 HD exposure (R2>0.9989) with preci-
min (ALB) at Cys-34 residue has become an important and sion of <4.50 % relative standard deviation (%RSD), accuracy
long-term retrospective biomarker of HD exposure. Here, a range between 96.5 % and 114 %, and a calculated limit of
novel, sensitive, and convenient approach for retrospective detection (LOD) of 0.532 ng•mL–1. The lowest reportable
quantification of HD concentration exposed to plasma was
limit (LRL) is 1.00 ng•mL–1, over seven times lower than that
established by detection of the HD-ALB adduct using ultra-
of the previous method. The entire method required only 0.1
high pressure liquid chromatography-tandem mass mL of plasma sample and took under 7 h without special
spectrometry (UHPLC-MS/MS) with a novel non-isotope sample preparation equipment. It is proven to be a sensitive,
internal standard (IS). The HD-ALB adduct was isolated from simple, and rugged method, which is easily applied in interna-
HD-exposed plasma with blue Sepharose. The adduct was tional laboratories to improve the capabilities for the analysis
digested with proteinase K to form sulfur- of biomedical samples related to verification of the Chemical
hydroxyethylthioethyl ([S-HETE])-Cys-Pro-Phe tripeptide Weapon Convention (CWC).
biomarker. The tripeptide adduct could be directly analyzed
by UHPLC-MS/MS without an additional solid phase extrac- . .
Keywords 2,2′-dichlorodiethyl sulfide Albumin adduct
tion (SPE), which was considered as a critical procedure in . .
Chemical warfare agent UHPLC-ESI-MS/MS HD exposure
previous methods. The easily available 2-chloroethyl .
[S-HETE]-CPF tripeptide adduct
ethylsulfide (2-CEES) as HD surrogate was first reported to
be used as IS in place of traditional d8-HD for quantification
of HD exposure. Furthermore, 2-CEES was also confirmed to Abbreviations
be a good IS alternative for quantification of HD exposure by ACN Acetonitrile
investigation of product ion spectra for their corresponding BCG Bromocresol green
tripeptide adducts which exhibited identical MS/MS fragmen- 2-CEES 2-chloroethyl ethylsulfide
tation behaviors. The method was found to be linear between CPF Cysteine-proline-phenylalanine
CWAs Chemical warfare agents
CWC Chemical Weapon Convention
* HaiLing Xi ESI Electrospray ionization
fhxihl@163.com GC Gas chromatography
* ShiLei Liu HD Sulfur mustard
liushilei402@263.net IS Internal standard
LOD Limit of detection
1
Laboratory of Analytical Chemistry, Research Institute of
LOQ Limit of quantitation
Chemical Defence, Beijing 102205, China LRL Lowest reportable limit
2
State Key Laboratory of NBC Protection for
MS Mass spectrometry
Civilian, Beijing 102205, China NICI Negative ion chemical ionization
7038
QC Quality control
QCL Quality control low
QCH Quality control high
QCM Quality control medium
RSD Relative standard deviation
[S-ETE]-CPF Sulfur-ethylthioethyl cysteine-
proline-phenylalanine
[S-HETE]-CPF Sulfur-hydroxyethylthioethyl
cysteine-proline-phenylalanine
SPE Solid phase extraction
TCA Trichloroacetic acid
UHPLC-MS/MS Ultra-high pressure liquid
chromatography-tandem mass
spectrometry
Introduction
Sulfur mustard (2,2-dichloroethyl sulfide, US military code: HD)
is a potent and persistent vesicant, causing ocular injury and
respiratory disorders, as well as erythema and blisters on the skin
[1]. HD was first developed as chemical warfare agents (CWAs)
during World War I, in which about 80 % of the chemical casu-
alties were generated by HD [2]. It was then extensively imposed
on the Chinese people by Japan with devastating casualties dur-
ing World War II. So far, there is still a large amount of aban-
doned chemical weapons containing HD in China, by which it
was estimated that over 2000 casualties were affected since
World War II [3]. It was again used as CWAs by Iraq during the
Iran–Iraq conflict [4]. Together with the recent occurrence of
some accidental human exposure during demilitarization of HD
munitions or containers [5], HD will continue to be of great
concern to public health risks in spite of the Chemical Weapons
Convention (CWC) prohibiting its production, stockpiling, and
use. Furthermore, potential threat to public safety will also be
aggravated if abused by terrorists because of its ease of prepara-
tion. In response to these possible health risks, the development
of reliable methods for long-term retrospective biomonitoring HD
exposure in clinical specimens is essential for improving medical
countermeasures and providing forensic evidence in the event of
alleged use of chemical weapons.
As a bifunctional alkylating agent, HD reacts rapidly with
numerous nucleophilic components in the human body, in-
cluding water, glutathione, DNA, and various amino acid res-
idues in proteins via an episulfonium ion intermediate to form
different metabolites. Based on these metabolites in vivo, HD
exposure to humans is diagnosed and assessed [4, 5]. The
hydrolytic metabolite thiodiglycol and its oxidation product
thiodiglycol sulfoxide, a series of β-lyase metabolite products,
and HD-DNA adducts at N7, O6 positions of guanine, N3
position of adenine had been diagnosed as biomarkers for
clinical samples of HD exposure in many cases, and provided
credible information for medical treatment [5–7]. However,
C.C. Liu et al.
one common disadvantage of these biomarkers is that they
predominantly occur within 2 wk after an exposure. As a
result, they cannot be considered as biological markers for
long-term retrospective detection.
