Enzyme-Linked Immunosorbent Assay Detection of Pyrrolizidine

Enzyme-Linked Immunosorbent Assay Detection of Pyrrolizidine Alkaloids: Immunogens Based on Quaternary Pyrrolizidinium Salts. David M. Roseman, Xiying...
0 downloads 0 Views 196KB Size
Bioconjugate Chem. 1996, 7, 187−195

187

Enzyme-Linked Immunosorbent Assay Detection of Pyrrolizidine Alkaloids: Immunogens Based on Quaternary Pyrrolizidinium Salts David M. Roseman, Xiying Wu, and Mark J. Kurth* Department of Chemistry, University of California, Davis, California 95616. Received August 10, 1995X

Polyclonal antibody-based enzyme-linked immunosorbent assays (ELISAs) were developed for the detection of retrorsine (1, 351 g/mol), monocrotaline (2, 325 g/mol), and retronecine (3, 155 g/mol) in the parts per billion (ppb) range. A set of three bifunctional linking arms (6-8) was synthesized. By N-alkylation of pyrrolizidine alkaloids (PAs) retrorsine, monocrotaline, and retronecine acetonide (9), six haptens (6.1, 6.2, 7.1, 7.2, 7.9, and 8.9) were synthesized and used to generate rabbit antisera. The resulting anti-retrorsine antiserum gave a 50% inhibition (I50) value of 0.9 ( 0.2 ppb for retrorsine with detection limits of 0.5-10 ppb. The same ELISA system also detected isatidine (4, retrorsine N-oxide) dihydrate (403 g/mol) with an I50 of 1 ppb and senecionine (5, 352 g/mol) with an I50 of 100 ppb. A second monocrotaline-based ELISA detected monocrotaline with an I50 of 36 ( 9 ppb 2 with detection limits of 5-500 ppb and shows no cross-reactivity with 1 or 5; this ELISA demonstrates the potential for the substrate-specific detection method. A third retronecine-based ELISA detects 3 with an I50 of 3000 ( 600 ppb (3 ( 0.6 ppm) and detection limits of 600-10000 ppb. None of these ELISAs cross-react with the structurally similar swainsonine (10) or lupinine (11) alkaloids. PAs were detected in extracts of Senecio vulgaris and Crotalaria retusa, but not in Lupinus spp., as a demonstration of the ELISA’s usefulness.

INTRODUCTION1

Pyrrolizidine alkaloids [cf. retrorsine (1), monocrotaline (2), and senecionine (5)] constitute a class of secondary plant metabolites of wide geographical and botanical (including Boraginaceae, Compositae, Gramineae, Leguminosae, Orchidaceae, Rhizophoraceae, Santalaceae, and Saptoaceae) distribution which are toxic to both humans and animals (Smith & Culvenor, 1981). Culvenor (1980) estimates that as many as 6000 species may be PAproducing species, accounting for 3% of all flowering plants. Many plants produce more than one PA, and these often co-occur with varying amounts of their corresponding N-oxide [cf. isatidine (4)]. Structural similarities in the >250 known PAs are striking; for example, the necine base retronecine (3) is found in alkaloids from 6 botanical families and 26 genera (Robins, 1982; Bull et al., 1968; McLean, 1970; Peterson & Culvenor, 1983). Collectively, PAs exhibit a broad range of cytotoxic/pathological actions, including hepatotoxic (Mattocks & Bird, 1983; Segall et al., 1985), pneumotoxic (Chesney et al., 1974), embryotoxic (Mattocks, 1986), mutagenic (Ames et al., 1973), carcinogenic (Kuhara et al., 1980; Allen et al., 1975; Svoboda & Reddy, 1972), and teratogenic (Keeler, 1983) effects. Chronic gastrointestinal (Hooper, 1974), cardiopulmonary (Chesney, 1973), and central nervous system (Hooper, 1974) disorders are * Author to whom correspondence should be addressed: Mark J. Kurth, Department of Chemistry, University of California, Davis, CA 95616. E-mail: [email protected]. Phone: (916)752-8192. Fax: (916)752-8995. X Abstract published in Advance ACS Abstracts, December 15, 1995. 1 Abbreviations: ELISA, enzyme-linked immunosorbent assay; ppb, parts per billion; PAs, pyrrolizidine alkaloids; I50, 50% inhibition; Tris, tris(hydroxymethyl)aminomethane; IgG-HRP, goat anti-rabbit IgG antibody-horseradish peroxidase conjugate; OPD, o-phenylenediamine; BSA, bovine serum albumin; OVA, chicken egg ovalbumin; KLH, keyhole limpet hemocyanin; DCC, dicyclohexylcarbodiimide; NHS, N-hydroxysuccinimide; DME, 1,2-dimethoxyethane; DIBAL-H, diisobutylaluminum hydride; PCC, pyridinium chlorochromate; TLC, thin-layer chromatography; PBST, phosphate-buffered saline-Tween.

1043-1802/96/2907-0187$12.00/0

Figure 1. PAs used as analytes in this study.

further manifestations of PA poisoning. Although the toxic response depends on both the chemical nature of the PA and the animal species involved, the development of liver disease (veno-occlusive) is a common occurrence. Because of these issues, there is an urgent need for a reliable, rapid, and sensitive method for identification/ quantification of PAs and PA metabolites in biological/ environmental samples. In previous reports, we detailed several strategies for detection of PAs. The first and more speculative strategy involved direct detection of the retronecine substructure in intact parent PAs (Bober et al., 1990; Kurth et al., 1992), while the second strategy involved a somewhat less direct assay but one which capitalized on the fact that hydrolysis of any retronecinebased macrocyclic diester PA liberates the necine base retronecine (e.g., saponification of monocrotaline yields monocrotalic acid and retronecine). This “class-specific” PA ELISA detects retronecine at an I50 value of 11 ( 3 ppb, with detection limits of 1.0-100 ppb (Kurth et al., 1989; Roseman et al., 1992). Free base PAs are readily N-alkylated by methyl iodide in ethereal solvent to produce the N-methylated salt as © 1996 American Chemical Society

188 Bioconjugate Chem., Vol. 7, No. 2, 1996

a precipitate (Leonard, 1960). In this paper, we report that primary functionalized alkyl halides can be used to N-alkylate PAs and that the resulting quaternary ammonium alkaloids can be conjugated to a carrier protein. The development of ELISAs based on these N-alkylated quaternary ammonium PAs is reported. EXPERIMENTAL SECTION

