Chiral Separation of G-type Chemical Warfare Nerve Agents via Analytical Supercritical Fluid Chromatography

Article abstract of DOI:10.1002/chir.22368

Chemical warfare nerve agents (CWNAs) are extremely toxic organophosphorus compounds that contain a chiral phosphorus center. Undirected synthesis of G-type CWNAs produces stereoisomers of tabun, sarin, soman, and cyclosarin (GA, GB, GD, and GF, respectively). Analytical-scale methods were developed using a supercritical fluid chromatography (SFC) system in tandem with a mass spectrometer for the separation, quantitation, and isolation of individual stereoisomers of GA, GB, GD, and GF. Screening various chiral stationary phases (CSPs) for the capacity to provide full baseline separation of the CWNAs revealed that a Regis WhelkO1 (SS) column was capable of separating the enantiomers of GA, GB, and GF, with elution of the P(+) enantiomer preceding elution of the corresponding P(-) enantiomer; two WhelkO1 (SS) columns had to be connected in series to achieve complete baseline resolution. The four diastereomers of GD were also resolved using two tandem WhelkO1 (SS) columns, with complete baseline separation of the two P(+) epimers. A single WhelkO1 (RR) column with inverse stereochemistry resulted in baseline separation of the GD P(-) epimers. The analytical methods described can be scaled to allow isolation of individual stereoisomers to assist in screening and development of countermeasures to organophosphorus nerve agents.

Full text of DOI:10.1002/chir.22368

CHIRALITY 26:817–824 (2014)
Chiral Separation of G-type Chemical Warfare Nerve Agents via
Analytical Supercritical Fluid Chromatography
1
2
2
3
1
*
*
SHANE A. KASTEN, STEVEN ZULLI, JONATHAN L. JONES, THOMAS DEPHILLIPO, AND DOUGLAS M. CERASOLI
1Physiology and Immunology Branch, Research Division, US Army Medical Research Institute of Chemical Defense, APG, MD
2Waters Corp., New Castle, DE
3Waters Corp., Milford, MA
ABSTRACT
Chemical warfare nerve agents (CWNAs) are extremely toxic organophospho-
rus compounds that contain a chiral phosphorus center. Undirected synthesis of G-type CWNAs
produces stereoisomers of tabun, sarin, soman, and cyclosarin (GA, GB, GD, and GF, respec-
tively). Analytical-scale methods were developed using a supercritical fluid chromatography
(SFC) system in tandem with a mass spectrometer for the separation, quantitation, and isolation
of individual stereoisomers of GA, GB, GD, and GF. Screening various chiral stationary phases
(CSPs) for the capacity to provide full baseline separation of the CWNAs revealed that a Regis
WhelkO1 (SS) column was capable of separating the enantiomers of GA, GB, and GF, with elu-
tion of the P(+) enantiomer preceding elution of the corresponding P(–) enantiomer; two
WhelkO1 (SS) columns had to be connected in series to achieve complete baseline resolution.
The four diastereomers of GD were also resolved using two tandem WhelkO1 (SS) columns,
with complete baseline separation of the two P(+) epimers. A single WhelkO1 (RR) column with
inverse stereochemistry resulted in baseline separation of the GD P(–) epimers. The analytical
methods described can be scaled to allow isolation of individual stereoisomers to assist in
screening and development of countermeasures to organophosphorus nerve agents. Chirality
26:817–824, 2014. © 2014 The Authors. Chirality published by John Wiley Periodicals, Inc.
KEY WORDS: organophosphorus (OP); tabun (GA); sarin (GB); soman (GD); cyclosarin (GF);
chiral stationary phase (CSP); Regis WhelkO1
INTRODUCTION
preparation has been achieved via either enzyme-mediated
purification5–9 or chiral chromatography.4–8,10,11 Stereoiso-
mers have been typically designated with respect to optical
activity (polarimetry) such that the levorotatory (P-) enantio-
mers were found to be more efficient inhibitors of AChE
in vitro and more toxic to mice in vivo.3
Chiral high-performance liquid chromatography (HPLC) is
often used to separate and isolate stereoisomers from race-
mic mixtures. Chromatographic separation and isolation of-
fers a process that is nondestructive, easily scalable, capable
of yielding highly efficient enantioseparation, and suitable
for a wide array of chemically related compounds. Our labora-
tory has adopted and made use of published chiral HPLC
methods for both the analytical and semipreparative scale
separation and/or isolation of GA and VX via a Chiral
Technologies Chiralcel OD-H column.10–12 Chiral stationary
phases (CSPs) applicable for full separation of GB, GD, and
GF stereoisomers by HPLC have not been well characterized.
Organophosphorus (OP) compounds such as pesticides
are typically symmetric around the phosphorus center, while
chemical warfare nerve agents (CWNAs) possess a chiral
center about the phosphorus. Lethal exposure to a CWNA
results in irreversible inhibition of the acetylcholinesterase
(AChE) enzyme, leading to cholinergic crisis in the central
and peripheral nervous systems, with death often resulting
from respiratory depression.1 CWNAs are a serious threat
to soldiers on the battlefield and to civilian populations from
potential terrorist or other activities (e.g., Aum Shinrikyo in
Tokyo, Japan 1995, or Syria 2013).2,3 Several decades of re-
search have been devoted to understanding the toxicology
of OP nerve agents as well as to developing safe, effective,
and broad-spectrum medical countermeasures, leading to
the currently fielded combination of atropine, an oxime, and
an anticonvulsant as a therapeutic drug regimen.