Compared with the aforementioned metabolites, HD ad-
ducts to protein are more stable and have a longer lifespan,
allowing for a longer window of exposure detection and a
more reliable identification of the involved HD exposure [8,
9]. HD forms adduct to hemoglobin at its N-terminal valine
residue, and the adduct is a long-lived biomarker that can
exist for about 120 d [5, 8]. The HD-hemoglobin adduct was
de-tected as hydroxyethylthioethyl-valine adduct using
Edman degradation combined with gas chromatography (GC)-
nega-tive ion chemical ionization (NICI)-mass spectrometry
(MS) in many studies [4, 5, 10, 11].
HD can also form stable sulfur-2-hydroxyethylthioethyl
([S-HETE]) adduct to human serum albumin (ALB) at its
reactive Cys-34 residue, of which -SH groups were nucleo-
philically attacked by episulfonium ion intermediate [4, 12].
[S-HETE]-ALB adduct has become another important bio-
marker for long-term retrospective detection of exposure in
spite of the shorter half-life of 20–25 d [9, 13]. Many plasma
samples collected from HD exposure accidents or conflicts
had been analyzed for the [S-HETE]-ALB adduct to verify
HD exposure [13–15]. The established assay methods for the
stable [S-HETE]-ALB adduct have been based on the fact that
enzymatic digestion of the adduct leads to the formation of [S-
HETE]-Cys-Pro-Phe ([S-HETE]-CPF) tripeptide ad-duct,
which can be analyzed by liquid chromatography-tandem
mass spectrometry (LC-MS/MS) [14]. Efforts have been made
for quantification of HD exposure in plasma by the response
of the measured [S-HETE]-CPF tripeptide, where d8-HD was
used as internal standard (IS) [9, 16].
However, the previous method involved complicated pro-
cedures for sample preparation and highly variable yields of
[S-HETE]-CPF tripeptide digested by pronase from different
batches that have varying enzyme composition over the years.
It appears that these problems can be addressed by a recently
developed and validated method for quantification of HD-
albumin adduct, in which there is a simplified procedure for
decreased analysis time, low usage amount of blood speci-
mens, and reproducible tripeptide yields produced by novel
selected proteinase K [17]. In spite of thie, the method could
only make quantification of [S-HETE]-CPF tripeptide instead
of HD concentration exposed, and the need for an isotope
labeling [S-HETE]-CPF as IS was not readily available for
many ordinary laboratories. In addition, all these established
methods still required one laborious solid phase extraction
(SPE) procedure prior to LC-MS/MS analysis. To address
this, herein an improved approach for retrospective quantita-
tion of HD exposure was developed using UHPLC-MS/MS
detection of the [S-HETE]-CPF tripeptide adduct produced by
proteinase K digestion of enriched ALB. The method allows
Improved albumin-adduct method for retrospective quantification of sulfur mustard-exposure in plasma
for the direct reporting of HD concentration exposed to
plas-ma, and has simplified procedure, decreased sample
usage, and increased sensitivity, as well as universality that
is readily extended to other laboratories in comparison
with the previ-ously reported methods.
Materials and methods
Safety considerations
Handle HD with great care. HD is a hazardous and highly
reactive alkylating vesicant agent. HD should be handled only
by trained personnel with individual stringent protective gear
in well-ventilated fume cupboards. After use, all HD-
containing solutions should be decontaminated in a designat-
ed decontamination container with bleach solution.
The analysis of [S-HETE]-CPF and [S-ETE]-CPF
tripeptide adducts are not expected to pose a risk greater
than general peptide analyses. Universal safety precautions
were followed for handling clinical specimens such as
blood products.
Chemicals and materials
Monobasic sodium phosphate, dibasic sodium phosphate,
so-dium chloride, ammonium bicarbonate, and Tris base
were purchased from Amresco (Solon, OH, USA). Amicon
Ultra-0.5 mL and 4.0 mL centrifugal filters and proteinase
K isolat-ed from Tritirachium album were purchased from
Merck Millipore (Darmstadt, Germany). Blue sepharose 6
FF was from GE Healthcare (Uppsala, Sweden) and
Bromocresol green (BCG) albumin assay kit from Sigma-
Aldrich (Steinheim, Germany). HPLC-grade water,
acetonitrile (ACN), and formic acid were purchased from
Honeywell Burdick and Jackson (Muskegon, MI, USA).
Zorbax Eclipse Plus C18 (50×2.1 mm, 1.8 μm) analysis
column and Zorbax Eclipse Plus C18 Narrow-Bore guard
column (12.5×2.1 mm, 5 μm) were obtained from Agilent
Technologies (Palo Alto, CA, USA). [S-HETE]-CPF
tripeptide, HD, and its surrogate 2-chloroethyl ethylsulfide
(2-CEES) were synthesized in-house. Plasma matrix from
healthy individuals without known exposure to HD was
purchased from Beijing Red Cross Blood Center.
Internal standard (IS)
IS was prepared by addition of 20 μL 20.0 μg•mL–1 2-
CEES solubilized in ACN to 2.0 mL unexposed plasma
with shaking at a speed of 120 r•min–1 incubated at 37 °C
for 4 h. The prepared IS was kept at 4 °C until it was
spiked into calibrators or samples.