Equipment. ELISAs were carried out on 96-well Immulon 2 microtiter plates (Dynatech) and read with a UVmax microtiter plates reader (Molecular Devices) at 490 nm. Nuclear magnetic resonance spectra were taken on a QE-300 NMR spectrometer (General Electric) with 1H spectra at 300 MHz and 13C spectra at 75.6 MHz. Materials. Senecionine, Senecio vulgaris whole plant, Crotalaria retusa seeds, and Lupinus spp. whole plant samples were a gift from Dr. H. J. Segall (Veterinary Medicine Pharmacology and Toxicology, University of California, Davis). (-)-Lupinine hydrochloride was purchased from Research Plus. Monocrotaline was purchased from Trans World Chemicals. Solvents and inorganic reagents were purchased from Fisher Scientific. Swainsonine, isatidine (retrorsine N-oxide), Tween-20, Tris base, citric acid, goat anti-rabbit IgG antibodyhorseradish peroxidase conjugate, o-phenylenediamine, Fruend’s complete and incomplete adjuvants, bovine serum albumin, chicken egg ovalbumin, and keyhole limpet hemocyanin were purchased from Sigma Chemical Co. Retrorsine, dicyclohexylcarbodiimide, and N-hydroxysuccinimide were purchased from Aldrich Chemical Co. Dialysis tubing was purchased from Spectrum Medical Industries. 1,2-Dimethoxyethane was distilled under a nitrogen atmosphere from sodium/potassium benzophenone ketyl immediately prior to use. Retronecine (3). In a 250 mL round bottom flask equipped with a reflux condenser and a magnetic stirrer, monocrotaline (6.00 g, 18.4 mmol), Ba(OH)2‚8H2O (15.00 g, 47.55 mmol), and water (125 mL) were combined and heated at reflux for 3 h (with constant bubbling of N2 through the solution). The mixture was cooled to room temperature and saturated with solid carbon dioxide. The barium carbonate was removed by suction filtration through Celite, and the clear filtrate was concentrated under vacuum to 25% of its original volume. The aqueous concentrate was basified (pH 10) with Na2CO3, saturated with NaCl, and continuously extracted with ether for 72 h. The organic extract was concentrated under reduced pressure and evacuated at 1 Torr overnight to give a light brown solid. Decolorization with activated carbon and recrystallization from acetone gave retronecine (2.05 g, 13.2 mmol, 72%) as white prisms: mp 120-121 °C; 1H NMR (300 MHz, CDCl3) δ 1.89-2.02 (2H, complex, NCH2CH2), 2.75 (1H, m, NCHHCH2), 3.22 (1H, m, NCHHCH2), 3.42 (1H, m, NCHHCHd), 3.85 (1H, m, NCHHCHd), 4.06-4.15 (2H, complex, NCH and CHHOH), 4.27-4.31 (2H, complex, CHOH and CHHOH), 5.70 (1H, s br, vinyl); IR (KBr) 3321 (OH) cm-1. N-[(4-Bromocrotonyl)oxy]succinimide (6). The NHS ester of 4-bromocrotonic acid was formed by the method of Anderson et al. (1964). Dry 3-bromocrotonic acid (0.3025 g, 1.83 mmol) was dissolved in freshly distilled DME (2.6 mL) in an oven-dried 10 mL round bottom flask equipped with a flea-size stirring bar, a septum, and a nitrogen inlet. The mixture was chilled in an ice bath, and a DME solution of DCC (0.24 M in DCC, 8.6 mL, 2.0 mmol) (Aldrich; caution, sensitizer!) was injected. After standing overnight at 4 °C, the dicyclohexylurea was removed by filtration (medium glass frit with Celite). The remaining liquid was reduced by rotary evaporation to reveal a yellow crystalline solid, which after drying in vacuo, afforded the crude NHS ester

Roseman et al.

(0.49 g) which recrystallized from 2-propanol (10 mL) to yield pale yellow needles (0.328 g, 1.25 mmol, 68%) of NHS ester 6: mp 129-131 °C; 1H NMR (300 MHz, CDCl3) δ 2.86 (s, 4H, COCH2CH2CO), 4.06 (d, 2H, J ) 7 Hz, CH2Br), 6.26 (d, 1H, J ) 15 Hz, CHCHCO), 7.31 (m, CHCHCO); 13C NMR (75 MHz, CDCl3 + DMSO-d6) δ 24.4 (COCH2CH2CO), 27.7 (BrCH2), 117.3 (CHCHCO2), 146.5 (CHCHCO2), 159.5 (CHCHCO2), 168.2 (COCH2CH2CO); IR (KBr) 1642 (NCdO), 1734 (OCdO) cm-1. Anal. Calcd for C8H8BrNO4: C, 36.67; H, 3.08; Br, 30.49; N, 5.34. Found: C, 36.60; H, 3.01; Br, 30.52; N, 5.28. N-[[p-(r-Bromomethyl)benzoyl]oxy]succinimide (7). N-Hydroxysuccinimide (4.33 g, 37.6 mmol) and p-(Rbromomethyl)benzoic acid (8.04 g, 37.4 mmol) were combined in a 500 mL round bottom flask, and the vessel was charged with N2. These solids were dissolved in dry DME (250 mL), and the solution was chilled in an ice bath. A DME solution (30 mL) of DCC (7.90 g, 38.3 mmol) was added by cannula, and the resulting solution was stirred overnight at 5 °C under N2. The solution was then warmed to room temperature, and the dicyclohexylurea was removed by filtration. The DME was removed under reduced pressure, and the crude solid was recrystallized from 2-propanol to provide rhombic white crystals (10.81 g, 34.6 mmol, 93%): mp 160-162 °C; 1H NMR (300 MHz, CDCl3) δ 2.86 (s, 4H, COCH2CH2CO), 4.59 (s, 2H, ArCH2Br), 7.60 (d, 2H, J ) 8.2 Hz, meta), 8.08 (d, 2H, J ) 8.2 Hz); 13C NMR (75 MHz, CDCl3 + CD3CN) δ 26.3 (COCH2CH2CO), 32.6 (ArCH2Br), 125.5 (ipso), 130.3 (ortho), 131.3 (meta), 141.6 (para), 162.2 (ArCO2N), 170.5 (COCH2CH2CO); IR (KBr) 1732 (CdO) cm-1. Anal. Calcd for C12H10BrNO4: C, 46.18; H, 3.23; Br, 25.60; N, 4.49. Found: C, 45.99; H, 3.12; Br, 25.43; N, 4.40. 4-(Bromomethyl)benzaldehyde (8). DIBAL-H (140 mL, 140 mmol) was added under argon over 10 min to a -70 °C solution of ethyl R-bromotoluate (12.07 g, 49.7 mmol) in toluene (20 mL). The reaction mixture was stirred for 1 h at -70 °C and at room temperature for 2 h, and then the reaction was quenched with the addition of 1 M HCl (10 mL). The mixture was taken up in ether (50 mL) and washed with 1 M HCl (3 × 30 mL), and the aqueous washes were extracted with ether (3 × 90 mL). The combined organic layers were dried over sodium sulfate and concentrated under reduced pressure. Drying under high vacuum delivered oily, low melting translucent plates of 4-(bromomethyl)benzyl alcohol (8.08 g, 40.2 mmol, 81%): mp 76-77 °C; 1H NMR (300 MHz, CDCl3) δ 3.24 (1H, s, OH), 4.43 (2H, s, CH2Br), 4.50 (2H, s, CH2OH), 7.20 (2H, ABq, J ) 8.1 Hz), 7.30 (2H, ABq, J ) 8.1 Hz); 13C NMR (75 MHz, CDCl3) δ 33.3 (CH2Br), 64.2 (CH2OH), 127.1 (Ar), 129.0 (Ar), 126.8 (CH2Br), 141.0 (CH2OH); IR (neat) 3250 (OH) cm-1. A slurry of PCC (4.37 g, 20 mmol, 1 equiv) in CH2Cl2 (40 mL) was added in 5 mL portions over 5 h to 4-(bromomethyl)benzyl alcohol (6.06 g, 30 mmol) in CH2Cl2 (100 mL) under argon. The reaction was monitored by silica TLC (20:80 EtOAc:hexane), and after 2 h, the mixture was filtered through Celite in a fritted glass funnel. Flash column chromatography gave 8 as a white crystalline material which sublimated on standing (used without further purification): mp 93.5-95 °C; 1H NMR (300 MHz, CD2Cl2) δ 4.51 (CH2Br), 7.56 (2H, ABq, J ) 8 Hz), 7.86 (2H, ABq, J ) 8.2 Hz), 10.01 (CHO); 13C NMR (75 MHz, CD2Cl2) δ 32.6 (CH2Br), 130.1 (CH, Ar), 130.4 (CH, Ar), 136.8 (CCH2Br), 144.8 (CCHO), 191.8 (CHO); FTIR (NaCl plates) 1693 (CdO) cm-1. N-[[(Succinimid-3-yloxycarbonyl)ethenyl]methyl]retrorsine, Bromide Salt (6.1). Retrorsine (408 mg, 1.16 mmol) was dissolved in dry DME (12 mL) and treated with a DME (10 mL) solution of bromoester 6 (355 mg, 1.35 mmol). After 2 min, stirring was discon-

