LIBRARY Michigan State University This is to certify that the dissertation entitled INVESTIGATION OF THE CHEMISTRY AND TOXICOLOGY OF CHLORINATED BORHANE (TOXAPHENE) RESIDUES ISOLATED FROM GREAT LAKES LAKE TROUT (Salvelinus namaycush) presented by Jay William Gooch has been accepted towards fulfillment of the requirements for Ph.D. degnwin Environmental Toxicology/ Fisheries and Wildlife flag/am Major professor Date JEN—193.6. Ilf'll-_ ALA. A A 1:. llL I: n L .1 0‘127’1 ill/WI Ill/ill fill/ll/l/l/Wl K 3 1293 00852 2207 IVISSI_J RETURNING MATERIALS: Place in book drop to LIBRARJES remove this checkout from ._:—. your record. FINES will be charged if book is returned after the date stamped below. igtfigxn INVESTIGATION OF THE CHEMISTRY AND TOXICOLOGY OF CHLORINATED BORNANE (TOXAPHENE) RESIDUES ISOLATED FROM GREAT LAKES LAKE TROUT (Sal vel inus namaycusiy By Jay Hilliam Gooch A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Center for Environmental Tbxicology/ Department of Fisheries and wildlife 1986 ABSTRACT INVESTIGATION OF THE CHEMISTRY AND TOXICOLOGY OF CHLORINATED BORNANE (TOXAPHENE) RESIDUES ISOLATED FRON GREAT LAKES LAKE TROUT (Sal vel inus namaycush) By Jay Hilliam Gooch Recent studies have demonstrated the presence of a complex pattern of chlorinated bornane residues . similar to the insecticide toxaphene, in Great Lakes fish. These residues, though structurally similar to toxaphene, are composed of a different mixture of compounds than technical toxaphene. This study provides concrete evidence for the presence of the major toxic components of toxaphene and toxicity associated with chlorinated bornane residues isolated from Great Lakes lake trout (Salvelinus namaycush). Chlorinated bornanes were isolated from tissues of Lake Michigan and Siskiwit Lake (Isle Royale) lake trout and purified using thin- layer. column and gas-liquid chromatography. The two major toxic constituents of toxaphene, Toxicants A and B, were detected and quantified as a nmre toxicologically appropriate reference point for residue levels. Capillary gas chromatography/mass spectrometry was used to confirm the identity of the toxic congeners and to demonstrate Jay William Gooch the structural similarities of the other components with components in toxaphene. The results show that chlorinated bornane compounds in lake trout from Lake Michigan and Siskiwit Lake on Isle Royale include the two major toxic components of toxaphene at concentrations l0-20 fold less than the estimated total toxaphene. Twenty-four hour. static, acute, bioassays were conducted using Aedes egypti mosquito larvae and demonstrated that the residues purified from lake trout are as toxic as technical toxaphene isolated irIa similar'manner. IUiaddition, experiments performed on the GABA (a-aminobutyric acidl-chloride ionophore complex binding site of the central nervous system, show that the toxaphene residues from Great Lakes fish are as potent as technical toxaphene at displacing 35S-t- butylbicyclophosphorothionate from its macromolecular target site in rat and lake trout brain. - Studies conducted on residues from different tissues and different collection years show no trend toward decreasing toxicity of the toxaphene residue with tissue or time. Pattern recognition techniques did not demonstrate any identifiable change in the composition of the residue over the past three years. Capillary GC-MS of purified residues revealed a relatively pure toxaphene residue composed of a greater degree of structures with nine chlorines than anticipated previously. ACKNOHLEDGEHENTS I would like thank a number of people who have been important to my efforts on this project. Thank you Jim Keller for supplying me with mosquitoes on many occasions and Brian Musselman for providing access to, and help with, a very busy mass-spectrometer. To Ms. Karen Obermeyer for invaluable assistance in the laboratory over the past two years - Thanks and Good Luck in Colorado. To Cheryl Burke, Alice Ellis, and Carol Fishery whose clerical support and social contact have been an important part of my day for the past several years - I'll miss you all. Thank you to Dr. Richard Leavitt for tolerance of my unscheduled, unannounced visits to his IBM-PCPS and to Mrs. Jackie Schartzer for preparation of this document in a congenial and professional nmnner. I want to express sincere gratitude to Dr. John Giesy for both the many social and scientific interactions we have shared over the past several years. Through both himself and the people in his laboratory, I have benefited in ways that will continue to grow for many years. I look forward to a future as peers and friends. The space here is totally inadequate to express the importance of Dr. Matthew Zabik in my graduate career. The things I have experienced through him will remain highlights in my life for many years. I will always be grateful. To attempt to encompass all of the impacts that Dr. Fumio Matsumura has had on my life would be impossible. He has been instrumental in shaping my knowledge, my ii experience. and my scientific identity. The scope of experiences I have had, both scientifically and socially; I am sure cannot be rivaled anywhere else. He has provided an atmosphere of encouragement, freedom, stimulation, and scientific inquiry that has been exciting. I will always be indebted to him for his support and training. I will always identify with my days in his laboratory. The friends I have developed in his laboratory will always be important to me. A special thanks to Dr. Niles Kevern, who found me orphaned into his obligations and has allowed me to work and grow unencumbered for the past several years. I have been in an in-between position departmentallyrand my ties to fisheries science are very important to me. Lastly, to my wife Cindi, who has had to put up with all of the emotional ups and downs that are part of the graduate student package - I Love You. This publication is a result of research funded by the Michigan Sea Grant College Program, project number R/TS- 24 with grant NABSAA-D—SGO45 from the National Sea Grant College Program, National Oceanic and .Atmospheric Administration (NOAA), U.S. Department of Commerce, and funds from the State of Michigan. TABLE OF CONTENTS PAGE LIST OF TABLES ..................... . . . . . vi LIST OF FIGURES. ......................... viii CHAPTER 1 - INTRODUCTION Introduction ......................... 2 Synthesis and Structural Characterization ........... 3 Physical Properties ...................... 4 Toxicity - General ...................... l2 Toxicity - Structure Activity ................. l2 Toxicity - Mechanistic . . .................. l4 Toxicity - Aquatic . . .................... l5 Environmental Dynamics ........... . ........ l7 Agricultural ........................ l7 Soils ........................... l7 Aquatic Systems ...................... l9 Global Residues ....................... 20 Chapter 2 - EVALUATION OF THE TOXIC COMPONENTS OF TOXAPHENE IN LAKE MICHIGAN LAKE TROUT Introduction ......................... 24 Experimental Section ..................... 26 Materials . . . . . .................... 26 MethOOs O O O O O O O O O O O ...... O O O ...... 26 TLC Purification of Toxicant A ............... 27 Gas Chromatography and Quantitation of Residues ...... 28 Results and Discussion ............ . ...... 30 CHAPTER 3 - TOXICITY 0F CHLORINATED BORNANE (TOXAPHENE) RESIDUES ISOLATED FROM GREAT LAKES LAKE TROUT (Salvelinus namaycush) Introduction ......................... 45 1. Acute Toxicity and Mammalian Neuroreceptor Binding. . . . . . 45 Materials and Methods . . . . ...... . . . . . . . . . 46 Extraction and Purification of Residues . . . . . . . . . . 46 Aggte Toxicity Tests .................... 50 I S] t-Butylbicyclophosphorothionate (TBPS) binding. . . . 50 iv TABLE OF CONTENTS CONTINUED: PAGE Results and Discussion .................... 52 ggxicity . . . . . .......... . ........ . 60 S'TBPS BIOdTng o o o o e o e o o e o o o o o o o o m o o 62 II. Inhibition of 35s-TBPs Binding to Lake Trout Brain Synaptic Membranes . . . . . . . . . . . . . . . . . . . . . . 68 Materials and Methods ................... 68 Results and Discussion ....... . . . . ...... . . 68 CHAPTER 4 - SUMMARY OF RESIDUE LEVELS AND STATISTICAL PATTERNS ASSOCIATED WITH LAKE MICHIGAN LAKE TROUT RESIDUES FROM 1982-1985 Introduction . . . ....................... . 76 Quantitative Results ....................... 77 Pattern Recognition. . . .................... . 79 Background ................... . ...... . 79 Data from Fish Analyses . . . ................. 90 Large Data Set. . . . ..... . ......... . . . . . . 97 CHAPTER 5 - CAPILLARY GAS CHROMATOGRAPHY - MASS SPECTROMETRY OF PURIFIED TOXAPHENE RESIDUES. Introduction . . . . . . ..................... 106 Materials and Methods. ...................... 107 Results and Discussion ...................... l08 CHAPTER 6 - SUMMARY ........................ l20 LIST OF REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . I30 APPENDICES Appendix A - Toxaphene Residue Concentrations in Individual Fish Analyzed from l982-l985. . . . . . . . . . . . . l40 Appendix B - Uptake and Metabolism Experiments .......... l46 TABLE 10 LIST OF TABLES PAGE Physical properties of toxaphene.l1) .......... ll Levels of the two primary toxic constituents and estimated total toxaphene found in lake trout from Lake Michigan. . . . . . . . . . . . . . . . . . . 42 Percent composition estimates for toxicants A and B in residues and standards used in this study. Data are expressed as mass 1 (area 1) where area % equals the integrated GC peak area of A or B/total peak area used in quantitation. . . . . . . . . . . . . 58 Acute toxicity of toxaphene residues purified from Lake Michigan lake trout to mosquito larvae (Aedes egypti)z Data are expressed as 24 hour . . LCSO in ug/l (X 1’sd (n)) . . . . . . . . . . . . . . . 6l Effect of toxaphene residues from Lake Michigan lake trout on 35S-TBPS binding to rat brain synaptic membranes.la) . . . . . . . . . . . . . . . . . . . . . 66 Inhibition of 355-TBPS binding to lake trout brain synaptosomes by residues isolated from Great Lakes lake trout. Data are expressed as % inhibition (2 i s.d. (n)). . . . . . . . . . . . . . . . ..... 69 Summary statistics for toxaphene residues in Lake Michigan lake trout. All data are expressed as ug/g wet weight. x :_s.d. (n) and range. . . ..... 78 Standard toxaphene data used for pattern recognition analysis. . . . . . . . . . . . . . . . . . 84 Clusters produced by disjoint clustering and canonical discriminant analysis of 20 different standards . . . . . . ............ 93 Large data set used for overall pattern recognition analysis of toxaphene residues. . . . . . . 98 vi LIST OF TABLES CONTINUED: TABLE PAGE ll Relative composition of standard toxaphene and purified lake trout toxaphene residue derived from ion chromatograms of the three major isotope clusters. The monitored ion represents the most intense ion in the spectrum from compounds with the indicated molecular formula . . . . . . . . . . . ........ ll4 A-l Summary of toxaphene concentrations in Lake Michigan lake trout collected in l982 (ug/g, wet weight). . . ..... . . . . ....... l4l A-2 Summary of toxaphene concentrations in Lake Michigan lake trout collected in l983 (ug/g, wet weight). . . . . . . . . . . ........ l42 A-3 Summary of toxaphene concentrations in Lake Michigan lake trout collected in l984 lug/g. wet weight). . . . . . . . . . . . . . . . . . .l43 A-4 Summary of toxaphene concentrations in Lake Michigan lake trout collected in l985 (pg/g, wet weight). . . . . . . . ........... l44 A-S Summary of toxaphene concentrations in Siskiwit Lake (Isle Royale) lake trout collected in l984 (pg/g, wet weight) .......... l45 lB Concentrations of 14C-toxaphene in different compartments after exposure in a model uptake system. . . . . . . ....... . .l53 FIGURE Figure Figure Figure Figure Figure Figure Figure LIST OF FIGURES PAGE Structures of toxaphene components and metabolites that have been identified . . . . . . . . 5 Capillary column GC-ECD traces of toxaphene (above) and the toxaphene fraction of an extract from a lake trout from Lake Michigan (below). Conditions are reported under Materials and Methods. Locations of toxicants A and B are indicated . . . . . . . . . . . 3l Capillary column GC-ECD traces of the toxaphene fraction of an extract of lake trout from Lake Michigan prior to (above) and after (below) TLC purification for toxicant A. Locations of toxicants A and B are indicated. . . . . 35 Electron impact mass spectra of toxicant A (below) and of toxicant A (above) co-eluting with p,p' DDT from a lake trout residue during a capillary GC-MS run. The spectrum of toxicant A (below) was acquired during a capillary GC-MS run of technical toxaphene ...... 37 Mass spectra of toxicant B eluting from a technical standard (below) and of toxicant B co-eluting with p,p' DDD in a lake trout residue (above) . . . . . . . . ........... 4O Capillary gas chromatogram of technical toxaphene (above) and purified lake trout residue toxaphene (below). Peaks designated (0) have been used for quantitative and comparative purposes. For conditions, see Materials and Methods . . . . . . . . . . . . . . . . 53 Normalized histogram of the peaks used for quantitation of the residues and standards compared in this study. For a description of abbreviations, see Table 3. Peak numbers can be established from designated (Q) peaks in Figure 6 by moving from left to right in sequence. . . . . . . . . . . . . . . . . 55 viii LIST OF FIGURES CONTINUED: FIGURE Figure 8 Figure 9 Figure Figure Figure Figure Figure 10 ll. l2. l3. l4. Inhibition curve of toxaphene (El) and the 19848 (A) lake trout residue for [3551-7395 binding to rat brain synaptic membranes. IC-SO values were 5.9 :_3.8 x lO'8 M for toxaphene and 2.9 + 2.1 x lO'8 M for the lake trout residue Tn = 3 experiments) . . . . . . . . . . . . . Concentration - response curve for toxaphene and chlordane inhibition of 35S-TBPS binding to lake trout brain synaptic membranes ........ Plot of the lst two principal components from analysis of the group of 26 standards outlined in Table 8. These two components explained 57% of the variance in the data . . . . . . . . . . . Canonical variate plot of the disjoint clusters produced from the analysis of technical standards. Disjoint clusters were produced from the propor- tional peak area data outlined in Table 8 . . . . . . Canonical variate plot of disjoint clusters produced from analysis of standards using the first six principal components as input into the clustering procedure. The first six principal components explained approximately 90% of the variability in the original data and are by definition jointly uncorrelated. . . . . . . . . . Principal components plot from analysis of toxaphene residues from belly flap of Lake Michigan lake trout collected in l983, l984, and 1985. Technical standards were also included in the analysis. Seventy-six percent of the variability is contained in the first two components. . . . . . . . . . . . . . . . . . . . Principal components plot of 101 observations from various years, tissues, and stages of purification. For an explanation of sample types see Table 10. 0- Belly; @— Fillet; 0- Eggs; U- Charcoal column; a)! - Pooled sample; 0 - Nitrated final purification;-+ - Lake Siskiwit; A- Technical Standard; X- Procedural spike . . . . ix PAGE 64 71 86 88 91 95 99 LIST OF FIGURES CONTINUED: FIGURE PAGE Figure 15. Canonical variate plot from a cluster and discriminant analysis of the first six principal components obtained from the 101 observations in Figure 14. All symbols are the same as Figure 14 . . . . . . . ......... 101 Figure 16. Chromatograms of total ion current for standard toxaphene (bottom) and a purified lake trout residue from 1985 (top) ............... 108 Figure 17. Ion current chromatograms from a toxaphene standard. TIC = total, RIC = 343, 377, and 413 for masses 343, 377, and 413 respectively. These masses correspond to the major ion in the chlorine isotope cluster from [M-Cll' fragments from structures with 7, 8, and 9 chlorines respectively. . . . . . . . . . . . . . . .110 Figure 18. Ion current chromatograms from a purified 1985 Lake Michigan lake trout residue. Profile descriptions are described in Figure 17 ....... 112 Figure 19. Electron (top) and negative chemical ionization (bottom) total ion chromatograms for technical toxaphene. Note the increased sensitivity for later eluting peaks ......... 117 Figure 1B Model uptake system used with juvenile lake trout ........................ 149 Figure 2B Toxaphene residue in juvenile lake trout after exposure in a model uptake system. a) toxaphene standard; b) sand spiked at 100 ppb-exposure 10 days; c) 100 ppb - 15 days; d) 200 ppb - 10 days; and e) control . . . .151 Figure 38 Extracts from in vitro incubation of toxaphene with lake trout liver post mitochordrial super- natant. a) + NADPH, 24°C; b) + NADPH, 4°C; c) - NADPH and d) boiled enzyme .......... 155 CHAPTER 1 INTRODUCTION INTRODUCTION Toxaphene is an insecticide which is a complex mixture of compounds made by chlorinating camphene to a 67-69% chlorine content. It has an average molecular formula of C10H10C13 with a corresponding molecular weight of approximately 414 (410 nominal). Technical formulations are an amber-colored waxy solid with a characteristic terpene odor (Hercules Chemical Co. Bulletin AP-103B). A similar product, strobane T, is synthesized from a-pinene and was manufactured mainly in Europe (Saleh and Casida, 1977). Originally developed by the Hercules Chemical Company, toxaphene was used from 1947 until 1982 before being banned by the LLS. Environmental Protection Agency fbr use in the United States. Hercules held the patent until 1971. By 1974, toxaphene was being manufactured by 186 companies producing 817 different registered products (Toxaphene Working Group. 1977). While production and use estimates are difficult to determine. it is clear that toxaphene was the most heavily applied chlorinated hydrocarbon insecticide ever used in the United States. Between 1964 and 1976, usage is estimated to have been 36 million pounds per year. In 1976, the USDA estimated that approximately 31 million pounds were applied to . approximately 4.9 million acres of cropland and pasture. Cotton received approximately 85% of the total. In addition, approximately 2.4 million pounds were applied to livestock, particularly cattle. At that time, toxaphene had 277 commodity and other site registrations. State and Federal agencies were recommending use of toxaphene for 2 control of 167 insect pests on 44 commodities. 40 of which had no alternative control strategy. All of the above statistics have been taken from a report submitted to the Environmental Protection Agency by a Toxaphene Assessment Team in response to the May 25, 1977 Federal Register notice of rebuttable presumption against registration (RPAR) issued by the EPA (USDA. 1978). While toxaphene was being reviewed. synthetic pyrethroids were registered and their use quickly began to replace toxaphene. In October of 1982, toxaphene was banned for most uses in the UJL Existing stocks can be used until the end of 1986. though only under limited circumstances. In the cancellation of toxaphene, the EPA relied heavily on hazard to, and persistence in, aquatic organisms (Federal Register, 1982). In hindsight. given the structural and physical properties of toxaphene and the immense quantities that were applied, it is not surprising that toxaphene residues have become problematic. A lack of adequate analytical techniques to deal with toxaphene residues and a gap in our knowledge about toxaphene pharmaco- dynamics in fish, no doubt contributed to our lack of ability to predict this problem. Synthesis and Structural Characterization Toxaphene is readily produced by passing chlorine gas through a UV-irradiated carbon tetrachloride solution of camphene until a chlorine content of 67-69% is obtained (Buntin. 1951). This process yields a complex mixture of nearly 200 different constituents which have never been completely characterized (Casida and Saleh. 1978). Saleh and Casida (1977) analyzed toxaphene lots from various sources around the world and demonstrated that the composition of'toxaphene from different manufacturers varied substantially in its capillary column gas chromatography profile. Batches from different years from Hercules. the major manufacturer in the LLSM. however, were very consistent. Despite the variable capillary-GC profiles, there was very little difference in toxicity to either mammals or insects. Four main research groups have taken part in the identification of different toxaphene components. The 17 structures that have been identified are shown in Figure 1. In addition to these structures. Holmstead et a1. (1974) and Saleh (1983) used gas chromatography/mass spectrometry to characterize nearly 200 components resolvable by both thin layer chromatographic pre-separation and capillary column GC. Saleh (1983) gives composition estimates of 75% isomeric polychlorobornanes. 18% polychlorobornenes, 2% polychlorobornadienes, 1% other chlorinated hydrocarbons and 3% nonchlorinated compounds. These estimates were derived from both electron impact and negative chemical ionization mass spectrometry. Nearly all studies to date have been conducted with technical or synthesized material; there has been few attempts to characterize the structures found in environmental samples. Physical Properties Table 1 is a list of the physical properties that have been reported for toxaphene. Because toxaphene is a complex mixture, each component will have its own unique physical properties and therefore, it is very difficult to arrive at good estimates of overall parameters. Different combinations of numbers have been used by several 5 Figure 1: Structures of toxaphene components and metabolites that have been identified. Structure Reference Casida et a1. 1974 2,2,5-8"d°,6-8X0,8,9,Io heptachlorobornane (toxicant B. I) Turner et al. 1975 Matsumura et a1. 1975 2,2.5‘endo.6-exo,8.8’9’]o octachlorobornane (toxicant Ac, B-chloro-I) Matsumura et a1. 1975 Turner et a1. 1975 2,2,5-endo.6-exo.8.9,9.10 octachlorobornane (toxicant AB, 9-chloro-I) Figure 1 continued: Structure Reference Anagnostopoulos et a1. 1974 2-exo,3-exo,5,5,6-endo 8,9,10,10 nonachlorobornane (toxicant C) Chandurkar et al. 1978 Z-exo,3-endo.5-exo,6-endo. 8,8,9,10,10-nonachlorobornane (toxicant Ac) Anagnostopoulos et a1. 1974 2-endo,3-endo.5.5.6-endo 8,8,9,10,10 decachlorobornane Figure 1 continued: Structure Reference Black 1974 Z-exo.3-exo.5.5.8.9.10.10 octachlorobornane Black 1974 2-exo.3-endo.6-exo 8.9.10 hexachlorobornane Landrum et a1. 1976 2.5.6-exo 8.8.9.10 heptachloro dihydrocamphene Figure 1 continued: Figure 2.2.5.5.6-ex08.9.10 octachlorobornane (S-exo-chloro-I) 2-exo.3-exo.5.5.6-endo 9.10.10 octachlorobornane C1 1 2.2.3-exo.S-endo.6-exo.8.9.10 octachlorobornane (3-exo-chloro-I) Reference Turner et a1. 1977 [from chlorination of toxicant B (1)] Anagnostopoulos et al. 1974 Turner et a1. 1977 Figure 1 continued: Structure Reference Turner et a1. 1977 [from chlorination of toxicant B (1)] 2.2.5-endo.6-exo 8.9.9.10.10-nonachlorobornane (8.10-dichloro-I) Turner et a1. 1977 [from chlorination of toxicant B (1)] 2.2.3-exo.5-endo 6-exo.8.9.10.10 nonachlorobornane Saleh and Casida 1978 (metabolite) 2-endo.S-endo.6-exo 8.9.10 hexachlorobornane 10 Figure 1 continued: Structure Reference Saleh and Casida 1978 (metabolite) 2-endo.5-endo.6-exo 8.9.10 hexachlorobornane ‘ Saleh and Casida 1978 ‘l (metabolite) 2-exo.5-endo.6-exo 8.9.10 hexachlorobornane 7 other di. tri and tetachlorobornanes were synthesized and identified by Parlar et a1. (1977). These have not been drawn here because they are anticipated to make up very little of the technical products. Most of these structures and more detailed descriptions of identification procedures are contained in Korte et a1. (1979). Casida and Saleh (1978) and Parlar and Michna (1983). Table 1. Physical properties of toxaphene.(1) Molecular 414 (CloHloClg) Weight Melting range 70-95°C(3) oint Boilin decomposes > 120°C oint l Vapor 0.2-0.4mm Hg(25°C) Pressure “"““ 1 x 10‘5mm Hg(25°C) 3 x 10'7mm Hg(20°C)(4) 0.3mm Hg(25°C) Agueous 3.0 mg/l Solubility 0.5 mg/l(5) 0.04 mg/l 0.74 mg/l 0.40 mg/l Octanol/Water 3.3 I 2.5 X 102 L— Partition Writ 8.3 x 10?- 6.4 x 104 2.7 x 106 2.0 x 103 52:2:2022 (2) Brooks 1974 Brooks 1974 Brooks 1974; Hercules AP-103B Korte et a1. 1979 Bidleman and Christenson 1979 Reinert et a1. 1982 Brooks 1974; Hercules AP-IOBB Paris et a1. 1977 Reinert et a1. 1982 EPA 1979 Sanborn et a1. 1976 Paris et a1. 1977 Sanborn et a1. 1975(6) Reinert et a1. 1982 Rice and Evans 1984 EPA 1982 (1) Often values are listed in the references with no Obvious source. (2) Most authors agree on the average molecular weight for the mixture. (3) Technical toxaphene is an amber waxy semi-solid at room temperature thus melting point is probably very subjective. (4) Value reported for Strobane T but listed under toxaphene and used in subsequent references by this author. (5) The only actual value that was measured by the authors. (6) Measured with C-14 toxaphene. This is probably an overestimate due to a greater abundance of polar constituents compared to unlabelled material (Saleh and Casida. 1977). 12 investigators in order to try to understand the environmental behavior of toxaphene. In light of the variability in the numbers and the complexity of the mixture. this approach to understanding the environmental fate of toxaphene is inadequate. Tbxicity-General Toxaphene is classified as a general convulsant which acts on the central nervous system. It causes stimulation of the cerebrospinal axis leading to chronic convulsions and tetanic muscular contraction. Death is usually'caused by respiratory failure (Hercules Technical Bulletin T-lOSBL. Barbiturates such as phenobarital and pentobarbital are generally effective in controlling convulsions (Stormont and Conley. 1952L The acute oral toxicity of toxaphene to various mammals ranges from 40 mg/kg (cat) to 270 mg/kg (Guinea pig) and thus would be classified as moderately toxic. Toxicity - Structure Activity As toxaphene is indeed a very complex mixture. attempts have been made to isolate and identify structures of some of the major toxic components. This process was dictated by following the toxicity of various chromatographic fractions. This method allowed investigators to dwell only on fractions that seemed to be responsible for the toxicity of the insecticide. The first toxic component identified was 2.2.Sgggggg 6:259. 8.9.10 heptachlorobornane (Casida et'al.. 1974). This compound would later become known as toxicant B (See Figure l for this compound and others to be discussed). Another major constituent was identified in the same 13 year and identified as 2-gxg.3-gxg.5.5.6ggggg.8.9.l0.10 nonachlorobornane (Anagnastopoulos et al.. 1974). This component is now known as toxicant C. The third major component identified. toxicant A. was found to consist of two components; 2.2.5-eggg.6- 352.8.8.9.10 octachlorobornane and 2.2.5-gggg.6-gxg.8.9.9.l0 octachlorobornane (Nelson and Matsumura. 1975. Turner et al.. 1975). There are a number of other minor components that have been identified including a chlorinated dihydrocamphene (Landrum et al.. 1976) and chlorinated bornanes with fewer than 7 or more than 9 chlorines (Parlar et al.. 1977 and Turner et al.. 1977). In terms of toxicity. toxicant A has been shown by a number of investigators to be the most potent of the three major toxic fractions to mice. goldfish. houseflies. (Turner et al.. 1977. Saleh et al.. 1977) brine shrimp. and mosquito larvae (Nelson and Matsumura. 