Natural Chlordecone Degradation Revealed by Numerous

Article Cite This: Environ. Sci. Technol. 2019, 53, 6133−6143

pubs.acs.org/est

Natural Chlordecone Degradation Revealed by Numerous Transformation Products Characterized in Key French West Indies Environmental Compartments

Downloaded via EASTERN KENTUCKY UNIV on August 4, 2019 at 23:29:33 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Marion L. Chevallier,†,@ Oriane Della-Negra,†,@ Sébastien Chaussonnerie,†,▽ Agnès Barbance,†,▽ Delphine Muselet,†,▽ Florian Lagarde,† Ekaterina Darii,† Edgardo Ugarte,† Ewen Lescop,‡ Nuria Fonknechten,† Jean Weissenbach,† Thierry Woignier,§,○ Jean-François Gallard,‡ Stéphane Vuilleumier,∥ Gwenaël Imfeld,⊥ Denis Le Paslier,*,† and Pierre-Loïc Saaidi*,† †

Génomique Métabolique, Genoscope, Institut François Jacob, CEA, CNRS, Univ Evry, Université Paris-Saclay, 91057, Evry, France Institut de Chimie des Substances Naturelles, CNRS - UPR, 2301 Bâtiment 27, 1 avenue de la Terrasse, 91198 Gif-sur-Yvette Cedex, France § Aix Marseille Univ, Univ Avignon, CNRS, IRD, IMBE, Avenue Escadrille Normandie Niemen, 13397 Marseille, France ○ IRD, UMR IMBE, Campus Agro Environnemental Caraïbes B. P. 214 Petit Morne, 97235 Le Lamentin, Martinique, France ∥ Génétique Moléculaire, Génomique, Microbiologie (GMGM), Université de Strasbourg, UMR 7156 CNRS, 4 allée Konrad Roentgen, 67000 Strasbourg, France ⊥ Laboratory of Hydrology and Geochemistry of Strasbourg (LHyGeS), Université de Strasbourg, UMR 7517 CNRS/EOST, 1 Rue Blessig, 67084 Strasbourg Cedex, France ‡

S Supporting Information *

ABSTRACT: Production and use of the insecticide chlordecone has caused long-term environmental pollution in the James River area and the French West Indies (FWI) that has resulted in acute human-health problems and a social crisis. High levels of chlordecone in FWI soils, even after its ban decades ago, and the absence of detection of transformation products (TPs), have suggested that chlordecone is virtually nonbiodegradable in the environment. Here, we investigated laboratory biodegradation, consisting of bacterial liquid cultures and microcosms inoculated with FWI soils, using a dual nontargeted GC-MS and LC-HRMS approach. In addition to previously reported, partly characterized hydrochlordecones and polychloroindenes (families A and B), we discovered 14 new chlordecone TPs, assigned to four families (B, C, D, and E). Organic synthesis and NMR analyses allowed us to achieve the complete structural elucidation of 19 TPs. Members of TP families A, B, C, and E were detected in soil, sediment, and water samples from Martinique and include 17 TPs not initially found in commercial chlordecone formulations. 2,4,5,6,7Pentachloroindene was the most prominent TP, with levels similar to those of chlordecone. Overall, our results clearly show that chlordecone pollution extends beyond the parent chlordecone molecule and includes a considerable number of previously undetected TPs. Structural diversity of the identified TPs illustrates the complexity of chlordecone degradation in the environment and raises the possibility of extensive worldwide pollution of soil and aquatic ecosystems by chlordecone TPs.

■

been commonly used as a fire retardant) represent a particular class of OC compounds obtained from hexachlorocyclopentadiene dimerization.1,2 They share several characteristics such as low molecular orbital energy, high hydrophobicity, and low Kow.3 Chlordecone is also known to be a transformation product (TP) of mirex4 and kelevan, another insecticide mainly used in

INTRODUCTION

Organochlorine (OC) compounds have been used extensively worldwide for decades, for insect control in agriculture. Their use has gradually been prohibited since the 1970s because of biological biomagnification, high toxicity, and their long persistence in the environment. Many OC compounds synthesized from hexachlorocyclopentadiene1 belong to the Stockholm convention list of persistent organic pollutants (POPs), which includes aldrin, dieldrin, endrin, chlordan, heptachlor, mirex, chlordecone, and endosulfans. Among them, the insecticides chlordecone and mirex (which has also © 2019 American Chemical Society

Received: Revised: Accepted: Published: 6133

November 9, 2018 May 6, 2019 May 13, 2019 May 13, 2019 DOI: 10.1021/acs.est.8b06305 Environ. Sci. Technol. 2019, 53, 6133−6143

