A Systematic Method for the Targeted Discovery of Chemical

Article pubs.acs.org/ac

A Systematic Method for the Targeted Discovery of Chemical Attribution Signatures: Application to Isopropyl Bicyclophosphate Production Carolyn L. Mazzitelli, Michael A. Re, Melissa A. Reaves, Carlos A. Acevedo, Stephen D. Straight, and Joseph E. Chipuk* Signature Science, LLC., 8329 North MoPac Expressway, Austin, Texas 78759, United States S Supporting Information *

ABSTRACT: Potential attribution signatures for the synthesis of a highly toxic bicyclophosphate, 4-isopropyl-2,6,7-trioxa-1phosphabicyclo[2.2.2]octane 1-oxide (Isopropyl Bicyclophosphate or IPBCP) were discovered using a trilateral synthetic, analytical, and statistical approach. Initially, five synthetic routes were confirmed to successfully produce IPBCP using a range of reaction solvents, reactant ratios, and reaction temperatures. Experimental design principles were subsequently used to guide a formal study specifically aimed at discovering attribution signatures that could be used to differentiate forensic samples. A comparison of three-dimensional scatter plots comprised of the detected ions, their relative retention times (RRTs) and intensities (from LC-MS analyses) identified: (1) signatures that were unique to a synthetic route; (2) signatures associated with a combination of synthetic route and reaction solvent; (3) signatures related to reaction solvent, and (4) signatures associated with reagent source. Top level analysis revealed that the majority of the signatures are related to the synthetic route or a combination of the synthetic route and reaction solvent. Deeper analysis utilizing high resolution mass spectrometry (HRMS) and MSn revealed that most of the signatures stem from impurities in the reagents or byproducts formed from incomplete reactions between the reagents used in a given synthetic route. Finally, a subsequent validation study was performed to assess the presence and absence of the key route dependent signatures. icyclophosphates (BCPs) are antagonists of γ-aminobutyric acid (GABA), and thereby exhibit severe toxicity to mammals.1,2 Upon entering the body, the rapid and potent disruption of chloride ion flow through GABAA receptors results in overstimulation of the central nervous system and uncontrollable and fatal convulsions within minutes of exposure.1 Sharing a site of action with other cage convulsants such as picrotoxin,2 tetramethylenedisulfotetramine (TETS),3 and bicyclothiophosphates,4 the mammalian toxicity of BCPs stems from the noncompetitive antagonism of the chloride ion channel of the GABAA receptor.5−8 Therefore, the structures of these antagonists are much different than the neurotransmitter GABA as their mode of action relies on reducing the flux of chloride ions via strong interactions with the interior of the channel rather than modulating the flow from an external binding site. An example of a potent BCP is 4-isopropyl-2,6,7-trioxa-1phosphabicyclo[2.2.2]octane 1-oxide (Isopropyl Bicyclophosphate or IPBCP - structure shown in Figure 1). In general, the synthesis of BCPs is conducted in two steps. A triol intermediate, such as isopropyl triol (Triol) shown in Figure 1, is first synthesized and isolated. A separate reaction with a phosphorus containing reagent is then conducted to complete the formation of the caged phosphoester. A total of five synthetic routes were explored in this study (Figure 1), three of which were previously reported in the literature9−11 along with

B

© 2012 American Chemical Society

two additional routes that were based on the mechanisms of those previously reported. Chemical attribution signatures are the impurities, byproducts, and degradation products that can be used to readily differentiate samples of a material and associate them with a particular source of reagents, type of equipment, synthetic pathway, or reaction condition. The ability to link recovered material to the methods and reagents used in the manufacturing process make chemical attribution signatures an important component of forensic investigations. For example, there are several published reports of attribution studies performed on illicit drugs,12−15 chemical weapon precursors,16,17 and toxic agents.18,19 In these reports, attribution signatures were largely identified during the investigation of material that was commercially obtained or seized, but not during manufacture. Consequently, most of the reported results did not provide definitive information about the source of the discovered signatures. The objectives of this study were to utilize a trilateral synthetic, analytical and statistical approach to discover attribution signatures for IPBCP production and to subReceived: April 26, 2012 Accepted: June 22, 2012 Published: June 23, 2012 6661