All CWNA compounds (e.g., tabun [GA], sarin [GB], so-
man [GD], cyclosarin [GF], and VX/VR/VS) possess a chiral
phosphorus center, with undirected synthesis pathways pro-
ducing racemic mixtures of enantiomers in roughly equal pro-
portions. GD possesses a second chiral center found in its
pinacolyl-R-group, giving rise to a diastereomer mix of four
epimers. The toxicity of the CWNAs has been studied exten-
sively utilizing individually isolated or stereospecifically syn-
thesized stereoisomers of each CWNA to better understand
the stereoselectivity of the AChE active site.4 It is possible
that the chirality of the CWNAs may be a major contributor
to the dramatically higher toxicity of CWNAs as compared
to most (structurally similar but achiral) OP pesticides. Isola-
tion of individual stereoisomers of CWNAs from a racemic
Supercritical fluid chromatography (SFC) is
a well-
established chiral chromatography technology that offers
several advantages over chiral HPLC.13–15 For this study,
the main advantages of SFC over HPLC include reduced reli-
ance on organic solvents (normal phase methods), increased
column efficiency, higher enantioseparation efficiency for
*Correspondence to: S. A. Kasten or D. M. Cerasoli, Physiology and
Immunology Branch, Research Division, US Army Medical Research Institute
of Chemical Defense, 3100 Ricketts Point Rd, APG, MD 21010. E-mail: shane.
a.kasten.ctr@mail.mil or douglas.m.cerasoli.civ@mail.mil
Received for publication 2 June 2014; Accepted 4 August 2014
DOI: 10.1002/chir.22368
Published online 9 October 2014 in Wiley Online Library
(wileyonlinelibrary.com).
© 2014 The Authors. Chirality published by John Wiley Periodicals, Inc.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

818
KASTEN ET AL.
Aldrich as Chromasolv-Plus reagents (≥99.9% pure). Bovine pancreas
α-Chymotrypsin was acquired from Sigma-Aldrich. Female rabbit (New
Zealand White) plasma was obtained from Bioreclamation (Westbury, NY).
Analytical scale chiral columns (4.6 mm x 250 mm, ID x L coated on
5 μm silica gel) were purchased from the following vendors: Chiral
Technologies, West Chester, PA (Chiralcel OD-H and OJ-RH, Chiralpak
AD-H and IB), Supelco, Bellefonte, PA (Astec CycloBond I, II, and II
AC), and Regis Technologies, Morton Grove, IL (Whelk-O1 [SS]). A
semipreparative Whelk-O1 (RR) column (10 mm x 250 mm, ID x L coated
on 5 μm silica gel) was also purchased from Regis.
numerous compounds, and low solvent consumption. Individual
stereoisomers can also be recovered in aqueous solvents,
allowing for biological assays to be performed immediately
after collection with some methods. We report here the
development of chiral SFC methods for the analytical scale
separation and isolation of GA, GB, GD, and GF enantiomers.
All G-type CWNAs were fully resolved via a Waters (Milford,
MA) SFC-MS Prep15 system using analytical Regis WhelkO1
(SS) or (RR) chiral columns in tandem with a mass spectrome-
ter to detect and direct open-bed fraction collection of the
individual stereoisomers.
SFC/MS Conditions
CWNA and GD stereoisomers were diluted by mixing 1 part nerve
agent with 3 parts hexane (GC-grade) prior to injection into an SFC sys-
tem. SFC was performed on a Waters SFC-MS Prep15 system modified
for the atypical very low cosolvent percentage chromatographic condi-
tions described herein (Fig. 1). Due to the very low cosolvent percentage
of the mobile phase in these separations, the sample injector was revised
from “modifier stream” injection mode (standard configuration), where
only the cosolvent flows through the injector, to “combined stream”
mode, where the CO2 and cosolvent premix under high pressure before
flowing through the injector. Combined stream injection mode results
in a faster flow rate through the injector, increasing the sweep rate of
the sample from the sample loop, resulting in less band broadening as
compared to the modifier stream injection technique. Novel hardware
and software modifications to the sample injector were implemented to
safely and accurately perform combined stream injections. Method con-
trol and data acquisition were managed via MassLynx (v. 4.1 SCN 846)
software. The SFC system was equipped for analyte detection with a photo-
diode array (PDA) containing a flow cell modified to contain analytical
lenses, increasing energy transmittance in the low UV regions of 190 nm
to 220 nm to improve UV detection because of the low absorbance charac-
teristics of the analytes, as well as a 3100 series single-quadrupole mass
spectrometer (MS) fed via passive sample split. Samples were injected
in combined-stream mode at 5 mL/min flow rate under 120 bar using
MATERIALS AND METHODS
CWNA was obtained from the Edgewood Chemical Biological Center
(ECBC, Aberdeen Proving Ground, MD) at >98% purity per nuclear
magnetic resonance (NMR) and gas chromatography mass spectrometer
(GC-MS) analysis. All experiments conducted with CWNAs were performed
in accordance with the strict safety standards established at the US Army
Medical Research Institute of Chemical Defense for handling of these highly
toxic compounds. Racemic ethyl N,N-dimethylphosphoramidocyanidate
(tabun, GA) and racemic O-isopropyl methylphosphonofluoridate (sarin,GB)
were diluted in ethyl acetate to 1.91 mg/mL (11.76 mM) and 1.88 mg/mL
(13.42 mM), respectively, prior to use. Likewise, a diastereomer mix of
pinacolyl methylphosphonofluoridate (soman, GD) and a racemic sample
of cyclohexyl methylphosphonofluoridate (cyclosarin, GF) were each diluted
in hexane to concentrations of 1.89 mg/mL (10.35 mM) and 1.86 mg/mL
(10.35 mM), respectively, prior to use. GD stereoisomer standards were
obtained from the TNO Laboratory (Rijswijk, The Netherlands) in ethyl
acetate, with concentrations listed for each as C-P-, 1.54 mg/mL; C + P-,
1.69 mg/mL; C-P+, 1.72 mg/mL; C + P+, 1.6 mg/mL. Ethyl acetate
(GC-grade) and hexane (GC-grade) were both purchased from Sigma-
Aldrich (St. Louis, MO). Cosolvents (methanol, ethanol, 1-propanol,
1-butanol, and isopropanol) and H2O were all purchased from Sigma-
Fig. 1. Block diagram of the modified Waters Prep15 SFC configuration. The Prep15 instrument was modified to run in combined-stream mode in which the
cosolvent and CO2 are mixed under high pressure prior to injection of sample.
