Biological Hazard Markers of Exposure to Chemical Welfare Agents and Analytics Method
The first Gulf war increased the attention given to cyclosarin (GF), previously considered a nerve agent of secondary importance, while further development work has been done on newer oxime nerve agent treatments, such as HI-6. In addition, a considerable amount of work has been carried out on the skin effects of sulphur mustard. The terrorist incidents using nerve agents, which took place in 1994 and 1995 in Japan, kindled a considerable amount of interest in other countries and gave rise to a number of symposia, such as the seminar on responding to the consequences of chemical and biological terrorism held at Bethesda, Maryland, in July 1995. Subsequent events, such as the 9/11 attacks in New York and Washington and the Bali, Madrid and London tube bombings, none of which involved chemical agents, have increased the attention given to the possibility of the use of chemicals by Al-Qaida and other groups. The possibility of the terrorist use of chemical weapons means that the management of civilian casualties has to be considered. Previously, management of chemical casualties has generally been in the context of military personnel, who may be protected physically and in some cases, by pharmacological preparations, against chemical warfare agents, and who will in any case usually be young and physically fit. Civilian casualties, by contrast, may include the infirm, the elderly and children. In addition, armed forces may have in place procedures for dealing with chemical attacks, whereas, until recently, that was not the case for civilians. Most western countries now have in place some procedures to deal with civilian casualties in the event of a terrorist attack using chemicals. However, many problems remain, including, for example, the need for mass decontamination after an incident. This and other topics receive special attention. It is now about ninety years since chemical weapons were used on a large scale during World War I. That these weapons still pose a threat to both civilians and military personnel says little for mankind’s socio-political progress. We hope this may stand as a small memorial to his work in this area.
Disclaimer: The use of chemical weapons is prohibited by a number of international treaties or conventions. This illegality is said to confirm and codify a customary prohibition based on concepts of ethics and morals. Despite these prohibitions, and unlike other weapon systems also prohibited, the number of countries which have acquired or which are acquiring chemical weapons continues to increase. It might be argued that this is explained, in part, by the lack of force of international agreements resulting from their being based upon false assertions and their being established largely as a result of vigorous propaganda exercises in the aftermath of a world war. Current international moves to abandon chemical weapons may well be brought to a successful conclusion but, with the lack of a sound and persuasive argument against the use of such weapons, it remains very unwise to assume that clandestine production of such weapons will not be attempted and that they will not be used during a war. In addition, the lack of a sound ethical compulsion to desist from using such weapons may very likely lead to their proliferation and use in the so called Third World conflicts, particularly among nations with no memory of WWI.
The analysis of biomedical samples, such as urine and blood, can provide qualitative and quantitative information on exposure to CW agents. Detection of metabolites and covalent adducts provides forensic evidence in cases of allegations of military or terrorist use of CW agents. The methodology may also be used for diagnostic purposes to ensure administra- tion of appropriate medical countermeasures, and for monitoring exposure in workers engaged in demilitarization and other defensive activities. Some of the simpler analytical methods have been applied in toxicokinetic studies. This chap- ter reviews the biological fate of CW agents, the metabolites and adducts that may be used as biological markers of exposure, and analyt- ical methods for their detection. The emphasis is placed on the biological fate of CW agents and their biological markers of exposure. Ana- lytical methods are summarized; they have been reviewed in greater detail elsewhere (Noort et al., 2002a; Black and Noort, 2005; Noort and Black, 2005), together with methods for unchanged agents in biological fluids. In cases of allega- tions of CW use, battlefield use of riot control agents (RCAs), which contravenes the Chemical Weapons Convention, may be an important is- sue. The biological fate of RCAs is therefore also included.
Biological Reactions of CW Agents
The first requisite in analyzing biomedical sam- ples is knowledge of the biological fate of CW agents. Most are reactive electrophiles, i.e. they have atoms which are electron-poor that react with electron rich, nucleophilic sites on other molecules. This chemical reactivity is a key com- ponent in the biochemical mechanisms of action of vesicants, nerve agents and phosgene. Because they are chemically reactive, CW agents gener- ally have short lifetimes in the body, although the more lipophilic agents may be partially protected by sequestration into fatty tissues. The major fraction of an absorbed dose is rapidly metabolized and eliminated as metabolites, predominantly in the urine but with a small fraction in the faeces. The remainder of the absorbed dose is accounted for by covalent reactions with nucle- ophilic sites on macromolecules, such as proteins and DNA.
The most abundant free nucleophiles in the body are water and the tripeptide glutathione (γ- Glu — Cys — Gly), which has a reactive thiol group associated with the cysteine residue. Reactions with these nucleophiles, either chemically or me- diated by enzymes, are the starting point for most metabolic pathways of electrophiles and account for most of the metabolites excreted in urine. A host of nucleophilic sites also exist on proteins,
DNA and other macromolecules (Van Welie et al., 1992). Examples are cysteine (SH), serine and tyrosine (OH), lysine and N -terminal valine (NH2), and aspartic and glutamic acid (CO2H) amino acid residues on proteins, and NH and PO−3 residues on DNA. Haemoglobin and albumin are the most abundant proteins in blood, and covalently bound residues on these proteins provide biological markers of exposure to many electrophiles, including CW agents. Nerve agents are somewhat more selective in that they specifically target a serine OH group in the enzymes acetylcholinesterase (AChE) and butyrylcholinesterase (BuChE). Sulphur and nitrogen mustards react with guanine NH residues in DNA, which are present in every tissue in the body.
As described below, urinary metabolites have been identified for vesicants, nerve agents, 3- quinuclidinyl benzilate (BZ), hydrogen cyanide and the RCAs, CS, CR and capsaicin. Protein adducts have been identified for vesicants, nerve agents and phosgene, and DNA adducts for sul- phur and nitrogen mustards. With the rapid ad- vances being made in proteomics and metabo- nomics, new biological markers of exposure will undoubtedly be identified in the near future.
Free Metabolites as Biological Markers
Free metabolites occur primarily in urine. Uri- nary excretion may account for up to ∼ 90% of an absorbed dose of agent. Unlike blood, urine does not require invasive collection and is often the easiest biomedical fluid to obtain. It is a sim- pler, though more variable, matrix than blood, and does not have the complications of coagula- tion or red cell lysis. It should be frozen as soon as possible after collection to minimize chemical or microbial degradation of metabolites. Most uri- nary metabolites are relatively simple molecules, the synthesis of analytical standards is not too de- manding, and in many cases analytical methods can be performed using benchtop instrumenta- tion, e.g. gas chromatography — mass spectrom- etry (GC — MS) or gas chromatography — tandem mass spectrometry (GC — MS — MS). The major disadvantage of urinary metabolites as biologi- cal markers of exposure is demonstrated by the typical excretion profile shown in Figure 1. Up to ∼ 90% of the total amount excreted is usually eliminated within the first 48 — 72 h following ex- posure (with percutaneous exposure, the curve may be somewhat flatter and broader). Depend- ing on the exposure level, the detection of urinary metabolites within this time period is usually not too demanding, as illustrated by the detection of hydrolysis products of sarin following terrorist attacks in Tokyo and Matsumoto (Minami et al., 1997; Nakajima et al., 1998). After about 72 h, there usually follows a prolonged elimination of low concentrations of metabolites, and more sensitive methods are required to detect them. Experience has shown that in cases of allega- tions of CW use, particularly in remote conflicts, samples may be collected days or even weeks after the alleged exposure. Urine collected 13 days after exposure is the longest period after which metabolites have been reported in hu- man CW casualties, and concentrations were < 1 ng/ml (Black and Read, 1995a). In some cases, e.g. the hydrolysis products of sulphur mustard, nitrogen mustard (HN-3) and tabun, metabo- lites may not be unequivocal indicators of ex- posure because of their occurrence from other sources. Although urine is the preferred sample, free metabolites may be detected in blood if col- lected within a few days of exposure. This was illustrated by samples collected after an assas- sination with the nerve agent VX (Tsuchihashi et al., 1998) and the terrorist release of sarin in the Tokyo subway (Noort et al., 1998). Saliva is another possible sample for detecting metabo- lites (e.g. Timchalk et al., 2004) but so far has received little attention in the context of CW agents.