dx.doi.org/10.1021/ac300859j | Anal. Chem. 2012, 84, 6661−6671

Analytical Chemistry

Article

J.T. Baker, and Acros Organics. Pyridine was purchased from Sigma-Aldrich. Hydrogen peroxide and acetone were purchased from Fisher Scientific, while ethyl acetate, acetonitrile, and water were purchased from Mallinckrodt Chemicals. Safety Considerations. IPBCP is highly toxic;1 therefore safety precautions were taken during its synthesis and when handling samples and solutions containing the compound. When setting up reactions, sampling and preparing analytical samples, and cleanup or decontamination activities, the following personal protective equipment (PPE) were worn: a full-face or half-face cartridge or air-supplied respirator, safety glasses or goggles, nitrile gloves, and lab coat or Tyvek disposable coveralls. Activities involving the isolation, purification, or handling of pure or concentrated IPBCP required the following PPE: a full-face or half-face cartridge or air-supplied respirator, safety glasses, double nitrile gloves, and a lab coat or Tyvek disposable coveralls. If a lab coat was used, disposable Tyvek sleeves were worn over the lab coat with the sleeves tucked into inner gloves, taped, and then covered with outer gloves. A full face or half face respirator and safety glasses or goggles were also worn by all occupants present in the laboratory room when the aforementioned operations were occurring. All activities involving IPBCP synthesis and sample preparation were performed in a designated laboratory room and ventilation hood which was marked with the appropriate signage. During operations involving IPBCP, at least two chemists were present in the laboratory. Following any operations involving the production, manipulation, or handling of samples that potentially contained IPBCP, all work surfaces and applicable equipment were decontaminated by wiping with 2 M sodium hydroxide, which was found to degrade IPBCP in previous experiments. Surfaces were further wiped using water, followed by acetone in-between reaction sets and after sample preparation. All reaction vessels, gloves, pipettes, and disposal equipment involved in the production, manipulation, or handling of samples that potentially contained IPBCP were fully submerged in a waste container of 2 M NaOH solution after use. Synthesis of 2-(Hydroxymethyl)-2-(propan-2-yl)propane-1,3-diol (Triol). A modification of the method described by Derfer et al. was used,20 and is described in the Supporting Information section. Synthesis of 4-Isopropyl-2,6,7-trioxa-1phosphabicyclo[2.2.2]octane 1-oxide (IPBCP). 100 mg (0.67 mmol) of Triol was added to an 8 mL reaction vial. 0.18 mL (2.2 mmol) pyridine was then added to the reaction vessel, followed by the addition of 60 μL (0.65 mmol) of phosphorus oxychloride and a stirbar. The reaction vessel was capped and heated at 100 °C in an aluminum heating block for 24 h with constant stirring and then cooled to ambient temperature. A 4 mL aliquot of ethyl acetate was added to the reaction vessel, followed by 1 mL of water and the reaction was vortexed for approximately 20 s. The ethyl acetate layer was removed and the solvent was evaporated to yield a white solid that was used without further purification. APCI-MS: 193, [M + H]+; HRMS: m/z calculated for C7H14O4P: 193.0630, found at 193.0624. Note: Procedures for the synthesis of IPBCP via the other synthetic routes are provided in the Supporting Information section. Sample Preparation. Table S-1 in the Supporting Information section summarizes the experimental design. A total of 180 samples were synthesized over multiple days, with a

Figure 1. Synthetic routes to produce IPBCP.

sequently validate their formation. The coupling of formidable toxicity and facile synthesis via multiple synthetic routes makes the discovery of chemical attribution signatures for IPBCP highly relevant. In this study, the experimental design was primarily aimed at the discovery of route-dependent signatures, but it also afforded exploration of the influence of reaction conditions (e.g., choice of reaction solvents, catalysts, concentrations, and temperatures), and commercial sources of the phosphorus containing reagent for each synthetic route. The signature discovery methodology utilized in this study follows the following steps: (1) Experimental design principles are used to guide a controlled synthetic chemistry study for the targeted discovery of source-dependent (i.e., influenced by the source of a reagent material), route-dependent (i.e., influenced by the synthesis pathway utilized in the reaction), and condition-dependent (i.e., influenced by specific reaction conditions such as choice of reaction solvents, catalysts, concentrations, and temperatures) signatures. (2) LC-MS analysis and data processing is used to generate scatter plots comprised of the detected ions and their relative retention times. (3) Potential signatures are compared to determine those that are unique to a particular raw material source, synthetic route, reaction condition or a combination thereof. (4) Structure elucidation of the key signatures is performed using complementary analytical techniques. The methodology presented here is demonstrated using IPBCP; however, this approach can be readily applied to other forensically relevant analytes.