Chirality DOI 10.1002/chir

SEPARATION OF G-TYPE CHEMICAL WARFARE NERVE AGENTS
819
either 0% or 1% cosolvent during isocratic method development. Sample
injection used sample sandwiching in which 80 μl hexane (presolvent)
was added to the injection syringe followed by 95 μl of sample and then
20 μl hexane (postsolvent) immediately prior to injection onto a 100 μl
wide bore loop (0.04″ i.d.). An in-line mobile phase mixer was installed
inside the column oven upstream of the column in the flow path. The
additional homogenization of the mobile phase resulted in lower
baseline noise in the chromatograms, increasing the signal-to-noise
(S/N) of the peaks. The column oven was held at a constant tempera-
ture of 40 °C. When tandem columns were used, a second column was
connected to the first column using 5 cm of 0.02” (i.d.) stainless steel
tubing. A sample fraction was split off the main line and infused with
MS makeup solvent (90/10/0.1; MeOH/H2O/Formic Acid, v:v) at a
flow rate of 0.4 mL/min prior to entering the electrospray ionization
(ESI) source. Main line flow was further supplemented with makeup
solvent, which was used to carry the sample into a passive heat
exchanger before entering the gas-liquid separator (GLS) following
CO2 evaporation. Make-up solvent was either filtered H2O (18 ohm,
0.2 μm filter) at a flow rate of 1 mL/min for method development or
Chromasolv-Plus H2O at a flow rate of 4 mL/min when individual stereo-
isomers were isolated. Test injections were made and used to simulate
fraction collection via the FractionLynx simulator prior to individual
stereoisomer collection. Collection of fractions was triggered by m/z
intensity thresholds with open-bed fraction collection of separated stereo-
isomers (collection parameters as per Waters recommendation). Opti-
mized parameters for ESI-MS were as follows: Positive ion mode;
capillary, 5.0 kV; Cone, 25V; Extractor, 1 V; RF lens, 0.1 V; Source Temper-
ature, 150 °C; Desolvation Temperature, 350 °C; N2 gas flow, Desolvation
(400L/hr) and Cone (60L/hr). Chromatographic analyses were performed
using either original extracted ion chromatograms or versions that were
smoothed via a Savtzky Golay algorithm (window size, 1.0; smoothing
iterations, 40) with background subtracted (polynomial order, 1; below
curve %, 0.1; tolerance, 0.01); no significant differences were observed in
the parameters derived from original and smoothed chromatograms. For
presentation purposes only smoothed chromatograms are shown.
most accurate quantitation of stereoisomers and provide the best yield
and enantiopurity when isolation is required.
Stereoisomer Elution Order
The elution order of stereoisomers was determined using various pub-
lished methods and designated with respect to optical activity by analogy.
The elution order of GA enantiomers was determined by incubating
α-chymotrypsin with racemic GA, which has previously been established
to preferentially bind the P(+) isomer of GA.6 Chymotrypsin was resus-
pended in 1 mM HCl/2 mM CaCl2 at a concentration of 100 mg/mL. Race-
mic GA was diluted 1/10 in 50 mM 4-morpholinepropanesulfonic acid
(MOPS) buffer (pH 7.4) with reaction initiated by mixing 1200 μl diluted
GA with 250 μl of chymotrypsin suspension and conducted by incubation
at room temperature. Several 200-μl aliquots were collected at various
times (0, 0.5, 1, 2, 5 min) and mixed with 1000 μl hexane to partition the
unreacted GA into the organic phase following vortexing for 30 s and
1 min of centrifugation. Approximately 900 μl of organic phase was re-
moved and placed in a 4-mL SFC vial prior to analysis.
Racemic GB was incubated with rabbit plasma to determine the elution
order of the enantiomers, as it has been shown that rabbit plasma prefer-
entially hydrolyzes the P(+) isomer of GB.4 A reaction was initiated by
mixing 50 μl of rabbit plasma with 1000 μl of racemic GB (diluted 1/10
in saline) and was then incubated at room temperature for 20 min. Four
200-μl aliquots were removed at various times (0, 2, 10, and 20 min),
and unreacted GB was extracted with 1000 μl of hexane using the same
procedure described above prior to injection onto the tandem WhelkO1
(SS) columns.