1975L Toxicant B is the next most toxic constituent. For structural derivatives of toxicant B. Saleh et a1. (1977) lists the toxicity as: 9-choro > 8-chloro > B > 3-exo-chloro > 5-exo-chloro > lO-chloro. Hexachlorobornane and hexachlorobornane metabolites of toxicant B were less toxic than most of the other structures. For many of the compounds tested. pretreatment of mice and houseflies with the mixed function oxidase inhibitor. piperonyl butoxide. increased toxicity 2 to 8-fold. indicating that metabolism plays a significant role in the pharmacologic action of many of the toxaphene components. Isolation and evaluation of all of the components of toxaphene would be virtually impossible. Together. toxicants A and B comprise approximately 8.2% of the technical mixture (Gooch and Matsumura. 1985). 14 Toxicity-Mechanistic Though the neurotoxic action of toxaphene is well known. the precise molecular mechanism for this effect has not been known. Recently. a number of investigators have found that toxaphene and cyclodiene type insecticides interact very potently at a binding site associated with the GABA-chloride ionophore complex (Ghiasuddin and Matsumura. 1982; Matsumura and Ghiasuddin. 1983; Lawrence and Casida. 1984). GABA (gamma-aminobutyric acid) is the major inhibitory neurotransmitter of the central nervous system of nearly all vertebrates and functions by stimulating Cl' flux across nerve membranes. The resultant decrease in membrane resistance inhibits the generation of excitatory post synaptic potentials (EPSPs) (Lund. 1985). Cyclodiene type insecticides have been shown to inhibit Cl‘ f1dx by binding with a receptor (the picrotoxinin receptor) proposed to be located at. 1"“ very near. the chloride channel (Matsumura and Ghiasuddin. 1983; Lawrence and Casida, 1984; Abalis et al.. 1985). By interfering with the ability of the inhibitory neurotransmitter system to regulate nervous impulses. these insectides produce excessive stimulatory activity. Two major ligands. 3H-picrotoxinin and 35S-t-butylbicyclophos- phorothionate (TBPS). have been shown to bind to a site at or near the chloride channel (Ticku and Olsen. '1979; Squires et al.. 1983; Ramanjaneyulu and Ticku. 1984). The ability of a compound to inhibit the binding of these ligands to the receptor site is highly correlated with toxicologic potency (Lawrence and Casida. 1984; Matsumura and Tanaka. 1984). This type of activity will be discussed in greater 15 detail in a later section. In addition to its acute effects. toxaphene. or some fraction thereof. has been shown to be mutagenic (Hooper et al.. 1979) and mildy carcinogenic in rats and mice (Reuber. 1979). It induces rat hepatic microsomal enzymes (Pollock et al.. 1983). alters hepatobiliary excretion (Mehendale. 1978). and affects learning behavior in rats (Olson et al.. 1980). Histopathologic lesions have been reported for the thyroid. adrenals. pituitary and manrnany glands (Reuber. 1979). Toxicity - Aquatic Toxaphene is highly toxic to nearly all aquatic organisms that have been tested. Fish are very sensitive to toxaphene. Indeed. toxaphene was once tested experimentally as a piscicide in rough fish control programs (Johnson et al.. 1966). Hughes (1970) wrote an excellent review of the literature on this topic. Henderson et a1. (1959) lists 48-hour LC-50's as: 7.5 ug/l for fathead minnows. 3.3 ug/l for bluegills. 6.8 ug/l for goldfish. and 24 ug/l for guppies. Katz (1961) lists 3.3 ug/l for juvenile chinook salmon. 10.5 ug/l for juvenile coho salmon and 8.4 ug/l for rainbow trout. Johnson and Finley (1980) give 96 hour LC-50's ranging from 2 ug/l to 18 ug/l for 12 major fish species (4 salmonids. 3 cyprinids. 2 ictalurids. 3 centrarchids. and l percid). In addition. Johnson and Finley (1980) noted that swim-up fry were the most sensitive and thatifliand water hardness did not affect the toxicity. Macek et a1. (1969) noted a decrease in tolerance to toxaphene with a decrease in temperature. Invertebrates are less sensitive than fish to the acute effects of toxaphene. though not immensely. Forty-eight hour LC-SO values have 16 been reported as: 15 ug/l for Daphnia pulex and 19 ug/l for Simocephalus serrulatus (Sanders and Cope. 1966). Johnson and Finley (1980) list 96 hour LC-SO values ranging from 1.3 - 4O ug/l for nine common invertebrate species. Nearly all of the chronic toxicity data available on toxaphene are from studies by the United States Fish and Wildlife Service. Fish- Pesticide Research Lab in Columbia. MO. On the basis of growth. reproduction. mortality and bone development. maximum acceptable toxicant concentrations (MATC's) were determined to be < 39 ng/l for brook trout. between 25 and 54 ng/l for fathead minnow and between 49 and 72 ng/l for channel catfish (Mehrle and Mayer. 1975a. 1975b). Collagen deposition and bone development were the most sensitive parameters and differences in brook trout fry were evident after only 7 days exposure. Later studies (Mayer et al.. 1978) demonstrated a relationship between vitamin C levels (depletion) and the bone anomaliesm High levels of vitanfiriC in the diet reduced whole-body residues of toxaphene and significantly increased the tolerance of fish to the chronic bone effects. The authors proposed a relationship between induction of the hepatic monooxygenase system by toxaphene and a functional vitamin C depletion due to excessive demand from the induction of the biotransformation enzymes. Bioconcentration factors for various life stages in these studies ranged from 5.000 to 76.000 for brook trout. 10.000 to 69.000 for fathead minnows and 17.000 to 50.000 for channel catfish. These studies demonstrate that small concentrations (parts per trillion) of toxaphene can cause significant effects on fish. Swain et a1. (1982) estimated open water concentrations of 1.5 ng/l for Lake l7 Huron in 1980 and 1981. Given the uncertainties in measuring low concentrations. these concentrations are not very far from the MATC for brook trout. Environmental Dynamics Agricultural - Nash et a1. (1977) studied the behavior of toxaphene and DOT in a model agro-ecosystem chamber. At the end of the experiment 24% of the applied toxaphene had volatilized. 20% was in the soil and the rest was on the plants and chamber. Volatilization half- lives were 15.1 days for toxaphene and lELB days for DDT. though losses were non-linear with a large percentage volatilizing in the first 24 hours. Seiber et a1. (1979) studied toxaphene movement in an agricultural field and found that 59% of the toxaphene applied to cotton leaves was lost in 28 days. Analysis of soil and air samples indicated that volatilization was a major loss mechanism with the early eluting. more volatile. GC peaks being lost preferentially. Willis et a1. (1982) found that only 5 to 10% of the toxaphene applied to mature cotton plants was washed off by simulated rainfall. Earlier studies by the same group (Willis et al.. 1980) had indicated that volatilization was a major loss mechanism. Soils - Hermanson et a1. (1971) calculated an approximate half- 1ife of'4 years for toxaphene repeatedly applied to a soil over 5 years. Nash et al. (1973) recovered 45% of the toxaphene that had been applied to soil at a high rate 20 years prior to extraction. Soils. however. had been stored at -5°C and would not reflect environmental conditions. Seiber et a1. (1979) found a topsoil concentration 18 decrease from 13.1 ppm to 504 ppm in 58 days and suggested that volatilization was the major cause. LaFleur et a1. (1973) found a half-life in top soil of about 100 days from a field plot treated with 100 kg/ha. a very high level. They also found a concentration of 1 ng/l in the underlying ground water. Degradation of toxaphene in soils occurs primarily under anaerobic conditions. Murthy et a1. (1977) demonstrated reductive dechlorination of l4C-toxaphene separated into nine fractions and applied to a metapeak silt loam. 14coz production and volatility losses were minimal. Parr and Smith (1976) found the order of degradation in a Crowley silt loam after 6 weeks as: flooded anaerobic (stirred) > moist anaerobic > flooded anaerobic (unstirred) > moist. No degradation was found in autoclaved soils. Toxaphene in anaerobic salt marsh sediments was degraded within a few days (Williams and Bidleman. 1978). Breakdown was also demonstrated in the presence of a Fe+2/Fe+3 redox couple operating at a -250 mv potential and pH = 5.6. Clark and Matsumura (1979) demonstrated significant aerobic degradation of toxaphene in aquatic sediments. They suggested that anaerobic dechlorination coupled with aerobic metabolism would probably be necessary for complete degradation of toxaphene. Gallagher et al. (1979) found that toxaphene accumulated in anaerobic salt marsh soils and was associated mainly with dead roots. Toxaphene spiked below the soil surface was taken up and translocated by marsh grass (Spartina alterniflora). 19 Aquatic Systems - Much of the work described in this section was done while toxaphene was being tested as a possible poison for use in rough fish control programs. The best review of the early work in this area is contained in Hughes (1970). Terriere et a1. (1966) studied toxaphene dynamics in two Oregon lakes treated with 40 and 90 ug/l. Five years after the application. the deep oligotrophic lake was still too toxic to support fish and had a water concentration of 0.4 ug/l (originally 40). Fish placed in live boxes readily accumulated 3-7 ug/l of residue in 10 days. Both plants and aquatic invertebrates accumulated toxaphene 500 to lOOO-fold over water concentrations. Toxaphene residues in the shallower lake. "rich in aquatic life“. were detoxified within one year. though fish. plants. and invertebrates still accumulated high residues. Extracts from fish were just as toxic to houseflies as technical material. Veith and Lee (1971) studied the detoxification of Wisconsin lakes and determined that adsorption to sediments and settling played a major role in the detoxification process. They also reported that greater than 50% of the toxaphene applied to the lakes may have been lost during application through evaporation 90%. Residue concentrations were not corrected for recovery. TLC Purification of Tbxicant A Toxicant A is at variable mixture of two octachlorobornane components which chromatograph with identical retention times on capillary GC (Saleh and Casida. 1978; Gooch and Matsumura. unpublished data). The A (0:) (Nelson and Matsumura. 1975) or A-l (Turner et al.. 1975) component is 2.2.5-endo.6-exo.8.9.9.10-octachlorobornane while the A (B) (B-l) component is 2.2.5-endo.6-exo.8.8.9.l0- octachlorobornane. Saleh (1983) provides.a good description of the basic structural framework of the chlorinated bornane. bornene and bornadiene components of toxaphene. Silica-gel 60 F-254 (250 u. E. Merck) plates were heated at 130°C prior to use. The final extracts from the procedure described above were spotted and plates developed 4 times using n-heptane (saturated chamber). Toxaphene chromatographs as a streak with several distinct spots under these conditions (Nelson and Matsumura. 1975). Authentic 28 toxicant A was chromatographed under identical conditions to determine the appropriate Rf region to isolate. A region corresponding to an Rf of 0.30 to 0543 was scraped. placed into a chromatographic tube and eluted with 25% dietbyl ether in hexane. This procedure separates some of the toxaphene components from p.p'DDT which migrates with an Rf > 0.60. Gas Chromatography and Quantitation of Residues Extracts were analyzed with a 30M x 0.25 mm i.d. DB-l (J and W Scientific) fused silica capillary column using helium as the carrier gas (fixed pressure. 140 kilopascals). Samples were injected (injection temperature 200°C) in the split mode (split ratio 3:1) with the column at 190 °C. The oven was immediately programmed to heat to 260°C at 2°/min. The Sc3H foil electron capture detector was operated with a nitrogen make up gas flow of 30 ml/min at 280°C. Data was collected using a Spectra Physics 4270 integrator. Technical toxaphene standards. originally obtained from the Hercules Chemical (KL. (Lot # x16189-49) were used for comparison with residues. Areas from 30-32 peaks (after p.0' DDE with the appropriate retention times. window 0.1 min) were used for quantitation of the residue. Retention times used were either absolute or relative to p.p' DDE. Quantitation of ”estimated total toxaphene" was done using methods employed by several other authors. whereby peak areas are summed for selected peaks with identical retention times to those of the analytical standard (Ribick et al.. 1982; Musial and Uthe. 1983). The same peaks are summed for several concentrations of a known quantity of standard for generation of a standard curve. This method assumes equal 29 peak response factors for analogous peaks in the residue and standard. a situation which may or may not be true. For this study. 30 peaks eluting after p.p' DDE were used for quantitation. Peak matching was done manually using a light table and overlaying residues on technical standards spiked with p.p' DDE. By using this method we were able to more accurately determine peak matches than by relying on consistent performance of our integrator. After viewing a large number of chromatograms. it is readily apparent that the residue profile. in terms of presence or absence of peaks. is very similar between fish. sites. and years. (this subject will be discussed in more detail in a future report). Quantification of toxicants A and B was done using peaks matched with NMR certified standards isolated in this laboratory (Nelson and Matusumurs. 