Article

Environmental Science & Technology Europe.5 Their highly stable perchlorinated bishomocubane structure renders such OC compounds extremely recalcitrant toward environmental conditions. Chlordecone formulations from the US (such as Kepone) and Brazil (such as Curlone) have mainly been used in the Caribbean area, Central America, and Africa,6 whereas the use of mirex has been reported in North America,7,8 Latin America, Europe,6 and more recently China.9 Because of its direct use or in situ formation, chlordecone is thus essentially present throughout the world and has even been detected in the coral reefs of French Polynesia.10 We thus studied chlordecone as a relevant and challenging pollutant of the hexachlorocyclopentadiene-based OC class. To date, chlordecone has been associated with two major environmental disasters. Environmental contamination near the Hopewell chlordecone production plant (US) in 1975 resulted in acute exposure of workers and massive pollution of the James River, extending over 100 miles, that lasted for decades.11 On a larger scale, extensive use of chlordecone in French West Indies (FWI) banana plantations from 1972 to 1993 has resulted in long-term pollution of environmental compartments and the local food chain (soil, water resources, farmed animals, and fish).6,12,13 Chronic exposure to chlordecone has also resulted in human health problems14−26 and subsequent socio-economic problems for the FWI and James River area.27−30 Following its application, chlordecone is absorbed by soil and sediment particles, especially soils and sediments of FWI with high organic content. The commonly accepted paradigm of the persistence of chlordecone for decades, and even centuries, in the FWI, based on a leaching model,31 was recently comforted by two studies that suggested only marginal degradation in tropical soils, if any.32,33 Traditionally, POP contamination is assessed and followed by environmental monitoring of the parent molecule, with only limited efforts dedicated to detecting TPs potentially formed in situ and assessing their impact. In the case of chlordecone, several laboratory studies on bacterial degradation of chlordecone suggested formation of two main TP families, as shown by GC-MS analysis: hydrochlordecones arising from reductive dechlorination (family A) and polychloroindenes formed after ring-opening and elimination steps (family B).34−37 With the exception of 8-monohydrochlordecone and 2,8-dihydrochlordecone,38 none of the structures of these TPs has been completely elucidated, i.e., all hydrogen and chlorine atoms have not been correctly positioned on core structures. In addition, the Archaeon Methanosarcina thermophila has been shown to convert chlordecone into unknown polar and nonpolar products, based on silica gel thin-layer chromatography analysis.39 A definitive assignment of these TPs to the hydrochlordecone and polychloroindene TP families could not be obtained. Despite extensive analytical studies on chlordecone contamination in the FWI,6 the only chlordecone derivatives detected in environmental samples so far (i.e., chlordecol and 8-monohydrochlordecone) turned out to be contaminants of commercial formulations of chlordecone. Moreover, the existence of other TPs cannot be excluded at present, as only targeted analyses have been applied. Although chlordecol and 8-monochlordecone are the only commercially available chlordecone derivatives, other TPs from families A and B have been reported following exposure of chlordecone to specific chemical degradation protocols. For example, UVirradiation can lead to the production of 8-mono- and 2,8dihydrochlordecones in very low yields,38 and reductive degradation of chlordecone in the presence of vitamin B12 and titanium citrate gives rise to TPs of both families A and B.37

Importantly, all other successful protocols for chemical degradation of chlordecone have required a reducing agent, either alone40 or in combination with vitamin B12.37,39,41,42 Here, we addressed the paradigm of the nonbiodegradability of chlordecone in the FWI environment through the prism of its TPs: (i) by extending the currently known library of chlordecone TPs,36 relying on anoxic microbial degradation and the use of untargeted GC-MS and LC-HRMS; (ii) through the development of chemical degradation protocols for synthesizing and purifying TPs at the milligram scale; (iii) by structural elucidation of isolated TPs using NMR spectroscopy combined with chemical derivatization; and (iv) by use of the newly elucidated TPs as analytical standards for TP detection in FWI soil, water, and sediment samples.