An analytical chiral GC-MS method for the baseline separation of enan-
tiomers of GF was duplicated in our laboratory and used to determine the
elution order and optical activity of the enantiomers separated on the
WhelkO1 (SS) column via the SFC.17 SFC-isolated enantiomers were
each injected into an Agilent 6890/5973 GC-MS fitted with
a
™
GammaDEX 225 (Supleco, Bellefonte, PA) column operated in EI mode
using the referenced method. Individual enantiomers of GF were isolated
via the SFC using a minimum intensity threshold to trigger collection of
peak fractions based on detection of the m/z 181 ion. Enantiomers were
isolated in Chromasolv-Plus H2O and extracted with ethyl acetate at a
volume ratio of 1:0.5 (enantiomer:ethyl acetate). Individually extracted
enantiomers at a concentration of roughly 100 μM were injected (1 μl)
into the GC-MS to determine relative retention times. The elution order
of the enantiomers from the SFC was correlated to the elution order from
the GC and by analogy used to assign optical activity to the enantiomers
based on the referenced method.17
Chromatographic Characterization
Chromatography parameters were determined by drawing lines tan-
gent to the leading and trailing edge of each peak to approximate the
four-sigma peak width to estimate the theoretical plates (N), retention fac-
tor (k), separation factor (α), and resolution (RS).16 Parameters were used
to verify both the enantioselectivity and the separation efficiency of vari-
ous CSPs for G-type CWNAs. Theoretical plates were assessed using
the following formula:
Stereoisomer standards of GD described above were provided with
defined optical activities. The individual isomers were diluted (1 part
isomer with 3 parts hexane, v:v), and 95 μl was injected onto the
WhelkO1 (SS) tandem columns to determine elution order and respec-
tive retention times for each. The stereoisomer standards had slight
contamination with the antipode epimers, with the C + P + isomer having
the most contamination.
ꢀ
ꢁ
2
tR
Wt
N ¼ 16
in which the peak width (Wt) was taken as the difference in time between
the tangent line intersections with the baseline for each peak. The aver-
age retention time, tR, was determined by averaging the time to peak max-
ima from subsequent injections. Calculation of k was carried out using the
following formula:
RESULTS
G-Agent Electrospray Ionization - Mass Spectrometry
tR ꢀ t0
k ¼
Detection of G-agent stereoisomers required a mass spec-
trometer since no spectral signature could be identified for
the G-agents even when using the PDA module modified to
afford enhanced sensitivity. ESI-MS parameters were
established for detection of GF ions using the Intellistart mod-
ule (Masslynx) via direct infusion. Both the protonated
molecular ion and the fragment ion(s) abundances were fur-
ther optimized by manual adjustment of ion source parame-
ters. Optimized parameters determined for GF also provided
suitable ionization and detection of GA, GB, and GD ions.
Ions detected in positive mode included the protonated mo-
lecular ion [M + H]+, characteristic fragment ions, and pro-
posed methanol adducts of molecular fragments (Fig. 2).
Fragment ions were typically the most abundant ions
Chirality DOI 10.1002/chir
t0
where t0 is the column void time (set equal to 1 min for all chromato-
grams). The α was calculated as:
k2
α ¼
k1
with k1 and k2 as the respective retention factors for two adjacent peaks.
Resolution (RS) between adjacent peaks was determined as:
!
pffiffiffiffi
ꢀ
ꢁꢀ
ꢁ
N
4
α ꢀ 1
k
Rs ¼
α
k þ 1
in which the N for the second peak was used for the equation. Methods
that give full baseline resolution between peaks (RS ≥ 1.6) allow for the

820
KASTEN ET AL.
could not be resolved with 0% or 1% cosolvent on the SFC with
a single OD-H column inline, nor could the enantiomers of
GB, GD, or GF be resolved under these conditions (data
not shown). Various other polysaccharide-type chiral
columns (Chiralpak AD-H, Chiralcel OJ-RH, Chiralpak IB)
were tested along with the OD-H, but none afforded detectable
separation of GF enantiomers. Slight separation (RS ~0.6) was
detected for two different unmodified cyclodextrin columns,
CycloBond I (β-cyclodextrin) and CycloBond II (γ-cyclodex-
trin). A diacylated γ-cyclodextrin HPLC column, CycloBond
II AC, was also screened primarily based on similarity to
the GammaDex 225 GC capillary column (stationary phase
also diacylated γ-cyclodextrin) capable of resolving the enan-
tiomers of GF by gas chromatography.17 No separation was
detected with the CycloBond II AC column inline in the
SFC. The only CSP found to be able to provide considerable
baseline resolution (RS = 1.0) of the GF enantiomers was the
Regis pirkle-type WhelkO1 (SS) column. Further chiral
method development centered on establishing full baseline
separation (RS ≥ 1.6) of the stereoisomers of GF followed by
separation methods for the stereoisomers of GA, GB, and
GD using the WhelkO1 (SS) column. Raising the cosolvent
percentage to >1% decreased the retention time of the GF
enantiomers on the WhelkO1 (SS) column but also drasti-
cally reduced the separation efficiency, such that method
development with isocratic mobile phases was restricted to ab-
normally low cosolvent percentages of either 0% or 1%. It should
be noted that the current version of Masslynx software only
supports programming whole-number cosolvent percentages;
as such, fractions of a percent were not assessed during
isocratic method development.
Fig. 2. Ions corresponding to the protonated molecular ion ([M + H]+),
fragment, and fragment-methanol adduct were detected for GA, GB, GD,
and GF in positive ion mode. Fragmentation and adduction patterns were sim-
ilar with m/z ratios specified in parentheses for each ion detected.
detected for each agent. Ion scans of GF ranging from m/z 50
to 200 included detection of the protonated molecular ion
(m/z 181), a fragment ion (m/z 99), and a proposed
methanol-fragment adduct ion (m/z 131). Adduction of frag-
ment ions with a solvent molecule is rare under ESI condi-
tions, but such adducts have been reported for acetonitrile
adducted to G-agent fragments.18 An ionization profile similar
to that obtained for GF was also detected for GA, GB, and GD.