1975; Matsumura et al.. 1974). Toxicant B was quantitated in the original extract prior'to‘TLC purification of the residue for toxicant A. Because of the limited amount of material available (micrograms). we have used our standards for peak identification purposes only. In order to generate standard curves for toxicants A and B we have used average fractional composition estimates based on available information in the literature. Using the mass spectrometric data of Casida et a1. (1974). Turner et al. (1975). Khalifa et al. (1974). Holmstead et a1. (1974) and Saleh (1983). we have arrived at estimates of 4.8% for toxicant A and 3.4% for toxicant B. Masses for the respective toxicants in the mixture can be derived by multiplying the mass of the standard times the proportional composition (0.48 or 0.34) estimate and using the corresponding mass/peak area relationship to generate a standard curve. Although this is an estimate. we feel that the proportional estimates for 30 composition are sound and thus can be used by other investigators when no technical standard is available. as long as identification of the appropriate peak can be made. Capillary gas chromatography/mass spectrometry was done using a Hewlett Packard 5985A quadrupole mass spectrometer operated in the electron impact mode (70 eV). A 30 M x 0.25 mm i.d. DB-l fused silica capillary column was used and operated essentially as described for routine ECD analysis although injections were done in the spl itless mode. Approximately 700 spectra were acquired during each run (approximately 17 scans/minute). Spectra of the individual peaks were analyzed manually for the appropriate characteristics. RESULTS AND DISCUSSION The pattern of toxaphene compounds in Lake Michigan lake trout samples is very different from technical toxaphene (Figure 2). This has also been noted in canadian east coast marine fish (Musial and Uthe. 1983) as well as seals from swedish waters (Jansson et al.. 1979L. Zell and Ballschmiter (1980) called attention to the widespread occurrence of toxaphene [polychlorinated camphenes (PCC)] in spawn from fish from several different areas. though they only used 4 key reference peaks for quantitation from the 177 different components that toxaphene contains. Part of the difficulty stems from the fact that many toxaphene components are readily degraded by microbes (Clark and Matsumura. 1979) and there are several unrelated compounds in the residue that co-elute and interfere with accurate quantitation of toxaphene components. Most of the compounds found in samples after Figure 2: 31 Capillary column GC-ECD traces of toxaphene (above) and the toxaphene fraction of an extract from a lake trout from Lake Michigan (below). Conditions are reported under Materials and Methods. Locations of toxicants A and B are. indicated. 1M .0 (I ll 2 w 8 IMF OCIO U illuflfllm , Illlllllfllu Ililllllxv w WW IIYCY a CCCCC imilv ”wily ooooooooooo Wu W ooooooooooo 11W Wm m. 11111 L L . O 33 silica gel purification are related to technical chlordane and the DOT complex. Nitration procedures are available for selective removal of the aromatic compounds. a procedure we feel is useful but not absolutely necessary for reliable “overall" toxaphene analysis. However. as we will show. some type of purification procedure must be used for accurate quantitation of the toxic congeners. In our samples the ubiquity of the DOT compounds was useful for determining relative retention times since they chromatograph in strategic regions. None of the compounds eluting prior to p.p' DDE were considered for quantitation due to interferences froulcnrlordane components. In our system this excludes very few components since the vast majority of the standard toxaphene elutes after these compounds (Figure 2). In general. we were able to match approximately 30 peaks of the residue to the toxaphene standard. Since this is only a fraction of the available peaks. other peaks must either be transformation products or compounds unrelated to toxaphene. The mass spectral data indicates that most of the peaks followfing p.p'DDE have a polychlorinated bornane type of structure. Since accurate quantitation relies on matched peaks having the same composition and response factor. quantifying this residue can only be considered a rough estimate. a concept agreed upon by many authors (Jansson and Wideqvist. 1983; Musial and Uthe. 1983). Since toxicants A and B exert substantial toxicity in all organisms investigated (Casida et al.. 1974; Nelson and Matsumura. 1975). it is important to quantify these components in relation to the overall total. This should provide a more reliable indication of the toxicologic potential of the residue. Intial 1y. we used thin-layer chromatrography of a lake trout residue and authentic toxicant A to look for this component. A capillary gas chromatogram of a lake trout 34 residue before and after being fractionated in this manner is shown in Fugure 3. When this procedure is carried out. a peak free of p.p' DDT remains with the same Rf as authentic toxicant A. It also possesses an identical retention time on capilary GLC and has the same retention characteristics on florisil and silica gel. Investigation of gas chromatograms of initial extracts for toxicant B suggested that no post silica-gel purification was necessary (Figure 2). To confirm the presence of toxicants A and B in lake trout samples. we examined the pesticide fraction from the silica-gel column with capillary GC-MS in the electron (EI) impact mode. The mass spectrum of a toxaphene component co-eluting with p4? 001 (top) is shown in Figure 4. The clusters at masses 352. 316. 235. and 116 are from p.p' DDT with masses 352 and 316 corresponding to the parent ion (M*) and a fragment from the loss of HCl. respectively; Toxicant A elutes incompletely resolved from p.p' DDT in this region. For identification of toxicant A we have relied primarily on the fragment clusters at masses 375. 339. and 303. The cluster at 375 corresponds to the [M-Cl]+ ion from a compound having a C10H10C18 lMN=4lOI formula. For comparison. a mass spectrum of toxicant A as it elutes from technical toxaphene under the same GC conditions employed for the residue is shown (Figure 4. bottom). ‘The lack of a parent ion near mass 410 is typical of toxaphene components (Saleh. 1983; Holmstead et al.. 1974). The results supporting the identification of toxicant A are. the mass spectrum indicating a bornane structure with 8 chlorines. matching Rf”s on thin layer chromatography. similar retention on florisil and silica-gel columns. and capillary gas chromatography with authentic toxicant A. 35 Figure 3: Capillary column GC-ECD traces of the toxaphene fraction of an extract of lake trout from Lake Michigan prior to (above) and after (below) TLC purification for toxicant A. Locations of toxicants A and B are indicated. Figure 4: 37 Electron impact mass spectra of toxicant A (below) and of toxicant A (above) co-eluting with p.p' DDT from a lake trout residue during capillary GC-MS. The spectrum of toxicant A (below) was acquired during a‘ capillary GC-MS run of technical toxaphene. 38 .. X B . F q . gé. : i . , ‘ 1“ ii}: :I ii : f "I! 1:“! ‘,' I"; i; 1: l ' '21, us' ’3: ii {1. W l“! u: a a; . :‘V' i; ‘ I' 3.1;; UNI i “I NU .‘U (”mill "‘lll'lilu gil (Ilka-m gr .}§; 1 fl. , N), ‘01:} .0 100 120 140 160 180 200 220 240 280 280 300 320 340 360 3'0 400 .0 100 120 140 180 1.0 200 220 340 260 280 300 330 340 380 300 400 39 As stated earlier. identification of toxicant B in the residue was achieved by matching retention times of an authentic standard to that of a corresponding peak in the extract prior to TLC purification for toxicant A. The mass spectra acquired in this region which support our identification as shown in Figure 5. Toxicant 8. labeled in Figure 2 and 3. can be seen to elute just prior to p.p' 000. With the small loss in resolution we encountered in the GC-MS system and the relatively small amount of toxicant 8 present. the spectra acquired show a co-elution with pdr 000. The masses at 318. 325. and 165 identify p.p' 000 with the mass at 318 corresponding to the parent ion (MT). The lower spectrum of toxicant B as it elutes during a capillary GC-MS run shows a weak M+ ion at mass 376 (C10H11c17) and complex characteristic fragments at masses 339. 325. and 303. Identical fragments can be seen in the peak that co-elutes with p.p' 000. If it is indeed true that these residues are derived from the use of toxaphene. we would expect both of these compounds to be present. The results of the analysis of 4 Lake Michigan lake trout for polychlorinated bornanes (estimated total) and more specifically for the toxic congeners. are presented in Table 2. The concentration of toxicant A is more than an order of magnitude less than the total residue while toxicant B is 2-3 times less than toxicant A. Since these are two of the most universally toxic congeners. and we have no other good reference point. this concentration may be more toxicologically relevant. We are currently conducting more detailed studies to further explore this suggestion. 40 Figure 5: Mass spectra of toxicant B eluting from a technical standard (below) and of toxicant B co-eluting with pdf 000 in a lake trout residue (above). 41 I ‘ ‘I l .l .1; n. ‘1 11"..) 1 1'1 iN‘l‘Ull) '1‘”!le .idiléllg’i; .f. 111‘") ll L III '11 rr .0 100 120 140 150 180 200 220 240 260 280 300 320 340 360 300 400 I d d - d - u‘ . 1:: t ‘ .ll ' ' i I" u i 0.1 :{I ‘ ‘ ll ('11? ! ' '1 1 1 - g' _ is l 1 I‘l. ‘ 7’: l I {i .' 1:. :I '.' z ‘ ‘ l I . ‘ .0 100 120 140 150 1.0 200 220 240 260 ISO 300 '320 340 380 380 42 Tab1e 2: Concentrations of the two primary toxic constituents and estimated tota1 toxaphene found in 1ake trout from Lake Michigan. Estimated(a) Tota1 Fish Toxaphene Toxicant A Toxicant B 1 10.7 0.70 (6.611!» 0.21 12.0) 2 1.9 0.08 (4.4) 0.08 (4.0) 3 1.6 0.12 (7.1) 0.06 (3.7) 4 2.9 0.12 (4.2) 0.04 (1.4) X :_S.E. 4.3 + 3.7 0.26 i 0.26 0.10 1 0.07 (3)A11 residue concentrations are in ug/g (ppm) wet weight. Estimated tota1 toxaphene 1eve1s have been derived on the basis of equa1 e1ectron capture response factors for a11 components (see text). (b)Numbers in parentheses indicate the percent composition of the toxic components in the estimated toxaphene tota1. 43 Estimates of the amount of toxicant A and toxicant B in technica1 toxaphene varies from 1.5 to 8.5% for toxicant A and 2.5 to 4.1% for toxicant B. On average. toxicant A is approximate1y 4.8% and toxicant B is approximate1y'3.4% (Casida et a1., 1974; Turner et a1.. 1975; Sa1eh, 1983; Kha1ifa et a1., 1974; Ho1mstead et a1., 1974) of the total mixture. Using these estimates and the mean va1ues from Tab1e 2, it appears that the environmenta1 residues of toxaphene have near1y the same re1ative amounts of the toxic components as the technica1 mixture. This is somewhat surprising, and perhaps coincidenta1, since the estimates for the tota1 apparent toxaphene is subject to such 1arge assumptions during quantitation. Indeed, since the toxaphene we measure as a residue has been a1tered by various processes (Musia1 and Uthe,1983;Jansson et a1., 1979; 2e11 and Ba11schmiter. 1980), this simi1arity of ratios is unexpected. Perhaps the rate of environmenta1 transport and metaboTic degradation of toxicant A and B may be simi1ar to the rest of toxaphene residues as a who1e, an observation suggested by C1ark and Matsumura (1979) when they demonstrated that the rate of degradation of toxicant A by microorganisms and in sediment was rough1y identica1 to the overa11 mixture. The presence of toxicants A and B suggests that the residue shou1d be examined for its toxico1ogic properties. CHAPTER 3 TOXICITY OF CHLORINATED BORNANE (TOXAPHENE) RESIDUES ISOLATED FROM GREAT LAKES LAKE TROUT (SALVELINUS NAHAYCUSH) INTRODUCTION 1. Acute T0xicity and Hal-a1ian Neuroreceptor Binding Ch1orinated bornane residues, most 1ike1y derived from the pesticide toxaphene, have now been found in aquatic biota from around the wor1d (Jansson et a1., 1979; Ze11 and Ba11schmiter, 1980; Musia1 and Uthe. 1983; Wideqvist et a1., 1984). Recent1y. residues were a1so found in the Great Lakes (Ribick et a1.. 1982; Rice and Evans, 1984; Gooch and Matsumura. 1985). Because these residues are found in p1aces far from any known U585 ch1orinated bornanes have joinedINTTand the po1ych1orinated bipheny1s.(PCBs) as a c1ass of g1oba1 po11utants (Ze11 and Ba11schmiter, 1980). Studying toxaphene residues is difficu1t because of the comp1ex mu1ticomponent composition of the materia1. Toxaphene is an incomp1ete1y defined mixture of at 1east 177 different components formed during the process of ch1orinating camphene (Turner et a1., 1977; Sa1eh et a1.. 1977). Unti1 recent1y, ana1ytica1 techniques were unab1e to detect toxaphene residues in an unambiguous fashion. The routine avai1ability of capi11ary co1umn gas chromatography (capi11ary GC) and capi11ary GC-mass spectrometry has provided great strides in our abi1ity to identify residues derived from such comp1ex mixtures. This increased capabiiity now a11ows us to ask more detai1ed questions about the nature of the changes that are occurring with these materia1s during movement to and in the aquatic ecosystem. In particu1ar, since 45 46 most investigators have noted the dissimi1arity between toxaphene residues and ana1ytica1 standards (Musia1 and Uthe. 1983; wideqvist et a1.. 1984),ii:is unc1ear whether-the toxaphene residues found in the environment sti11 possess a broad spectrum of toxico1ogica1 activity. Since the toxicity of toxaphene is not distributed even1y throughout the mixture (Ne1$on and Matsumura, 1975; Sa1eh et a1., 1977; Isensee et a1.. 