7039
Calibration curve and quality controls
Plasma matrices were centrifuged at 13,000×g for 5 min and then
filtrated using 0.45 μm microporous membranes (0.45 μm pore
size) to remove particulates as much as possi-ble. A series of
standard solutions of HD solubilized in ACN were, respectively,
spiked into these unexposed plasma mate-rials, followed by
incubation at 37 C° for 4 h with shaking at a speed of 120 r•min–
1
. The final eight-point calibration curve ranged at three orders of
magnitude of HD-spiked concentra-tions from 1.00 to 250
ng•mL– (1.00, 5.00, 10.0, 20.0, 50.0, 100, 200, and 250 ng•mL–
1
) in unexposed plasma. Similarly, one 1.00 μg•mL– of HD-
1 1
spiked plasma sample was also generated. A blank sample was
produced using unexposed plasma spiked with the same amount
of ACN as those used in the HD-spiked plasma materials. The
quality control (QC)-low (QCL), QC-medium (QCM), and QC-
high (QCH) sam-ples were prepared to 2.00, 28.0, and 80.0
ng•mL– by dilution of a high concentration HD-spiked sample
1
(1.00 μg•mL– ) with unexposed plasma, respectively. Calibrators
1
and QC plas-ma samples were aliquoted and stored at –20 °C
until use.
Sample preparation
The ALBs were isolated from human plasma using affinity
chromatography on Cibacron blue Sepharose. One hundred
microliters of plasma were diluted with 400 μL Buffer A (15
mM Na2HPO4, 5 mM NaH2PO4, pH 6.7) and spiked with 5
μL IS (20.0 μg•mL–1 2-CEES-spiked plasma). The sample
was then loaded onto a blue Sepharose column, which had
been manually packed with 4 mL blue Sepharose 6 FF (GE
Healthcare) in an empty 6 mL SPE column (Agilent Technol-
ogy) and preconditioned with 10 mL Buffer A. The cartridge
was washed with 20 mL Buffer A at a flow rate of
5 mL•min–1, and eluted with 5 mL Buffer B (15 mM
Na2HPO4, 5 mM NaH2PO4, 2 M NaCl, pH 6.7) using a
syringe fitted to an adaptor on the column. The eluate (0.5–
5.0 mL) was collected and concentrated to about 100 μL using
a 4.0 mL 10-kDa molecular weight cut-off filter by two
successive centrifugation steps at 13,000×g. The albu-min of
the samples was quantified using a BCG albumin assay kit
(Sigma-Aldrich). The percent recovery of albumin was
calculated by the albumin amount extracted from sam-ples
divided by that of the original samples.
For effective digestion, solvent of the residue sample in the
filter was changed to Buffer C (20 mM Tris HCl, pH 8.0) by
addition of 400 μL Buffer C and filtration with a 0.5 mL 10-
kDa molecular weight cut-off filter at 13,000×g centrifuga-
tion for 6 min. The filter was inverted in a blank collection
tube and centrifuged at 13,000×g for 2 min. The poured so-
lution was transferred to a 1.5 mL Eppendorf tube and set the
volume to 400 μL with Buffer C, followed by addition of
7040
100 μL 20.0 mg•mL–1 proteinase K in Buffer C. The
sample was digested at 50 °C for 4 h.
After digestion, the samples were size-fractionated
using a 0.5 mL 10-kDa cut-off filter by centrifugation at
13,000×g for 20 min to remove any remaining proteinase
K. The resulting filtrate containing peptides was
transferred to glass autosampler vials (Agilent) for
UHPLC-electrospray ioniza-tion (ESI)-MS/MS analysis.
UHPLC-ESI-MS/MS
Peptide samples were analyzed using a 1290 Affinity series
UHPLC system connected to an Agilent Technologies model
6460 triple quadruple mass spectrometer. Twenty microliters
of the digested samples was separated by a symmetric C18 5-
μm, 2.1-mm×12.5-mm pre-column and a Zorbax Eclipse Plus
C18 1.8-μm, 2.1-mm×50-mm reversed-phase column. The
mobile phases were (A) 100 % H2O/0.2 % formic acid, and
(B) 100 % ACN/0.2 % formic acid. The following opti-mized
gradient elution program was used at a constant flow rate of
300 μL•min–1 and a constant column temperature of 40 °C:
0–0.8 min 5 % B; 0.8–1.3 min linear gradient to 13 % B; 1.3–
4.3 min 13 % B; 4.3–6.3 min linear gradient to 70 % B; 6.3–
6.8 min linear gradient to 100 % B; 6.8–9.0 min 100 % B.
The column was re-equilibrated with 5 % B for 2 min before
the next run.
Eluted peptides were determined on the triple quadruple
MS fitted with an ESI interface in the positive mode with the
following optimized parameters: gas temperature = 350 °C,
gas flow =10 L•min–1, nebulizer=35 psi, and capil-lary
voltage=4000 V. Multiple reaction monitoring (MRM) was
utilized with a 200 ms dwell time for each of four transi-tions,
segmental monitoring two transitions of [S-HETE]-CPF
tripeptide adduct before 5.95 min and [S-EES]-CPF after 5.95
min. Quantitation transitions for [S-HETE]-CPF and [S-EES]-
CPF were, respectively, monitored as m/z 470.2→ 104.8 and
m/z 454.2→88.8 at 22 V of collision energy. Their
confirmation transitions were, respectively, m/z 470.2→136.8
and m/z 454.2→120.8 at 18 Vof collision energy. In addition,
fragmentors and cell accelerator voltages of the four transition
ions were set as 135 V and 3 V, respectively.