Protein and DNA Adducts as Biological Markers
Blood is the most important source of protein adducts, but requires medically trained personnel for its collection and stringent safety precautions in handling. The major advantage of protein adducts is that they are potentially much longer- lived biological markers than urinary metabolites. Provided the adduct is stable to chemical or metabolic degradation, it may persist for the lifetime of the protein. Approximate half- lives of albumin, haemoglobin and BuChE are 21 days, 42 days and 5 — 16 days, respectively. As shown schematically in Figure 1, although the concentrations of adducts in blood are initially much lower than those of urinary metabolites, after several days most adducts remain close to their initial concentration while those of urinary metabolites have declined substantially. The major disadvantages of protein adducts are that acquisition of analytics standards can be costly and demanding, and many of the methods re- quire liquid chromatography — tandem mass spectrometry (LC — MS — MS), which is generally more costly and experimentally more challenging than GC — MS and GC — MS — MS.
Haemoglobin and albumin are the two most abundant proteins in blood, and for reactions that are purely chemical (i.e. non-enzymatic) usually provide the most abundant biomarkers. In the case of nerve agents, enzymatic reactions with the enzymes AChE and BuChE provide sensitive biomarkers. Haemoglobin and albumin are water-soluble proteins that have a large number of hydrophilic and nucleophilic amino acid residues on the periphery of the molecule. An important difference is that an agent must penetrate the red cell membrane to react with haemoglobin, and this will depend partially on its physical properties. Albumin has a cysteine residue (34) which is much more accessible than those in haemoglobin. Ideally, blood should be separated into red cell and plasma fractions and stored frozen shortly following collection, but this is not always possible.
Many electrophilic toxicants, including ethy- lene oxide, methyl bromide and acrylamide, react with haemoglobin and albumin (To ̈rnqvist et al., 2002). It is therefore not surprising that one or both of these proteins form adducts with vesicants, nerve agents and phosgene, at least in vitro. Adducts with haemoglobin have been iden- tified for sulphur mustard (Noort et al., 1996; Black et al., 1997b), lewisite (Fidder et al., 2000) and phosgene (Noort et al., 2000). Adducts with albumin have been identified for sulphur mustard (Noort et al., 1999) and nitrogen mustard HN-2 (Noort et al., 2002b), some nerve agents (Black et al., 1999; Harrison et al., 2006) and phosgene (Noort et al., 2000). Adducts with AChE and BuChE have been identified for the range of nerve agents (Fidder et al., 2002; Elhanany et al., 2001). Another possible source of protein adducts is skin. Sulphur mustard forms adducts with keratin aspartic and glutamic acid residues (Van der Schans, G.P. et al., 2003). Hair is also a possible source of protein adducts but, unlike albumin and haemoglobin, most of the nucleophilic sites are on the interior of the protein with hydrophobic residues on the exterior.
DNA adducts provide biomarkers of exposure for sulphur and nitrogen mustards. However, turnover of these adducts appears to be much shorter than the lifetime of natural DNA due to the intervention of repair mechanisms.
Analytical methods for biomedical samples require careful sample preparation, i.e. selective concentration of the analyte from the matrix, and a sensitive and selective method for detection. Analytes may be extracted directly from blood, plasma or urine using liquid — liquid extraction, but more commonly by solid-phase extraction (SPE) using a variety of solid phases. Early methods tended to use C18 — or C8 bonded silica cartridges, or anion exchange for acidic analytes such as phosphonic acids. In recent years, there has been an increasing tendency to use hydrophobic — hydrophilic polymeric cartridges, which are often more efficient for extracting par- tially polar analytes such as thiodiglycol (TDG). In a few examples, solid-phase microextraction has been used, which in favourable cases may provide very low limits of detection (Wooten et al., 2002). Affinity SPE using molecularly im- printed polymers has been applied to phosphonic acids in urine (Meng and Qin, 2001).
Mass spectrometry is generally the only spectrometric technique that universally provides the requisite combination of sensitivity and specificity for analysis at low parts per billion (ppb) in urine and blood. Nuclear magnetic resonance (NMR) spectrometry, although a very specific technique, still does not have the sensitivity required for trace analysis in complex matrices. Mass spectrometry is usually used in combina- tion with gas or liquid chromatography, or less commonly with capillary electrophoresis. For analysis down to mid — low ppb levels, single stage GC — MS or LC — MS may provide adequate limits of detection, but for detection at low-sub ppb levels with a high degree of confidence tandem mass spectrometry is required. Immunoassays are useful for the rapid screening of multiple samples but, unless antibodies for the analyte are available, they require a considerable effort to develop.
Most metabolites, particularly those derived from hydrolysis, are polar and require derivatization prior to analysis by GC — MS. A disadvantage of this approach is that derivatization can be a major source of error in trace analysis. The metabolites normally require isolation from aqueous media before derivatization, and this can result in loss of analyte. In addition, the presence of large amounts of extraneous material in an extract, plus residual traces of water, may suppress derivatization. An advantage of derivatization is that it can be used to enhance detection. Conver- sion to perfluorinated derivatives and detection by negative-ion chemical ionization mass spectrometry (NICI-MS) provides the lowest limits of detection for TDG and phosphonic acids. LC — MS (usually LC — MS — MS) provides an alter- native that can be applied directly to concentrated aqueous solutions of metabolites, in most cases without the need to derivatize. LC — MS — MS can provide very low limits of detection, but for many of the simpler polar metabolites, such as TDG and phosphonic acids, lowest limits of detec- tion have been obtained with derivatization and GC — MS — MS. However, LC — MS instrumenta- tion is still improving, particularly with the wider use of capillary columns.
A number of strategies can be applied to the detection of protein adducts. Some proteins can be selectively isolated from blood by affinity SPE; alternatively, blood may be fractionated using precipitation techniques. Detection of the en- tire protein adduct is possible with modern MS techniques, e.g. using electrospray or desorption ionization methods, but limits of detection are usually modest with considerable chemical back- ground. A common approach is to selectively di- gest the protein with enzymes, such as trypsin or pepsin, to produce short-chain peptides, and de- tect the alkylated or phosphylated peptide using LC — MS — MS. Alternatively, the protein may be digested to its constituent amino acids using the protease from Streptomyces griseus (Pronase) or 6 M hydrochloric acid, although these methods tend to produce a large chemical background. In some cases, particularly where ester linkages are formed, the bound moiety may be displaced from the protein by hydrolysis, nucleophiles such as fluoride ion, or chemical derivatization, and detected using a simpler methodology, usually GC — MS.
An important aspect of trace analysis is that rigorous quality control must be included in an- alytical protocols if the results are to withstand international scrutiny. In cases of allegations of CW use, a chain of custody must be maintained and any positive analysis must be preceded by
a negative control sample taken through the entire analytical procedure to demonstrate that no cross-contamination of equipment has occurred. Acceptable criteria for the trace-level detection of biological markers of CW agent poisoning are currently being discussed with the Organization for the Prohibition of Chemical Weapons.
Sulphur mustard remains one of the CW agents of most concern because of its ease of synthesis, advantageous physical properties and the dual hazard that it presents from skin contact and lung and eye damage from exposure to vapour. More biomarkers have been identified for sulphur mus- tard than for any other CW agent, in part due to its relatively indiscriminate reactions with nucleophiles. Research into retrospective identification of poisoning was stimulated by the extensive use of sulphur mustard in the Iraq — Iran conflict and against the Kurdish population in Iraq.
Distribution and Metabolism
Studies with 35S-sulphur mustard showed that radioactivity was distributed rapidly throughout the tissues following intravenous (IV) administration in the rat (Maisonneuve et al. 1993). Typically, 50 — 80% of an absorbed dose is elimi- nated in urine following IV, intraperitoneal (IP) or percutaneous (PC) administration in the rat (Hambrook et al., 1992). Much lower amounts of radioactivity are excreted in faeces, typically 5 — 15% of the dose. Radioactivity persisted in the blood for > 6 weeks following IV adminis- tration, associated mainly with the haemoglobin, indicative of covalent adduct formation (Hambrook et al., 1993). A small percentage of radioactivity was retained in the skin following percutaneous exposure, presumably also bound to macromolecules.