1979), more detai1ed ana1yses and studies need to be conducted in order to c1arify the potentia1 significance of toxaphene derived residues. Our recent work (Gooch and Matsumura, 1985) demonstrated the presence of toxicants A and B, two of the most toxic constituents of toxaphene, in residues from Lake Michigan 1ake trout (Sa1ve1inus namaycush). The presence of these two components suggests that the residue may possess significant toxico1ogica1 properties. This study was conducted to examine the toxicity associated with toxaphene residues derived from Great Lakes 1ake trout using acute toxicity and neuromo1ecu1ar receptor site binding affinity as endpoints. HMJERIALSIMHJFEIHOOS Extraction and Purification of Residues The 1ake trout examined in this study were taken from the southern ha1f of Lake Michigan. In 1982, fish (50-60 cm tota1 1ength) were obtained as incidenta1 catch from commercia1 fishing nets near Muskegon. Michigan, on August 6th. In subsequent years, fish were obtained from 1ate summer (September) samp1ing surveys conducted by the ILS. Fish and Hi1d1ife Service near Saugatuck. Michigan (approximate1y 47 45 mi1es south of Muskegon). Lake trout were a1so obtained from Siskiwit Lake on Is1e Roya1e in Lake Superior in Ju1y of 1984. Fish were kept on ice for approximate1y 10 hours before being frozen in individua1 p1astic bags at -20° C. For ana1ysis. fish were partia11y thawed and a samp1e of the be11y-f1ap region was removed from an area immediate1y anterior to the pe1vic fin between the 1atera1 1ine and the ventrum. This region is rich in adipose tissue and was chosen since it was expected to contain higher 1eve1s of 1ipophi11ic contaminants than other portions of tissue. Concurrent1y, a samp1e of musc1e tissue (fi11et) was removed from an area immediate1y dorsa1 to the be11y-f1ap and frozen for 1ater ana1ysis. A11 tissues were examined with the skin removed. Individua1 10-20 gm samp1es were weighed. minced, and ground with 4-6 vo1umes of anhydrous sodium su1fate using a mortar and pest1e unti1 a uniform preparation was obtained. Extraction. ge1 permeation 1ipid c1ean-up and pre1iminary fractionation for toxaphene was done using the mu1ti-residue procedure described by Ribick et a1. (1982). The fina1 nitration procedure was not done at this point. This protoco1 uses f1orisi1 and si1ica-ge1 to separate toxaphene fronlthe PCBs and the ubiquitous cyc1odiene insecticides (die1drin, endrin, eth. The appropriate si1ica-ge1 co1umn fraction was analyzed for toxaphene components using capi11ary BC with e1ectron capture detection. 66 conditions were as fo11ows: 30 m DB-1 co1umn (J and w Scientific).25 mm idJL25 u fi1m, head pressure 130 kpasca1s, preco1umn sp1it 5:1. injector temp 190°C, detector 270°C, and the oven was temperature programmed immediate1y fo11owing injection from 190-260°C 2°/min. Data co11ection and integration was done using a Spectra-Physics 4270 48 integrator. Chromatograms were checked and a1'l qua1ity contro1 comparisons made manua11y. Comparisons and quantitation were done using technica1 grade toxaphene (origina11y obtained from the Hercu1es Chemica1 Co" Lot # X16189-49) with methods described previous1y (Gooch and Matsumura, 1985L. Samp1es at this point contain severa1 major components of ch1ordane and some of the components of the DDT comp1ex. These compounds prec1ude simp1e interpretation of toxaphene derived residues and are not desirab1e for toxicity testing. For separation of ch1ordane and toxaphene, samp1es were app1ied to a charcoa1 co1umn simi1ar to that origina11y described by Underwood (1978) and further modified by Farre11 (1985). For this procedure, 2 g of a 1:1 mixture of Super A activated carbon (AX-21. Anderson Deve1opment Co.. Adrian, MI 49221) and Ce1ite 545 (Supe1co Inc.) was p1aced into a 22 mm id X 500 mm id chromatographic co1umn p1ugged with g1ass woo1 and 1 cm of anhydrous sodium su1fate. Another 1 cm of sodium su1fate was carefu11y added to the top of the charcoa1 mixture and the co1umn was rinsed with 1 1iter of pesticide grade benzene 'fl311owed by 200 m1 of diethy1 ether (2% ethano1). Samp1es (1-5 m1) were p1aced on the co1umn and e1uted with 80 m1 of diethy1 ether (ch1ordane fraction) fo11owed by 200 m1 of benzene (toxaphene fraction). The ch1ordane fraction contained approximate1y 5-10% of a 14C toxaphene spike and the toxaphene fraction contained cis-nonach1or and the DDT comp1ex as major contaminants. The charcoa1 co1umn can be washed with 200-300 m1 of ether and used again unti1 an excess of charcoa1 penetrates the sodium su1fate 1ayer. Charcoa1 co1umn purified toxaphene samp1es were reana1yzed using the same peaks that were used in the first ana1ysis. Samp1es were usua11y poo1ed before the fina1 nitration purification for removing 49 compounds re1ated to DDT. Nitration was done using a modification of the methods deve1oped by K1ein and Link (1970) and Ho1drinet (1979). So1vent evaporated samp1es were treated with a 1:1 nfixture of concentrated sulfuriczfuming nitric acids at 70°C for 1/2 hr. The mixture was quantitative1y transferred to a 500 m1 separatory funne1 with 3x 10 m1 portions of methy1ene ch1oride. The acidic water 1ayer was extracted twice with 50 m1 of methy1ene chToride and the so1vent was poo1ed into another separatory funne1. One-hundred m1 of a 5% sodium bicarbonate so1ution was added and the mixture was shaken. The organic 1ayer was fi1tered over anhydrous sodium su1fate and evaporated to near dryness using a rotary evaporator. The samp1e was p1aced on a 5 g f1orisi1 co1umn simi1ar to that used in the first part of the procedure and e1uted with 40 m1 of 6% diethy1 ether in petro1eum ether. The final extract was reana1yzed using the same peaks 'as previous1y with the addition of toxicant A (now removed from p.p' DDT interference). This samp1e contains a1most exc1usive1y ch1orinated bornanes as judged by capi11ary GC negative chemica1 ionization mass spectrometry. The major notab1e exception is the presence of cis- nonach1or, a minor component of technica1 ch1ordane, but a major component of ch1ordane residues in fish (Ribick and Zajicek, 1983). This fina1 fraction contains both toxicants A and B, the two major toxic toxaphene components found previous1y in 1ake trout from Lake Michigan (Gooch and Matsumura, 1985). Recovery studies were done using 14C toxaphene spikes (= interna1 standard) and were genera11y greater than 75% after the charcoa1 co1umn. Reagent b1anks were periodica11y checked for possib1e system contamination. Fina11y. a spike of technica1 toxaphene was added to 50 hatchery reared 1ake trout tissue and carried through the entire procedure in order to determine the changes that may be occurring due to the procedure aTone (= procedura1 standard). Acute TOxicity Tests Acute toxicity tests were conducted using 3rd and 4th instar Aedes ggypti mosquito 1arvae (Rockefe11er strain) reared in this Taboratory. Tests were conducted in 16 x 100 mm disposab1e cu1ture tubes containing 5 m1 of disti11ed water and 10 1arvae per treatment. Technica1 toxaphene positive controTs were run for a11 tests and ana1yzed separate1y for statistica1 heterogeneity. Because no statistica1 difference was detected among tests for the positive contro1s. the data were poo1ed and used as a separate treatment for ana1ysis of variance PPOCEOUVBS- L050 ca1cu1ations were done using probit ana1ysis on data conforming to the distribution with at 1east a 50% probabi1ity. [35$]-t-Buty1bicyclophosphorothionate (TBPS) binding The ability of toxaphene residues extracted from 1ake trout to inhibit the binding of 3SS--TBPS (New Eng1and Nuc1ear, Lot #2101-118, origina1 specific activity 6943 Ci/mmo1e) to the picrotoxinin receptor site in the GABA (gamma-aminobutyric acid)-ch1oride ionophore comp1ex of the centra1 nervous system. was examined using rat brain microsomes (synaptic membranes + endop1asmic reticu1um) prepared by the method of Ramanjaneyu1u and Ticku (1984L.The cerebe11a and cerbra1 cortices from 3 ma1e Sprague-Daw1ey rats were removed. weighed, and homogenized in 25 vo1umes of co1d 5 mM Tris. 1 mM EDTA (pH 8 7.1) using 14-16 strokes of a tef1on-g1ass homogenizer. The homogenate was centrifuged at 1000 g 51 for 10 min. The supernatant was then centrifuged at 140,000 g for 30 min yie1ding a mitochondria1 p1us microsoma1 pe11et. The pe11et was homogenized in 25 vo1umes of 5 mM Tris, 1 mM EDTA (pH = 7.1) and dia1yzed overnight against 50 vo1umes of doub1e disti11ed water at 4°C. The dia1yzed preparation was centrifuged at 140,000 g for 30 min afer which the pe11et was either resuspended in .32 M sucrose or 5 mM Tris/1 mM EDTA and frozen in 3 m1 a1iquots at -80 C. For binding assays, tubes were thawed, membranes precipitated at 140,000 g twice and homogenized in 5 mM Tris/1 mM EDTA/0.2 M KBr (ph = 7.5). Binding assays were conducted essentia11y as described by Aba1is et a1. (1985). One-hundred 111 of the membrane preparation was incubated in Tris/EDTA/KBr (pH = 7.5) containing 2 nM 355-13115 in a fina1 vo1ume of 1 m1. Non-specific binding was determined in the presence of 50 11M co1d t-buty1bicyc1ophosphate (TBP) or picrotoxinin (pm. Specific binding was determined as the tota1 12 nM 35S-TBPS a1one) minus non-specific (2 nM 35S-TBPS + 50 11M TBP or PTX) and was genera11y 40-70%. Residues and toxaphene were tested for their abi1ity to inhibit 35S-TBPS binding by incubation for 10 min before the addition of radioactive 1igand. A concentration/inhibition curve was p1otted to determine the binding characteristics and the optimum concentration for comparisons among residues. Tubes were incubated for 90 min at room temperature and the reaction was terminated via di1ution with ice co1d reaction buffer and rapid fi1tration through whatman GF/B g1ass fiber fi1ters. Fi1ters were rinsed twice and counted using 1iquid scinti11ation. Quench correction was done using curves generated for 14C standards since the beta energies are near1y identica1 for the two isotopes. A11 statistica1 ana1yses were done using Statistica1 Ana1ysis System (SAS) software (version 4). RESULTS AND DISCUSSION Residues were purified from Lake Michigan 1ake trout and compared to technica1 toxaphene (Figure 6). The 1ake trout residue is essentia11y free of interferences from ch1ordane and DDT re1ated compounds; cis-nonach1or3 a minor component of technica1 ch1ordane (retention time near 18 min) is the major notab1e exception. The 1arge peak with a retention time near 23 minutes has not been identified. Mass spectra1 data is consistent with a ch1orinated bornane with 8 or 9 ch1orines. Because this peak is disproportionate1y 1arge compared to the rest of the chromatogram, it has not been used for quantitation. Carefu1 inspection of other avai1ab1e 1iterature on purport-ed atmospherica11y deposited toxaphene (Ze11 and Ba11schmiter, 1980; Kramer et a1., 1984; Vaz and B1omkvist) revea1s a 1arge peak in the residues with a retention time simi1ar to the one mentioned here. In a11 cases. this peak is characteristic of a toxaphene component, through no structura1 identification has been made. This conformation is apparent1y quite stab1e compared to the rest of the components. The peaks that were used for quantitation have been noted in Figure 6. A further description of the characteristics and 1eve1s of these residues wi11 be the subject of a future pub1ication. The peaks used for quantitation of the purified residues were p1otted as norma1ized histograms (Figure 7) in order to faci1itate visua1 comparisons. Norma1ization was to peak with the greatest area in each case. While it is difficu1t to genera1ize, interpretation of 52 Figure 6: 53 Capi11ary gas chromatogram of technica1 toxaphene (above) and purified 1ake trout residue toxaphene (be1ow). Peaks designated (0) have been used for quantitative and comparative purposes. For conditions, see Materia1s and Methods. Figure 7: 55 Norma1ized histogram of the peaks used for quantitation of the residues and standards compared in this study. For a description of abbreviations, see Tab1e 3. Peak numbers can be estab1ished from designated (Q) peaks in Figure 6 by moving from 1eft to right in sequence. 'aiii-AI IliadoloBEI-Ou IIlIbI i anal-in IIIIIIII IIIIIIIIIIIIIII EEEEEEEEEE EEEEEEEEEE are lllllllllllllll 57 Figure 7 1eads to the conc1usion that the purified residues iso1ated from 1ake trout are remarkab1y simi1ar to both technica1 toxaphene and the toxaphene procedura1 spike. We emphasize purified since it is un1ike1y that this conc1usion wou1d be reached without using the purification procedures we have described. with the number of compounds present in Great Lakes fish, protoco1s designed for mu1tip1e residue quantitation do not provide adequate separation for detai1ed ana1ysis of toxaphene residues. The additiona1 steps we have app1ied are necessary to conduct more specific tests. 0f particu1ar importance toxico1ogica11y3 is the re1ative composition of toxicants A and B which are represented by peaks 7 and 3, respective1y in the extracts from 1ake trout. We have demonstrated previous1y1(Gooch and Matsumura, 1985) that these two constitutents, which are two of the most toxic, were present in 1ess purified 1ake trout residues. Tab1e 3 is a comparison of the composition of the residues and the standards with respect to these two components. These va1ues on1y represent estimates, since the 1eve1s are derived from approximate standard curves made by taking fractiona] composition averages for toxicants A and B in technica1 materia1 (4c8% and 3.4% respective1y), mu1tip1ying by a known mass of materia1. and using the resu1ting mass/peak area re1ationship. In addition, the mass va1ues obtained for A and B are then divided by the mass va1ue obtained for tota1 toxaphene. an estimate as we11 since we assume that each peak summed for quantitation is the same as that in the technica1 standard. The percentages for A and B in the standards have been determined the same way as the residue in order to make comparisons more direct. The fish residues have s1ight1y more toxicant A and less B than the 58 Tab1e13: Percent composition estimates for toxicants A and B in residues and standards used in this study. Data are expressed as mass % (area %) where area % equa1s the integrated GC peak area of A or B/tota1 peak area used in quantitation. 