Detected ions included the [M + H]+ ion and fragment ion(s)
for GA (m/z 135), GB (m/z 99), and GD (m/z 99; also m/z 85
resulting from fragmentation of the alkoxy R-group but not
depicted in Fig. 2) along with proposed methanol-fragment
adducts of m/z 167, 131, and 131 for GA, GB, and GD, respec-
tively. Chiral separation of the G-agent stereoisomers was
monitored under positive mode using a scan range from
m/z 50 to 200 with chiral chromatography parameters
determined using extracted ion chromatograms (EIC) for
fragment ions only. Fraction collection of individual stereo-
isomers was directed by mass detection of either the pro-
tonated molecular ion or characteristic product ions.
The separation efficiency of racemic GF was evaluated on
the WhelkO1 (SS) column with either 0% or 1% of various
cosolvent alcohols (Table 1) to determine the best separation
method for the isolation of individual stereoisomers.
Cosolvents commonly used in chiral SFC method develop-
ment include methanol, ethanol, and isopropanol because of
polarity differences that can affect separation efficiency and
analyte retention for a given CSP. Two other cosolvents,
n-propanol and n-butanol, were also included with polarity
ranging from high to low: methanol > ethanol > n-propanol
n-butanol > isopropanol for the whole group. Full baseline
TABLE 1. Separation efficiency of GF using different
cosolvents
Cosolventa
None
WhelkO1 (SS)b
k1
k2
α
RS
Chiral Column Screening for the Separation of GF
Enantiomers by SFC
x1
x2
x1
x2
x1
x2
x1
x2
x1
x2
x1
x2
15.3
39.0
6.2
13.1
7.5
15.2
8.5
22.6
7.8
17.3
43.7
6.8
14.2
8.4
16.8
9.5
24.6
8.6
1.13
1.12
1.10
1.08
1.12
1.11
1.12
1.09
1.11
1.09
1.13
1.12
1.7
2.2
0.7
1.0
0.9
1.2
1.1
1.2
0.8
1.3
1.0
1.7
Methanol
Ethanol
Several analytical SFC columns were screened based on
published results and recommendations from column manu-
facturers (for safety reasons, CWNA could not be sent out
for vendor screening services). Parameters that influence chi-
ral separation include flow rate, cosolvent composition and
percentage, cosolvent modifier (basic/acidic), column tem-
perature, column length, system back pressure, and CSP
type. Preliminary screening of various CSPs was initiated
using only 1% isopropanol (IPA) cosolvent until a column ca-
pable of at least partial resolution (RS ~1.0) of the enantiomers
of GF was discovered. As mentioned above, a single Chiralcel
OD-H was found to be capable of resolving the enantiomers of
GA with 0.3% cosolvent via HPLC, although GA enantiomers
Chirality DOI 10.1002/chir
n-Propanol
n-Butanol
Isopropanol
18.0
8.7
21.9
19.7
9.9
24.6
aMethanol has the highest cosolvent polarity, with isopropanol having the
least (Methanol > Ethanol > n-Propanol > n-Butanol > Isopropanol).
b1x indicates a single column inline or 2x for analysis utilizing a second col-
umn connected in series.

SEPARATION OF G-TYPE CHEMICAL WARFARE NERVE AGENTS
821
GD
TABLE 2. Separation parameters of G-agents
resolution (RS = 1.7) was achieved with 0% cosolvent when a
single WhelkO1 (SS) column was inline. The addition of 1%
cosolvent reduced the retention factor ~2–fold, allowing for
shorter retention times. However, GF enantiomers were no
longer baseline resolved with any of the cosolvents screened.
Since inclusion of cosolvent decreased resolution, a second
WhelkO1 (SS) column was connected to the first column to
determine how doubling the column bed length might en-
hance the separation of the enantiomers. The resolution of
the GF enantiomers increased to an RS value of 2.2 with 0%
cosolvent utilizing two WhelkO1 (SS) columns in tandem
(Fig. 3). Only the isopropanol cosolvent allowed full baseline
separation (RS = 1.7) on tandem columns, whereas the
structural isomer, n-propanol, did not allow full resolution.
Resolution increased as cosolvent polarity decreased when
two columns were connected in series. The RS of 2.2 with
0% cosolvent is more than adequate to allow for the isolation
of pure enantiomers of GF and is also very amenable to a
semipreparative scaling of the process. Despite full base-
line separation of the enantiomers with 1% isopropanol,
recovering isolated stereoisomers devoid of alcohol cosolvents
was determined to better allow easy sample manipulation
postisolation and also to provide for a more biologically
relevant background for immediate isomer-specific enzyme-
based assays.
1xa
2xa
WhelkO1(SS)
G-Agent
GA
GB
GD
GA
GB
k1
k2
k3
k4
α1
α2
α3
RS12
RS23
RS34
11.4
12.9
7.8
9.6
7.6
9.1
11.1
11.1
1.2
1.22
1.0
1.4
1.2
0
25.7
28.5
17.7
21.6
16.5
19.6
23.2
24.3
1.18
1.2
1.05
2.2
2.7
1.13
0.8
1.23
1.1
1.11
1.9
1.22
2.0
0.7
a1x indicates a single column inline and 2x for analysis utilizing a second col-
umn connected in series.