■

EXPERIMENTAL SECTION Reagents and Materials. Phosphorus oxychloride was purchased from Alfa Aesar, Strem, and Acros Organics. Phosphorus pentachloride and phosphorus trichloride were purchased from Alfa Aesar, Strem, and Sigma-Aldrich. Triethyl phosphite was purchased from Alfa Aesar, Strem, and TCI America. Phosphoric acid was purchased from Sigma-Aldrich, 6662



■

subset of randomized samples being synthesized and analyzed on each day. Triplicate samples for each reaction were synthesized over the course of the study, thereby increasing the confidence in the presence or absence of signatures. Descriptions of the preparation of the sample, method blank, and solvent blank solutions are described in the Supporting Information section. UPLC-MS. Ultra-high-pressure liquid chromatography−mass spectrometry (UPLC-MS) using an Accela ultrahigh performance liquid chromatograph (Thermo Fisher, San Jose, CA) coupled to a LTQ linear ion trap mass spectrometer (Thermo Fisher, San Jose, CA) via an atmospheric pressure chemical ionization (APCI) source was used for the signature discovery studies. UPLC was performed using a reversed-phase Hypersil Gold UPLC column, 50 × 2.1 mm, 1.7 μm (Thermo Scientific, San Jose, CA) with a gradient program. The mobile phase was comprised of 0.1% formic acid in water (v/v) and 0.1% formic acid in acetonitrile. Chromatographic RRTs were calculated by dividing the retention time of the signature peak (in minutes) by the retention time of the IPBCP peak (in minutes). Targeted collision induced dissociation (CID) and MSn, as well as data dependent MSn experiments were performed on select samples for structure elucidation and confirmation of identified signatures. Specific instrument conditions are included in the Supporting Information section. To aid in the identification of unknown signatures, high resolution mass spectrometry (HRMS) experiments were also performed on select samples using an Accela UPLC (Thermo Fisher, San Jose, CA) with a high resolution Exactive Orbitrap mass spectrometer (Thermo Fisher, San Jose, CA). The chromatographic and APCI conditions were identical to those described in the Supporting Information section for work performed on the LTQ UPLC-MS instrument. Injection Sequence and System Suitability. To allow for meaningful comparisons of the LC-MS results to be made during the study, a standardized injection sequence was used for all sample analyses as follows: solvent blanks (at least two to ensure equilibration of the column), method blanks (five replicate injections), samples (one injection per sample, up to five consecutive samples), method blank (one injection). Up to five samples were bracketed by method blanks. Furthermore, a set of system suitability criteria were established for the LC-MS analyses and are detailed in the Supporting Information section. Data Conversion and Analysis. The LC-MS data files, acquired as *.raw files using the Thermo Fisher XCalibur instrument control software, were loaded into a database prior to further manipulation as described in the Supporting Information section. A description of the full data analysis process and the generation of the three-dimensional scatter plots as well as data filters established to prioritize potential signatures are also included in the Supporting Information section. An additional requirement for signature identification was the presence of the compound in at least two out of the three reaction replicates for each set of samples. Conversely, for the signature to be considered absent, it must not be present in more than one of the three reaction replicates. Upon identification of a potential attribution signature, its presence in all samples included in the experimental design was also assessed to ensure it was unique to a specific synthetic route, source, condition, or combination thereof.