1932313) 1983B 1984B Sisktib) Technica1 Standard Procedura1 Standardic) Toxicant A 2.80 (9.20) 2.82 (9.12) 2.58 (10.21) 2.44 (8.54) 2.39 (8.57) 2.77 (9.06) Toxicant B 1.21 (5.24) 1.19 (5.13) 1.19 (5.11) 0.83 (3.17) 1.70 (8.09) 1.35 (5.95) Tota1 4.01 (14.44) 4.01 (14.25) 3.77 (15.32) 3.27 (11.71) 4.12 (15.01) 4.09 (16.66) 13’ B signifies adipose tissue rich 'be11y f1ap" and F denotes edib1e fi11et. (b) SiskL - Lake trout from Siskiwit Lake on Is1e Roya1e in Lake Superior. (c) Procedura1 standard is technica1 toxaphene iso1ated with the same procedure used for regu1ar fish samp1es. 59 technica1 standard, the difference being due to the iso1ation procedure since the procedura1 toxaphene standard has a simi1ar bias. The concentrations of toxicants A and B is highest for the standards. The qua1itative characteristics of the residues appear to be simi1ar from year to year and from different tissues and 1ocations. To date. their has been no systematic assessment of the changes that may be occurring with time in fish from the Laurentian Great Lakes. Since the most wide1y accepted hypothesis for how toxaphene enters the Great Lakes is primari1y via atmospheric transport from the cotton-be1t regions of the southern United States (Rice and Evans, 1984), and because toxaphene use decreased significant1y after the 1ate 1970's, 1eve1s and patterns of toxaphene residues may be expected to continuous1y change as biotic and abiotic environmenta1 forces act on the mixture. Toxaphene is degraded by some nficroorganisms and mamma1ian species (C1ark and Matsumura, 1979; Parr and Smith, 1976), whi1e 1itt1e is known of its metabo1ism in aquatic organisms. The patterns observed in fish. however, do not demonstrate dramatic change. The spectra in Figurer7 do not show any great changes in the residue from 1982 through 1984, an observation which a1so extends to toxicants A and 8. Recent studies suggest that current inputs.of'toxaphene are 1ow (Strachan, 1985). Thus, toxaphene is apparent1y not being degraded in the Great Lakes system. If inputs are 10w and degradation is significant. we shou1d expect to see a shift in the composition of the residue with time. Our data do not show this. A simi1ar observation was made by fisheries bio1ogists in the 1ate 1950's and ear1y 1960's when toxaphene was being tested as a possib1e piscicide. In many 60 cases, 1akes poisoned with toxaphene remained toxic for a number of years (Johnson et a1., 1966). We conc1ude that toxaphene wi11 be present for a number of years and wi11 probab1y degrade or dissipate in the biota very s1ow1y. Tbxicity Acute toxicity bioassays were conducted wininmsquito 1arvae to demonstrate whether'the toxaphene residues iso1ated from 1ake trout exhibited different toxicities than technica1 or procedura1 standards. At the time each test was conducted a positive contro1 with technica1 toxaphene was performed in order to monitor any differences that might occur due to uncontro11ed variab1es. After the data had been co11ected, a separate ana1ysis of variance was conducted on the positive contro1s. iSince there was no significant difference among tests (Tukey“s HSD a = 0.10), the contr01 data was poo1ed and used as a separate treatment. Statistica1 ana1ysis of the overa11 data (Tukeyks H50 0 = 0.10) revea1ed that the residues from 1982 and 1984 were s1ight1y 1ess toxic than the technica1 standard (Tab1e 4). 'There were, however, no differences among the different years when compared to each other or the procedura1 standard. This 1eads to the conc1usion that the acute toxicity of toxaphene residues iso1ated from Lake Michigan 1ake trout is essentia11y equiva1ent to standard materia1. This supports our identification of toxicants A and B in these samp1es and suggests that though the residue appears quite different from the ana1ytica1 standard. the bio1ogic activity of the materia1 is not great1y different.Though ana1ysis of variance procedures are genera11y robust with respect to non-norma1ity (Stee1 and Torrie, 1980), a 61 Tab1e 4: Acute toxicity of toxaphene residues purified from Lake Michigan 1ake trout to mosquito 1arvae (Aedes egypti). Data are expressed as 24 hour LC50 in ug/1 (X 1 s.d. (n)). 198231a) 751 1 213 (3MP) 1983B 673 1135 (3) 1984B 733 i.203 (3)* Technica1 toxaphene 441 1 140 (8) Procedura1 toxaphene 603 :_139 (6) (a) For description see Tab1e 1. (b) * = Significant1y different from technica1 toxaphene (p < 0.1) Tukey's HSD. 62 nonparametric ana1ysis of the data (Kruska1-wa11is) yie1ded simi1ar resu1ts. The degree of variabi1ity we encountered in this test system was great and prec1uded any further comparisons from being statistica11y significant. 355-1395 Binding Acute toxicity bioassays are integrated measures of the activity of a11 of the components in the test mixture and wi11 ref1ect the summation of a number of factors inc1uding uptake, metabo1ism, excretion, partitioning to the active site. etc. Thus, morta1ity is a very nonspecific endpoint which measures the tota1 activity of the test mixture irrespective of the mode of action. Since we measured residues from environmental samp1es, it is possib1e that some unknown factor, possib1y toxic. cou1d have been introduced by the procedure. Because our ana1ytica1 endpoint is predominant1y capi11ary BC with e1ectron capture detection, anything not detected by this procedure is a potentia1 interference. When the mode of action of a c1ass of chemica1s is known and a putative target site has been identified, it is possib1e to conduct more specific tests in order to gain a greater degree of understanding about the toxico1ogy of the test materia1. For cyc1odiene type insecticides inc1uding toxaphene, the most p1ausib1e target site appears to be the GABA-ch1oride ionophore comp1ex (Ghiasuddin and Matsumura, 1982; Matsumura and Ghiasuddin. 1983). GABA (gamma- aminobutyric acid) is the major inhibitory neurotransmitter of the centra1 nervous system of near1y a11 vertebrates and functions by stimu1ating C1‘f1ux across nerve membranes. Cyc1odiene type 63 insecticides have been shown to inhibit this f1ux by binding at the picrotoxinin receptor. This receptor is proposed to be 1ocated at, or very near, the ch1oride channe1 (Matsumura and Ghiasuddin. 1983; Lawrence and Casida, 1984; Aba1is et a1., 1985). The resu1t is an finba1ance in excitatory and inhibitory signa1s.and 1ess stimu1ation causes an overproduction of excitatory activity. Two major 1igands, picrotoxinin and TBPS, have been found to bind to a site at or near the ch1oride channe1 (Ticku and 01sen, 1979; Squires et a1., 1983; Ramanjaneyu1u and Ticku, 1984). The abi1ity of a compound to inhibit the binding of these 1igands to the receptor site is high1y corre1ated with toxico1ogic potency (Lawrence and Casida, 1984; Matsumura and Tanaka. 1984). Inhibition curves for technica1 toxaphene and a residue iso1ated from the be11y-f1ap region of fish co11ected in 1984 were compared (Figure 8). Both inhibitors yie1ded simi1ar curves with 1050 va1ues of 5.0 1 3.8 x 10--8 M for toxaphene and 2.9 i 2.1 x 10-8 M for the 1ake trout residue (n=3 experiments). This simi1arity agrees with the toxicity data obtained ear1ier and the 1C50 va1ues are simi1ar to those obtained by Co1e et a1. (1984). 10'8 M was se1ected as an appropriate 1eve1 to use for further point comparisons among residues. Residues from different years. sources, and tissues inhibited the binding of 35-TBPS to the receptor site in rat brain membranes (Tab1e 5). Extracts from be11y-f1ap samp1es from 1982 and 1983 were the 1east potent and were not significant1y different from one another. The 1982 sample was, however, different from the others. The 1983 samp1e was significant1y different from the procedura1 standard. It is interesting to note that the samp1es from fi11et tissue and from Figure 8: 64 Inhibition curve of toxaphene (EJ) and the 1984B ([5) 1ake trout residue for [3551-TBPS binding to rat brain synaptic membranes. IC-50 va1ues were 5.9 i 3.8 X 10'8 M for toxaphene and 2.9 1 2.1 X 10'8 M for the 1ake trout residue (n = 3 experimentsk % Speci ic Binding ‘lOoD I 80.0 I 60.0 I 30.0 I 65 E] Technical Sidnddrd A Lake Troui Residue 0-0 9.0 T l 8.0 7.0 6.0 5.0 -log [inhibii‘or] 66 Tab1e 5: Effect of toxaphene residues from Lake Michigan 1ake trout on 35S-TBPS binding to rat brain synaptic membranesfia) 181: 2.1.1 1982B 31.4 (11b 31.3 1983B 38.7 (1.2) 38.7 1984F 43.6 (2.3) 43.5 SiskL 44.1 (2.3) 44.0 Technica1 standard 40.8 (2,3) 42.0 Procedura1 standard 49.5 (3) 49.4 (a) Data are expressed 5 mean % inhibition from 3 experiments. [Inhibitor] =1 x10‘ M. Specific binding was determined using both co1d t-buty1bicyc1ophosphate (TBP) and picrotoxgnin (PTX) at 10’4 M as disp1acing 1igands. The concentration of 5S-TBPS was approximate1y 2 nM. For further detai1s see materia1s and methods. (b) Means with the same number are not significant1y different, Tukey”s H50 01 = 0.05. 67 Siskiwit Lake on Isle Royale tended to be more inhibitory than the more fatty belly-flap samples and the technical standard. though the difference was not statistically significant. The most inhibitory sample was the procedural standard. If this data is compared to the relative composition estimates for the toxic constituents in Table 3, it is found that inhibitory potency only rough1y follows the total level of these two constituents. This suggests that toxicity is mediated by components other than just toxicants A and B, an observation noted previously (Sa1eh et al.. 1977). By considering the acute toxicity data, the 35S-TBPS binding data, and the toxicant A and B composition estimates in concert we conclude: (1) Residues derived from toxaphene. isolated from Great Lake fish, possess toxicologic potency similar to technical materia1 (i.e. toxaphene in the Great Lakes ecosystem has not been appreciably detoxified; (2) Composition estimates of toxicants A and B are not sufficient to predict toxicologic potency direct1y. Toxicity is probably attributed to many of the components of the toxaphene residue; (3) The spectruniof activity and the composition of the residues has changed very little from 1982 through 1985; (4) Edib1e portions of tissue (fi11et) and fish from a remote island in Lake Superior, contain residues as toxic as standard toxaphene (though total levels are less than that found in fatty samples from lake trout from Lake Michigan) and (5) Neuroreceptor binding studies are useful for measuring activity of environmentally derived residues. These studies require much less material and provide more specificity for the target site. 68 II. Inhibition of 35S—TBPS Binding to Lake Trout Brain Synaptic Membranes. NATERIALS AND METHODS Lake trout were originally obtained as fingerlings from the Jordan River National Fish Hatchery, Elmira, Michigan. Fish were reared for two years on Biodiet pellets (Biodiet Products, Warrenton, Oregon) prior to use. Brains from two fish were removed (1.0 9 total brain tissue), homogenized and processed as described for earlier studies with rat brain. All binding assays etc. were also conducted with the same methods as described previously. RESULTS AND DISCUSSION Since only one report has appeared in the literature regarding TBPS binding in fish brain (Cole et a1. 1984), preliminary experiments were conducted to determine the degree and nature of binding. These experiments demonstrated 140-168 fmoles/mg protein of 92-98% specific binding in the presence of 2 nM TBPS. Binding was 70% inhibited by 10'6 M GABA (Table 6), a behavior consistent with interaction at the GABA-chloride ionophore. The binding characteristics are also similar to those demonstrated for the blackfish (Orthodon microlepidotus) by Cole et a1. (1984). 69 Table 6: Inhibition of 35S-TBPS binding to lake trout brain synaptosomes by residues isolated from Great Lakes 1ake trout. Data are expressed as % inhibition 1; i s.d. (n)). Residue [M] % Inhibitionc 84B toxaphenea 1 x 10-7 74.6 i 10.9 (3)9ab 85B toxaphene 1 x 10‘7 54.9 i 18.8 (3)a’b 84F toxaphene 1 x 10‘7 87.3 1 7.3 (3)a 85F toxaphene 1 x 10-7 50.5 i 8.6 (3)b 85E toxaphene 1 x 10"7 70.6 i 7.8 (3)9:b L.Sisk 1 x 10-7 61.6 113.7 (3)9»b GABA 1 x 10-6 69.5 1 5.5 mil-b 1,111. Hzob 1 01/50 01 36.0 i 9.1 (4) 3B, F, and E designations correspond to belly, fi11et, and egg samples as previously described. Numbers are year designations, L. Sisk is fish from Siskiwit on Isle Roya1e. bLake Michigan water extract. For a more complete description see the results section. cMeans with the same letter are not significantly different (Tukey's HSD 01:0.05). The minimum significant difference for any two means is 32.5%. 70 Inhibition curves were plotted for toxaphene and chlordane (Figure 9). Both of these complex mixture insecticides were found to inhibit 35S-TBPS binding to lake trout brain membranes in a sensitive concentration dependent fashion. Approximate IC50 values of 5 and 8 X 10‘8 M were determined graphically and correlate well with the sensitivity of fish to these two insecticides. Chlordane is slightly less toxic to salmonids (96 hr LC-50 23.5 i 14.0 ug/l) than toxaphene (96 hr LC-50 7.3 1 4.0 ug/l) (Johnson and Finley, 1980). The ICso values determined for 1ake trout are similar to values obtained for rat brain in this work and that of others (Cole et al.. 1984; Aba1is et al.. 1985; Lawrence and Casida, 1984). while fish are very sensitive to the effects of cyclodience insecticides (Podowski et al.. 1979), there is no exceptional sensitivity at the putative target site demonstrated here or by Cole et al. (1984). This suggests that target site sensitivity is not a major reason for the high degree of toxicity associated with these compounds in fish. The sensitivity of the lake trout GABA-chloride ionophore complex to various toxaphene residues isolated from Great Lakes fish is shown in Table 6. All of the toxaphene residues potently inhibited 35S-TBPS binding to lake trout brain membranes, a behavior consistent with the sensitivity of lake trout to this class of compounds. The only significant difference (at o=0.05) noted among the residues was between fi11et samples from 1984 and 1985. The reasons for this are not clear. The inhibition for both of the 1985 samples (belly and fillet) is less than was typically found. The relative amount of toxicants A and B (5.3% total for both samples) is similar to the samples reported in 71 Figure 9: Concentration - response curve for toxaphene and chlordane inhibition of 35S-TBPS binding to lake trout brain synaptic membranes. 