■

MATERIALS AND METHODS Chemicals and Analytics. All used chemicals and analytical and purification methods are described in detail in the Supporting Information, Supporting Methods. Analytical Methods. GC-MS Analysis. GC-MS analyses were carried out using a Thermo Fisher Focus GC coupled to a single-quadrupole mass spectrometer (Thermo Fisher DSQ II). The instrument was equipped with a nonpolar 30 m × 0.25 mm × 0.25 μm DB-5MS column (Agilent J&W) and a split/splitless injector. Ionization conditions and GC program were described elsewhere.36,37 Detailed chromatographic methods are provided in the Supporting Information. LC-HRMS Analysis. LC-HRMS analyses were carried out using a Dionex Ultimate 3000 LC system coupled to an LTQOrbitrap Elite mass spectrometer (Thermo Fisher Scientific) fitted with a heated electrospray ionization source (HESI) operating in negative ionization mode. Voltage optimization was described elsewhere.36 Chromatographic separation was achieved using a Thermo Fisher Syncronis C18 column (50 mm length, 2.1 mm inner diameter, 1.7 μm particle size). Detailed chromatographic methods are provided in the Supporting Information. Anoxic Microbial Incubations. Bacterial cultures were incubated at 24 °C in a glovebox (Unilab mBraun) under anaerobic conditions in an N2/H2 (98/2; V/V) atmosphere in modified MM+ medium36 containing Na2S (0.4 g/L) as a reducing agent. Bacterial Degradation of Chlordecone in the Presence of Consortium 86 and Citrobacter sp. 86. Modifications of the previously described MM+ medium36 were as follows: MgCl2 and NH4Cl were replaced by MgSO4 and (NH4)2SO4, respectively. NaCl, CaCl2, and KCl were left out. Two liters of freshly prepared medium were inoculated with an active consortium 86 culture (1/100 v/v). After 8 h of incubation, three glass bottles were filled with 650 mL growing bacterial culture and another with medium alone as a negative control. Chlordecone (162.5 μL of a solution of 200 mg chlordecone in 1 mL dimethylformamide) was added to a final chlordecone concentration of 50 mg/L in each bottle. The same procedure was carried out with Citrobacter sp. 86. Cultures were incubated and monitored for 250 days by GC-FID, GC-MS, LC-HRMS, and chloride analysis. Chlordecone Biodegradation in Soil/Liquid Microcosms from Guadeloupe Island. Soil microcosms were prepared from 0.5 g dry weight (dw) andosol or 0.5 g dw nitisol sampled from Guadeloupe Island. Each sample was inoculated under an oxygen-free nitrogen atmosphere containing 5% H2, at 25 °C, into 12 mL M9 mineral medium or M9 mineral medium 6134

DOI: 10.1021/acs.est.8b06305 Environ. Sci. Technol. 2019, 53, 6133−6143

Article

Environmental Science & Technology

Figure 1. TP formation during bacterial chlordecone (CLD) degradation by Citrobacter sp. 86, monitored by (A) GC-MS full scan analysis [CLD (green), A1 (blue), B1 (pink), and B3 (light pink)], (B) LC-HRMS analysis corresponding to extract ion chromatograms for quasimolecular ion [M − H]− of m/z = 506.6797, 472.7187, 296.8852, and 260.9271 [CLD (green), A1 (blue), C1 (purple), C2 (purple), C3 (magenta) and C4 (magenta), respectively].

compound A1 (46.4 mg; 9.8 × 10−5 mol; 50%) as a white solid. All NMR, GC-MS, and LC-HRMS analyses for A1 are provided in Figures S17, S26, and S30−S31. Preparation of TPs B1, B2, and B3−B4 (2,4,5,6,7Pentachloroindene, 4,5,6,7-Tetrachloroindene, 2,4,6,7-Tetrachloroindene, and 2,4,5,7-Tetrachloroindene, Respectively). Titanium(III) citrate (50 mL, 3.3 × 10−3 mol, 8.4 equiv) basified to pH 12.7 with NaOH (3 M) was added to a solution of chlordecone (200 mg, 3.9 × 10−4 mol, 1 equiv) and vitamin B12 (60 mg, 5.8 × 10−5 mol, 0.15 equiv) in degassed H2O/MeOH 64:36 (250 mL). The reaction mixture was stirred under N2 at room temperature for 80 min and quenched by exposure to O2. Extraction with pentane (5 × 250 mL), followed by concentration under reduced pressure, gave rise to a white crude solid. TPs B1, B2, and B3−B4 were purified by preparative HPLC. Isocratic elution (MeCN/H2O 7:3; v/v) was applied at a flow rate of 25 mL/min. Fractions containing B1 (retention time of 42 min), B2 (retention time of 28 min), and B3−B4 (retention time of 32 min) were pooled separately, extracted three times with pentane, and concentrated under reduced pressure. Each compound was then purified by PLC (preparative layer chromatography; Merck, PLC silica gel, 1 mm, F254, 20 × 20) (cyclohexane/EtOAc 9:1); the B1, B2, and B3−B4 retardation factors were 0.78, 0.68, and 0.88, respectively. B1 (32.6 mg; 1.2 × 10−4 mol; 30%), B2 (3.0 mg; 1.2 × 10−5 mol; 3%), and B3−B4 (4.1 mg; 1.3 × 10−5 mol; 4%) were obtained as white solids. All NMR, GC-MS, and LCHRMS analyses for B1, B2, and B3−B4 are provided in Figures S19−S20, S26, and S32−S48. Preparation of 13C-enriched B1 is described in the (Supporting Information, Supporting Methods). Preparation of TPs C1−6 and D1−4 is described in the Supporting Information, Supporting Methods. All NMR, GCMS, and LC-HRMS analyses are provided in Figures S22−S23, S26−S28, and S49−S69. Analyses of FWI Environmental Samples. Field Sites and Soil Sampling. Soil samples were collected on Martinique Island. Soil (andosol, nitisol, and ferralsol) from the vicinity of the “Montagne Pelée” volcano and bed sediments from Galion bay were sampled from the 0−30 cm layer and conserved in a glass box. River and mangrove water samples were collected from 0−30 cm below the water surface in glass bottles. Samples were stored in the dark at 4 °C until chemical extraction. Chemical Extraction Procedure. Each sample was processed in duplicate. For soils and sediments (4 g), 15 mL of Milli-Q