Article

RESULTS AND DISCUSSION

Preliminary Studies. To ensure a properly targeted and thorough experimental design capable of assessing potential route, condition, and source-dependent signatures of IPBCP, preliminary studies were executed to determine the key synthetic routes and experimental conditions (e.g., time, temperature, and reaction solvent) for IPBCP production. All attempted synthetic routes in this study used Triol as a common starting material, and various phosphorus containing reagents as summarized by the general routes shown in Figure 1. The synthetic routes employing phosphorus oxychloride (POCl3),11 phosphorus trichloride with hydrogen peroxide (PCl3/H2O2),9 and triethyl phosphite with hydrogen peroxide P(OEt)3/H2O210 were previously reported and confirmed to produce IPBCP. In addition to the previously reported routes, two additional routes were also evaluated. Phosphorus pentachloride (PCl5) was evaluated as a phosphorus containing reagent due to its similarity in reactivity to PCl3 and the proposed mechanism for BCP formation. Reactions involving phosphoric acid (H3PO4), were designed to utilize the phosphate anion instead of the chlorinated precursors utilized in the literature routes. Both the PCl5 and H3PO4 routes yielded IPBCP and were therefore included in the experimental design. For each of the five verified synthetic routes, Triol was mixed with the phosphorus containing reagent with or without reaction solvent, and the reaction mixture was heated to a specified temperature. For Routes 4 and 5 an oxidation step using hydrogen peroxide was required to form IPBCP from a phosphite intermediate. Reaction conditions including solvent, temperature, and reaction time were varied in the preliminary investigation. A reaction time of 24 h was found to be adequate for all synthetic routes. For Routes 1 and 3−5, a reaction temperature of 50 °C was sufficient, while Route 2 required heating to 100 °C to form the desired product. These preliminary studies also revealed that IPBCP was formed in solvents covering a range of polarities in addition to reactions run without solvent. The analytical method for the detection of IPBCP and its related attribution signatures was also developed during the preliminary studies. Reversed-phase LC-MS using positivemode APCI was found to be ideal for the detection of IPBCP as well as the Triol starting material. This analytical method was chosen for the chemical attribution study since compounds that were structurally similar to IPBCP are likely to be detected by the method, and to maintain a focused scope for the study. While it is expected that other signatures that are better suited to detection by GC-MS are also present, LC-MS was used to detect larger molecular weight signatures that are not only more likely to persist in recovered forensics material, but are also expected to be specific to IPBCP production. Experimental Design. The targeted discovery of route, condition, and source-dependent attribution signatures followed the construction of a full factorial experimental design (Supporting Information Table S-1). For the determination of route-dependent signatures, each of the five synthetic routes were defined as variable parameters, while the reagent source and reaction conditions were held constant. Reagent sourcedependent signatures were assessed by using three independent suppliers for each of the phosphorus containing reagents. Finally, condition-dependent signatures were determined by altering the solvents used for the reactions. Reactions were 6663



Article

Table 1. Summary of Representative Signatures of IPBCP Production with Observed RRT, m/z, High Resolution MS Exact Mass Data, Possible Molecular Formula(e), and Proposed Structures When Available

6664



Article

Table 1. continued

6665



Article

Table 1. continued

a

Not expected to be present since acetonitrile was used for validation experiments. bRRTs were calculated by dividing the chromatographic retention time of the signature peak, in minutes, by the retention time of the IPCBP peak, in minutes.

performed in three solvents: ethyl acetate, acetone, and acetonitrile. In addition, reactions were performed without solvent. While other conditions such as reaction time and temperature could have been examined, the impact of solvent on the signature profile was determined to be the most relevant to a forensic investigation. To ensure further control over other experimental factors that could affect the signature profile of the material, Triol was synthesized as a single batch and used for all reactions in the design matrix. Pyridine and hydrogen peroxide were procured from a single supplier and one lot of each reagent was used for the entire study. The solvents employed in the study (ethyl acetate, acetone, and acetonitrile) were also obtained from single supplier with one lot of each used for the study. A synthesis schedule was developed based on the experimental design and randomized so that no single set of variables for a reaction was completed on a single day, therefore, lowering any possible confounding factors. Reaction blanks containing only reaction solvents were also prepared in the same manner as the other reaction samples (i.e., heating, stirring, sample preparation and analysis procedures). These control samples were analyzed to evaluate potential contamination and/or identify impurities associated with the reaction solvent. Signature Discovery. The primary objective of this work was to discover attribution signatures indicative of the synthetic route used for IPBCP production. In addition, the experimental design also facilitated the evaluation of potential conditiondependent signatures through the effect of reaction solvent on the signature profile, and reagent source-dependent signatures specific to the manufacturer of the phosphorus containing

reagent. The investigation revealed many more potential signatures than could be addressed in this report. Thus, to discuss the application of the methodology in a concise fashion only a few key signatures are highlighted in Table 1, while the breadth of signatures discovered in this study is demonstrated in the representative three-dimensional scatter plot shown in Figure 2. In this plot, potential signatures of Route 5 are shown, with relative retention time and m/z plotted on the x- and yaxes, and sum of ion intensity plotted on the z-axis. As discussed in the Supporting Information section, ion intensity was used to prioritize the signatures that were evaluated in this study. For example, the potential signature highlighted in the plot at m/z 383 and RRT 1.05 (Signature-7) had the greatest intensity and was one of the potential route-dependent signatures that was identified in the study. In addition to determining the m/z and RRT of unique signatures, experiments were also undertaken to elucidate the structures of the compounds using exact mass data determined from HRMS, and MSn fragmentation spectra. Tentative structures of key signatures are also presented in Table 1, and where possible, assignments were confirmed with analysis of those compounds that were commercially available or could be independently synthesized. It is important to note that for many signatures, especially those with molecular weights greater than 400 and those with multiple isobaric peaks, multiple structures are possible. For these compounds, Table 1 shows a plausible structure that is consistent with the data. Due to the scope of this study, structures could not be proposed for all discovered signatures, so exact mass data and likely molecular formulas are presented for those potential signatures with no proposed structures. As expected, most of the 6666