Method validation
Each analytical run consisted of a plasma blank, three QC
samples (QCL, QCM, and QCH), and a complete eight-point
calibration standard set. Peak area ratios of [S-HETE]-
CPF/[S-ETE]-CPF were plotted against the theoretical HD
concentration exposed to plasma to construct calibration
curves from a series of eight-point calibrators. QC character-
ization was completed over the course of 4 wk during method
validation (n=19) and performed by three analysts. Linearity,
intra-day/inter-day precision (relative standard deviation,
C.C. Liu et al.
%RSD), and accuracy were calculated by the data of 19 ana-
lytical runs. The limits of detection (LOD) were calculated using
the standard deviation approach [18]. Stability experi-ment was
performed by determining the influence of temper-atures and
freeze-thaw cycles on the response signal of targets.
Data acquisition and processing
Data acquisition was performed on MassHunter workstation
software, LC/MS data acquisition for 6400 series triple qua-
druple v. B.05.00, build 5.0.5027.0. Spectral analysis and
quantitation were carried out applying MassHunter worksta-
tion software quantitative analysis v. B.05.00, build 5.0.291.0,
by which some extracted chromatograms were assessed for
correcting peak shape and retention time, and manually inte-
grated when needed.
Results and discussion
Isolation of ALB using affinity chromatography
on Cibacron blue Sepharose
Affinity chromatography of albumin on Cibacron blue Sepha-
rose is an very effective isolation strategy and the first step for
detection of [S-HETE]-ALB adduct [19]. However, most of
the previously reported methods for HD adduct isolation re-
quire special automated equipment for high throughput, which
is very expensive for many laboratories. To improve univer-
sality of the method available for more laboratories, we man-
ually established a simple and convenient system for ALB
isolation including one column manually packed with
Cibacron blue Sepharose and one syringe fitted to the adaptor.
Optimization of washing and elution conditions showed that
washing with 20 mL of Buffer A and elution with 5 mL
Buffer B were necessary for increasing the purity, recovery,
and re-producibility of ALB. As a result, the built purification
system allows for effective isolation of ALB from 100-μL
plasma (containing about 4 mg ALB) with a high and
consistent re-covery (96.0 %±3.5 %). Furthermore, the
method reproduc-ibility was also confirmed to be satisfied
with a stable recovery of 96.5 %±5.0 % by five analysts’
operation in triplicate, suggesting its good robustness.
Proteinase K digestion of ALB forming stable tripeptide
adduct
To improve digestion efficiency, gel filtration and trichloro-
acetic acid (TCA)/acetone precipitation method were ap-plied
for desalting the eluate containing albumin enriched by
Cibacron blue Sepharose prior to digestion in reported studies
[13, 16, 20]. However, these methods are not easy to operate
because of the relatively large volume of the
Improved albumin-adduct method for retrospective quantification of sulfur mustard-exposure in plasma
eluate. For example, followed by gel filtration, eluate vol-
ume will increase at least two times, resulting in not
enough concentration of ALB for digestion. The volume of
TCA/acetone is at least two times as large as that of eluate
for effective precipitation of proteins. To address this,
ultra-filtration method was used in our study. The 4.5 mL
of collected eluate was conveniently concentrated and
changed into enzymatic buffer using 10-kDa cut-off filters.
As a result, recovery of over 90 % was obtained and
relative short time was spent during this procedure.
Proteinase K isolated from Tritirachium album has been
confirmed to be more reliable and effective than currently
available pronase for digestion of total proteins into the [S-
HETE]-CPF tripeptide adduct [17]. Therefore, in this study,
proteinase K was chosen to produce the tripeptide. To fur-ther
improve the performance, conditions including diges-tion
time, temperature, pH, and enzyme concentration were,
respectively, re-optimized since the target proteins are puri-
fied ALBs instead of total proteins. Each level of all factors
were assessed against yields of the produced tripeptide ad-
duct [S-HETE]-CPF using the QCM plasma samples. The
QCM samples were run in triplicate under each test condi-
tion. Three pH factor levels of enzymatic buffer, including pH
7.2 of 50 mM PBS, pH 8.0 of 20 mM Tris HCl, and pH 8.8 of
50 mM NH4HCO3, were investigated. The re-sults indicated
that pH value of enzymatic buffer had sig-nificant influence
on the yield of adducted tripeptide with variation of over 50
%, and the highest signal response was from pH 8.0 of 20 mM
Tris HCl, which had not previously been reported. The
investigated digestion reaction at 40, 50, and 60 °C showed
that temperature was a critical factor, and there were higher
signals at 50 and 60 °C, which is consistent with that of
previous study [17]. Conditions in enzyme concentration
(2.00, 4.00, 6.00 mg•mL–1) and digestion time (1.0, 2.0, 4.0,
8.0, 12.0 h) were also optimized and investigat-ed. It was
found that the two parameters had insignificant influence on
the adducted tripeptide with the standard of var-iation at
values ±20 %. However, the level of 4.00 mg•mL-1 enzyme
and digestion time ≥4.0 h had relatively high and stable yields
of the tripeptide adduct. Obviously, longer diges-tion time
was needed in our study than 1.5 h of the previous study [17],
which could be well explained by the fact that the digestion
mixture was stirred in the previous study, while it was not
stirred during digestion in our study. Interestingly, longer time
for digestion (16, 24, and 48 h) was also investi-gated with the
result that the yields of the tripeptide adduct were still stable
to 48 h, suggesting that proteinase K digestion of ALB can
form stable tripeptide adduct. In addition, besides the [S-
HETE]-CPF tripeptide adduct, one dipeptide adduct [S-
HETE]-CP could be formed by pronase digestion in our
previous study. It is worth noting that only the tripeptide
adduct has been found in all the investigated reactions
digested by proteinase K.