The chemical and metabolic reactions of sulphur mustard are dominated by reactions with nucleophiles at its two electrophilic carbon atoms, plus oxidation of the electron-rich sulphur atom. Nucleophilic reactions proceed by an internal SN1 type mechanism, via the episulphonium ion shown in Figure 2.
The formation of the episulphonium ion is rate limiting and occurs very rapidly in polar solvents (Bartlett and Swain, 1949). In a competitive environment, the episulphonium ion reacts preferentially with ‘soft’ nucleophiles such as thiols, although it will react with a broad range of soft and hard nucleophiles. Under physiological conditions, reaction with the cysteinyl thiol function in glutathione (probably chemical rather than enzyme-mediated) competes with hydrolysis and reactions with nucleophilic sites on macromolecues. Metabolism studies in the rat with 35S- and 13C-labelled sulphur mustard showed that the initial reaction products with glutathione are metabolized by two divergent pathways to produce mercapturic acids (N-acetylcysteine conjugates) and β-lyase metabolites (methylthio/methylsulphinyl conju- gates) (Black et al., 1992a). The mustard sulphur atom in most metabolites is oxidized to sulphox- ide or sulphone. The various permutations of
reactions with nucleophiles, on one or both electrophilic carbon atoms, combined with three pos- sible oxidation states of the sulphur atom, produce a large number of metabolites (at least 20 by HPLC using radioactivity detection). Figure 3 shows 5 of the 9 metabolites identified (1 — 5), which are the ones that have so far been investigated as biological markers. Not shown are three monomercapturic acid conjugates and mustard sulphoxide, which is much less reactive than mustard and was a very minor metabolite. These metabolites were identified by offline mass spectrometry following isolation by preparative HPLC, before the introduction of modern LC — MS systems that use atmospheric pressure ionization. Many more metabolites (e.g. derived from the partial hydrolysis product, hemi-mustard) could be identified using current instrumentation. Metabolism by the mercapturic acid pathway to the bis-N-acetylcysteine conjugate of mustard sulphone (3) is consistent with early studies by Roberts and Warwick (1963), in which TLC and dilution assays were used for tentative identification. The β-lyase metabo- lites 4 and 5 are formed through cleavage of the S — C bond in an intermediate bis-cysteinyl conjugate by the enzyme β-lyase, followed by
S-methylation (Bakke and Gustafsson, 1984). This metabolic pathway is believed to be mediated predominantly by gut flora. Of the metabo- lites derived from simple hydrolysis, thiodigly- col sulphoxide (TDGO) (2) was the major one, with much lower amounts of TDG (1). A separate metabolism study of TDG in the rat showed that > 90% is excreted as TDGO (Black et al., 1993). An early metabolism study using IV administration in the rat reported excretion of the bis-glutathione conjugate of sulphur mustard (Davison et al., 1961).
In a quantitative study of urinary excretion in rats following PC exposure to sulphur mustard, 3.7 — 13.6% of the dose was excreted as products of hydrolysis, mainly as TDGO, and 2.5 — 5.3% as β-lyase metabolites, but with considerable variation between animals (Black et al., 1992b). The excretion of β-lyase metabolites showed a sharper decline than that of hydrolysis products, suggesting that the prolonged excretion of the latter results from TDG being slowly liberated from adducts with macromolecules, e.g. from esters formed with aspartic and glutamic acid residues. A satisfactory analytical method for the bis-mercapturic acid conjugate (3) was not then available.
Two other urinary metabolites have been reported in animal studies (Figure 4). N7- (2-Hydroxyethylthioethyl)guanine (6), derived from the breakdown of alkylated DNA, was detected in rats (Fidder et al., 1996). The un- usual metabolite (7), assumed to result from re- action with a histidine residue, was identified in the pig following percutaneous administration (Sandelowsky et al., 1992).
Reactions with proteins
Incubation of human blood with 35S-sulphur mustard, and analysis of tryptic digests by LC — MS — MS, identified alkylation on 6 histidine residues, 3 glutamic acid residues, and both of the N-terminal valines (Noort et al., 1996; Black et al., 1997b). N1 and N3 histidine adducts were the most abundant. Alkylated cysteine, aspartic acid, lysine and tryptophan were also detected in Pronase digests, although the cysteine, which is somewhat hindered in haemoglobin, may have originated from glutathione. In all cases, and with albumin and keratin, the adducts possess a CH2CH2SCH2CH2OH residue resulting from a single alkylation with hydrolysis at the second electrophilic carbon atom.
Sulphur mustard alkylates the cysteine-34 residue in human serum albumin (Noort et al., 1999, 2004b), which is known to react with a number of electrophiles. This residue is much more accessible than the cysteines in haemoglobin, and its reactivity is promoted by its relatively low p K a , resulting from intramolecular stabilization of the thiolate anion. Aspartic and glutamic acid residues on albumin also react with
sulphur mustard to give 2-hydroxyethylthioethyl esters (Capacio et al., 2004). Adducts with albu- min histidine residues have not been reported.
The skin is a primary target of sulphur mustard. Exposure of human callus to 14C-sulphur mustard showed that a significant part (15 — 20%) of the radioactivity was covalently bound to keratin (Van der Schans, G.P. et al., 2003). Approximately 80% of the bound radioactivity could be released by alkaline hydrolysis, indicating it to be bound as esters of glutamic and aspartic acid residues.
Reaction with DNA
It has long been known that the predominant interaction of sulphur mustard with DNA is alkylation of N7 of deoxyguanosine residues (Brookes and Lawley, 1960; Fidder et al., 1994). This and crosslinking actions are assumed to be responsible for the carcinogenic activity of sulphuric mustard. Depurination of the resulting N7- (2-hydroxyethylthioethyl)-2′-deoxyguanosine releases N7-(2-hydroxyethylthioethyl)guanine (6), which can be detected in tissue, blood and urine samples (Fidder et al., 1996). Sulphur mustard may also react with the phosphate groups in DNA, although no evidence for this has been presented.
Sensitive analytical methods have been developed for TDG, TDGO, the two β-lyase metabolites (4, 5) and the bis-N-acetylcysteine conjugate (3). TDG is best analyzed by GC — MS
or GC — MS — MS after derivatization. The bispentafluorobenzoyl derivative in combination with NICI-MS (Black and Read, 1988,1995a), and the bis-heptafluorobutyryl derivative in com- bination with electron ionization or positive ion chemical ionization (Jakubowski et al., 1990; Boyer et al., 2004; Riches et al., 2007), provide the most sensitive methods, with limits of detection (LODs) down to ∼ 0.1 ng/ml. Conversion to the bistrimethylsilyl or tertbutyldimethylsilyl derivative is widely used in environmental analy- sis but these give higher LODs in urine (Ohsawa et al., 2004). TDGO is most easily analyzed after reduction to TDG with titanium trichloride (Black and Read, 1991). This is because of the difficulty of isolation of this very polar metabolite from aqueous media, and the different reactions involving the sulphoxide function that may occur on derivatization (Black and Muir, 2003). TDG and TDGO can be analyzed by LC — MS but detection limits have been too high for biomedical sample analysis. The β-lyase metabolites are readily analyzed by GC — MS and GC — MS — MS, provided that the sulphoxide groups, as with TDGO, are reduced with tita- nium trichloride (Black et al., 1991; Black and Read 1995a; Young et al., 2004). This produces the single analyte O2 S(CH2 CH2 SCH3 )2 , which is easily extracted and analyzed (LOD ∼ 0.1 ng/ml). Thus, TDG, TDGO and the two β-lyase metabolites can be analyzed as two analytes, TDG and O2 S(CH2 CH2 SCH3 )2 , from the same aliquot of urine treated with titanium trichloride (Black and Read, 1995a; Boyer et al., 2004). Some TDG and TDGO is excreted as glucuronides, as indicated by increased levels after treatment of urine with glucuronidase. LC — MS — MS, using positive electrospray ionization (ESI), also provides a sensitive method for the β-lyase metabolites, detecting them individually as the original metabolites (Read and Black, 2004a). No satisfactory GC — MS method has been developed for the bis N -acetylcysteine conjugate (3), probably because of thermal instability, but LC — MS — MS using negative electrospray ionization provides acceptable detection limits (Read and Black, 2004b).