Article

Figure 2. Three-dimensional interactive scatter plots showing potential signatures of Route 5. The colors are indicative of the percentile range of the sum of the ion intensity values for the different points relative to all of the points in the plot: Red: greater than 90%; yellow: 81% to 90%; green: 71% to 80%; blue: 51% to 70%; and black: less than 50%. Figure 2A shows the plot with m/z 383 at RRT 1.05 and Figure 2B shows the same data without m/z 383 at RRT 1.05 to better visualize lower intensity signatures.

available triethyl phosphate was obtained from Sigma-Aldrich, and run in a side-by-side LC-MSn experiment with a sample containing the signature from the study. As demonstrated in Figures 4, the observed retention time, m/z, mass spectrum and MSn spectrum for the signature in the sample matched that of the commercially available triethyl phosphate, further confirming the identification of the signature. Diethyl isopropyl phosphate is another potential Route 4 signature that is suspected to result from an impurity in the triethyl phosphite starting material. This compound was confirmed to elute at RRT 0.69 with an m/z of 197. HRMS and MSn results (data not shown) are consistent with the structure shown in Table 1. This compound is expected to form from the oxidation of diethyl isopropyl phosphite, a probable impurity in the triethyl phosphite starting material. A number of the potential route-specific attribution signatures discovered in this study were structurally similar byproducts that resulted from the incomplete reaction between Triol and the phosphorus containing reagent. An example of this type of signature is the Route 4-specific compound present with an m/z of 239 and an RRT of 0.48 (Signature-2). The structure shown in Table 1 was proposed based on expected reaction intermediates as well as the molecular formula determined from

signatures identified from this experimental design were specific to a synthetic route or a combination of synthetic route and reaction solvent, with most signatures attributed to Routes 3 and 4. Interestingly, Routes 1 and 5 yielded the fewest number of potential signatures. Not only are both synthetic routes wellestablished in the literature,9,11 but in preliminary studies these routes were found to generate the highest conversion of starting material to IPBCP. Logically, synthetic routes with higher conversion to product are less likely to form the byproducts and degradation of byproducts that could result in potential attribution signatures. An ion with an m/z of 183 and RRT of 0.65 was found to be specific to Route 4. Figure 3A shows the presence of the signature peak in a representative extracted ion chromatogram of a sample synthesized via Route 4 in ethyl acetate. The same peak is absent in the extracted ion chromatograms of samples synthesized in ethyl acetate via Routes 1, 2, 3, and 5 (Figures 3B-E). The MSn spectra for this compound (Figure 4B and C) as well as the molecular formula generated from HRMS data (summarized in Table 1) suggested that the signature was triethyl phosphate, which would be expected to form exclusively in Route 4 reactions from the oxidation of unreacted triethyl phosphite by hydrogen peroxide. Commercially 6667



Article

to form the dichloro ethyl phosphate intermediate followed by addition of one molar equivalent of Triol using a mixture of dichloromethane and pyridine as the solvent system. However, in this case the mixture was not heated. This alteration facilitated the formation of the proposed signature, since the more reactive chlorine atoms would be easily displaced by Triol and the reaction of Triol with the ethoxy group would be less favorable at lower temperature. As shown in Figure 5, the retention time and fragmentation pathway of the synthesized molecule was found to match the signature when analyzed in a side-by-side LC-MSn experiment. An example of a reaction byproduct that was found to be a signature of Route 3 was m/z 229 present at RRT 1.16 and 1.18 (Signature-1). The presence of two, isobaric peaks with similar retention times in the extracted ion chromatogram for m/z 229 suggests the presence of isomers. The structure shown in Table 1 was proposed based on the HRMS exact mass data as well as the MSn spectra, both shown in Figure S-1. The use of HRMS in this study was especially useful as many potential signatures could contain chlorine atoms that produce characteristic isotopic patterns. As shown in Figure S-1A, the isotopic pattern for m/z 229 was indicative of one chlorine atom. The signature is thought to form from the reaction between PCl5 and a hydroxyl group of Triol forming an alkyl halide with POCl3 produced as a byproduct. This hypothesis is supported by the fact that PCl5 is commonly used to form alkyl halides from alcohols.21 The remaining hydroxyl groups of Triol react with PCl5 or POCl3 to form the Triol - phosphorus adduct. Upon dilution of the sample for LC-MS analysis or exposure to trace levels of water in the reaction mixture, any remaining P−Cl bonds are hydrolyzed to produce the suspected signature. Signature-3 (m/z of 457 at RRT 0.79), is another potential Route 3 signature. The structure is suspected to be similar to