supplemented with vitamin B12 (2 mg/L).43 Duplicate samples were taken over 36 months, using a sacrificial approach. A microcosm pair (duplicate samples) was retrieved from each series and analyzed at t = 0, t = 10 months, and t = 36 months. Two negative-control series were conducted for each series (again in duplicate, for t = 0 and t = 36 months) and consisted of (a) aerobically incubated soil samples and (b) irradiated soil samples (30 ± 1.5 kGy). Duplicate samples from t = 0, t = 10, and t = 36 months were first basified to pH 12 with NaOH (1 M) and extracted with pentane (6 × 15 mL) after vortexing and decanting. Aqueous phases were then acidified to pH 1 with HCl (1 M) and extracted with CH2Cl2 (12 × 15 mL) after vortexing and decanting. Organic layers were pooled, concentrated in vacuo, and analyzed in duplicate by GC-MS and LC-HRMS (Figure S3 and Tables S2−S3). Chemical Access to Chlordecone TPs. All chemical chlordecone transformation experiments were monitored by GC-MS and LC-HRMS. Preparation of TP A1 (10-Monohydrochlordecone). Sodium sulfide (2.2 g, 2.8 × 10−2 mol, 140 equiv) and vitamin B12 (40 mg, 2.9 × 10−5 mol, 0.15 equiv) were added to a solution of chlordecone (100 mg, 2.0 × 10−4 mol, 1 equiv) in degassed water (300 mL). The reaction was carried out under a N2 atmosphere at room temperature (rt) for 30 h, quenched with HCl (6 M) to pH 4.0, and degassed with N2 for 1 h to evacuate hydrogen sulfide. The aqueous reaction mixture was extracted with DCM (3 × 200 mL) and the combined organic phases concentrated in vacuo, resulting in a brown viscous residue. A first purification step was performed on the crude residue using a Combi Flash Companion Elution column at a flow rate of 40 mL/min using heptane as solvent A and a mixture of DCM/ (CH3)2CO (1:1; v/v) as solvent B. Elution started with 0% B for 7 min, followed by a linear gradient, reaching 50% B within 5 min, a second linear gradient, reaching 100% B within 3 min, and further elution for 15 min with 100% B. Fractions containing A1 (from 8 to 29 min) were pooled and concentrated under reduced pressure. A second purification step was performed using a preparative HPLC system. Isocratic elution using a tetrahydrofuran/MeCN/(NH4)2CO3 buffer (10 mM, pH 9.5) (29:29:42; v/v/v) was applied at a flow rate of 20 mL/min. Fractions containing A1 (retention time of 9 min) were pooled, acidified to pH 3 with HCl (6 M), extracted three times with DCM, and concentrated under reduced pressure to give the title 6135


Article


Figure 2. Chlordecone TPs and their conditions of formation. (A) Chlordecone, chlordecol, and generic structures of the five TP families and (B) TP profiles and associated release of chlorine atoms in microbial or chemical treatment of chlordecone. 8-Monohydrochlordecone was assigned to A2. A3 and A4 correspond to dihydrochlordecones, and A5 to a trihydrochlordecone.