7041
Use of alkylated albumin by 2-CEES as IS
Plasma exposed to a defined amount of d8-HD has been well
used as an IS for retrospective quantifying detection of un-known
samples probably exposed to HD [9, 16]. However, d8-HD is
very expensive and not readily available or synthe-sized; thus, the
method might be difficult to be extended to other laboratories. To
address this, the known HD surrogate 2-CEES, sharing similar
structure characterization and mecha-nism with HD for alkylation
of albumins in plasma, was in-troduced as a new IS instead of
d8-HD. The IS samples were prepared by spiking 2-CEES to the
unexposed plasma, bind-ing with reactive Cys residues of
albumins and forming an-other stable sulfur-ethylthioethyl ([S-
ETE])-albumin adduct. The 2-CEES-exposed plasma samples
digested with protein-ase K following albumin affinity
enrichment by blue Sepha-rose were applied to UHPLC-ESI-
MS/MS (MRM) for detec-tion of the [S-ETE]-CPF tripeptide
adduct. As expected, [S-ETE]-CPF was produced extensively,
confirming that alkyl-ation of ALB by 2-CEES is similar to that
of HD. Subsequent-ly, the product ion spectra of [S-ETE]-CPF
and [S-HETE]-CPF were, respectively, produced by collision-
induced disso-ciation (CID) in UHPLC-ESI-MS/MS (Fig. 1). As
Fig. 1a and b show, [S-ETE]-CPF possesses consistent MS
fragmentation behavior through the entire mass range with that of
[S-HETE]-CPF, especially for its predominant product ion
+ +
[ETE+H] (m/z 89.1) and adjacent product ion [ETE-S + H]
(m/z 120.9), which correspond with the two product ions of [S-
+ +
HETE]-CPF, [HETE +H] (m/z 105.0) and [HETE-S+H] (m/z
137.0). This result clearly shows that 2-CEES-exposed plasma
samples are suitable to be IS in place of the traditional d8-HD for
retrospective quantification of HD exposure.
UHPLC-MS/MS method optimization
Fragmentation of [S-HETE]-CPF and [S-ETE]-CPF in the
CID of MS/MS results in the predominant product ions m/z
105.0 and 89.1, respectively (Fig. 1). The MRM parameters of
[S-HETE]-CPF and [S-ETE]-CPF were optimized by
MassHunter software (Agilent) and shown in the Materials
and methods section. Quantification for [S-HETE]-CPF was
based on the transition m/z 470.2→104.8, representing the
cleavage of [HETE+H]+ (m/z 104.8) from Cys residue sulfur
atom of the protonated adduct [S-HETE]-CPF (m/z 470.2).
Quantification transition m/z 454.2→88.8 of [S-ETE]-CPF
represented the similar cleavage as [S-HETE]-CPF. The con-
firmation transitions of m/z 470.2→136.8 for [S-HETE]-CPF
and m/z 454.2→120.8 for [S-ETE]-CPF, respectively, repre-
sented the cleavage of [HETE-S+H]+ (m/z 136.8) and [ETE-
S+H]+ (m/z 120.8) from their individual protonated adduct
molecule.
In addition to MS method, LC conditions were also opti-
mized for good separation of [S-HETE]-CPF and [S-ETE]-
7042
Fig. 1 Chemical structures and product ion mass spectra of the precursor
ions of (a) [S-HETE]-CPF ([M+H]+ m/z 470.2), and (b) [S-ETE]-CPF
([M+H]+ m/z 454.2). Spectra were collected at a collision energy of 20 V
over the mass range m/z 20–500. The dashed lines in each chemical
structure indicate the fragmentation sites of quantification and
confirmation transition. These fragment ions monitored by MRM are
denoted as the m/z values on structures; in (a) the transition m/z
470.2→105.0 was selected for quantitation of HD-albumin adduct,
representing the cleavage of [HETE+H]+ from the Cys sulfur atom of
CPF tripeptides from the matrix, and for further decreasing
matrix interference in MS analysis. In previous studies, the
short run-time for analysis of [S-HETE]-CPF tripeptides was
the focus instead of good separation, resulting in the necessity
of an additional SPE procedure after digestion [9, 16, 17].