Sensitive methods have been reported for sulphur mustard adducts with N-terminal valine and histidine, and for TDG released from aspartic acid and glutamic acid residues. In the case of N-terminal valine, the alkylated amino acid can be selectively cleaved using a method developed for other alkylating agents (To ̈rnqvist et al., 1996). Reaction of haemoglobin with pentafluorophenyl isothiocyanate (a fluorinated Edman reagent) releases the alkylated amino acid as the hydantoin (8) (Figure 5), which is further derivatized to its heptafluorobutyryl derivative and analysed by GC — MS/GC — MS — MS (Noort et al., 1996, 2004a; Black et al., 1997a). This provides a relatively simple method. Alkylated N-terminal valine could be detected up to 90 days following administration of sulphur mustard (4.1 mg/kg IV) to a marmoset (Noort et al., 2002a; Benschop et al., 2000). TDG can be cleaved from aspartic and glutamic acids residues on haemoglobin and analyzed by GC — MS (Capacio et al., 2004). TDG released from aspartic and glutamic residues in whole blood proteins could be detected up to 45 days following administration of sulphur mustard (1 mg/kg IV) to an African Green monkey. Alkylated histidine is more problematic; haemoglobin is digested to its constituent amino acids with 6 M HCl and the alkylated histidine converted to its fluorenyl- methoxycarbonyl (Fmoc) derivative for analysis by LC — MS — MS (Noort et al., 1996; Black et al., 1997a).
A sensitive method was developed for detecting the alkylated cysteine residue based on affinity isolation of the protein, digestion with Pronase, and detection of the alkylated tripeptide Cys* — Pro — Phe (Noort et al., 1999, 2004b). This method is more sensitive than those for alkylated haemoglobin, based on incubations of plasma with sulphur mustard.
N7-(2-Hydroxyethylthioethyl)guanine (6) is analyzed in urine using LC — ESI — MS — MS. The adduct was detected in the urine of guinea pigs exposed to sulphur mustard, although concentrations declined rapidly after 36 — 48 h (Fidder et al., 1996). HPLC (Ludlum et al., 1994), 32P-postlabelling (Niu et al., 1996) and ELISA (Van der Schans, G.P. et al., 1994, 2004) methods have also been developed.
A limited number of biomedical samples became available from CW casualties of sulphur mustard poisoning during the Iraq — Iran conflict, and from attacks on Kurdish communities in Iraq. Addi- tional samples have since been obtained from accidental exposures in the laboratory and from old munitions.
A summary of positive analyses is shown in Table 1. Results for β-lyase metabolites from Iranian and Kurdish casualties (Black and Read, 1995a) clearly show the advantages of having sub-ng/ml detection limits when samples are collected several days after the exposure. In the case of two Kurdish casualties, where urine was collected 13 days after exposure, levels of β-lyase metabolites (0.3 and 0.1 ng/ml) were close to the LOD. The excretion of TDG was monitored for 14 days following an accidental laboratory exposure to sulphur mustard (Jakubowski et al., 2000). The casualty developed blisters on hands and arms (< 1% of body area) and erythema on his face and neck (< 5% of body area). A maximum excretion of 20 μg per day was observed between days 3 and 4 (maximum concentration 65 ng/ml), with < 10 ng/ml after 7 days. A problem with TDG and particularly TDGO is the occurrence of very low background levels in human (and animal) urine (Wils et al., 1985, 1988; Black and Read, 1988, 1991, 1995a). Recent screening of 105 human samples from subjects with no known exposure to sulphur mustard indicated detectable levels of TDG normally < 2.5 ng/ml. When combined with TDGO analysis, the geometric mean was 3.4 ng/ml, and as high as 20 ng/ml (Young et al., 2004). The source of these background levels is unknown, but TDG is used in inks and dyes for fabrics; there may also be a dietary source. For this reason, β-lyase metabolites are regarded as the more definitive biological markers of sulphur mustard poisoning, because these have not been detected in non-exposed individuals (Black and Read, 1995a; Young et al., 2004). The relative amounts of β-lyase and hydrolysis products ap- pear to be quite variable. In two casualties accidentally exposed to a WWI munition, hydrolysis products were more abundant in the urine of one casualty and β-lyase metabolites were more abundant in the other (Black and Read, 1995b). The bis-N-acetylcysteine conjugate (3) was detected in the same samples, but only at levels close to the LOD, and 15 years following collection, when a sensitive method had been developed (Read and Black, 2004b). Further analyses for this metabolite in human urine are required to determine its importance. In the rat it is a major metabolite.
Protein and DNA Adducts
Adducts with N-terminal valine, histidine and as partic/glutamic acid residues have been detected in samples from human casualties. N-Terminal valine adducts were detected in blood from Iranian casualties collected up to 26 days after exposure (Benschop et al., 1997; Black et al., 1997a), and in blood collected 2 — 3 days after exposure from the two subjects (see above) accidentally exposed from a WWI munition (Black et al., 1997a). In the case of a sample collected 26 days after exposure, adduct concentration corresponded with that found in human blood after incubation with 0.9 μM sulphur mustard. The
results were corroborated by immunochemical analysis for the DNA adduct in lymphocytes from the same blood samples (Benschop et al., 1997). The histidine adduct was detected at slightly higher concentrations than the valine adduct in those samples analyzed for both (Black et al., 1997a). In samples collected up to 10 days after exposure, signal to noise ratios for β-lyase metabolites were generally greater than those obtained for adducts, but after a longer period adducts are the superior biomarkers.
Non-Metabolized Sulphured Mustard
In most circumstances, absorbed sulphur mustard should be fully metabolized. It is, however, highly lipophilic and can partition into fatty tissues. High concentrations of sulphur mustard were reported in various organs and tissues, particularly abdominal fat, removed post mortem from a deceased casualty of severe sulphur mustard poisoning (Drasch et al., 1987). Hair may also be a source of unchanged sulphur mustard. A United Nations (1986) investigation reported the detection of sulphur mustard in hair removed from a CW casualty.
Three nitrogen mustards, HN-1, HN-2 and HN- 3 (Figure 6), are included in Schedule 1 of the CWC. Most of the limited information on metabolism relates to HN-2, which has been used as an anti-cancer drug (mechlorethamine).
In vitro studies of HN-2 incubated with rat and rabbit liver homogenates indicated N -demethylation to be a significant pathway (up to 7% in 120 min), as judged by the generation of formaldehyde (Trams and Nadkarni, 1956). Loss of one of the CH2 CH2 X substituents also oc- curred, as indicated by the generation of acetalde- hyde. In aqueous media, nitrogen mustards are hydrolysed to the corresponding ethanolamines (9) (Figure 6) and, like TDG, these are excre- tion products in rodents following exposure. No metabolites derived from conjugation of nitro- gen mustards with glutathione have yet been reported.
Ethanolamines were determined quantitatively following PC administration of nitrogen mustards in the rat. Excretion in urine up to 48 h accounted for < 0.1% of the applied doses of HN-1andHN-2,andupto∼0.3%ofHN-3 (absorbed doses were not determined) (Lemire et al., 2004). The ethanolamines appeared to be excreted unconjugated, as treatment of the urine with β-glucuronidase had no effect.
Metabolism studies with ethanolamines have been undertaken in the context of their use as industrial chemicals. Following PC and IV administration of N-methyldiethanolamine in the rat, a major fraction of the absorbed dose was excreted as unidentified urinary metabolites, with some unchanged N-methyldiethanolamine (Leung et al., 1996). Triethanolamine was excreted predominantly unmetabolized in mice following both IV and percutaneous administration (Stott et al., 2000).