72 7352:; mo. [I n — n b n — '4 a 052030 «Id 022398.. mlm 5U!PU!8 ouioeds % 73 Chapter 2 and therefore, a possible explanation for the difference is that there was an error in quantitating the 1985 test samples. Since the in vitro binding assay is specific, sensitive and requires very little material, a preliminary experiment was conducted to determine if GABA-chloride ionophore comp1ex specific neuroactive substances could be detected in Lake Michigan water. It requires 4.1 ng of toxaphene in a 1 ml test volume (1 x 10"8 M) to produce a readily detectable inhibition of 35S-TBPS binding. Assuming an extraction efficiency of 100% and a concentration factor of 40,000, a water concentration of 102 ng/l toxaphene equivalents should be easily detected. A concentration factor of 40,000 is fairly conservative, therefore, a concentration of 50 ng/l should be readily detectable. This assumes that the only specific neuroactive substance in the water is toxaphene and that there are no interferences in the assay. When a 2 1 sample of Lake Michigan water (open water, subsurface) was extracted with methylene chloride, fractionated with florisil. and concentrated as above, it produced 36% inhibition of binding. This corresponds to a toxaphene equiva1ent solution concentration of approximately 2 x 10"8 M. With a concentration factor of 40,000, this corresponds to a water concentration of 205 ng/l. Reported toxaphene concentrations in Great Lakes waters are nearly 100 fold less than this (Sullivan and Armstrong, 1985L. Since the Great Lakes are a repository for a large number of chlorinated cyclodienes, all of which interact at this binding site, the calculations and assumptions used here have little real meaning. It is interesting that a simple water extract possessed significant activityu Since no blank controls were done. this activity requires further verification. The specificity of the 74 system, however, in theory. should make it possible to take this approach for measuring the response of chloride channel antagonists in environmental samples. This experiment with one water sample and no blank controls is admittedly preliminary and hopefully; will serve as a stimulus for further investigation. CHAPTER 4 SUNNARY OF RESIDUE LEVELS AND STATISTICAL PATTERNS ASSOCIATED HITH LAKE MICHIGAN LAKE TROUT RESIDUES FRM 1982-1985 INTRODUCTION Residues of toxaphene or toxaphene-like compounds have been measured in Great Lakes fish for the past several years. The data of Schmitt et al. (1981) from the National Pesticide Monitoring Program suggests that these residues have been detectable since the early 1970's. Subsequent reports (Schmitt et al.. 1983; Schmitt et al.. 1985) have shown an increase in concentrations and frequency of occurrence for samples from 1976-1981. Excellent sumnaries of this data can be found in Rice and Evans (1984) and Sullivan and Armstrong (1985). Whole lake trout taken near Saugatuck, Michigan, from 1977-1979, contained a mean concentration near 7.0 ug/g (wet weight) (Sullivan and Armstrong. 1985). Schmitt et al. (1 985) reported a downward trend at some stations in Lake Michigan and Lake Superior in 1980-81 (2-5 ug/g) compared to 1978-79 (5-10 ug/g). These data were generated using whole fish samples from a narrow size range (approximately 60-70 cm). Trends presented by Schmitt et a1. (1985) and Devault et a1. (1986) suggest that many contaminant concentrations in Great Lakes, 1ake trout are decreasing. Since toxaphene has been banned for most uses in the United States (Federal Register, 1982), it is important to determine if concentrations of toxaphene are decreasing with time. In addition. it is important to study the differences, if any, that might be present in the profiles of the components of the toxaphene residue 76 77 from different years and from different tissues, particularly as these profiles might be related to toxicity. Since the results reported earlier (Gooch and Matsumura. 1986) demonstrated that toxaphene residues are at least as toxic as technical material, it is important to determine whether or not any statistically identifiable differences exist in the various toxaphene residues that have been studied. Quantitative Results Toxaphene concentrations that were found in belly flap and fillet samples from lake trout analyzed in this study were summarized (Table 7). Appendix.A.contains all of the individual information for each fish and the results of replicate analyses of selected samples. Overall, values in belly flap samples ranged from 0.59-10.76 ug/g, whi1e fillets ranged from 0.46 to 3341 ug/g. The ratio of belly flap/fi11et varied from 1.89 to 9.29. Lipid normalized residues may have been helpful in narrowing this range (Hughes, 1970L Concentrations found in fish collected from Siskiwit Lake on Isle Royale are similar to those found in the fillet samples from the Lake Michigan fish. The data presented in Table 7 are the result of a number of factors regarding the type of samples that were analyzed. The most important is the range of the size of fish that were analyzed. No attempt was made in this study to segregate fish samples on the basis of size. Many of the largest fish caught on the day of sampling were intentionally selected in order to maximize the amount of residue obtained per unit weight of tissue for toxicology studies. This 78 «oo._ i_hm.ov Luv m~.o + mm.— .o~._ -.w_.o. «A. em.o + ~_._ mxwg uwszmwm «ow.e -_h~.~. Am~.a -.Wm.~. .ow._ - we.o, .e~.o_ ni~4.mv .m. mm._ + Fm.m .mv ce.~ + .e.m .me mm.o + -._ .m. Fm.~ + om.e mam. .mo.m - mm.~. .NN.m - mm._. ._e.m - em.o. .o_.~ -_Ww.~. «__v wo.o + .w.m .__. o_._ + o~.m ._.V om.o + _e._ .._. Fe._ + em.e «no, «mo.m - mn.~. .oN.N -_pm.mv ._~._ -.we.o, .mm.~ - mm.o. .~_. oa.o + so.m .mv om._ + oe.m «my o~.o + om.o LN_. ~_.~ + No.8 mam, .mw.m .ne.mv .Nm_ .ue._v «mm.m -_m~.mg - .Nv oe.m .N. um._ «m. o_.~ + «4.8 Nam. .mx. d\m eem_oz o_eem eop_ed »__om mew open F~< .uaocu .35.. E; 256:... H x .232. um: 39. we ummmmcaxm mxc— cem_;u_z 8x84 cm mm=u_mmc mcmsqexou cow mu_umvumum accee:m "e opeae 79 sampling protocol..whileenot considered optimal for studying trends (Devault et al.. 1986), will more realistically'represent the degree of variability that is present in residues from larger fish. This data does not establish any apparent trend in the range of concentrations present in lake trout from 1982-1985. This suggests that, though the agricultural use of toxaphene has been banned and new inputs appear to be low (Strachan, l985), toxaphene is not being rapidly eliminated in the Great Lakes or that the existing toxaphene reservoir in the Great Lakes ecosystem is so large that the cessation of toxaphene input not immediately impacted the residue levels. Pattern Recognition Background - Each gas chromatographic profile for a sample can be considered a single observation defined by measures of several quantitative variables. As such, the values of the variables and their relationship to one another'and to other samples defines that sample. In many cases, this relationship is similar for similar types of samples. Multivariate statistica1 pattern recognition techniques have been developed which are useful for defining or visualizing these types of relationships (Kowalski and Bender, 1972), and in recent years these techniques have been increasingly applied to complex chemical data (Derde and Massart, 1982). A full review of the philosophy and mathematics behind pattern recognition is beyond the scope of this work. In addition to the two works cited, excellent reviews and introductions can be found in Cooley and Lohnes (1971). Massart and Kaufman (1983), Delaney (1984), Hold and Sjostrom (1977) and Kowalski 80 and Bender (1973). Often, papers published where the authors have used a specific technique, will contain a parsimonious review of that particular technique. Examples of this include Dunn et a1. (1984), Stalling et a1. (1985), Carey et a1. (1975), Heiner and Parcher (1973) and Pino et a1. (1985). A recent American Chemical Society Symposium Volume (Breene and Robinson, 1985) contains numerous examples of these techniques applied to environmental chemical data. As with any statistical technique, there are a number of assumptions behind the use of multivariate algorithms. In general, it is usually required that the variables be distributed as multivariate normal with common variance and covariance. This also applies between classification groups when they are present. To avoid chance patterns in the data, it is also required that the number of observations be greater than the number of variables being studied. As the simplest example, two points define a line whether there is any linear relationship among the two observations or not. Hold and Dunn (1983) contains an excellent discussion of this requirement. The following is a brief discussion of each of the techniques used in this analysis. All of these techniques are available as part of the SAS statistical analysis package available on the IBM 4381 computer. PrinConp - The Princomp procedure performs a principal components analysis and is used to examine relationships among several quantitative variab1es. It is often used for summarizing data and detecting linear relationships among variables that explain or help interpret some other classification. Because of the ability to explain the major portion of the variability in data by using linear combinations of the original variables, it is often used as a data 81 reduction technique prior to further statistical analysis. Briefly; given p numeric variables, princomp constructs p principal components as linear uncorrelated combinations of the original variables. The first principal component, the first synthetic variable created from some combination of the original,. explains the largest variance of any of the p principal components. Each successive principal component explains the next largest amount of variability via a combination of variables that is orthogonal to the rest. Generally, the first few1principal components will explairithe majority of the variance. The more principal components it takes to explain the data, the greater the likelihood of many unique observations in the data. Principal components can be generated from both correlation and covariance matrices. A plot of the first two principal components is often useful for visualizing patterns in the data since these two components genera11y explain a large portion of the variance. The eigenvectors generated from the matrix manipulations used to do this analysis are useful since they are the scoring (weighting) coefficients for the variables on the respective principal components. Inspection of these coefficients can often give insight into the major variables that are important in describing the variability in the data (TJL which variables are most important for explaining differences). It is important to emphasize that this technique assumes no underlying structure in the data. Principal components analysis is classified as an unsupervised pattern recognition technique, in contrast to other techniques to be described. It is therefore often useful for visualizing patterns in the data that are not readily evident. 82 Fastclus - The Fastclus procedure belongs to a family of techniques generally known as cluster analysis. Massart and Kaufman (1983) is an excellent review of the philosophy and mathematics behind these techniques. Specifically, the fastclus procedure performs a non-hierarchical (disjoint) cluster analysis of observations composed of several quantitative variab1es. It is most often used with large data sets and clusters observations based on a sorting algorithm which uses Euclidean distances between the quantitative variables for each observation. The Euclidean distance is a geometric construct described by the square root of the sum of all of the paired squared distances between corresponding variables among observations. The user generally' designates the maximum number of clusters to be generated and how similar observations must be to be considered in the same cluster. Very simply, this means that observations with similar Euclidean distances will group together. Typically, severa1 analyses are done using different numbers of clusters. The procedure generates several statistical parameters, which relate to the quality of the clustering. Candisc - The Candisc procedure performs a canonical discriminant analysis on data characterized by a classification variable (in contrast to the other techniques) and several quantitative variables. This technique can also be considered a dimension reduction technique and is related to principal components analysis. In this case, we seek to derive canonical variables (similar to principal components) which are linear combinations of the original variables that summarize between-c1ass variation in contrast to the total variation that princomp seeks to explain. The canonical variables (i.e. the synthetic one number variables) have the greatest multiple correlation '83 with the groups. The first linear combination of the variables that has the greatest multiple correlation with the groups is cal led the first canonical variable (component). The weighing coefficients for the variables are similar to the weighing coefficients of principal components analysis. Canonical variables are generated in descending order of their ability to explain multiple correlation with the groups in much the same way as principal components explains overall variabi1ity. A plot of the first two canonical variables is generally used to visualize the relationships between the classification variables. This particular technique is often used if some preliminary treatment of the data has suggested a classification scheme, or if one was known a prior'. It is also used as a method for visually displaying data generated by some other classification system. Standard Toxaphene Data - In order to determine if any artificial patterns could potentially appear in later data analysis that were not due to meaningful differences, a set of 26 standards was chosen for preliminary pattern examination. Because these observations are all for the same material, this analysis should reveal any underlying patterns due to factors including, amount (ng injected), date, column conditions, etc. The standards were chosen to cover a broad range of conditions and are listed in Table 8. The peaks used are noted in Chapter 3, Figure 6. Raw area data was entered for each of 20 peaks in the gas chromatographic profiles. All data was then scaled so that each peak was expressed as a fractional proportion of the total peak area for those twenty peaks. From this set of 20 peaks, a subset of 19 peaks 84 Table 8: Standard toxaphene data used for pattern recognition ana1ysis. Quantity Analysis Injected Total Peak Std Date (ng) Area A 4-26-85 9.0 217476 B 4-22-85 22.5 456645 C 5-2-85 9.0 218411 0 5-31-85 22.5 519793 E 7-25-85 22.5 200139 F 7-25-85 18.0 158688 G 8-21-85 22.5 210547 H 8-21-85 13.5 66460 I 9-21-85 24.0 265475 J 9-21-85 9.6 69313 K 10-14-85 24.0 247542 L 10-14-85 14.0 127388 M 10-17-85 24.0 611792 M 10-17-85 9.6 277760 0 10-28-85 24.0 633607 P 10-28-85 9.6 98452 0 10-31-85 24.0 400626 R 10-31-85 9.6 198947 S* 11-7-85 24.0 216996 T 11-7-85 24.0 181192 U 11-7-85 9.6 84871 V 11-7-85 9.6 73225 N 1-13-86 9.6 132756 X 1-13-86 13.6 365750 Y 5-28-85 54.4 158755 2 5-28-86 27.2 84635 *S. T; U. V, are replicate injections of the same standard. 