Here, effective separation was the main consideration, and
was achieved by combined isocratic with gradient elution pro-
gram utilizing a shorter C18 reversed-phase column (50 mm
length). As a result, the optimal chromatographic conditions
with good separation of the two targets from background
C.C. Liu et al.
protonated [S-HETE]-CPF; the transition m/z 470.2→137.0 was
selected for confirmation, representing the cleavage of [HETE-S+H]+
from the Cys sulfur atom of protonated [S-HETE]-CPF; in (b) the
transition m/z 454.2→89.1 was selected for IS quantitation,
representing the cleavage of [ETE+H] + from the Cys sulfur atom of
protonated [S-ETE]-CPF; the transition m/z 454.2 →120.9 was
selected for IS confirmation, representing the cleavage of [ETE-
S+H]+ from the Cys sulfur atom of protonated [S-ETE]-CPF
matrix were obtained as stated above. It was found that a 10
%–15 % B isocratic elution over 3 min prior to a gradient
eluent program was pivotal for removing matrix interference
as much as possible (Fig. 2), which is beneficial to MS anal-
ysis of the two adducted tripeptides of [S-HETE]-CPF and [S-
ETE]-CPF. The two targets would be eluted earlier in the case
of more than 20 % B isocratic elution for 3 min, resulting in
decreasing the sensitivity because of co-elution of matrix in-
terference and targets. Surprisingly, it was found that the good
separation of [S-HETE]-CPF from matrix by the optimized
Improved albumin-adduct method for retrospective quantification of sulfur mustard-exposure in plasma
Fig. 2 UHPLC-MS/MS MRM
chromatograms of [S-HETE]-
CPF determined from a
100 ng•mL-1 HD-spiked plasma
sample. (a) [S-HETE]-CPF
quantitation transition (m/z
470.2→104.8) at 5.74 min and
IS quantitation transition (m/z
454.2→88.8) at 6.09 min, (b)
[S-HETE]-CPF confirmation
transition (m/z 470→136.8) at
5.74 min and IS confirmation
transition (m/z 454.2→120.8) at
6.09 min
LC parameters and by the SPE after digestion contributed
almost equally to the increase of signal response. Thus, in the
current study, we could directly inject 20 μL of ultra-filtered
sample with the smallest peak width and the best peak shape
without the additional SPE procedure, which had been
considered to be very important for purification of [S-HETE]-
CPF to increase MS response of the target after digestion [9,
16, 17]. Figure 2 shows a typical chromatogram for the four
transitions acquired. The retention time of the two adducted
[S-HETE]-CPF and [S-ETE]-CPF tripeptides remained at 5.7
min and 6.1 min, respectively.
Specificity
The transition (m/z 470.2→104.8) was previously reported
for quantification of HD adduct to albumin with high speci-
ficity [16]. However, in this study, the specificity of method
was further investigated by confirmation ion ratio (CIR, the
area of confirmation ion peak divided by that of the quantifi-
cation ion peak). The value of CIR was proven to be repro-
ducible with a range from 0.33 to 0.45 over all plasma
samples exposed to HD, which was markedly consistent with
the pre-vious report [16]. The specificity of the method was
also in-vestigated by determining the presence or absence of
background/interference peaks for the [S-HETE]-CPF. Anal-
ysis of 24 plasma specimens from healthy individuals known
to be unexposed to HD showed that all samples were negative
for the [S-HETE]-CPF, and no background/interference peaks
7043
were present within ±0.1 min of the corresponding
retention time of the [S-HETE]-CPF.
Linearity and LOD
The peak area ratios of [S-HETE]-CPF/[S-ETE]-CPF were
found to be linearly proportional to the nominal concentration
of HD in plasma ranged from 1.00 to 250 ng•mL–1. Based on
19 analytical runs, the average calibration line equation was
calculated to be y=(0.0280±0.00054)x+0.0072±0.0033 with a
coefficient of determination of R2 >0.9989. The higher con-
centration outside the linearity range was not investigated
since the validated concentration range covered previous mea-
surements of the exposure extent from some HD-exposed in-
dividuals [13, 14].
The limit of detection (LOD) was estimated by quantifica-
tion transition using the standard deviation (SD) approach
[18]. The peak area of the four lowest calibrators (1.00, 2.00,
5.00, and 10.0 ng•mL–1) was reanalyzed for SD and plotted
for another linear curve against their respective HD-exposed
concentration. LOD was calculated as 3.3 times SD of the
responses in 1.00 ng•mL–1 calibrator divided by the slope of
the resulting linearity curve. LOD was determined to be 0.532
ng•mL– 1, which was slightly lower than 0.005 μM (0.790
ng•mL–1) of a recent study [16], whereas the lowest
reportable limit (LRL), which was the lowest cali-brator of
1.00 ng•mL–1 HD exposure utilizing our improved method,
was far below the LRL value of 0.05 μM
7044 C.C. Liu et al.

Table 1 Calculated concentration, accuracy, and precision for calibrators and QC samples over 19 measurements

Sample ID Theoretical Intra-day (n=5) Inter-day (n=19)

concentration
(ng•mL–1)a Mean calculated concentration Accuracy (%) RSD (%) Mean calculated concentration Accuracy RSD
(±SD) (ng•mL–1)a (±SD) (ng•mL–1)a (%) (%)

Calibrator 1 1.00 1.11±0.04 111 3.83 1.13±0.05 114 4.06

Calibrator 2 5.00 5.54±0.36 111 6.57 5.47±0.06 109 1.19
Calibrator 3 10.0 – – – 10.1±0.40 101 3.99
Calibrator 4 20.0 22.6±0.41 113 1.83 22.0±0.53 110 2.40
Calibrator 5 50.0 – – – 49.3±0.48 98.7 0.97
Calibrator 6 100 97.2±1.12 97.2 1.20 96.5±1.10 96.5 1.14
Calibrator 7 200 – – – 201±5.25 100 2.62
Calibrator 8 250 – – – 251±4.08 100 1.63
QCL 2.00 – – – 2.05±0.08 103 3.80
QCM 28.0 – – – 27.2±0.72 97.2 2.66
QCH 80.0 – – – 83.2±1.29 104 1.55

a
The theoretical and calculated concentration units Bng•mL–1 ^ are based on in vitro exposure of plasma to HD

(7.90 ng•mL-1) in a recent study [16]. It was also indicated respectively (n=19). The inter-day %RSD of these QC mate-
that lower LOD and LRL referring to HD concentration of in rials over a 4-wk period was calculated as 3.80 % for QCL,
vitro exposure to human plasma had been acquired in some 2.66 % for QCM, and 1.55 % for QCH (Table 1). The
earlier reports with 1.5 nM (0.240 ng•mL–1) of LOD and 4.5 method’s precision and accuracy follow the FDA guidance for
nM (0.710 ng•mL–1) of LRL [9, 13]. However, these methods
bioanalytical method validation and, thus, show good applica-
bility for the analysis of clinical samples.