ELISA Detection of Pyrrolizidine Alkaloids

tinued, and the solution was allowed to stand overnight. A fine white powder of 6.1 appeared at the bottom of the vessel which was collected by filtration and washed with fresh DME (6 mL), dried under high vacuum in a heated vessel, and used without further purification (670 mg, 1.09 mmol, 94%): mp 115-120 °C dec; 1H NMR (300 MHz, CD3CN) δ 0.84 (3H, d, J ) 6.5 Hz, H-19 CH3), 1.73 (1H, m, H-13 CHCH3), 1.79 (1H, m, H-14b CHH), 1.85 (3H, dd, Jvic ) 7.1 Hz, Jallyl ) 1.3 Hz, H-21 CH3), 2.27 (1H, d, Jgem ) 13.2 Hz, H-14a CHH), 2.80 (2H, m, H-6 CH2), 2.87 (4H, s, 3′′- and 4′′-COCH2CH2CO), 3.65 (2H, s br, H-18 CH2), 3.84 (1H, m, H-5b CHH), 4.92 (1H, m, H-5a CHH), 4.30 (1H, d, Jgem ) 12.3 Hz, H-9b CHH), 4.51 (1H, d, J ) 9 Hz, H-3b NCHH), 4.53 (1H, m, H-3a NCHH), 4.54 (2H, s br, NCH2CHdCHCO2), 5.25 (1H, m, H-8 NCH), 5.46 (1H, d, Jgem ) 12.1, H-9a CHH), 5.48 (1H, m, H-7 OCH), 5.90 (1H, dq, Jq ) 6.7 Hz, Jallyl ) 0.8 Hz, H-20 CHCH3), 6.33 (1H, s br, H-2), 6.77 (1H, d, J ) 15.5 Hz, H-2′ CH2CHdCHCO2), 7.28 (1H, dt, Jd ) 15.5 Hz, Jt ) 7.2 Hz, H-3′ CH2CHdCHCO2); IR (KBr) 1559 (NCdO), 1653 (CCdO), 3746 (OH) cm-1. Anal. Calcd for C26H33BrN2O10: C, 50.91; H, 5.42; Br, 13.03; N, 4.57. Found: C, 50.80; H, 5.40; Br, 12.58; N, 4.57. N-[[2-(Succinimid-3-yloxycarbonyl)ethenyl]methyl]monocrotaline, Bromide Salt (6.2). Monocrotaline 2 (253.7 mg, 0.7806 mmol) was dissolved in dry DME (8 mL) and treated with a DME (6 mL) solution of bromo ester 6 (224.2 mg, 0.8558 mmol). After 2 min, stirring was discontinued, and the solution was allowed to stand overnight. A fine white powder of 6.2 appeared at the bottom of the vessel which was collected by filtration and washed with fresh DME (4 mL), dried under high vacuum in a heated vessel, and used without further purification (403.3 mg, 0.6871 mmol, 88%): mp 120-123 °C dec; 1H NMR (300 MHz, CD3CN) δ 1.21 (3H, d, J ) 7.3 Hz, H-17), 1.30 (3H, s, H-18), 1.44 (3H, s, H-14), 2.40 (1H, m, H-6 trans), 2.68 (1H, m, H-6 cis), 2.86 (4H, s, H-3′′ and H-4′′), 3.23 (1H, q, J ) 7.1 Hz, H-19), 3.89 (1H, m, H-5 trans), 4.47 (1H, m, H-5 cis), 4.55 (2H, m, H-4′), 4.59 (1H, m, H-3 trans), 4.74 (1H, m, H-3 cis), 4.90 (2H, m, H-8 and H-9 cis), 5.19 (1H, d br, J ) 7.4 Hz, H-7), 5.36 (1H, q br, J ) 6.8 Hz, H-9 trans), 6.28 (1H, s br, H-2), 6.71 (1H, d, J ) 16 Hz, H-2′), 7.28 (1H, dt, Jvinylic ) 15 Hz, Jvicinal ) 7.1 Hz, H-3′); IR (KBr) 1655 (NCdO), 1730 (CCdO), 3391 (OH) cm-1. Anal. Calcd for C24H31BrN2O10: C, 49.07; H, 5.32; Br, 13.60; N, 4.77. Found: C, 48.82; H, 5.32; Br, 13.08; N, 4.62. N-[[(Succinimid-4-yloxycarbonyl)phenyl]methyl]retrorsine, Bromide Salt (7.1). Retrorsine (396 mg, 1.13 mmol) was dissolved in dry DME (12 mL) and treated with a DME (10 mL) solution of bromo ester 7 (530 mg, 1.31 mmol). After 2 min, stirring was discontinued, and the solution was allowed to stand overnight. A fine white powder of 7.1 appeared at the bottom of the vessel which was collected by filtration and washed with fresh DME (6 mL), dried under high vacuum in a heated vessel, and used without further purification (688 mg, 0.912 mmol, 81%): mp 170-174 °C dec; 1H NMR (300 MHz, CD3OD) δ 0.82 (3H, d, J ) 6.5 Hz, H-19), 1.71 (1H, m, H-13), 1.82 (1H, m, H-14b), 1.85 (3H, dd, Jvicinal ) 7.0 Hz, Jhomoallylic ) 1 Hz, H-21), 2.22 (1H, d, J ) 12.7 Hz, H-14a), 2.68 (2H, m, H-6), 2.93 (4H, s, H-3′′ and H-4′′), 3.65 (2H, s, H-18), 3.85 (1H, m, H-5a), 4.20 (1H, d, Jgem ) 12 Hz, H-9b), 4.23 (1H, m, H-3b), 4.46 (1H, d, J ) 17 Hz, H-3a), 4.80 (1H, s, H-6′ NCH2Ar), 4.83 (1H, d, J ) 8 Hz, H-7), 5.17 (1H, d, Jgem ) 12 Hz, H-9a), 5.43 (2H, dt, Jd ) 10 Hz, H-8), 5.90 (1H, dq, Jvicinal ) 6.7 Hz, Jallylic ) 0.6 Hz, H-20), 6.12 (1H, s br, H-2), 7.95 (2H, ABq, J ) 8.3 Hz, H-4′), 8.23 (2H, ABq, J ) 8.3 Hz, H-3′); IR (KBr) 1740 (CdO), 1772 (CdO), 3377 (OH) cm-1. Anal. Calcd