Separation Efficiency of GA, GB, and GD Stereoisomers on
WhelkO1 (SS)
Based on the results obtained with GF, chiral separation
methods were evaluated and developed for GA, GB, and
GD using single and tandem WhelkO1 (SS) columns. Chiral
separation parameters for both single and tandem columns
are listed in Table 2, with representative fragment EICs
shown in Fig. 4 (A, GA; B, GB; C, GD) for tandem column
separation only. Enantiomers of GA, GB, and GD could not
be fully resolved (RS ≤ 1.4) on a single WhelkO1 (SS) col-
umn. Cosolvents screened with GF were also screened with
GA, GB, and GD (data not shown); as was the case with GF,
the best resolution was achieved with 0% cosolvent. Enantio-
mers of GA (RS = 1.9) and GB (RS = 2.0) were completely
baseline resolved on tandem columns. The first two stereo-
isomers of GD to elute were baseline resolved, while the
remaining two stereoisomers to elute were only partially
separated (Fig. 4C).
Elution Order of Stereoisomers on SFC WhelkO1 (SS)
Fig. 4. Separation efficiency profiles for GA (A), GB (B), and GD (C).
Stereoisomers were separated using 2x WhelkO1 (SS) columns with
0% cosolvent.
Optical activity has often been used to distinguish stereo-
isomers from one another based on rotation of plane
polarized light. Stereoisomers that rotate the light counter-
clockwise are called levorotatory or (–), while those that
rotate the light clockwise are called dextrorotatory or (+).
While the individual stereoisomers were not assessed by po-
larimetry directly in this study, several previously developed
methods were used here to determine the elution order of
the stereoisomers on the WhelkO1 (SS) column, with optical
activity assigned by analogy.
Bovine pancreas chymotrypsin has been shown to be
enantioselective for the hydrolysis of the P(+) enantiomer of
GA, allowing for the isolation of the P(–) enantiomer at up
to 99% optical purity.7 Racemic GA was incubated with α-chy-
motrypsin to determine the rate of hydrolysis of each enantio-
mer. By analogy, the enantiomer with the higher rate of
hydrolysis can be assigned as the P(+) enantiomer of GA.
Several aliquots were removed from the reaction mixture
Chirality DOI 10.1002/chir
Fig. 3. EIC showing the separation profile of GF enantiomers on 2x
WhelkO1 (SS) columns with 0% cosolvent.

822
KASTEN ET AL.
over the incubation period, and unreacted GA was extracted
into hexane prior to injection onto the SFC with two WhelkO1
(SS) columns in tandem. Extracted ion chromatograms for
each aliquot were overlaid to demonstrate enantioselective
degradation of the P(+) enantiomer. Based on the hydrolysis
profile seen in Fig. 5, the enantiomer with the shorter reten-
tion time on the WhelKO1 (SS) column was exclusively hy-
drolyzed over the observed time course. By analogy the
enantiomer that was first to elute from the WhelkO1 (SS)
column was the P(+) enantiomer (light gray fill), and the
second eluting enantiomer was the P(–) enantiomer (dark
gray fill) of GA.
The elution order of GB enantiomers from the WhelkO1
(SS) SFC column along with optical activity assignment was
determined in a manner similar to that described above for
GA. It has been demonstrated that rabbit plasma is highly
enantioselective, with a marked preference for degrading
the P(+) enantiomer of GB.7 To determine the enantiospecific
elution order of GB enantiomers, racemic GB was incubated
with rabbit plasma and various reaction endpoints were col-
lected. Extracted samples were injected onto the WhelkO1
(SS) column, and the results indicated enantioselective hy-
drolysis (Fig. 6). The enantiomer eluting first (light gray fill)
was hydrolyzed preferentially, such that by analogy this enan-
tiomer is the P(+) GB enantiomer followed by elution of the P
(–) enantiomer (dark gray fill).
Fig. 6. GB enantiomer elution order was determined by incubating race-
mic GB with rabbit plasma. Several aliquots were removed and injected onto
2x WhelkO1 (SS) columns. The first peak to elute was determined to be the
P(+) enantiomer (light gray shading) followed by elution of the P(–) enantio-
mer (dark gray shading) of GB based on the enantiopreference of
rabbit plasma.
Chiral separation of GF enantiomers has been published
where a Supleco GammaDex 225 GC capillary column was
used and optical activity was correlated with isomer elution
order.17 In this study, the P(–) enantiomers of GF were found
to elute prior to the P(+) enantiomers on the GammaDex 225
column. Individual enantiomers of GF were separated on the
SFC using the WhelkO1 (SS) column, with each enantiomer
isolated by open-bed fraction collection. The isolated enantio-
mers were then extracted into ethyl acetate and injected onto
a GammaDex 225 GC column (see Fig. 7) using a method vir-
tually identical to that in the work referenced above. Based on
the published elution order, the isolated enantiomer found to
elute first from the WhelkO1 (SS) SFC column was the P(+)
enantiomer of GF, as it was the second enantiomer to elute
from the GammaDex 225 GC column (light gray fill). As with
GA and GB, the P(+) enantiomers of GF have shorter reten-
tion on the WhelkO1 (SS) column, while the P(–) enantio-
mers are more strongly retained.