were not widely applied because of the shortcoming in robust-
ness [9, 13]. The lowest calibrator gave reproducible CIR
value (0.42±0.04) and confirmation ion signal with an aver-
age accuracy of 114 % and precision of 4.1 % RSD (Table 1). Stability
In addition, the lowest calibrator peak with a signal-to-noise
(S/N) ratio of 9.5±1.3 at 5.7 min was clearly distinguishable To assess the stability, the storage temperature and freeze/thaw
from that of the matrix blank sample (Fig. 3). cycle for QC samples were evaluated. They were stored at – 20
°C, 4 °C, room temperature, and 37 °C, and analyzed weekly for
4 wk. As a result, all QC samples were stable within ±10.0 % of
Accuracy and precision (RSD) initial values for 4 wk at all temperatures. In addition, all QC
samples were also found to be stable within
The intra-day/inter-day precision and accuracy of the calibra-
tors for quantification of HD-spiked concentrations were de-
termined as a part of the method validation. To calculate
them, HD-spiked plasma samples of four calibrator levels
(1.00, 5.00, 20.0, 100 ng•mL–1) were re-prepared according
to the aforementioned method and analyzed independently
(n=5) (Table 1). The intra-day accuracy was between 97.2 %
and 112.9 % with precision of ≤6.6 % RSD (Table 1). The
inter-day precision and accuracy of the calibrators were
recorded over the course of 4 wk, and calculated as the mean
HD-spiked concentrations from each of the three inter-day
batch runs. The inter-day accuracy was, respectively, 96.5 %
with corresponding precision of ≤4.1 % RSD (Table 1).
The QC materials were run in conjunction with each cali-
bration curve, and characterized for accepting or rejecting an- Fig. 3 Chromatogram comparison of [S-HETE]-CPF tripeptide
adducts from HD-exposed plasma samples. The MS/MS spectrum of
alytical runs according to the FDA’s QC rules [21]. The QCL, 10.0 ng•mL–1 of HD-exposed plasma is in purple line, the MS/MS
QCM, and QCH materials were calculated to be 2.05 ± 0.08 spectrum of the lowest 1.00 ng•mL-1 HD-spiked plasma and the
ng•mL–1, 27.2±0.72 ng•mL–1, and 83.2±1.29 ng•mL–1, matrix blank sample is in pink line and blue line, respectively
Improved albumin-adduct method for retrospective quantification of sulfur mustard-exposure in plasma
±10.0 % of the theoretical values after four freeze-thaw
cycles from –20 °C to room temperature.
Conclusion
In summary, a simple and sensitive method for retrospective
quantification detection of HD exposure was developed by
analysis of HD adduct to ALB in plasma samples using
UHPLC-MS/MS. The reported method is based on the blue
Sepharose affinity isolation, digestion of the purified ALB
adduct with proteinase K, and direct UHPLC-MS/MS analy-
sis of the resulting [S-HETE]-CPF tripeptide adduct. The
modifications have decreased the required sample volume and
simplified the sample preparation procedure without la-
borious SPE procedure prior to UHPLC-MS/MS analysis.
Dependent on thorough studies on digestion of purified ALB,
the [S-HETE]-CPF tripeptide adduct is effectively and
reliably produced by proteinase K with improvement of ro-
bustness and sensitivity of the method over previously report-
ed approaches. The HD surrogate 2-CEES was first intro-
duced as IS in the study, which obviously increased the uni-
versality of the method. No occurrence of background/
interference peak was confirmed by investigation of 24 plas-
ma specimens from known healthy individuals unexposed to
HD. Upon optimization, this improved approach was validat-
ed on investigation of reproducibility, accuracy, precision, lin-
earity, and stability. The validation results showed that our
improved method met all requirements in the FDA guidance
for bioanalysis. Furthermore, the sample preparation proce-
dure can be carried out manually and easily without any spe-
cial equipment or expertise. This study provides a very useful
method for the international laboratories to improve the capa-
bilities for the analysis of biomedical samples related to veri-
fication of CWC.
Acknowledgments The work was supported by the State Key Labora-
tory of NBC Protection for Civilian (grant no. SKLNBC2014-06).
The authors declare no conflicts of interest.