Bioconjugate Chem., Vol. 7, No. 2, 1996 189

for C30H35BrN2O10: C, 54.31; H, 5.32; Br, 12.04; N, 4.22. Found: C, 53.83; H, 5.28; Br, 11.81; N, 4.05. N-[[(Succinimid-4-yloxycarbonyl)phenyl]methyl]monocrotaline, Bromide Salt (7.2). Monocrotaline 2 (411 mg, 1.26 mmol) was dissolved in dry DME (12 mL) and treated with a DME (10 mL) solution of bromo ester 7 (533 mg, 1.32 mmol). After 2 min, stirring was discontinued, and the solution was allowed to stand overnight. A fine white powder of 7.2 appeared at the bottom of the vessel which was collected by filtration and washed with fresh DME (6 mL), dried under high vacuum in a heated vessel, and used without further purification (751 mg, 1.03 mmol, 82%): mp 177-179 °C dec; 1H NMR (300 MHz, CD3OD) δ 1.21 (3H, d, J ) 7.0 Hz, 3H, C-19 Me), 1.29 (3H, s, C-17), 1.43 (3H, s, C-18), 2.38 (1H, m, H-6b), 2.66 (1H, m, H-6a), 2.90 (4H, s, H-3′′ and H-4′′ COCH2CH2CO), 3.23 (1H, q, J ) 7.2 Hz, H-14), 3.82 (1H, m, H-5a), 3.94 (1H, m, H-5b), 4.61 (1H, d, J ) 12.1 Hz, H-9a), 4.34 (1H, d, J ) 12.1 Hz, H-9b), 4.45 (1H, d, J ) 17.2 Hz, H-3b), 4.71 (1H, d, J ) 16.6 Hz, H-3a), 4.87 (2H, s, H-6 NCH2Ar), 5.36 (1H, t, J ) 7.4 Hz, H-8), 5.41 (1H, q, J ) 7.5 Hz, H-7), 6.11 (1H, s br, H-2), 7.93 (2H, d, J ) 8.3 Hz, H-4′ Ar, meta), 8.23 (2H, d, J ) 8.3 Hz, H-3′ Ar, ortho); IR (KBr) 1739 (CdO), 1779 (CdO), 3370 (OH) cm-1. Anal. Calcd for C28H33BrN2O10: C, 52.76; H, 5.22; Br, 12.53; N, 4.39. Found: C, 52.35; H, 5.23; Br, 12.23; N, 4.32. Retronecine Acetonide (9). Retronecine (209 mg, 1.35 mmol) was partially dissolved in acetone (4.7 g, 81 mmol), and 2,2-dimethoxypropane (12.6 g, 121 mmol, predistilled from sodium) and methanesulfonic acid (180 mg, 1.87 mmol, 1.4 equiv) were added. The mixture was stirred overnight under argon and the reaction quenched with potassium carbonate (990 mg, 7.15 mmol). Acetone and dimethoxypropane were removed under reduced pressure, and the resulting crude product was taken up in water (120 mL), extracted with equal volumes of ether (3 × 120 mL), and dried over MgSO4. Concentration under reduced pressure and then high vacuum delivered 9 as a pale yellow oil (190.2 mg, 975 mmol, 72%): 1H NMR (300 MHz, CDCl3) δ 1.29 (3H, s, CH3), 1.32 (3H, s, CH3), 1.72 (1H, dd, J ) 13 Hz, J ) 5 Hz, NCH2CHH), 1.88 (1H, m, NCH2CHH), 2.50 (1H, m, NCHHCH2), 3.02 (1H, t, J ) 8 Hz, NCHHCH2), 3.37 (1H, m, NCHHCHd), 3.64 (1H, m, NCHHCHd), 4.21 (1H, d, Jgem ) 15 Hz, CHHOC(CH3)2), 4.30 (2H, s br, CHOC(CH3)2), 4.38 (2H, t, J ) 5 Hz, NCH), 4.47 (1H, d, Jgem ) 15 Hz, CHHOC(CH3)2), 5.32 (1H, s br, vinyl); FTIR (NaCl plates) 507, 515, 521, 542, 580, 617, 668, 733, 768, 876, 885, 951, 995, 1283, 1314, 1447, 1619, 3054 cm-1. N-[[(Succinimid-4-yloxycarbonyl)phenyl]methyl]retronecine Acetonide, Bromide Salt (7.9). Acetonide 9 (100 mg, 0.513 mmol) was dissolved in dry DME (2 mL), and bromo ester 7 (161 mg, 0.516 mmol) in DME (2 mL) was injected with rapid stirring. After 2 min, stirring was discontinued, and the solution was allowed to stand overnight. Fine white needles appeared at the bottom of the vessel which were collected by filtration, washed with fresh DME (2 mL), and dried under high vacuum in a heated vessel to give 7.9 (248 mg, 0.489 mmol, 95%) (used without further purification): mp 142-147 °C dec; 1H NMR (300 MHz, CD3CN) δ 1.33 and 1.38 (2C, CH3CCH3), 2.21 (2H, s), 2.50 (1H, septet, J ) 7.0 Hz), 2.86 (2H, d, J ) 0.9 Hz), 3.28 (2H, d, J ) 1.4 Hz), 3.45 (2H, d, J ) 1.3 Hz), 3.93 (1H, dt, J ) 0.n Hz, J ) 0.n Hz), 4.19 (2H, m), 4.46 (1H, d, J ) 16 Hz), 4.66 (1H, dm, J ) 15 Hz), 4.90 (1H, m), 4.98 (2H, s), 5.39 (1H, s br), 5.50 (1H, d, J ) 5.4 Hz), 7.92 (2H, ABq, J ) 7.5 Hz, Ar, meta), 8.20 (2H, ABq, J ) 7.3 Hz, Ar, ortho); IR (KBr) 1738 (NCdO), 1771 (CCdO) cm-1.