Fig. 7. GF enantiomers were separated on 2x WhelkO1 (SS) columns, col-
lected, and individually injected onto a GammaDex 225 GC capillary column to
determine the relative elution order of the isolated enantiomers. An EIC of the
elution of each enantiomer on the GammaDex 225 GC column is shown. By
analogy, the first enantiomer of GF to elute from 2x WhelkO1 (SS) columns
was the P(+) enantiomer followed by the P(–).
isomer standards (nearly enantiopure), as previously used
in our laboratory to determine the GD isomer elution order
on a TA GC capillary column.19 As described in the preceding
section, two analytical-scale WhelkO1 (SS) columns were
joined together, giving ample separation of the GD diastereo-
mers for individual identification. The retention times of the
isolated standard epimers were correlated with the retention
times of the individual stereoisomers separated from a diaste-
reomer mix of GD, as shown in Fig. 8. The P(+) isolated stan-
dards were retained the least on the WhelkO1 (SS) columns,
with the C(–)P(+) isomer (solid line, light gray fill) eluting
first followed by the C(+)P(+) isomer (dashed line, light gray
fill) as seen in the Fig. 8B overlay (both P(+) isomers had
modest contamination with the antipode P(–) isomers) . Both
P(+) isomers were completely baseline resolved from one an-
other and from the two P(–) isomers (Table 2). The P(–)
enantiomers were more tightly retained on the WhelkO1
(SS) column, with the C(+)P(–) isomer (solid line, dark gray
fill) eluting immediately prior to the C(–)P(–) isomer (dashed
line, dark gray fill), resulting in incomplete resolution even
when tandem columns were used. The elution order of the
four GD isomers was C(–)P(+) followed by C(+)P(+), then C
(+)P(–) followed shortly thereafter by C(–)P(–), based on
the retention times of the isolated isomer standards
(Fig. 8A). Three semipreparative WhelkO1 (SS) columns
were connected in series, but the P(–) epimers of GD still
could not be fully resolved (data not shown). As an alterna-
The elution order of the four GD diastereomers was also
determined for the WhelkO1 (SS) column using individual
Fig. 5. Racemic GA was incubated with bovine pancreas chymotrypsin, and
aliquots were removed over time and injected onto 2x WhelkO1 (SS) columns.
The first peak (light gray shading) to elute from the SFC was preferentially de-
graded by chymotrypsin, while the second peak (dark gray shading)
remained intact over the 5-min incubation period. Based on chymotrypsin
enantioselectivity the elution order of GA enantiomers from the WhelkO1
(SS) column was P(+) followed by the P(–) enantiomer.
tive,
a single semipreparative WhelkO1 column was
purchased from Regis with the reverse stereochemistry
Chirality DOI 10.1002/chir

SEPARATION OF G-TYPE CHEMICAL WARFARE NERVE AGENTS
823
separation of the enantiomers of GA, GB, GF, and the P
(+) enantiomers of GD, while the WhelkO1 (RR) column
separates the P(–) enantiomers of GD. Separation of the
enantiomers of VX was also possible using the WhelkO1
(SS) column (data not shown). Chiral SFC methods are
currently being developed for VX and other
dialkylphosphonothiolate compounds and will be de-
scribed in a subsequent article. Changes in column tem-
perature and system back pressure were two parameters
that were not tested during method development but are
currently being evaluated as ways to further optimize
chiral separation.
Preliminary scale-up and method development on the
semipreparative scale has already proven successful,
potentially allowing for isolation of pure enantiomers on
the milligram scale. We are currently developing
semipreparative methods for such isolation. We are also
attempting to confirm the assignment of individual stereo-
isomers by using vibrational circular dichroism to deter-
mine absolute configuration (R/S assignment) of each
stereoisomer in solution. These efforts will allow assess-
ment of racemization rates of CWNAs in different buffer
conditions, as well as determination of spontaneous and
oxime-mediated reactivation rates of AChE inhibited by
defined individual CWNA stereoisomers.
Understanding the stereochemistry of CWNAs is criti-
cal for understanding the biology and biochemistry of
these compounds in vivo. Countermeasures such as stoi-
chiometric or catalytic bioscavengers, or protein-based
drugs designed to intercept and destroy nerve agents
in the circulatory system, require activity against the
more toxic stereoisomers of CWNAs to be effective. The
capacity to easily produce enantiomerically pure isomers
of CWNAs will aid in the development of medical counter-
measures against these highly toxic compounds by
allowing high-throughput screening of candidate enzymes
for activity against the more toxic CWNA isomer(s).
These reagents will also allow rapid determination of the
absolute kinetic parameters of candidate bioscavenger
enzymes for isolated stereoisomers, accelerating the
down-selection process of these enzymes prior to in vivo
protective efficacy studies.
Fig. 8. (A) Separation profile of diastereomic GD on 2x WhelkO1 (SS) col-
umns. Sets of enantiomers, (B) P(+) and (C) P(–), of GD were injected onto 2x
WhelkO1 (SS) columns to determine the relative elution order of the four GD
stereoisomers. The elution order based on retention time was determined for
each stereoisomer and represented in panel A. (D) Elution order of GD ste-
reoisomers from a single WhelkO1 (RR) column showing the reverse
enantioselectivity of the antipode stationary phase resulting in elution of the
P(–) enantiomers prior to the P(+) enantiomers.
(WhelkO1 (RR)) and was tested to determine how reversing
the stereochemistry of the functional group would influence
the elution order of the four GD diastereomers. Chiral sepa-
ration on a single (RR) column resulted in a reversed order
of elution but almost identical elution profile for the GD epi-
mers (Fig. 8D). As with the single (SS) column, the first two
stereoisomers eluted with near baseline resolution, but on
the (RR) column these two stereoisomers were the C(+)P
(–) followed by the C(–)P(–) isomer. Addition of a second
(RR) column in tandem is anticipated to result in the full
resolution of the two P(–) stereoisomers of GD. Based on
these results, the four diastereomers of GD can be isolated
to enantiopurity using tandem WhelkO1 (SS) columns for
isolation of the P(+) isomers and tandem WhelkO1(RR)
columns to isolate the P(–) isomers.