■

RESULTS AND DISCUSSION Search for New TPs Formed during Microbial Degradation of Chlordecone. To date, two families of chlordecone TPs have been identified during bacterial degradation of chlordecone.34−37 On the basis of GC-MS analysis, the detected TPs were assigned to hydrochlordecones (family A), with A1 as a monohydrochlordecone, and polychloroindenes (family B), with B1 (pentachloroindene) as main TP, and B2 and B3 (tetrachloroindenes) as minor TPs.36 Here, anoxic bacterial degradations of chlordecone in the presence of consortium 86 and Citrobacter sp. 8636 were replicated and analyzed using a set of complementary techniques (chloride titration, LC-HRMS, GC-FID, and GC-MS) to search for new TPs of chlordecone

water was added, followed by acidification to pH 1 with HCl (1 M) and vortexing. After decanting, the supernatant was extracted with DCM (12 × 15 mL) and the pellet washed twice with DCM (15 mL). For river and mangrove water, 0.75 L water sample was acidified to pH 1 with HCl (1 M) and extracted with DCM (12 × 350 mL). Organic layers were pooled, concentrated in vacuo, and analyzed in duplicate injections by GC-MS (in hexane/acetone 85:15) and LCHRMS (in 10 mM NH4OAc buffer/MeCN 4:1) (Supporting Information, Supporting Methods). Soil samples from Martinique taken at locations known not to be contaminated with chlordecone (Nitisol 926) were used as negative controls and were prepared and treated as mentioned above. 6136


Article

Environmental Science & Technology Chloride concentration increased by 18.8 ± 0.9 mg/L (i.e., 5.5 ± 0.3 Cl atoms per chlordecone molecule) in consortium 86 liquid cultures and by 19.7 ± 1.7 mg/L (i.e., 5.8 ± 0.5 Cl atoms per chlordecone molecule) in Citrobacter sp. 86 liquid cultures after 250 days (Figure S15), suggesting chlordecone dechlorination, which was confirmed here by both GC-MS and LCHRMS (Figure 1 and Figures S13 and S15). Although previous GC-MS analyses suggested the predominance of TP B1 (C9Cl5H3) and, to a minor extent, the presence of A1 (C10Cl9H3O2),36 chloride release clearly suggested the formation of other so far undetected chlordecone TPs containing four (or less) chlorine atoms. GC-FID monitoring also confirmed the presence of volatile TPs A1 and B1 in all experiments (Figure S14). In addition, untargeted LC-HRMS analysis revealed the formation of four previously undetected polar chlorinated compounds (Ci, i = 1, 2, 3, 4) of generic formula C10Cl4−nO2H4+n (n = 0, compounds C1 and C2; n = 1, compounds C3 and C4). Their low chlorine content matched the observed chloride release in these experiments (Figure 1 and Figures S15 and S27−S28). In parallel, laboratory microcosms with contaminated FWI soils (typically containing between 0.1 and 30 mg chlordecone/ kg of dry soil) were incubated in the dark for 36 months at room temperature under a N2/H2 atmosphere to evaluate the capacity of native FWI soil microbiota to degrade residual chlordecone (Figure S3). For each condition (soil/liquid medium), two replicates were collected over time by a sacrificial approach, to evaluate the produced chlordecone TPs using the same analytical protocols as for liquid cultures. GC-FID turned out to be inadequate due to high background from soil samples and low chlordecone concentration. Untargeted GC-MS analyses (Table S2) demonstrated predominance of TP B1, sporadic detection of A1, and the presence of B2 and B3 previously reported for bacterial degradation of chlordecone.36 In addition, a number of other chlorinated compounds were detected: B0 (19.10 min), B5 (10.00 min), D1 (18.78 min), D2 (18.81 min), D3 (16.54 min), D4 (16.57 min), E1 (19.85 min), E2 (19.88 min), and E3 (17.60 min and E4 (17.63 min). LC-HRMS analyses (Table S3) demonstrated the presence of Ci (i = 1, 2, 3, and 4), already observed during degradation of chlordecone in the presence of consortium 86 and Citrobacter sp. 86. Chlordecol, a known chlordecone contaminant in commercial formulations and present in FWI soils at low levels,44 was also found in all samples. Mass Spectrometric Analysis of Newly Detected Chlorinated Compounds. The combined approach of GCMS and LC-HRMS from in-source fragmentation chosen in this study was sufficient to classify the newly detected chlorinated compounds into four distinct families (the known polychloroindene family B, and three new families C, D, and E), and to postulate a generic core structure for each of the families C, D, and E newly identified in this work (Figure 2). On the basis of GC-MS data, we first assigned compounds B0 and B5 to hexachloroindene and trichloroindene, respectively. Indeed, their mass spectra contained ion series analogous to those previously reported for pentachloroindene B1 and tetrachloroindenes B2 and B3, 36 i.e., [C 9 Cl 6−n H 2+n] +• , [C9Cl5−nH2+n]+, [C9Cl4−nH2+n]+•, [C9Cl4−nH1+n]+, with n = 0 (B0), n = 1 (B1), n = 2 (B2 and B3), and n = 3 (B5) (Figures S18−S21). The other chlorinated compounds detected in GC-MS were grouped into two new families (D and E), with respective generic formulas of C 11 Cl 4−n O 2 H 6+n (n = 0, 1) and