190 Bioconjugate Chem., Vol. 7, No. 2, 1996

N-[(4-Formylphenyl)methyl]retronecine Acetonide, Bromide Salt (8.9). Acetonide 9 (300 mg, 1.53 mmol) was dissolved in dry DME (3 mL), and (bromomethyl)benzaldehyde 8 (379 mg, 1.90 mmol) in DME (3 mL) was injected with rapid stirring. After 2 min, stirring was discontinued, and the solution was allowed to stand overnight. Fine white needles appeared at the bottom of the vessel which were collected by filtration, washed with fresh DME (2 mL), and dried under high vacuum in a heated vessel to give 8.9 (248 mg, 0.489 mmol, 95%) (used without further purification): mp 187190 °C; 1H NMR (300 MHz, CDCl3) δ 1.33 and 1.40 (6H, s, CH3CCH3), 2.06 (2H, complex, NCH2CH2), 2.75 (1H, m, NCHHCH2), 3.55 (1H, m, NCHHCH2), 4.20 (1H, d, Jgem ) 17 Hz, CHHOC(CH3)2), 4.21 (1H, s br, CHOC(CH3)2), 4.40 (1H, d, Jgem ) 16 Hz, CHHOC(CH3)2), 5.01 (2H, m, NCH2CHd), 5.35 (1H, s br, vinyl), 5.57 and 5.58 (2H, s br, ArCH2), 5.61 (1H, d, J ) 5 Hz, NCH), 7.85 (2H, d, J ) 8.2 Hz, Ar, meta), 7.98 (2H, d, J ) 8.2 Hz, Ar, ortho), 10.01 (1H, s, CHO); IR (KBr) 1742 (CdO) cm-1. Protein Conjugation (General Procedure). Protein was dissolved in phosphate-buffered saline (PBS, pH 7.4) in an oven-baked round bottom flask equipped with a stirring bar. The protein solution was cooled in an ice bath, quaternary ammonium ester was added and the mixture stirred overnight at 5 °C. The solution was dialyzed (molecular mass cutoff 100 000 kDa) repeatedly against PBS (1000 mL) at 5 °C to remove any uncoupled hapten. Dialysate aliquots of hapten-protein conjugate were stored in glass vials at -20 °C until required for immunochemical studies. Following this general procedure, retrorsine-BSA conjugate (6.1bsa) was prepared from BSA (60 mg), PBS (12 mL), and quaternary ammonium ester 6.1 (65 mg, 106 µmol), monocrotaline-BSA conjugate (6.2bsa) was prepared from BSA (40 mg), PBS (12 mL), and quaternary ammonium ester 6.2 (45 mg, 76.3 µmol), retrorsine-OVA conjugate (7.1ova) was prepared from OVA (120 mg), PBS (12 mL), and quaternary ammonium ester 7.1 (82 mg, 109 µmol), and monocrotaline-OVA conjugate (7.2ova) was prepared from OVA (30 mg), PBS (12 mL), and quaternary ammonium ester 7.2 (78.7 mg, 108 µmol). Retronecine-KLH Conjugate (7.3klh). KLH (15 mg) was weighed into an oven-dried 10 mL round bottom flask equipped with an oven-dried stir bar and dissolved in 0.2 M pH 8.5 borate buffer (5 mL). Quaternary salt 7.9 (25 mg, 49 µmol) was dissolved with gentle stirring, and the reaction was quenched after 5 h by the dropwise addition of 0.1 M pH 3.0 phosphate buffer. The pH was adjusted to 3 with 0.1 M HCl, and the reaction mixture was stirred at room temperature for 5 h. Equilibration to pH 7 with 0.1 N NaOH and dialysis at 5 °C in a 100 kDa molecular mass cutoff dialysis tubing against PBS (9 × 1000 mL, 8 h each) produced the 7.3klh protein conjugate which was removed (8 mL, 1.9 mg/mL), aliquotted, and stored in a freezer for later use in immunochemical studies. Retronecine-OVA Conjugate (8.3ova). OVA (50 mg) was weighed into an oven-dried 10 mL round bottom flask equipped with an oven-dried stir bar and dissolved in PBS (3 mL) plus methanol (1 mL). Quaternary salt 8.9 (21 mg, 55 µmol) was dissolved with gentle stirring, and the mixture was stirred for 5 h at room temperature. NaBH4 (4 mg, 100 µmol) was added and the mixture stirred for 1 h, at which time the reaction was quenched with the dropwise addition of 0.1 M pH 3.0 phosphate buffer. Stirring was continued at room temperature at pH 3 for 5 h. The mixture was equilibrated to pH 7 with 0.1 N NaOH and dialyzed in a 100 kDa molecular mass cutoff dialysis tubing against PBS (9 × 1000 mL, 8 h each) at 5 °C. Protein conjugate 8.3ova was removed (6.5

Roseman et al.

mL, 3.2 mg/mL), aliquotted, and stored in a freezer for later use in immunochemical studies. Rabbit Immunization. Three-month-old New Zealand white rabbits were immunized subcutaneously with 100 µg of immunogens 6.1bsa (rabbits 1323 and 2206), 6.2bsa (rabbits 2203 and 2205), or 7.3klh (rabbits 2376 through 2380) emulsified in Freund’s complete adjuvant (1:1), followed later by three injections of 100 µg of immunogen in Freund’s incomplete adjuvant at 3-week intervals. Blood samples were collected 10 days after each booster injection and evaluated by ELISA titration against a homologous coating and analyte. Serum was stored at -20 °C. General ELISA. Basic solid phase immunoassay principles were followed (Laurent, 1988) in performing ELISAs in 96-well microtiter plates (Immulon 2). Thus, microtiter plates were coated with OVA conjugates [100 µL/well of a 0.05 M carbonate buffer (pH 10) solution containing 1 µg/mL of 7.1ova, 7.2ova, or 8.3ova] by storing the plates at 4 °C for 2 h. The plates were washed (4×) with washing buffer (9 g/L NaCl and 0.5 mL/L Tween20 with 2 mM pH 7.4 phosphate buffer) and blotted dry. Each washing was accomplished with a hand-held 12channel plate washer. Appropriate antisera dilutions in PBST (PBS containing 120 mM NaCl, 2.7 mM KCl, and 10 mM inorganic phosphate at pH 7.4 plus 0.045% v/v Tween-20) were prepared from previously frozen samples of rabbit antisera, and after preincubation with standards at 20 °C for 2 h, 50 µL aliquots were pipetted into each well and the plates were allowed to stand covered at 20 °C for 2 h. Washing 4 times with washing buffer was followed by the addition of 50 µL/well of commercial IgGHRP conjugate (1.0 mg/mL) diluted 1:2000 with PBST. The plates were then incubated for 1 h at 20 °C and washed with washing buffer. A 50 µL/well solution consisting of OPD (1 mg/mL) and hydrogen peroxide (30% diluted 1:1000) in a Tris-citrate buffer (24.2 g/L of Tris base and 12.6 g/L of citric acid monohydrate adjusted to pH 6.0) base was added to the plate. The color change was monitored with an automated plate reader. When using the kinetic format, the rate of color change (mOD/min) at 450 nm was recorded over a 15 min period. When employing the end point format, the reaction was stopped after 5 min by the addition of 50 µL/well 2 N H2SO4 and the final optical density (OD490) was measured at 490 nm. Some initial screenings were performed with the kinetic format, while final characterizations were performed with the end point format. Nonspecific Binding Controls. As a control, each microtiter plate had 12 wells free of coating antigen, having contained only the coating buffer during the coating procedure. To the first 3 wells was added only OPD, to the second 3 wells were added both IgG-HRP and OPD, and to the last six wells were added antisera, IgG-HRP, and OPD. Screening of Antisera by Testing Their Titers: Checkerboard Titrations. Optimum dilution factors for both the antiserum and coating antigen used in each assay were determined by checkerboard titrations. Microtiter plates were coated by overnight incubation with coating antigen solutions (7.1ova, 7.2ova, or 8.3ova at 100 µL/well) ranging from 0.1 to 18.85 µg/mL. Next, solutions of antisera (50 µL/well) with dilution factors ranging from 1:250 to 1:32000 were added to each well. Those coating and antisera concentrations giving optical densities at 490 nm (OD490’s) between 0.9 and 1.3 after 5 min were selected for use in initial competitive ELISA screens. Competitive ELISA with PA Standards. For standard curves with less sensitive antisera (serum numbers 1323 and 2205), a primary standard of 10.00 mg/mL of