85 with coefficients of variation of less than 25% was chosen for subsequent analysis. The peak that was removed was highly variable (c.v. = 41%) and was from a region of the chromatogram where resolution was poor and integrator performance highly variable. A principal component analysis of the covariance matrix of this data, followed by a plot of the first two principal components, yielded the pattern shown in Figure 10. The first two principal components explained approximately 57% of the variance. The symbols plotted are the unique alpha character associated with each standard. The first principal component weighted heavily on peaks 3 and 7, which coincidently are toxicants A and B. The second component is weighted on peaks 7 and 14, again two of the more prominent peaks. A careful examination of the GC traces reveals a leading shoulder on peak 7, which may or may not be resolved and detected by the integrator on eveny run. This may explain why this peak is so important in explaining variability in the data. The principal component plot suggests approximately four different kinds of patterns in the data (outlined in Figure 10). Since the first two principal components only explained 57% of the variability in the data, the same data was subjected to a disjoint cluster analysis, followed by a canonical discriminant analysis of the clusters. A preliminary screen using various numbers of clusters from 2 to 10 demonstrated that 4 clusters fit the data. The principal components analysis also suggested 4 groups. Figure 11 is a plot of the first two canonical variables for each sample. As with the principal components analysis, the disjoint cluster analysis produced 4 groups that are remarkably similar to the groupings suggested by 86 Figure 10: Plot of the lst two principal components from analysis of the group of 26 standards outlined in Table 8. These two components explained 57% of the variance in the data. 87 0¢w.0 ._.Zm_ZOn=200 4 ._<0_zoz ._ w a. 70.”!- E a . EWBVIHVA "IVOINONVO PUZ 93 Table 9: Clusters produced by disjoint clustering and canonical discriminant analysis of 20 different standards. Mass Injected Total Peak 10 Date (ng) Area Cluster 1 R 10-31-85 9.6 198947 G 8-21-85 22.5 210547 K 10-14-85 24.0 247542 3 11-7-85 24.0 216996 H 8-21-85 13.5 66460 X 1-13-86 13.6 365750 N l-13-86 9.6 132756 Cluster 2 C 5-2-85 9.0 218411 E 7—25-85 22.5 200139 0 10-28-85 24.0 633607 A 4-26-85 9.0 217476 N 10-17-85 9.6 277760 M 10-17-85 24.0 611792 I 9-21-85 24.0 265475 Q 10-31-85 24.0 400626 U 11-7-85 9.6 84871 D 5-31-85 22.5 519793 B 4-22-85 22.5 456645 T 11-7-85 24.0 181192 Cluster 3 F 7-25-85 18.0 158688 Y 5-28-85 54.4 158755 2 5-28-86 27.2 84635 J 9-21-85 9.6 69313 Cluster 4 V 11-7-85 9.6 73225 P 10-28-85 9.6 98452 L 10-14-85 14.0 127388 94 degradative forces, the composition of the residue should shift with time. A lack of change in lake trout residues with time would suggest that the exposure variables (i.e. food and water) are not being altered substantially. To address this question, 6 fish of similar size were selected from 1983, 1984, and 1985. The fish weighed 3.30 1 0.31, 3.63 i 0.26, and 3.48 i 0.26 kg for each group respectively. The residue profile from the adipose tissue rich belly flap section was compared among the years and to 6 randomly selected standards. The 17 peaks used in this analysis are the same as those from the previous analysis with standards excluding the 4th, 7th, and 16th peaks. Figure 13 is a plot of the first two principal components from an analysis of the covariance matrix for this data. Seventy-six percent of the variability in the data is contained in these first two components. The standards are evident as a tight cluster distinctly separate from the rest of the samples. None of the samples from different years consistently group together, though samp1es do tend to clump into 3 different groups. A review of the dates from the analyses suggests that for some cases, there may be patterns to the residue that are attributed to an analytical error term that encompasses operating conditions, purification procedure etc. and is unique to a particular ana1ysis. Giesy et a1. (1986) demonstrated a large analytical variability relative to other sources of variability using this same system. If this variability is large relative to differences present in residues, it would be very difficult to demonstrate consistent patterns with these techniques. Therefore, this data set does not suggest that there is any difference between the three years that is Figure 13: 95 Principal components plot from analysis of toxaphene residues from belly flap of Lake Michigan lake trout collected in 1983, 1984, and 1985. Technical standards were also included in the analysis. Seventy-six percent of the variability is contained in the first two components. 96 Fzmzomzoo 1_ 4<0_zoz<0 0m — 00.0 00. _. 00.01 00.51 00. _. T... p . _ p b . _ . 00.“! O D D O a C 3093 n— .m 0.8.7 0 O a O O 9%9 0 n1 a On. DD 69. D 0% at... + D N joo. —. . *w W N N S * + ++ 4 N 1 + x 1.00.0 BWBVIBVA 'WOINONVO PUZ 103 Royale all tend to be very similar to one another as do samples that have been through the entire procedure. Samples from different years, tissues, silica gel, and charcoal column purification do not vary in any systematic fashion. Individual egg samples are similar to other tissues, while the pooled and purified samples are similar to the other samples of the same manipulation. All of the field samples are considerably different from the standards. The tentative conclusion from this analysis is that the samples are fairly hetenogeneous in the original state and that as more purification procedures are applied and samples are pooled for toxicologic testing, the residues become increasingly simi1ar. A second analysis was conducted using principal components, cluster and canonical discriminant analysis sequentially. Initially; a principal components analysis was done and the first six principal components, explaining 90% of the variation in the data, were entered into the clustering routine specifying a maximum of six clusters. A canonical discriminant analysis was done to maximize the between group correlations among the six clusters. Figure 15 is a plot of the first two canonical variables for each of the observations. The same observations that were made with the principal components analysis are evident here. The Lake Siskiwit samples are all similar and lie closest to the pooled test samples. Two of the observations are probable outliners from the rest. One is a pooled test sample of fillets from 1983 fish and one is an individual belly sample from a 1983 fish. Two charcoal column samples from fillets of 1984 fish also tend to be different from the rest. 104 It is interesting to note that both ana1yses provide essentially the same results. Many of these techniques can be sensitive to correlations among the variables so that using the first six principal components, which by definition are uncorrelated, for further analysis and obtaining the same results as with the original data, validates the assumption of non-significant correlation among the variables. In summary, the statistical pattern recognition techniques applied to this data provides the following conclusions: 1) The analytical system used for this study has a statistically identifiable variability among replicate injections of the same material, the source of which was not identified. 2) No differences in peak patterns were evident in belly flap samp1es, standardized for size and tissue type, from 1983-1985. 3) Samples from different tissues and different years were indistinguishable. 4) As samples were purified and pooled, they became increasingly similar. 5) Samples from Lake Siskiwit on Isle Roya1e were all simi1ar’to one another and 6) All of the types of field samples were different from the standards. CHAPTER 5 CAPILLARY GAS CHROMATOGRAPHY-MASS SPECTROMETRY OF PURIFIED TOXAPHENE RESIDUES INTRODUCTION Because of the complexity and dissimilarity between toxaphene and toxaphene like residues found hienvironmental samp1es,it;has been recommended that all suspected toxaphene residues be confirmed by mass spectrometric methods (Ribick et al., 1982; Jansson and Wideqvist 1983). While there have been a number of published reports on the mass spectrometry of toxaphene (Sa1eh, 1983; Budde and Eichelberger, 1977; Holmstead et al., 1974), it is clear that methane enhanced chemica1 ionization with negative ion detecton is the most sensitive (Johnson and wideqvist, 1983; Vaz and Blomkvist. 1985; Ribick et al., 1982). Under these conditions, toxaphene produces very simple spectra that are composed predominantly of ions and chlorine isotope clusters resulting from [M-ClJ- fragments. Swackhamer et al. (1986) and Jansson and Wideqvist (1983) have suggested using several specific ions related to toxaphene components with 5-10 chlorines as a sensitive, specific technique for detection of toxaphene. While this is a useful approach for confirmation and/or quantitation of residues, it suffers from a lack of fragmentation and thus provides little structural information. For isolation and identification of specific components, it is necessary to use the less sensitive, though fragmentation rich, electron impact mode. This chapter examines the characteristics of the purified toxaphene residues that were generated and toxicologically tested during this study. 106 MATERIALS AND METHODS Capi11ary GC-MS was performed using a Nermag R-lO-lOC quadrupole GC-Mass spectrometer operated under the following conditions: Column - DB-l (J and w Scientific) 30 m, 0.25 mm id, 0.25 u film. Splitless injection, temperature programmed from 50°C (1 min) to 190°C at 30°/min followed by 2°/min up to 260°C. Spectra were acquired in the negative ion detection chemica1 ionization mode (Methane 0.3 torr, source temperature 150°C) at the rate of approximately 2.5 scans/second. RESULTS AND DISCUSSION A comparison was made between the total ion current for technical toxaphene and a purified 1ake trout residue (Figure 16). With the exception of the peak at scan 867 due to cis-nonachlor (a chlordane component), there is a distinct similarity between the standard and the residue. Figures 17 and 18 are the ion chromotograms and total ion current for the standard and the lake trout residue, respective1y. Masses 343, 377, and 413 were chosen as representative of the major ion clusters resulting from toxaphene components with 7, 8, and 9 chlorines, respectively (See Table 11) (Swackhamer et al., 1986; Jansson and Wideqvist, 1983). The validity of this approach was verified by examining a mass spectrum of authentic toxicant A (C18) available in 107 108 Figure 16: Chromatograms of total ion current for standard toxaphene (bottom) and a purified lake trout residue from 1985 HOP). 109 00v“ 8"“ OOON 80a 80a oovu a Gown 08¢ 80 80 00v 8“ — . p . p p — p — p — p — p L n - . 1 «new nun ONO. H O MON— v '3‘.“ non“ 800" _ _ - n u u Nuuu 00I0~h VNN'QNn— INDO— 0'. n v ND—N. flumu O k , f uwka Dan“ M . OaNu 0 0 K8. now. I kuu nkflu k 0 OIU—P OOUflKuK' I300“ rl Figure 17: 110 Ion current chromatograms from a toxaphene standard. TIC = total, RIC = 343, 377, and 413 for masses 343, 377, and 413 respectively. These masses correspond to the major ion in the chlorine isotope cluster from [M-ClJ‘ fragments from structures with 7, 8, and 9 chlorines respective1y. 111 DO?" OONN 000" 000— 000. OOv— OONu 000— 000 000 00' DON _ b P P11 p p .L [b1 — P. . 1. . 1|- . 1h. — 1. P - I‘ll-ll“ ( .I 1 ‘1 1 11 111 I11 111 (ill a\ mafia NNO- '0"— 7 nth“ 1 06¢. ¢ a n—VIu—u 0,4411 I (1 m.n. nuhz\sti/\Jr\w11. O onhk “an. “N — .00“ I 0.". f mama N. — hunlU—I . , NON-NNN INOOu I 1 ‘1 1 -11 1 \\/\ «14 < N ..«_ .. n~.. .Hma « w . n h-h 97D 1 0.0 nvnlU—m Oflkvkn- INOO- I . 0&0 .nOu O 0 fl OON¢ monu N, — 'NNVONna IxOO— 112 Figure 18: Ion current chromatograms from a purified 1985 Lake Michigan 1ake trout residue. Profile descriptions are described in Figure 17. 113 009$ SN“ OOON 00” u 000 a 00'— OON — 000 a 000 80 00‘ DON P IIJWV-«Inu. 1L1 _ p r . 1—.<.