C 12 Cl 4−n O 2 H 8+n (n = 0, 1). The presence of ions [C10Cl4−nH3+nO2]+, [C10Cl4−nH3+nO]+, [C10Cl4−nH2+nO]+•, [C9Cl4−nH3+n]+• (n = 0, 1) in the mass spectra of D1-D2 (n = 0) and D3-D4 (n = 1), displaying the same chromatographic shape and probably corresponding to in-source losses of CH3•, CH3O•, CH4O, and C2H3O2•, suggested the presence of a methyl ester moiety in D1-D2 and D3-D4 (Figures S22 and S23). Similarly, for E1-E2 and E3-E4, presumed in-source losses of C2H4, C2H5O•, C2H6O, and C3H5O2• were indicative of the presence of an ethyl ester moiety in E1-E2 and E3-E4 (Figures S24 and S25). In addition, the series of in-source fragment ions (C9ClxHy+/ C9ClxHy+•, x = 6, 5, 4, 3, and 2; y = 4, 3, 2, and 1) were common to compounds Bi (i = 0, 1, 2, 3, and 5), Dj (j = 1, 2, 3, and 4), and Ek (k = 1, 2, 3, and 4). This suggested a shared polychloroindene aromatic ring, with methyl- and ethylpolychloroindenecarboxylate structures for families D and E, respectively (Figure 2). Finally, the structure of chlorinated compounds Ci (i = 1, 2, 3, 4) only detected using LC-HRMS was investigated in more detail. The ions [C9Cl4−nH3+n]− (n = 0, 1) observed in the negative mass spectra of C1−C2 (n = 0) and C3−C4 (n = 1), displayed the same chromatographic shape in the LC-HRMS run as the quasi-molecular ions [C10Cl4−nO2H3+n]− (n = 0, 1) (Figure S27 and S28). Observed fragment masses correspond to the loss of CO2, likely indicating in-source decarboxylation of C1−C2 and C3−C4. This observation, associated with similar UV−visible absorption to reported indene core-ring profiles41 (Figure S2), suggested a polychloroindenecarboxylic acid structure for the new family C of chlorinated compounds (Figure 2). Chemical Reductive Degradation of Chlordecone. We then investigated reductive degradation of chlordecone by chemical treatment in order to (i) demonstrate the possibility of chlordecone to be transformed into the 14 newly detected chlorinated compounds, (ii) complete the structural elucidation of chlordecone TPs using NMR technique, and (iii) confirm the proposed generic structures for families C, D, and E. Chlordecone was first subjected to a selected set of 21 different chemical treatments, based mainly on literature protocols.37,39−42,45 This included treatment with a reducing agent alone, or in the presence of a metal complex (Table S1). Dual GC-MS and LC-HRMS monitoring of chemical reactions was applied to select the most suitable conditions, which were then further refined for milligram-scale recovery of the major detected chlordecone TPs. Three conditions (Table S1, entries a1, a4, and a6) expected to selectively produce polychloroindenes 41,42 resulted in the formation of chlorinated compounds of families B, C, and D or E depending on the used alcohol cosolvent (Figure S7). Titanium citrate and zerovalent iron without vitamin B12, in contrast, oriented chlordecone degradation toward hydrochlordecone formation (Table S1, entries e5 and e7). Polychloroindenes and chlorinated compounds of family C were formed concomitantly at the highest levels upon addition of vitamin B12 in the presence of titanium citrate and zerovalent iron (Figure 2 and Table S1, entries a5 and a7). Further, a combination of sodium sulfide and vitamin B12 specifically led to the production of monohydrochlordecone A1 (up to a 50% yield after purification). From the same series of experiments, compounds B1, B2, and B3 were isolated in yields of up to 30%, 3%, and 4%, respectively. Using zerovalent iron and vitamin B12 (Table S1, entry a5), compounds C1−C2 (5% yield), C3−C4 (3% yield), and C5−C6 (traces) were obtained 6137


Article


Figure 3. Chemical derivatization of Ci (i = 1, 2, 3, and 4). (A) Formation of commercial standard 4-carboxyindane from both C1−C2 and C3−C4 (LC-HRMS detection) (Figure S8), (B) formation of D1-D2 and E1-E2 from C1−C2 (GC-MS detection) (Figure S6), and (C) formation of D3-D4 and E3-E4 from C3−C4 (GC-MS detection) (Figure S6).