ELISA Detection of Pyrrolizidine Alkaloids Scheme 1. Antigens Prepared in This Studya

a

For a note regarding compound numbering, see footnote 2.

PAs was used to make a series of diluted standards in PBST buffer with concentrations ranging from 500 µg/ mL (500 ppm) to 0.1 ng/mL (0.1 ppb). For more sensitive antisera (2203 and 2206), PA stock solutions in 10 mM PBST were diluted from 100 000 to 0.01 ppb in 1:10 dilution steps. An antisera solution of an appropriate dilution for the particular coating antigen to be used was prepared by dilution in PBST. Samples of each PA standard (1 part) were diluted 1:10 with antisera solution (9 parts) in new, uncoated microtiter plates and preincubated at room temperature for 2 h. One standard, containing only antisera and buffer (1:10), was used as the control to determine the maximum kinetic or end point OD reading. Another sample containing only buffer was used as a blank. Inhibition curves were analyzed by a four-parameter logarithmic curve-fitting procedure which calculated I50 values (inhibitor concentration giving 50% inhibition) and the slope of the linear portion of the curve. Below 5% solvent content, methanol did not show any effect on antibody activity. Comparative Inhibition Calculations. Three coating antigens (7.1ova, 7.2ova, or 8.3ova) and six antisera combinations were evaluated in assays using PAs 1-5 as analytes. Initial screens were performed using PA standards decreasing by factors of 10 from a concentration value of 100 000 ng/mL (100 ppm) to 0.1 ng/mL (0.1 ppb). For those combinations where retronecine showed a promising inhibition range, additional standards were prepared and run in the range where inhibition curves displayed nearly linear behavior. I50 values were determined for these systems. Detection of PAs in Plant Samples. S. vulgaris and C. retusa were selected as potential positive samples with Lupinus spp. selected as a negative control. Whole Senecio and Lupinus plants where ground in a blender with methanol and refluxed for 1 h with nitrogen bubbled through. Crotalaria seeds were pulverized and refluxed with methanol as above. Solids were removed by filtration through glass wool, and the methanolic extract was dialyzed through a 10 000 molecular mass cutoff tubing into 2 volumes of PBS. The dialysates were assayed with the ELISA systems A-C. Cross-Reactivity Studies with Swainsonine and Lupinine. Retronecine standard curves were also run in the presence of commercially prepared swainsonine

Bioconjugate Chem., Vol. 7, No. 2, 1996 191 Scheme 2. Preparation of Retronecine Antigensa

a Protein conjugation conditions: (i) 7.9 f 7.9klh; KLH + borate buffer, pH 8.5; (ii) 8.9 f 8.9ova; OVA + PBS, pH 7.4, 4 h, 20 °C; 2 equiv of NaBH4, MeOH.

and lupinine hydrochloride at working concentrations of 2500 ppb 10 or 11. RESULTS AND DISCUSSION

Hapten Preparation. The N-4 free base of the necine-based PA is a good nucleophile and has been used in the past to form methiodides (Leonard, 1960). While N-4 is unreactive toward straight chain primary halides such as 6-iodohexanoate in DME, we find that necinebased PAs are reactive toward allylic and benzylic bromides as well as R-bromo esters in DME. Thus, treating a PA with an active electrophile such as 6, 7, or 8 in DME produces a pyrrolizidinium salt as a fine white (flocculant) precipitate which is easily recovered on a fine frit (Scheme 1). Crotyl NHS ester 6 was selected as our first linker target for two reasons. First, 4-bromocrotonate 6 proved to be sufficiently reactive to N-alkylate PAs. Second, hapten carrier tether length is an important variable in small molecule immunoreactivity since it can profoundly affect the hapten’s presentation to the immune system. In this regard, the four-carbon crotyl tether2 in immunizing antigen 6.Ypro seemed ideal (Goodrow et al., 1990; Harrison et al., 1990). However, in order to avoid “tether recognition” in the ELISA, benzoate hapten 7.Y, the N-alkylation product from reaction of NHS ester 7 with PAs, was targeted because the resulting coating antigen 7.Ypro would present a nonhomologous tether (methylbenzoate) from that of immunogen 6.Ypro (crotyl). Like crotyl NHS ester 6, NHS ester 7 proved sufficiently reactive to N-alkylate the parent PAs. 2 Note regarding compound numbering. Composite compound numbers presented as X‚Y are used to convey the fact that a linker moiety (X ) linker compound number) and a pyrrolizidine alkaloid (Y ) PA compound number) are covalently attached via N-4 of the PA. Thus, for example, 7.9 is the hapen derived from linker 7 and PA analogue 9. Likewise, superscripted pro, bsa, klh, or ova following a composite compound number X‚Yzzz conveys the linker (X), PA (Y), and carrier protein (zzz; protein, bovine serum albumin, keyhole limpet hemocyanin, or chicken egg ovalbumin, respectively).

192 Bioconjugate Chem., Vol. 7, No. 2, 1996 Table 1.

13C

Roseman et al.

NMR Data for PA Systemsa retrorsines

monocrotalines

retronecines

carbon

1b

6.1

7.1

2b

6.2

7.2

3b

9

7.9

8.9

1 2 3 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 1′ 2′ 3′ 4′ 5′ 6′ 2′′ 3′′

132.4 134.7 62.7 52.9 37.9 75 77.4 61 175.7 81.3 35.7 34.7 131.2 167.3 66.9 11.6 136.6 14.9 -

125.8 132.4 72 64.4 38.5 74 87.6 63.6 175.3 82.3 36.5 33.3 130.7 167 67.6 11.7 137 15.5 58.5 140.6 131.9 161.2 170.8 26.3

127.7 134.3 70.9 67 38.5 74.3 86.5 63.5 175.6 82.7 36.6 33.5 130.9 167.3 67.9 11.8 136.4 15.5 59 137.7 131.6 134.4 131.4 162.5 171.1 26.5

132.85 134.39 60.58 53.7 33.57 75.15 76.88 61.37 173.56 78.79 76.8 44.38 174.04 21.96 17.71 13.66 -

122.5 132.6 70 64.4 31.2 77 86.2 62 174.2 80.2 72.9 43.3 175.6 22.4 17.8 14.2 58.9 140.1 126.3 161.4 170.9 26.4

127.8 132.9 68.7 66.3 30.7 73.3 85.3 60.8 174.2 80.2 76.9 43.2 158.8 22.4 17.8 14.3 59.2 136.5 131.6 134.1 131 162.5 171.1 26.5

138.4 126 59.5 53.6 36.3 71.5 77.3 62.3 -

140.8 118.2 60.4 53.8 32.8 71.3 77.3 60.5 101.3 23.7 24.9 -

140.3 115.1 72.3 64.8 31.3 71.3 87.6 62.4 103.3 24.1 25.2 162.5 127.5 134.1 131.5 137.5 58.8 26.6 171.2

139.7 113.6 69.9 64.1 30.7 70.2 86.6 61.5 102.5 23.4 24.7 191.3 134.9 133.1 129.9 137.2 58.8 -

a See Figure 1 and Schemes 1 and 2 for numbering. Solvent CD CN/DMSO-d . b Literature values [CDCl solvent, see Jones et al. 3 6 3 (1990) and Roeder (1990)].