Similar to sulphur mustard, HN-2 binds covalently with the cysteine-34 residue of human serum albumin in vitro. Although no detection following in vivo exposures has been reported, analogous albumin adducts have been demonstrated in cancer patients being treated
with the anti-cancer nitrogen mustard drugs melphalan and cyclophosphamide (Noort et al., 2002b). HN-2 alkylation of histidine residues in haemoglobin has been indicated (Fung et al., 1975) but no definitive studies have been reported. Nornitrogen mustard, HN(CH2CH2Cl)2, reacts with N-terminal valines in haemoglobin in vitro (Thulin et al., 1996).
The major monoalkylated adduct of HN-2 with DNA is N-[2-(hydroxyethyl)-N-(2-(7- guanyl)ethyl]methylamine, analogous to the adduct formed with sulphur mustard. Alkylation of DNA in water with HN-2 gave four principal products, derived from mono-alkylation of guanine at N-7 and adenine at N-3, and from crosslinking of guanine to guanine or guanine to adenine (Osborne et al., 1995). The ratio of alkylation at N-7 of guanine to N-3 of adenine was 86:14.
The diand triethanolamine hydrolysis products of nitrogen mustards are best analyzed in urine by LC — ESI-MS — MS (Lemire et al., 2003, 2004), with preconcentration using strong cation exchange SPE (LODs between 0.4 and 3 ng/ml). Application of this method to 120 human urine samples found no background levels above the LODs for N-methyl and N — ethyldiethanolamines, but a high occurrence (47%) of triethanolamine, ranging from < 3 ng/ml to ∼ 6500 ng/ml. Triethanolamine is therefore not an appropriate biological marker of exposure for HN-3. Triethanolamine is widely used in commercial products, including household detergents and cosmetics.
Protein and DNA Adducts
The albumin adduct with HN-2 can be detected by LC — MS — MS using a similar methodology to that used for the sulphur mustard adduct (Noort et al., 2002b). HPLC with UV detection (LOD, 10 ng/ml) has been reported for the HN-2 N7 — guanine adduct (Sperry et al., 1998), but does not appear to have been applied to human samples.
LEWISITE Biological fate
Weapons grade lewisite consists of lewisite I, (CHCl — CH)AsCl2 (90%) and lewisite II, (CHCl — CH)2 AsCl (∼10%), with very small amounts of the non-vesicant lewisite III, (CHCl — CH)3As (which imparts the characteristic geranium-like odour). Biomedical sample analysis has been directed only at products derived from lewisite I.
No detailed metabolism studies have been reported for lewisite I. It is rapidly hydrolysed to 2-chlorovinylarsonous acid (CVAA), which is excreted in the urine of experimental animals (Jakubowski et al., 1993; Logan et al., 1999). It appears to be excreted over a relatively short period.
Trivalent arsenic has a high affinity for thiol groups, a characteristic that is exploited in derivatization for GC — MS analysis. Consistent with this reactivity, lewisite I reacts in vitro with cysteine residues in haemoglobin, forming a crosslink. On incubation of human blood with 14C-lewisite I, 93% of the total radioactivity was found in the erythrocytes, with 25 — 50% associated with globin (Fidder et al., 2000). The residual radioactivity in the erythrocytes was probably bound to glutathione. LC — MS — MS of tryptic digests indicated the presence of several binding sites, and specifically identified a crosslink between the cysteine-93 and cysteine-112 residues of β-globin. Binding to albumin was not observed.
CVAA is analyzed by GC — MS after derivatization (Figure 7); Like lewisite I, CVAA reacts readily with mono and dithiols at ambient
temperature. 1,2 Ethanedithiol (Jakubowski et al., 1993; Logan et al., 1999), 1,3 propanedithiol (Wooten et al., 2002) and 2,3 dimercaptopropan-1-ol (British Anti- Lewisite, BAL) (Fidder et al., 2000) have been used for biomedical sample analysis. Unlike most derivatizing reagents, which are reactive electrophiles, thiols can derivatize in situ in the biomedical fluid or aqueous extract. In the case of BAL as a derivatizing agent, the free hydroxyl is subsequently converted to its heptafluorobutyryl derivative. CVAA can be concentrated from urine by C18 SPE, either before or after derivatization, or by solid phase microextraction after derivatization. The latter provides a very sensitive method. LC — MS of CVAA gives ill-defined peaks unless it is oxidized to the pentavalent arsonic acid (Black and Muir, 2003).
Lewisite bound to haemoglobin cysteine residues is displaced by conversion to the BAL derivative shown in Figure 7 (Fidder et al., 2000). Exposure of guinea pigs to lewisite (0.25 mg/kg, subcutaneous) could be demonstrated by whole blood analysis (i.e. adducts + CVAA) up to at least 240 h.
There have been no reported human exposures to lewisite for which biomedical samples have been analyzed.
The three types of nerve agent known to have been weaponized are typified by sarin (GB), VX and tabun (GA). Soman (GD) and cyclosarin (GF) are less volatile phosphonofluoridate analogues of sarin, and RVX or R-33 is a Russian analogue of VX with broadly similar properties. Tabun differs from the other nerve agents in that it does not possess a P-methyl substituent, which has important implications for retrospective identification. A large number of human biomedical samples were collected following terrorist attacks with sarin in Matsumoto City in 1994 and the Tokyo subway in 1995, and these have contributed substantially to the development of analytical methods.
Distribution and Metabolism
Following systemic administration, nerve agents are rapidly distributed throughout the tissues, partly as unchanged agent, partly as metabolites and partly bound irreversibly to serine esterasetype enzymes and other proteins. Approximately 90% of subcutaneous doses (0.075 mg/kg) of sarin and cyclosarin in the rat were excreted in the urine within 24 h, as phosphonic acids resulting primarily from enzymatic hydrolysis (Shih et al., 1994). Soman was eliminated more slowly with a biphasic elimination curve; approximately 50% was excreted within the first 24 h, rising to 62% after 7 days. The first phase of elimination is due to rapid enzymatic hydrolysis of the inactive P(+) isomers by phosphorylphosphatases; the second phase is from slower hydrolysis of the active P(-) isomers (Benschop and De Jong, 2001). VX, which is a poor substrate for phosphorylphosphatases, disappears from the blood at a much slower rate than sarin and soman. Toxicologically relevant levels of VX were present in guinea pig blood 10 — 20 h following IV administration (2 × LD50) compared to < 2 h for GB or GD (Van der Schans, M.J. et al., 2003). Following PC exposure, which is the main battlefield hazard of VX, blood levels of agent may be present for significantly longer.
The metabolism of nerve agents is dominated by hydrolysis. This occurs to a small extent by chemical reaction with water, but is predominantly mediated by
phosphorylphosphatases. The main hydrolytic pathways that occur in the body are shown in Figure 8. Alternative hydrolytic pathways are possible for VX (through P — O and O — C cleavage) but these have not been reported in animal studies. For phosphonofluoridates (sarin, soman and GF) and V agents, the major metabolites are alkyl methylphosphonic acids (11). Further hydrolysis to methylphosphonic acid (MPA) may proceed slowly, as indicated by analyses of victims of the Tokyo subway attack (Nakajima et al., 1998). In rats, only traces of MPA were observed in urine (Shih et al., 1994).
Tabun hydrolyses by two pathways, through P — N and P — CN cleavage. It has not yet been es- tablished which of these predominates in animals or humans. Further hydrolysis produces ethyl phosphoric acid and eventually phosphate, both of which are ubiquitous from other sources.
Hydrolysis of VX produces 2-diisopropy- laminoethanethiol, HSCH2 CH2 N(i Pr)2 . A meta- bolite (12), derived from enzymatic S- methylation of this hydrolysis product, was identified in human plasma following an assassination with VX (Tsuchihashi et al. 1998). Experiments in rats confirmed the rapid metabolic formation of (12) from HSCH2 CH2 N(i Pr)2 (Tsuchihashi et al., 2000). This metabolite has not been reported in urine. By analogy with β-lyase metabolites of sulphur mustard, it might be excreted as a sulphoxide.