Figure 4. 1H NMR spectra regions of selected indene-based TPs (600 MHz). Data were recorded in (CD3)2CO excepted for B1 in CDCl3.

demonstrating an equilibrium between isomer pairs. Finally, supplementation of the reaction mixtures producing B and C TPs with either methanol or ethanol led to the production of

as nonseparable pairs of isomers. Indeed, when a single Ci (i = 1, 2, 3, 4) compound was isolated using preparative HPLC, partial interconversion to its isomer occurred upon evaporation, 6138


Article


We thus identified B3 and B4 as 2,4,6,7-tetrachloroindene and 2,4,5,7-tetrachloroindene, respectively. Complete Elucidation of C3−C4 Structures. Formation of 4-carboxyindane through hydrogenation of C3−C4 validated trichloroindene-4-carboxylic acid and trichloroindene-7-carboxylic acid as the reactants (Figure 3A). This was supported by the NMR spectra of B3−B4, which were highly similar to those of C3−C4 (Figure 4C). A strong HMBC cross-correlation peak between the less-deshielded protons and the carboxyl carbons of C3 and C4 confirmed the structures of 2,5,7-trichloro-1Hindene-4-carboxylic acid and 2,4,6-trichloro-1H-indene-7-carboxylic acid for C3 and C4, respectively (Figure S53). Complete Elucidation of B2, C5−C6, D1-D2, D3-D4, E1-E2, and E3-E4 Structures. Compounds B2 and C5−C6 showed the same 1H NMR pattern, allowing us to distinguish them from B1, C1−C2, B3−B4, and C3−C4 (Figure 4). All 1H NMR signals were assigned to the five-membered ring of the indene core. B2, C5, and C6 were thus identified as 4,5,6,7tetrachloro-1H-indene, 5,6,7-trichloro-1H-indene-4-carboxylic acid, and 4,5,6-trichloro-1H-indene-7-carboxylic acid, respectively. Finally, chemical derivatization of Ci (i = 1, 2, 3, and 4) (Figure 3B,C) proved that Dj (j = 1, 2, 3, and 4) and Ek (k = 1, 2, 3, and 4) were simply methylated and ethylated forms of the Ci (i = 1, 2, 3, and 4) carboxylic acids (Figure 3B,C), respectively. NMR data collected for Dj (j = 1, 2, 3, and 4) comforted these conclusions (see detailed NMR interpretation in the Supporting Information. Discovery and Significance of New TPs of Chlordecone. Investigation of the diversity of TPs formed during chemical reductive degradation of chlordecone demonstrates that the 14 new chlorinated compounds previously identified during microbial experiments (B0 and B5; Ci, i = 1, 2, 3, and 4; Dj, j = 1, 2, 3, and 4; Ek, k = 1, 2, 3, and 4) clearly represent TPs of chlordecone. NMR analyses combined with chemical derivatization allowed us to structurally elucidate these TPs and the most significant TPs from families A and B (Figures S4). Among them, the previously described “unknown nonpolar and polar products” reported by Jablonski et al. in 199639 were assigned here according to thin-layer chromatography analysis to TPs from families B and C, respectively (Figure S1). The present work thus significantly expands the list of fully characterized chlordecone TPs, which hitherto only comprised chlordecol, 8-monohydrochlordecone, and 2,8-dihydrochlordecone,38 and also corrects previous findings.40−42 In total, four of the five TP families (B, C, D, and E) exhibit a rare indene aromatic bicycle resulting from the opening of the chlordecone bishomocubane structure. Formation of the indene ring correlated with the presence of vitamin B12 in chemical degradation experiments. It is noteworthy that the only two isolated bacteria capable to transform chlordecone into such indene-based TPs, i.e., Citrobacter sp. 86 and Citrobacter sp. 92, encode the full anaerobic corrinoid biosynthetic pathway.36 Indeed, corrinoids including vitamin B12 are known to act as cofactors for reductive dehalogenases.48−50 In their free form, they could also mediate reductive dehalogenation under both biotic and abiotic conditions.41,51,52 However, similarities between the chemical and microbial ring-opening pathways could not be confirmed based on carbon isotope fractionation.37 This may be explained by the conditions used in chemical degradation experiments, which are not relevant to those prevailing in microbial experiments. Natural Chlordecone Degradation on Martinique Island. Here, we assessed the presence of hydrochlordecones