Figure 3. LC-MS extracted ion chromatograms for m/z 183 from samples prepared in ethyl acetate via (A) Route 4, (B) Route 1, (C) Route 2, (D) Route 3, and (E) Route 5. The signature of interest is indicated by the star and is specific to samples synthesized by Route 4.

the HRMS data and the MSn fragmentation pathway. To confirm the structure, the proposed molecule was independently synthesized via the following method: dilute phosphorus oxychloride was reacted with one molar equivalent of ethanol

Figure 4. LC-MS extracted ion chromatograms and APCI-MSn spectra for m/z 183 from an attribution study sample (RDEN0132) synthesized by Route 4 with ethyl acetate (Figure 4A−C), and triethyl phosphate from Sigma-Aldrich (Figure 4D−E): (A) Extracted ion chromatogram for m/z 183 in RDEN0132, (B) MS/MS of m/z 183 in RDEN0132, (C) MS3 spectrum for the m/z 183 → 155 [M − ethyl]1+ product ion in RDEN0132, (D) Extracted ion chromatogram for m/z 183 in triethyl phosphate, (E) MS/MS of m/z 183 in triethyl phosphate, (F) MS3 spectrum for the m/z 183 → 155 [M − ethyl]1+ product ion in triethyl phosphate. 6668



Article

Figure 5. LC-MS extracted ion chromatograms and APCI-MSn spectra for m/z 239 from an attribution study sample (RDEN0132) synthesized by Route 4 (Figure 4A−C), and an independent synthesis of the proposed impurity structure (Figures 4D−E): (A) Extracted ion chromatogram for m/ z 239 in RDEN0132, (B) MS/MS of m/z 239 in RDEN0132, (C) MS3 spectrum for the m/z 239 → 127 [M − b]1+ product ion in RDEN0132, (D) Extracted ion chromatogram for m/z 239 in the independent synthesis sample, (E) MS/MS of m/z 239 in the independent synthesis sample (F) MS3 spectrum for the m/z 239 → 127 [M − b]1+ product ion in the independent synthesis sample.

8 reflects the structural similarities with Signature-1 and Signature-3. The Route 3 signatures observed with an m/z 587 at RRTs of 1.15, 1.18, and 1.19 are examples of the many oligermeric byproducts that could form from the reaction of excess Triol with multiple equivalents of PCl5 or POCl3. The structure shown in Table 1 is an example of one of the possible signatures; however, many related isomers are believed to be present due to the presence of multiple isobaric peaks observed in the LC-MS chromatogram. It is expected that larger oligomeric products with different amounts of the various components are also produced as byproduct in this reaction and are potential route-dependent signatures. In addition to attribution signatures that were discovered to be specific to a synthetic route, the data also revealed signatures that were dependent on both synthetic route and reaction solvent. For example, Signature-7 was found only in reactions produced by Route 5 in which acetone was used as the reaction solvent. The proposed structure shown in Table 1 is suspected to form from the incomplete reaction between two equivalents of Triol with one equivalent of PCl3. One of the hydroxyl groups of Triol also reacts with the excess acetone to form a hemiketal. This hemiketal would be in an equilibrium likely favoring the two unbonded constituent functional groups; however, elimination of a hydroxyl group and the excess acetone present would serve to shift this equilibrium enough to produce a small of amount of the suspected impurity. The exact mass data from HRMS as well as the MSn fragmentation pathway of the molecule further supports the proposed structure.