Table 2. Haptens and Immunogens Prepared pyrrolizidine alkaloid 1 V retrorsine immunogens 2 V monocrotaline immunogens 3f9 V retronecine immunogens

linker 6

linker 7

hapten 6.1 V 6.1bsa

hapten 7.1 V 7.1ova

hapten 6.2 V 6.2bsa

hapten 7.2 V 7.2ova

-

hapten 7.9 V 7.3klh

linker 8 -

-

hapten 8.9 V 8.3ova

NHS-activated crotyl ester 6 was prepared by DCCmediated condensation of 4-bromocrotonic acid3 and NHS in DME. NHS-activated benzoate ester 7 was prepared in a like manner from R-bromo-p-toluic acid and NHS. Subsequent PA N-alkylation resulted in DME insoluble pyrrolizidine salts 6.Y and 7.Y, respectively, which were easily isolated by filtration. Unreacted reagents were removed by open air washing with DME. Evaporative removal of solvent from these pyrrolizidine salts delivered water, methanol, and DMSO soluble haptens which were recrystallized from acetonitrile to give analytical samples characterized by 1H (D2O or CD3OD solutions) and 13C NMR (CD3CN + DMSO-d6 solution) analysis. Benzaldehyde 8 was prepared as a third linker alternative from ethyl p-(R-bromomethyl)benzoate. Attempts at DIBAL-H reduction of this ester to 8 were frustrated by the production of p-(R-bromomethyl)benzyl alcohol and recovery of unreacted starting material. In light of this, 3 Methyl 4-bromocrotonate was purchased from Aldrich and hydrolyzed in 6 M HCl, THF, and acetone for 2 h. The mixture was evaporated and extracted into ether, and white crystals of 4-bromocrotonic acid were obtained upon evaporation of the ether.

reduction with 2 equiv of DIBAL-H (Bookser & Bruice, 1991) yielded p-(R-bromomethyl)benzyl alcohol as a low melting yellow solid which was immediately subjected to PCC oxidation in dichloromethane over 4 Å molecular sieves (Herscovic et al., 1982). Column chromatography produced p-(R-bromomethyl)benzaldehyde as a white, crystalline solid which sublimes upon standing and readily oxidizes in air to R-bromo-p-toluic acid. Benzaldehyde 8 was used to prepare hapten 8.Y from which coating antigen 8.Ypro was prepared. Model quaternary ammonium salts of retronecine were obtained by N-alkylation of 3 with the methyl esters of linkers 6 and 7. Unfortunately, the fine flocullant powders obtained dissolved into viscous pools upon contact with air, and neither compound delivered a recognizable NMR spectum in acetonitrile-d3. Since the two free hydroxyls of retronecine no doubt contribute to the hygroscopic nature of these salts, retronecine was protected as acetonide 9 by treatment with 2,2-dimethoxypropane in acetone (plus ≈1.1 equiv of methanesulfonic acid, Scheme 2). Unlike retronecine which is practically insoluble in ether, acetonide 9 was easily extracted into ether. Subsequent N-alkylation of 9 with linker 7 produced stable hapten 7.9 which was crystalline, readily characterizable, and yet freely soluble in aqueous buffer. A 1H NMR time course of 7.9 in D2O containing deuterated phosphate buffer4 established that the acetonide moiety was hydrolyzed within a few hours time in aqueous buffer of pH e3. Hapten 8.9 was also prepared from acetonide 9 by N-alkylation with linker 8. 13C NMR spectral resonances for quaternary ammonium PAs observed in CD3CN plus DMSO-d6 (Table 1) closely match published values (Jones et al., 1990; 4 Deuterated phosphate buffer was formed most inexpensively by quenching a small amount of P2O5 in D2O, adjusting the pH with drops of K2CO3 dissolved in D2O, and monitoring solution pH with pH paper.

Bioconjugate Chem., Vol. 7, No. 2, 1996 193

ELISA Detection of Pyrrolizidine Alkaloids Table 3. Homologous Immunochemical Systems: Screening of Antibodies system A immunizing antigen coating antigen analyte serum no. antiserum titer antiserum dilution I50 (ppb) slopea

B

C

6.1bsa

6.2bsa

7.3klh

7.1ova retrorsine 1323 980 2000 0.9 ( 0.2 0.7 ( 0.1

7.2ova monocrotaline 2203 1200 1000 39 ( 9 0.59 ( 0.03

8.3ova retronecine 2380 3480 1000 3000 ( 600 1.0 ( 0.1

a The slope b and I 50 were determined from the best-fit fourparameter curve acording to the equation y ) (a - d)/[1 + (x/c)b] + d; see text for a detailed explanation.

Figure 3. System B monocrotaline standard curve. Antibody inhibition is expressed as %B/Bo, the percentage of bound antibody relative to the observed maximum, versus known monocrotaline concentration. The composite standard curve of ten runs reveals a 50% inhibition (I50) of 36 ( 9 ppb in a linear range of 5-500 ppb. The error bars reflect the standard deviation in %B/Bo between ten runs. The curve has a slope of 0.59 ( 0.03 within the linear range.

Figure 2. System A retrorsine standard curve. Antibody inhibition is expressed as %B/Bo, the percentage of bound antibody relative to the observed maximum, versus known retrorsine concentration. The composite standard curve of eight runs reveals a 50% inhibition (I50) of 0.9 ( 0.2 ppb in a linear range of 0.5-10 ppb. The error bars reflect the standard deviation in %B/Bo between eight runs. The curve has a slope of 0.7 ( 0.1 within the linear range.