either D or E family compounds. Purification of D1-D2 and D3D4 isomeric pairs was achieved in 5% and 3% yield, respectively. Complete Elucidation of A1 Structure. Monohydrochlordecone A1, also formed after in situ chemical reduction daramend treatment of a FWI field,46 was previously identified as either 9- or 10-monohydrochlordecone, based on GC-MS and LC-MS fragmentations.40 Replacement of a chlorine atom by a hydrogen atom in chlordecone led to four different regioisomers, i.e., 6-, 8-, 9-, or 10-monohydrochlordecone (Figure S5). Here, we unequivocally identified TP A1 as 10monohydrochlordecone. Indeed, the six distinct signals detected in its 13C NMR spectrum require a conserved vertical plane of symmetry which is only compatible with this specific regioisomer. Complete Elucidation of B1 Structure. In our previous study, chemical derivatization allowed us to identify an indene aromatic ring in TP B1.37 Here, we took advantage of the purified B1 to elucidate its structure using NMR. 1H and COSY spectra (Figures S32 and S34) indicated an allylic domain at δ 3.7 ppm (d, J = 1.4 Hz, 2 H), coupled with an aromatic proton at δ 6.87 ppm (t, J = 1.4 Hz, 1 H). Neither the 1.4-Hz value, compatible with a coupling constant of 3J- and 4J-type in the case of indene (Figure S29), nor additional 1D- and 2D- NMR experiments (Figures S32−S36) allowed us to unequivocally assign the position of the aromatic proton. Then, a 13C−13C COSY experiment on 13C-enriched B1 synthesized from commercially available 13C8-chlordecone (Figure S38) allowed us to link every carbon atom to its direct carbon neighbors and thereby to unequivocally assign B1 to 2,4,5,6,7-pentachloro-1Hindene by way of a HSQC experiment (Figure S35). Complete Elucidation of C1−C2 Structures. Hydrogenation of the isolated C1−C2 mixture yielded, among other products, commercially available 4-carboxyindane (Figure 3A and Figure S8). Taking into consideration the known indene isomerization equilibrium,47 we assigned C1 and C2 as tetrachloroindene-4-carboxylic acid and tetrachloroindene-7carboxylic acid, respectively. Comparison of 1H and 13C spectra highlighted the similarity between TP B1 and C1−C2 (Figure 4A and Figures S32 and S49), and all protons of C1−C2 were unequivocally positioned on the indene five-membered ring. Interestingly, the HMBC NMR experiment of C1−C2 showed only one cross-correlation peak between the most deshielded allylic protons and one of the two carboxylate carbon atoms (Figure S53). We thus assigned this set of 1H and 13C signals to 2,4,5,6-tetrachloro-1H-indene-7-carboxylic acid, i.e., C2, and the other set to 2,5,6,7-tetrachloro-1H-indene-4-carboxylic acid, i.e., C1. Complete Elucidation of B3−B4 Structures. Tetrachloroindene B3 appeared as a single compound in GC-MS, but its 1 H NMR spectrum showed two similar but distinct sets of signals (Figure 4C). This suggested two regioisomers, which were arbitrarily called B3 and B4. By analogy with 2,4,5,6,7pentachloroindene B1, the methylene groups and the lessdeshielded aromatic protons of B3−B4 were placed on carbons 1 and 3 (Figure 4A,C). Additional coupling of 0.6−0.7 Hz, observed only for B3 between protons from carbon 1 and the most deshielded aromatic proton, was compatible with either a 2,4,5,7- or a 2,4,5,6-tetrachloro substitution pattern. Of note, indene isomerization formally transferred substituents from carbon 4 to carbon 7 and from carbon 5 to carbon 6. HMBC and HSQC experiments indicated that both B3 and B4 featured a chlorine substituent at carbon position 4 (Figures S47−S48). 6139


Article


Figure 5. Location of soil, water, and sediment samples in Martinique and chlordecone and TP concentrations.

soils, depending on both TP and matrix, as shown for chlordecone and metabolite B1 in andosol and nitisol soils (Figure S10). Thus, we provide only uncorrected concentration ranges here. Concentrations of chlordecone in the Galion River were in the range of previous studies (0.1−2 μg/L).29 Of note, the concentration of TP B1 was in the same order of magnitude as that of chlordecone itself (0.1−2 μg/L) in both the Galion River and nearby mangroves (Figure 5 and Table S6). Moreover, we detected TP B1 in bed sediments, in which chlordecone was not found. Chlordecone concentrations in soils were approximately 10-fold lower than previously reported typical values of dried samples (0.01−1 mg/kg compared to 0.8−5 mg/kg (dry weight46,54). Chlordecone TPs were easily found in solid matrices, with B1 and chlordecol detected in all soil samples, except in chlordecone-free nitisol 926. Concentration of TP B1 varied between 0.05 and 5 mg/kg. In contrast, chlordecol concentration was systematically around 0.05 mg/kg, in agreement with previous studies.44 We only observed significant levels of TP A1 in the two andosol soils (0.05−1 mg/kg). TPs C3−C4, less chlorinated than C1−C2, were more frequently detected, and at higher concentrations, in one nitisol soil (above 1 mg/kg). The low concentrations of 8monohydrochlordecone detected here (

Natural Chlordecone Degradation Revealed by Numerous

Recommend Documents