Acetyl and butyrylcholinesterase
The biochemical target for nerve agents is the enzyme AChE. Nerve agents inhibit the en- zyme by catalysed phosphylation of a serine hy- droxyl group within the active site. The phos- phylated enzyme regenerates extremely slowly, unless reactivation is accelerated by particular nucleophiles such as oximes or fluoride. Nerve agents react similarly with the related enzyme BuChE (Figure 9), which acts as a scavenger of nerve agents; its main physiological func- tion has yet to be elucidated. In rodent species and the rabbit, carboxylesterases (which are also serine proteases) act as additional scavengers. Adducts with the range of nerve agents have been identified by LC — MS — MS (Fidder et al., 2002; Elhanany et al., 2001).
AChE and BuChE, which have half-lives of 5 — 16 days, provide excellent biomarkers, but with one disadvantage. With certain nerve agents, particularly soman, a rapid secondary reaction oc- curs within the active site, in which the phosphyl moiety is dealkylated. This process, known as ‘ageing’, leads to loss of structural information on the inhibitor, for example with soman the pinacolyl group is lost (Figure 9). It also results in a negatively charged phosphyl moiety, which is resistant to reactivation by oximes and fluoride ion. Half-times quoted for ageing of human red blood cell AChE in vitro are: soman, 2 — 6 min;
sarin, 3h, 5h; tabun, 13h, >14h; GF40h, 7.5 h; VX, 48 h (Dunn et al., 1997).
Adducts with albumin
Nerve agents also react with a tyrosine residue associated with the albumin fraction in blood (Black et al., 1999) (Figure 10). Analysis of tryptic digests from plasma incubated with sarin identified a phosphylated tripeptide, MeP(O)(Oi Pr) — Tyr — Thr — Lys, consistent with the protein being albumin (tyrosine residue 411), although this sequence is common and occurs in other proteins. Before the advent of modern mass spectrometry, diisopropyl fluorophosphate was reported to bind to a tyrosine residue in bovine serum albumin (Murachi, 1963). The reaction with tyrosine is assumed to be purely chemical, and therefore at low exposure levels a catalytic reaction with ChE should predominate. At higher exposure levels, tyrosine adducts are formed at significant concentrations, both in human blood in vitro and in guinea pigs in vivo. Detectable levels of tyrosine adducts occur at BuChE inhibition lev- els of 10 — 20% for soman, GF and tabun, and at 70% for sarin in incubates with human plasma in vitro (Harrison et al., 2007). For example, when tabun was incubated with human plasma, levels of adduct remained below the limit of detection until ∼15% of the BuChE was inhibited; there
then followed a steep rise in the concentration of the adduct. Adducts with all four agents have been detected in guinea pigs exposed to 0.5 or 2 × LD50 doses. The tyrosine adducts do not appear to age rapidly and can still be detected after therapeutic treatment with oximes (which should substantially reduce the amounts of non-aged phosphylated AChE and BuChE). Non-aged adducts with soman and tabun were detected 7 days after exposure to 5 × LD50 doses after treatment with oxime, atropine and anticonvulsant. Consistent with the reaction being entirely chemical, a tyrosine adduct with the less reactive VX was observed in blood in vitro only at high con-centrations. The probable formation of tyrosine adducts in a non-human primate is supported by the fluoride regeneration of soman (2.4 ng/ml) from plasma 3 days after an exposure to 2 × LD50 (intramuscular), when all of the inhibited BuChE should have aged (Adams et al., 2004). These studies also showed that agent could be regenerated from incubates of agent with human serum albumin. Albumin was recently shown to bind a biotinylated organophosphorus agent in mice, with diisopropyl phosphate and some organophosphorus pesticides competing for the same site (Peeples et al., 2005).
There are numerous methods for the analysis of alkyl methylphosphonic acids in urine and blood, mostly using GC — MS and GC — MS — MS. These have been recently reviewed (Black and Muir, 2003; Black and Noort, 2005) and only a representative selection is summarized here. Isolation from urine is usually achieved by hydrophobic SPE (C18, C8 or polymeric) at low pH, or by anion exchange SPE. Phosphonic acids require derivatization for GC — MS analysis, and at least four different derivatives have been applied to biomedical samples. Silylation (trimethylsilyl or tertbutyldimethylsilyl) are commonly used for environmental analysis and were used in the analyses of samples from casualties of the Matsumoto and Tokyo terrorist attacks (e.g. Minami et al., 1997; Nakajima et al., 1998). In these cases, the first samples were collected within hours of the exposure and the detection limits obtainable with silylation were sufficient. A GC — MS — MS method using methyl esters has been de- veloped for high throughput analysis (Baar et al., 2004). The lowest limits of detection, down to at least 0.1 ng/ml, have been obtained by con- version to pentafluorobenzyl esters and detection by negative-ion chemical ionization (Shih et al., 1991; Fredriksson et al., 1995; Miki et al. 1999). Although these methods were initially developed using research-grade mass spectrometers, simple ion trap instruments can achieve comparable detection limits (Riches et al., 2005). LC — MS — MS is less sensitive but LODs of 1 — 4 ng/ml were achieved for isopropyl methylphosphonic acid in serum and applied successfully to samples from Japanese casualties (Noort et al., 1998). The metabolite (12) derived from VX can be simply extracted from serum and analysed by GC — MS (Tsuchihashi et al., 1998).
Historically, the inhibition of BuChE in plasma has been used to monitor exposure to nerve agents using the classical Ellman colorimetric method (Ellman et al., 1961), or modifications thereof (Worek et al., 1999). This method is used routinely in occupational health monitoring, and could be used to rapidly screen casualties. Its disadvantages are that detection of low-level exposure requires previous baseline measurements, and the assay is non-specific with regard to the inhibitor.
Mass spectrometric detection of phosphylated BuChE provides a much more specific biomarker of exposure. BuChE is usually preferred to AChE because it is much more abundant in blood plasma than AChE is in erythrocytes. A versatile method of detection involves isolation of BuChE from plasma using affinity SPE, digestion with pepsin, and LC — MS — MS detection of a phosphylated nonapeptide encompassing the active site serine (Fidder et al., 2002). In the case of aged residues, this method will only identify part of the structure of the nerve agent although, provided it is a phosphonofluoridate or V agent, a methylphosphonyl MeP(O)-residue will be a clear indication that a nerve agent has been used (there are no pesticides in use that include a Me — P(O) moiety; fonofos has Et-P(O)).
An alternative, and experimentally less demanding method, displaces the organophosphorus moiety as a fluoridate (e.g. with sarininhibited BuChE the original nerve agent is regenerated), either from plasma BuChE (Polhuijs et al., 1997) or red cell AChE (Jakubowski et al., 2004). The alkyl methylphosphonyl fluoridate is readily extracted and detected using GC — MS (or GC — FPD). Fluoride reactivation is currently the most sensitive method for detecting ChE inhibition by nerve agents, enabling detection in blood samples with < 1 % BuChE inhibition (Degenhardt et al., 2004). The method has been demonstrated to be effective for GA, GB, GF and VX. Figure 11 shows detection up to 56 days following exposure of atropinized rhesus monkeys to a dose producing an initial 40% inhibition of BuChE (Van der Schans et al., 2004).
In the case of sarininhibited ChE, the phosphyl moiety has also been displaced as isopropyl methylphosphonic acid using trypsin digestion and alkaline phosphatase (Nagao et al., 1997; Matsuda et al., 1998).
Albumin adducts can be detected by LC-MS- MS as phosphylated tyrosine, after digestion of the albumin fraction with Pronase and SPE clean- up (Harrison et al., 2000b).
The only incidents of nerve agent poisoning where biomedical samples have been reported are those resulting from terrorist disseminations of sarin in Matsumoto City (1994) and the Tokyo subway (1995), plus an assassination using VX. Positive analyses are summarized in Table 2. In contrast to the CW incidents involving sulphur mustard, most of the biomedical samples were collected within hours of the incidents.