Signature-1 in that it is likely formed from incomplete condensation of Triol by PCl5. In this case, a dimer is formed with trace water present in the reaction mixture partially hydrolyzing the remaining P−Cl bonds, forming the theorized structure. The extracted ion chromatogram for m/z 457 in Figure S-2A shows three isobaric peaks, likely due to the multiple oligomeric products that could form from this reaction. The proposed structure is further substantiated by the MSn data shown in Figure S-2B and S-2C, which is reminiscent of the MSn spectrum of Signature-1. One of the major products in the MS/MS spectrum of m/z 457 is m/z 229. Further, the MS3 spectrum of 457 → 229 matches the MS/MS spectrum of m/z of 229 as shown in Figure S-1B. The similarity of the MSn spectra as well as the fact that both are observed as Route 3 signatures support the theory that Signature-1 and Signature-3 form from related mechanisms and contain shared structural components. An additional signature of Route 3, m/z 395 at RRT 1.04 (Signature-8), also shares a common fragmentation pattern with Signature-1 and Signature-3. A product ion with m/z 229 is formed in the MS/MS spectrum of 395, and the subsequent MS3 spectrum of 395 → 229 (data not shown) matches the MS/MS spectrum of Signature-1, shown in Figure S-1B as well as the MS3 spectrum of 457 → 229 shown in Figure S-2C. Furthermore, m/z 395 is formed as a product ion in the MS/ MS spectrum of m/z 457, and the MS3 spectrum of this ion matches the MS/MS spectrum of 395. The similarity of the CID product ions for signatures associated with Route 3 highlights the potential for using parent ion scans to identify structurally similar signatures that may be associated with a particular reaction route. The proposed structure for Signature6669



Article

A signature with m/z of 247 and RRT of 0.69 (Signature-6) was also found to be specific to synthetic route and reaction solvent, specifically occurring in Route 2 reactions using acetone as the reaction solvent. The HRMS and MSn data are consistent with the proposed structure shown in Table 1. Mesityl oxide is a common acetone impurity formed by the selfaldol condensation of acetone. The proposed structure would result from the 1,4 addition of one of the hydroxyl groups of Triol with mesityl oxide.22,23 Signature Validation Study Design. A subsequent study was performed to assess the utility of the key route-dependent signatures. To limit the scope of the validation study, two of the five synthetic routes to IPBCP were evaluated: Route 1 was selected since it was found in the signature discovery studies to be the optimal synthetic route for producing IPBCP. Route 3 was selected as it generated the most route-dependent signatures in the signature discovery study. A new batch of Triol was synthesized and purified for the validation experiments using the same procedure described in the Experimental section. Both synthetic routes were used to prepare triplicate samples of IPBCP using a single source of either P(OCl)3 (Strem Chemicals) or PCl5 (Alfa Aesar) following the same procedures described in the Experimental section. For this study, the reaction solvent was also a fixed parameter, with one batch of acetonitrile being used for all reactions. The presence of the previously identified Route 1 and Route 3 signatures (Table 1) was assessed in all validation samples. For a signature to be considered validated, it had to be present in all three replicate samples from the appropriate synthetic route, and absent in samples synthesized by the other route. As summarized in Table 1, all Route 3-dependent signatures were detected in the appropriate validation samples with the exception of the unknown signatures with a m/z 270 at RRTs 0.75 and 0.77, and m/z 340 at RRT 1.08. For Route 1, the unknown signature with a m/z of 469 at RRT 1.05 was not consistently detected in all three P(OCl)3 reaction samples. The absence of these compounds in the Route 1 and Route 3 validation samples suggest that they do not reliably form or they are typically present at levels close to the detection limits of the analytical method. Furthermore, the absence of several key signatures in the validation samples was consistent with the results of the initial signature discovery study. For example, the compound with a m/z 429 at RRT 1.31 was determined to be only present in Route 3 samples synthesized with no solvent in the discovery study. As expected, this signature was not detected in the validation study samples since acetonitrile was used as the reaction solvent. Likewise, two of the Route 1 signatures, m/z 273 at RRT 0.52 and m/z 461 at RRT 0.50 were only observed in samples synthesized in acetone and were not present in the validation samples. While more extensive experiments are needed to fully validate all of the IPBCP signatures discovered in this investigation, the results presented here clearly demonstrate the utility of the methodology.

revealed that many of the attribution signatures were associated with byproducts formed from incomplete reactions between the phosphorus containing reagents or impurities in the phosphorus containing reagents and the isopropyl-triol (Triol), or from impurities common to the particular type of phosphorus containing reagent. The signatures attributed to impurities in the phosphorus containing reagents showed no specificity to reagent manufacturer, and in general few source-dependent signatures were identified, suggesting either homogeneity in the starting materials or processes used to manufacture the reagents. While this study establishes that potential attribution signatures are present and can be efficiently discovered using a targeted methodology, future work will be focused on collection and full validation of these signatures. In addition, experiments to discover potential attribution signatures for other bicyclophosphate analogs and an evaluation of the impact of the purity of Triol on the signature profile of bicyclophosphates are also in progress.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in the text. This information is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Funding from The U.S. Department of Homeland Security (DHS) under Contract # HSHQDC-10-C-00099 is gratefully acknowledged. The authors also wish to thank Hector Flores for his statistical data analysis, Christopher Simpson for his work with database management and Lydia Pavlos for her assistance with the LC-MS analyses.