References
1. Borak J, Sidell FR (1992) Agents of chemical warfare: sulfur
mus-tard. Ann Emerg Med 21(3):303–308
2. Dishovsky C, Pivovarov A, Benschop H (2006) Medical
treatment of intoxications and decontamination of chemical
agent in the area of terrorist attack. In: Noort D, Van Der
Schans M, Benschop H (eds) Biomonitoring of exposure to
chemical warfare agents. Springer, Dordrecht, pp 21–26
3. Zhu JX, Gao CT (2010) On chemical warfare in war of Japan
aggression against China and post-war Issues. The Publishing
House of Ordnance Industry, Beijing
4. Benschop HP, van der Schans GP, Noort D, Fidder A, Mars-
Groenendijk RH, de Jong LP (1997) Verification of exposure to
7045
sulfur mustard in two casualties of the Iran-Iraq conflict. J Anal
Toxicol 21(4):249–251
5. Xu H, Nie Z, Zhang Y, Li C, Yue L, Yang W, Chen J, Dong Y,
Liu Q, Lin Y (2014) Four sulfur mustard exposure cases:
overall anal-ysis of four types of biomarkers in clinical samples
provides posi-tive implication for early diagnosis and treatment
monitoring. Toxicol Rep 1:533–543
6. Li C, Chen J, Liu Q, Xie J, Li H (2013) Simultaneous
quantification of seven plasma metabolites of sulfur mustard by
ultra high perfor-mance liquid chromatography-tandem mass
spectrometry. J Chromatogr B 917:100–107
7. Zhang Y, Yue L, Nie Z, Chen J, Guo L, Wu B, Feng J, Liu Q,
Xie J (2014) Simultaneous determination of four sulfur
mustard–DNA adducts in rabbit urine after dermal exposure by
isotope-dilution liquid chromatography-tandem mass
spectrometry. J Chromatogr B 961:29–35
8. Black R, Clarke R, Harrison J, Read RW (1997) Biological fate
of sulphur mustard: identification of valine and histidine
adducts in haemoglobin from casualties of sulphur mustard
poisoning. Xenobiotica 27(5):499–512
9. Noort D, Fidder A, Hulst A, Woolfitt A, Ash D, Barr J (2004)
Retrospective detection of exposure to sulfur mustard: improve-
ments on an assay for liquid chromatography-tandem mass
spec-trometry analysis of albumin/sulfur mustard adducts. J
Anal Toxicol 28(5):333–338
10. Nie Z, Liu Q, Xie J (2011) Improvements in monitoring the N-
terminal valine adduct in human globin after exposure to sulfur
mustard and synthesis of reference chemicals. Talanta 85(2):
1154–1159
11. Lawrence RJ, Smith JR, Boyd BL, Capacio BR (2008)
Improvements in the methodology of monitoring sulfur
mustard exposure by gas chromatography-mass spectrometry
analysis of cleaved and derivatized blood protein adducts. J
Anal Toxicol 32(1):31–36
12. Noort D, Hulst AG, Trap HC, de Jong LP, Benschop HP (1997)
Synthesis and mass spectrometric identification of the major
amino acid adducts formed between sulphur mustard and
haemoglobin in human blood. Arch Toxicol 71(3):171–178
13. Smith JR, Capacio BR, Korte WD, Woolfitt AR, Barr JR (2008)
Analysis for plasma protein biomarkers following an accidental
human exposure to sulfur mustard. J Anal Toxicol 32(1):17–24
14. Noort D, Hulst AG, de Jong LP, Benschop HP (1999)
Alkylation of human serum albumin by sulfur mustard in vitro
and in vivo: mass spectrometric analysis of a cys-teine adduct
as a sensitive biomarker of exposure. Chem Res Toxicol
12(8):715–721
15. Newmark J, Langer JM, Capacio B, Barr J, McIntosh RG (2007)
Liquid sulfur mustard exposure. Mil Med 172(2):196–198
16. Andacht TM, Pantazides BG, Crow BS, Fidder A, Noort D,
Thomas JD, Blake TA, Johnson RC (2014) An enhanced
through-put method for quantification of sulfur mustard adducts
to human serum albumin via isotope dilution tandem mass
spectrometry. J Anal Toxicol 38(1):8–15
17. Pantazides BG, Crow BS, Garton JW, Quiñones-González J,
Blake TA, Thomas JD, Johnson RC (2015) A simplified
method for quantifying sulfur mustard adducts to blood pro-
teins by ultra-high pressure liquid chromatography-isotope
dilution tandem mass spectrometry. Chem Res Toxicol
28(2):256–261
18. Lee Y (2004) Method validation for HPLC analysis of relat-ed
substances in pharmaceutical drug products. In: Chan CC, Lam
H, Lee YC, Zhang XM (eds) Analytical method vali-dation and
instrument performance verification. John Wiley and Sons,
New Jersey, pp 27–49
19. Young CL, Woolfitt AR, McWilliams LG, Moura H, Boyer AE,
Barr JR (2005) Survey of albumin purification methods for the
7046
analysis of albumin-organic toxicant adducts by liquid
chromatography-electrospray ionization-tandem mass
spectrome-try. Proteomics 5(18):4973–4979
20. Noort D, Fidder A, Degenhardt-Langelaan C, Hulst A (2008)
Retrospective detection of sulfur mustard exposure by mass
C.C. Liu et al.
spectrometric analysis of adducts to albumin and hemoglobin:
an in vivo study. J Anal Toxicol 32(1):25–30
21. US Department of Health and Human Services: Food and Drug
Administration: Washington DC (2001) Bioanalytical method val-
idation. In: Guidance for industry