Very high levels of isopropyl methylphosphonic acid (iPrMPA) and methylphosphonic acid (MPA) were detected in the urine of a Matsumoto casualty with low blood AChE activ- ity and rendered unconscious (Nakajima et al., 1998). Urine was collected over a 7-day period. Estimated concentrations of iPrMPA declined from 760 ng/ml to 10 ng/ml and MPA from 140 ng/ml to below the LOD. A crude estimate of the total exposure was 2.8 mg. iPrMPA was also detected in urine collected over 7 days from casualties of the Tokyo attack (Minami et al., 1997). Concentrations were not reported but the estimated exposures were 0.13 — 0.25 mg
of sarin in a comatose patient and 0.016 — 0.032 mg in less severely intoxicated patients. i PrMPA was detected by LC — MS — MS in serum collected within 2 h of hospitalization at concentrations of 3 — 136 ng/ml in 4 casualties of the Matsumoto incident and 2 — 100 ng/ml in 13 casualties of the Tokyo attack (Noort et al., 1998). High levels of iPrMPA correlated with low levels of BuChE activity. Ethyl methylphosphonic acid and 2-diisopropylaminoethyl methyl sulphide (12) were detected by GC — MS and GC — MS — MS in the serum of a subject assassinated by application of VX from a disposable syringe to the neck (Tsuchihashi et al., 1998, 2000).
Application of fluoride reactivation to serum samples of casualties of the Matsumoto and Tokyo incidents yielded sarin concentrations in the range 0.2 — 4.1 ng/ml serum (Polhuijs et al., 1997). Hydrolytic displacement of the phosphonyl residue identified iPrMPA at levels sufficient for full scan mass spectra to be obtained from samples collected from casualties who died (Nagao et al., 1997); MPA was also identified. MPA was detected in formalin fixed brain tissues some two years later using a similar procedure (Matsuda et al., 1998). The phosphonylated peptic nonapeptide from BuChE was identified in serum samples from several casualties of the subway attack (Fidder et al., 2002).
QUINUCLIDINYL BENZILATE, BZ
BZ, 3-quinuclinidinyl benzilate, is an antagonist at central and peripheral muscarinic cholinergic receptors. It produces both mental and physical incapacitation in humans.
No metabolism studies have been reported but it has been assumed that its hydrolysis products, benzilic acid and 3-quinuclidinol (Figure 12), would be excreted in urine.
Unlike the vesicants and nerve agents, BZ is not a reactive electrophile (other than with regard to hydrolysis) and no adducts are known to be formed either directly or indirectly through reactive metabolites.
Little attention has been paid to the analysis of biomarkers of BZ. A method for detecting BZ, benzilic acid and 3-quinuclidinol in urine, iso- lated the metabolites (at different pH values) by SPE (C18 for BZ and benzilic acid, Florisil for 3- quinuclidinol) (Byrd et al., 1987, 1988). They were converted to their trimethylsilyl deriva- tives and analyzed by GC — MS. Urine was pre- treated with glucuronidase, although there was no evidence for the formation of glucuronide conjugates. LODs were in the range 0.5 — 5 ng/ml. LC — MS — MS methods are currently being developed.
No detailed metabolism studies of phosgene have been reported, although some relevant information is available from its occurrence as an active metabolite (through P450-dependent oxidation) of chloroform and carbon tetrachloride. Phosgene is a highly reactive electrophile with a half life < 1 s in water. It reacts chemically with two molecules of glutathione to form diglutathionyl dithiocarbonate, GluS — C(O) — SGlu, (Fabrizi et al., 2001) and with cysteine to form 2-oxothiazolidine-4-carboxylic acid (Kubic and Anders, 1980). Diglutathionyl dithiocarbonate has been detected in bile following IP administration of chloroform and carbon tetrachloride to phenobarbital treated rats (Pohl et al., 1981). However, it has not been established if significant amounts of these or related compounds are excreted in urine following exposure to phosgene. As an active metabolite of chloroform, phosgene reacts with the polar heads of phospholipids (Di Consiglio et al., 2001).
Consistent with its broad reactivity with nucleophiles, 14C-phosgene binds to haemoglobin and albumin upon in vitro exposure of human blood (Noort et al., 2000). Following Pronase digestion of globin, LC — MS — MS identified the pentapeptide O — C — (Val — Leu) — Ser — Phe — Ala, apparently derived from hydantoin formation between the N-terminal valine and leucine residues of the 1 — 5 N-terminal amino acids of α-globin. This adduct was rejected as a suitable biomarker of exposure because a peptide with similar properties was detected at trace levels in control samples, possibly formed by reaction of α-globin with carbon dioxide. LC — MS — MS of tryptic and V8 protease digests of human serum albumin identified crosslinking (via a urea linkage) of the lysine residues 195 and 199. C-Chloroform has been shown to bind in vivo to histones, which are lysine rich nuclear proteins (Diaz Gomez and Castro, 1980). In a model chemical system, representing amino acid residues 1 — 23 of human histone H2B, phosgene was shown to bind predominantly to lysine residues leading to cross linkages (Fabrizi et al., 2003).
Micro-LC — MS — MS enabled the adducted tryptic fragment from albumin to be detected in human blood exposed to ≥ 1 μM phosgene in vitro. The biomarker has not been demonstrated in animals exposed in vivo.
Distribution and Metabolism
Hydrogen cyanide (p K a , 9.22) is distributed in the body as hydrogen cyanide rather than cyanide ion. Inhaled hydrogen cyanide passes immediately into the systemic circulation and is distributed rapidly throughout the tissues (World Health Organization, 2004). The major proportion of cyanide in blood concentrates in the red cells bound to methaemoglobin, which acts as a cyanide sink. Cyanide is metabolized predominantly in the liver by conversion to thiocyanate (−SCN), through the action of the mitochondrial enzyme rhodanese, which catalyses the transfer of a sulphur atom from thiosulphate. Cyanide is excreted primarily as thiocyanate in urine. Approximately 15% of absorbed cyanide reacts with cystine to form 2-iminothiazolidine-4-carboxylic acid, which is in equilibrium with its tautomer 2-aminothiazoline-4-carboxylic acid (13) (Figure 13).
A small percentage reacts with vitamin B12, ap- pearing as cyanocobalamin in urine.
Cyanide can be regenerated from methaemo- globin by treatment with acid. There are many methods for detecting free and bound cyanide, and thiocyanate in blood; these have recently been reviewed (Black and Muir, 2003; Black and Noort, 2005). They have been applied mostly to the determination of cyanide levels in smokers and fire victims rather than cases of deliberate poisoning. 2-Aminothiazoline-4- carboxylic acid has been analysed in urine using LC with fluorescence detection after conver- sion to N-carbamylcysteine (Lundquist et al., 1995), and by GC — MS after conversion to its tristrimethylsilyl derivative (Logue et al., 2005). The analyte was concentrated from urine on a cation exchange resin.
Retrospective identification of cyanide poisoning in a CW context would be complicated by exposure from other sources, which include cigarette smoke, smoke from fires and some foods, e.g. cyanogenic glycosides in bitter almonds, fruit seeds and a number of plants. Quoted blood concentrations in non-smokers vary from a few ng/ml to >100 ng/ml. In nine fire victims, the con- centrations determined were 687 ± 597 ng/ml (Ishii et al., 1998). In smokers, cyanide lev- els in blood may rise to ∼ 500 ng/ml. 2-Aminothiazoline-4-carboxylic acid was detected in the urine of moderate cigarette smokers at concentrations between < 44 — 162 ng/ml (Lundquist et al., 1995).
Riot Control Agents
The major riot control agents, or ‘aids to arrest’, in current use are 2-chlorobenzylidene malononitrile (CS), 1-chloroacetophenone (CN) and capsaicin, N-(4-hydroxy-3-methoxybenzyl)-8-methyl-6-nonenamide (or pepper spray); the potent and persistent irritant dibenz[b,f]1:4-oxazepine (CR) has rarely been used (Olajos and Salem, 2001). Methods for the retrospective identification of exposure to RCAs have not yet been developed; the known metabolic pathways are summarized below.
Chemically, CS is a moderately reactive electrophile due to the presence of two electron withdrawing nitrile groups attached to the olefinic bond. The site of reaction with nucleophiles is the olefinic carbon adjacent to the aromatic ring. Reactions are SN2-like, i.e. CS reacts with nucleophiles directly in a bimolecular fashion. CS reacts quite rapidly with water when in solution (half-life ∼ 14 min, 25◦C, pH 7.4) to give 2-chlorobenzaldehyde and malonitrile. Reactions are much faster with thiols and amines. CS reacts rapidly with glutathione and plasma protein, although the reaction products have not been characterized (Cucinell et al., 1971).