■

REFERENCES

(1) Bellet, E. M.; Casida, J. E. Science 1973, 182, 1135−1136. (2) Ticku, M. K.; Olsen, R. W. Neuropharmacology 1979, 18, 315− 318. (3) Esser, T.; Karu, A. E.; Toia, R. F.; Casida, J. E. Chem. Res. Toxicol. 1991, 4, 162−167. (4) Squires, R. F.; Casida, J. E.; Richardson, M.; Saederup, E. Mol. Pharmacol. 1983, 23, 326−336. (5) Bowery, N. G.; Collins, J. F.; Hill, R. G. Nature 1976, 261, 601− 603. (6) Casida, J. E.; Eto, M.; Moscioni, A. D.; Engel, J. L.; Milbrath, D. S.; Verkade, J. G. Toxicol. Appl. Pharmacol. 1976, 36, 261−279. (7) Milbrath, D. S.; Engel, J. L.; Verkade, J. G.; Casida, J. E. Toxicol. Appl. Pharmacol. 1979, 47, 287−293. (8) Bowery, N. G.; Price, G. W.; Hudson, A. L.; Hill, D. R.; Wilkin, G. P.; Turnbull, M. J. Neuropharmacology 1984, 23, 219−231. (9) Verkade, J.; Reynolds, L. J. Org. Chem. 1960, 25, 663−665. (10) Wadsworth, W. S.; Emmons, W. D. J. Am. Chem. Soc. 1962, 84, 610−617. (11) Ozoe, Y.; Eto, M. Agric. Biol. Chem. 1982, 46, 411−418. (12) Gröger, T.; Schäffer, M.; Pütz, M.; Ahrens, B.; Drew, K.; Eschner, M.; Zimmermann, R. J. Chromatogr. A 2008, 1200, 8−16. (13) Kunalan, V.; Nic Daéid, N.; Kerr, W. J.; Buchanan, H. A. .; McPherson, A. R. Anal. Chem. 2009, 81, 7342−7348. (14) Waddell-Smith, R. J. H. J. Forensic Sci. 2007, 52, 1297−1304.

■

CONCLUSIONS This report demonstrates a methodology that uses a rigorous synthetic, analytical and statistical approach for targeted signature discovery. The production of a potent neurotoxin, IPBCP, was used as a model to demonstrate the utility of the technique. Attribution signatures indicative of IPBCP synthetic route, reaction solvent, and a combination thereof were discovered. Subsequent structural elucidation experiments 6670



Article

(15) Lociciro, S.; Esseiva, P.; Hayoz, P.; Dujourdy, L.; Besacier, F.; Margot, P. Forensic Sci. Int. 2008, 177, 199−206. (16) Fraga, C. G.; Clowers, B. H.; Moore, R. J.; Zink, E. M. Anal. Chem. 2010, 82, 4165−4173. (17) Hoggard, J. C.; Wahl, J. H.; Synovec, R. E.; Mong, G. M.; Fraga, C. G. Anal. Chem. 2010, 82, 689−698. (18) Fraga, C. G.; Farmer, O. T.; Carman, A. J. Talanta 2011, 83, 1166−1172. (19) Fraga, C. G.; Acosta, G. A.; Crenshaw, M. D.; Wallace, K.; Mong, G. M.; Colburn, H. A. Anal. Chem. 2011, 83, 9564−9572. (20) Derfer, J. M.; Greenlee, K. W.; Boord, C. E. J. Am. Chem. Soc. 1949, 71, 175−182. (21) March, J. Advanced Organic Chemistry, 3rd ed.; John Wiley and Sons: Hoboken, NJ, 1985. (22) Hauser, M. Chem. Rev. 1963, 63, 311−324. (23) Lorette, N. J. Org. Chem. 1958, 23, 937−937.

6671


A Systematic Method for the Targeted Discovery of Chemical

Recommend Documents