ACKNOWLEDGMENTS
We would like to thank John Richard Smith for insightful
discussions on chiral analysis, for kindly sharing reagents,
and for previously establishing initial conditions for chiral
separations of CWNAs at the USAMRICD. We also thank
Jon Oyler, USAMRICD, for thoughtful discussion and review
of this manuscript. Also we thank Dan Rolle and Lakshmi
Subbarao from Waters Corporation for technical assistance
important for making this study possible. This work was
supported by the Defense Threat Reduction Agency-Joint
Science and Technology Office, Medical S&T Division.
S.A.K. was supported in part by an appointment to the
Internship/Research Participation Program for the U.S.
Army Medical Research Institute of Chemical Defense,
administered by the Oak Ridge Institute for Science and
Education through an agreement between the U.S. Department
of Energy and the USAMRICD. The views expressed in this
article are those of the author(s) and do not reflect the official
policy of the Department of Army, Department of Defense, or
the U.S. Government.
CONCLUSION
The methods presented in this study demonstrate the
analytical chiral separation of GA, GB, GD, and GF via
an SFC-MS using the Regis WhelkO1 type column. The
Regis WhelkO1 (SS) column was capable of full baseline
Chirality DOI 10.1002/chir

824
KASTEN ET AL.
10. Tenberken O, Thiermann H, Worek F, Reiter G. Chromatographic prepa-
LITERATURE CITED
ration and kinetic analysis of interactions between tabun enantiomers and
acetylcholinesterase. Toxicol Lett 2010;195:142–146.
11. Reiter G, Mikler J, Hill I, Weatherby K, Thiermann H, Worek F. Chro-
matographic resolution, characterisation and quantification of VX enantio-
mers in hemolysed swine blood samples. J Chromatogr B Analyt Technol
Biomed Life Sci 2008;873:86–94.
1. Ballantyne B, Marrs TC. Clinical and experimental toxicology of
organophosphates and carbamates. Oxford, Boston: Butterworth
Heinemann; 1992.
2. Dolgin E. Syrian gas attack reinforces need for better anti-sarin drugs. Nat
Med 2013;19:1194–1195.
3. Okumura T, Takasu N, Ishimatsu S, Miyanoki S, Mitsuhashi A, Kumada
K, Tanaka K, Hinohara S. Report on 640 victims of the Tokyo subway
sarin attack. Ann Emerg Med 1996;28:129–135.
4. Benschop HP, Dejong LPA. Nerve agent stereoisomers — analysis,
isolation, and toxicology. Acc Chem Res 1988;21:368–374.
5. Benschop HP, Konings CA, Van Genderen J, De Jong LP. Isolation,
anticholinesterase properties, and acute toxicity in mice of the four
stereoisomers of the nerve agent soman. Toxicol Appl Pharmacol
1984;72:61–74.
6. Benschop HP, Konings CA, van Genderen J, de Jong LP. Isolation, in vitro
activity, and acute toxicity in mice of the four stereoisomers of soman.
Fundam Appl Toxicol 1984;4:S84–95.
7. Degenhardt CEAM, Vandenberg GR, Dejong LPA, Benschop HP,
Vangenderen J, Vandemeent D. Enantiospecific complexation gas-
chromatography of nerve agents — isolation and properties of the enantio-
mers of ethyl N,N-dimethylphosphoramidocyanidate (tabun). J Am Chem
Soc 1986;108:8290–8291.
8. Benschop HP, Bijleveld EC, Otto MF, Degenhardt CEAM, Vanhelden
HPM, Dejong LPA. Stabilization and gas-chromatographic analysis of
the 4 stereoisomers of 1,2,2-trimethylpropyl methylphosphonofluoridate
(soman) in rat-blood. Anal Biochem 1985;151:242–253.
12. Smith JR. Analysis of the enantiomers of VX using normal-phase chiral liq-
uid chromatography with atmospheric pressure chemical ionization-mass
spectrometry. J Anal Toxicol 2004;28:390–392.
13. Zhang C, Jin L, Zhou S, Zhang Y, Feng S, Zhou Q. Chiral separation of
neonicotinoid insecticides by polysaccharide-type stationary phases using
high-performance liquid chromatography and supercritical fluid chroma-
tography. Chirality 2011;23:215–221.
14. Pirzada Z, Personick M, Biba M, Gong X, Zhou L, Schafer W, Roussel C,
Welch CJ. Systematic evaluation of new chiral stationary phases for super-
critical fluid chromatography using
Chromatogr A 2010;1217:1134–1138.
a standard racemate library. J
15. Subramanian G. Chiral separation techniques: a practical approach.
Weinheim: Wiley-VCH; 2007.
16. Technologies A. The LC Handbook. Santa Clara, CA; 2011.
17. Reiter G, Koller M, Thiermann H, Dorandeu F, Mikler J, Worek F. Devel-
opment and application of procedures for the highly sensitive quantifica-
tion of cyclosarin enantiomers in hemolysed swine blood samples. J
Chromatogr B Analyt Technol Biomed Life Sci 2007;859:9–15.
18. D’Agostino PAaCCL. Analysis of chemical warfare agents: general overview,
LC-MS review, in-house LC-ESI-MS methods and open literature bibliogra-
phy: Defence Research and Development Suffield (Alberta); 2006.
19. Yeung DT, Smith JR, Sweeney RE, Lenz DE, Cerasoli DM. A gas
chromatographic-mass spectrometric approach to examining
stereoselective interaction of human plasma proteins with soman. J
Anal Toxicol 2008;32:86–91.
9. Harvey SP, Kolakowski JE, Cheng TC, Rastogi VK, Reiff LP,
DeFrank JJ, Raushel FM, Hill C. Stereospecificity in the enzymatic
hydrolysis of cyclosarin (GF). Enzyme Microb Technol 2005;
37:547–555.
Chirality DOI 10.1002/chir