The metabolism of CS is dominated by two pathways, the major one resulting from initial hydrolysis to 2-chlorobenzaldehyde and malononitrile (Figure 14), and the minor one from reduction to 2-chlorobenzylmalononitrile (dihydro- CS) (Figure 15) (Brewster et al., 1987). 2-Chlorobenzaldehyde (14) is further metabolized by hepatic oxidation to 2-chlorobenzoic acid (15), which is conjugated with glycine to form the major urinary metabolite 2-chlorohippuric acid (16). A minor metabolic pathway of 2-chlorobenzaldehyde is through reduction to 2-chlorobenzyl alcohol (17), which is excreted as glucuronide (18) and the N -acetylcysteine conjugate (19). The formation of the latter is postulated to occur via a sulphate intermediate (Rietveld et al., 1983). In these metabolic pathways, three of the carbon atoms of CS are lost as malononitrile, which is further metabolized to cyanide and thiocyanate. In the second metabolic pathway (Figure 14), CS is reduced to the hydrolytically more stable dihydro-CS (20), which is metabolized by hydrolysis with decarboxylation and excreted as the glycine conjugate (21), or by simple hydrolysis to the corresponding carboxamide
(22) and carboxylic acid (23). In this pathway, the carbon skeleton is either fully retained or loses two carbon atoms. No protein adducts of CS have been reported.
There have been no reported cases where exposure to CS has been confirmed by biomedical sample analysis. In human volunteer trials, with low concentrations of aerosolized CS, neither CS or 2-chlorobenzaldehyde were detected in the blood of six volunteers shortly after termination of the exposure (Leadbeater, 1973). A disadvan- tage of the major metabolic pathway is that three carbons are lost in the initial hydrolysis. It has yet to be shown if background levels of metabolites from this pathway exist in non-exposed individuals, resulting from environmental, dietary or drug exposure, e.g. chlorobenzoic acid is used as a preservative for glues and paints, and as an intermediate in the manufacture of fungicides and dyes. The alternative pathway, originating from reduction to dihydro-CS, should produce more definitive biomarkers but they account for a very low percentage of the dose in the rat.
CN is a less reactive electrophile than CS. Its reaction rate constant with phosphate buffer at pH 7.2 is approximately two orders of magnitude less than that for CS (Cucinell et al., 1971). It
reacts faster with sulphur based nucleophiles than with water, but still at a slower rate than CS. The reaction products of CN with nucleophiles have been poorly characterized. Although it has been in use much longer than CS, no metabolism studies appear to have been reported. It will inhibit a number of sulphydryl based enzymes such as lactic dehydrogenase (Mackworth, 1948), and so it may form covalent adducts with blood proteins.
Chemically CR is much less reactive than CS or CN, and is hydrolyzed relatively slowly. In the rat, following IV and intragastric administration of 14C-CR, most of the radioactivity (59 — 93%) was excreted in the urine, of which > 90% was excreted in the first 24 h (French et al., 1983a). Similar patterns were observed in the guinea pig and rhesus monkey after intragastric administration. The major metabolic pathway (Figure 16) is initiated by oxidation of the azomethine moiety to lactam (24). This is followed by hydroxylation at aromatic positions 4, 7 or 9 to give hydroxylactams (25 — 27), with excretion predominantly as sulphate conjugates in the rat and guinea pig, and as the non-conjugated hydroxy lactams in the rhesus monkey. C-7 is the major site of hydroxylation in all three species, the 7-hydroxy-lactam and its sulphate conjugate accounting for up to ∼ 60% of the administered dose. Deuterium labelling studies indicated that hydroxylation occurs via an arene epoxide intermediate (cf. benzene) (Harrison et al., 1978), but no mercapturic acid conjugates have been identified derived from reaction of a putative epoxide intermediate with glutathione.
Minor pathways arise from hydrolytic or oxidative cleavage of the azomethine moiety and reduction to dihydro-CR (French et al., 1983b). No samples from human exposures have been reported.
Capsaicin is the major pungent component of Oleoresin Capsicum (OC), commonly known as pepper spray. OC is extracted from dried ripe chilli peppers and is a variable mixture of many compounds. Related irritants (capsaicinoids) present in the mixture include dihydrocapsaicin, nordihydrocapsaicin, homocapsaicin, homodihydrocapsaicin and nonivamide. The latter is used as a synthetic substitute for pepper spray.
The metabolism of capsaicin and dihydrocapsaicin has been studied in the context of food and their use in skin creams for the treatment of arthritic pain and inflammation. Capsaicin (Figure 17) is a much more complex molecule than other RCAs, and offers more functional groups and other sites for metabolism. ω- Hydroxycapsaicin (28) was detected in the urine
of rabbits following IV administration of capsaicin (Surh et al., 1995). Following oral administration of dihydrocapsaicin (20 mg/kg) in the rat, ∼ 75% of the dose was eliminated in the urine as unchanged dihydrocapsaicin plus eight metabolites (Kawada and Iwai, 1985). The metabolites were all derived from an initial hydrolytic cleavage of the carboxamide function and consisted of free metabolites (14.5% of the total dose) and glucuronide conjugates (60.5%). The following were identified: dihydrocapsaicin (8.7%), vanillamine (29) (4.7%), vanillin (30) (4.6%), vanillyl alcohol (31) (37.6%) and vanillic acid (32) (19.2%) (Figure 18). Dihydrocapsaicin hydrolyzing enzyme activity was found in various organs of the rat, but particularly in the liver and gut.
The other product of hydrolysis is initially 8- methylnonanoic acid, but it is not clear to what extent this is metabolized.
One of the concerns for capsaicins as food components is the potential for the aromatic phenolic moiety to be oxidised by P450 enzymes to potentially carcinogenic epoxide, phenoxy radical or quinone type electrophilic intermediates. Detailed studies of the metabolism of capsaicin in vitro, by recombinant cytochrome P450 enzyme and hepatic and lung microsomes, were reported by Reilly et al. (2003) and Reilly and Yost (2005). Metabolites were identified that were derived from aromatic and alkyl hydroxylation, aryl O-demethylation, alkyl dehydrogenation and an additional ring oxygenation. Addition of GSH to microsomal incubations with capsaicin trapped several reactive intermediates as GSH adducts, although these have not been reported as metabolites in animal studies.
Confirming an exposure to capsaicin is likely to be complicated by the presence of background levels of metabolites. Capsaicin and dehydrocapsaicin are commonly ingested as hot
chilli spices. Vanillin, an intermediate in the hydrolytic pathway, is used extensively as a food additive.
Free metabolites and protein adducts for sulphur mustard and nerve agents, and DNA adducts for sulphur mustard, have been identified in experimental animals. Their validity as biomarkers of exposure has been demonstrated in samples collected from human casualties of accidental or deliberate exposure. Metabolites derived from hydrolysis have been detected in animal studies for nitrogen mustards and lewisite, plus adducts with either albumin or haemoglobin. These have not been demonstrated in human samples. In the case of HN-3, high and variable levels of the hydrolysis product, triethanolamine, in normal human urine precludes its use for confirming an exposure. No suitable biomarker for phosgene has yet been identified in animal studies although a protein adduct has been identified in vitro. Background levels of cyanide from sources such as cigarette smoke, fire smoke and certain food constituents make confirmation of cyanide exposure difficult.
Detailed metabolism studies have been re- ported for the RCAs, CS and CR, and to a lesser extent, capsaicin, but sensitive analytical meth- ods for the metabolites have yet to be developed. The formation of covalent adducts with proteins has been little studied, although observations have suggested that CS and CN react with proteins. In the case of CS and capsaicin, major metabolites are derived from an initial hydrolysis with loss of some of the carbon skeleton, and it needs to be established if background levels of these metabolites occur in non-exposed individuals.
Analytical methods continue to be improved, and in some cases adapted for less costly instrumentation. At present, expertise in biomedical sample analysis for CW agents is restricted to a small number of laboratories. Recent interest shown by the OPCW, and the concern for terrorist use of CW, may encourage a larger number of laboratories to acquire expertise.