Roeder, 1990) for their corresponding free bases taken in CDCl3 except for those carbon atoms attached to N-4. These carbons are shifted upfield by about 10 ppm. 1H NMR spectra taken in CD3OD are somewhat ambiguous for macrocyclic PAs and were assigned with the aid of published spectra (Segall & Dallas, 1983). 13C NMR spectral assignments for quaternary compounds were made with reference to published spectra for the parent free base alkaloids (Jones et al., 1990; Roeder, 1990). The water solubility of the haptens listed in Table 2 obviated the need for cosolvents in the protein conjugation step. Thus, BSA conjugates 6.1bsa and 6.2bsa were prepared by treating the appropriate hapten (retrorsine hapten 6.1 or monocrotaline hapten 6.2) with BSA in phosphate-buffered saline (PBS)5 at pH 7.4. The corresponding OVA conjugates 7.1ova, 7.2ova, and 8.3ova were prepared from OVA plus retrorsine hapten 7.1, monocrotaline hapten 7.2, and retronecine hapten 8.9, respectively. Since the hygroscopic nature of free (unprotected diol moiety) retronecine pyrrolizidinium salts precluded characterization, we were pleased to find that retronecine antigens were available from haptens 7.9klh and 8.9ova. Stirring these protein-hapten conjugates in a pH 3 5 PBS was used, although a borate buffer at pH 8.5-9 is preferrable.

Figure 4. System C retronecine standard curve. Antibody inhibition is expressed as %B/Bo, the percentage of bound antibody relative to the observed maximum, versus known retrorsine 5 concentration. The composite standard curve of ten runs reveals a 50% inhibition (I50) of 3000 ( 600 ppb in a linear range of 600-10000 ppb. The error bars reflect the standard deviation in %B/Bo between ten runs. The curve has a slope of 1.0 ( 0.1 within the linear range.

phosphate buffer presumably (vide infra) removes the acetonide moiety, and subsequent dialysis in a pH 7.4 buffer provides the targeted immunogens. In this way, KLH conjugate 7.3klh was prepared from hapten 7.9klh (Li et al., 1991) and OVA conjugate 8.3ova was prepared by reductive amination of hapten 8.9ova. With these immunizing (BSA or KLH) and coating (OVA) antigens in hand, three-month-old New Zealand white rabbits were immunized with conjugates 6.1bsa, 6.2bsa, or 7.3klh. Sera were collected 10 days following each booster injection and titrated against the appropriate coating antigens 7.1ova, 7.2ova, or 8.3ova, respectively.

194 Bioconjugate Chem., Vol. 7, No. 2, 1996

Roseman et al.

Table 4. Competitive ELISA for PA Detection in Homologous Systems A-Ca system A retrorsine 1 monocrotaline 2 senecionine 3 isatidine 4 retronecine 5

system B

I50 (ppb)b

slopec

0.9 ( 0.2 3 × 105 113 0.6 420

0.7 ( 0.1 1.52 0.37 0.2 0.64

I50 (ppb)b +++ 36 ( 9 +++ +++ 10.9

system C slopec

I50 (ppb)b

slopec

≈0 0.59 ( 0.03 ≈0 ≈0 0.62

+++ +++ +++ +++ 3000 ( 600

≈0 ≈0 ≈0 ≈0 1.0 ( 0.1

a PAs were serially diluted and assayed. b The notation +++ indicates that meaningful I ’s were not available at these slopes. c The 50 slope b and I50 were determined from the best-fit four-parameter curve according to the equation y ) (a - d)/[1 + (x/c)b] + d; see text for a detailed explanation.

Table 6. Detection of PAs in Plant Samplesa

Table 5. Reagent Grade Alkaloids Tested for Cross-Reactivity with Systems A and B

S. vulgaris (stem, leaf, and flower) C. retusa (seed) Lupinus spp. (seeds)

system A PA standarda 2500 ppb swainsonineb 2500 ppb lupininec

system B

I50 (ppb)

slope

I50 (ppb)

slope

0.61 0.95 1.65

1.1 0.66 0.76

30 29 11

0.44 0.64 0.50

a

PA standards used were as follows. Retrorsine was used for system A; monocrotaline was used for system B. b PA standard spiked to 2500 ppb with reagent grade swainsonine. c PA standard spiked to 2500 ppb with reagent grade lupinine.

The cycle of booster shots and small bleeds was repeated until adequate titers and PA detection were obtained. ELISA Assay Development. Checkerboard titration screening of the resulting antisera was used to establish coating conditions (typically 2 µg/mL coating protein) and serum dilution (typically 1:1000 working concentration). As one might rationally expect, system A (immunogen 6.1bsa + coating antigen 7.1ova) has excellent sensitivity for retrorsine and system B (immunogen 6.2bsa + coating antigen 7.2ova) for monocrotaline (see Table 3). System C (immunogen 7.3klh + coating antigen 8.3ova) showed parts per million level detection for retronecine, sensitivity less than that found with our previously reported retronamine method (Roseman et al., 1992). Anti-retrorsine antibodies (anti-6.1bsa sera 2203, diluted 1:1000) assayed against coating 7.1ova (immunogen/ microtiter plate coating system A, Figure 2) using retrorsine as the analyte afforded the most sensitive antiretrorsine titer of I50 ) 0.9 ( 0.2 ppb. System A had a useful linear range of 0.5-10 ppb as defined by the B/Bo range within 15-80% of the maximum control Bo. Likewise, anti-monocrotaline antibodies (anti-6.2bsa sera 1323, diluted 1:1000) assayed against coating 7.2ova (immunogen/microtiter plate coating system B, Figure 3) using monocrotaline as the analyte afforded the most sensitive anti-monocrotaline titer of I50 ) 36 ( 9 ppb with a useful linear range of 5-500 ppb. Finally, antiretronecine antibodies (anti-7.3klh sera 2380, diluted 1:1000) assayed against coating 8.3ova (immunogen/ microtiter plate coating system C, Figure 4) using retronecine as the analyte afforded an anti-retronecine titer of I50 ) 3000 ( 600 ppb with a useful linear range of 600-10000 ppb. Color development in these assays was measured using an end point optical density (OD490) format on a microtiter plate reader and plotted as a function of PA concentration. A best-fit four-parameter curve was calculated by Softmax software according to the equation y

system Ab (ppb, w/w)

system Bc (ppb, w/w)

system Cd (ppb, w/w)

10000

8800

0

64000 1900

35000 0

0 0

a The plant material was pulverized in methanol, filtered, and dialyzed into PBS. S. vulgaris is known to be rich in senecionine. C. retusa seeds are rich in monocrotaline. Lupinus spp. plants are thought to be PA-free but do contain lupinine. b Determined using retrorsine as the standard. c Determined using monocrotaline as the standard. d Determined using retronecine as the standard.

) (a - d)/[1 + (x/c)b] + d, where a and d are the upper and lower asymptotes, respectively, b is the slope of the linear portion of the curve, and c is the midpoint of the linear portion of curve, interpreted as the I50. The software calculated PA concentration values for individual wells on the basis of optical density with reference to this four-parameter standard curve; hence, unknowns must be diluted or concentrated so they fall within this working range. With the objective of developing “PA-specific” immunoassays, senecionine and isatidine were also serially diluted and assayed with all three systems. Senecionine and isatidine are structurally quite similar to retrorsine, and as a consequence, system A detects these two PAs at comparably low levels. In contrast, monocrotaline is structurally quite distinct from retrorsine and senecionine and is uniquely detected by system B. That both ELISA systems A and B provide very good retronecine detection is apparently a consequence of in vivo macrocycle diester hydrolysis of immunogens 6.1bsa and 6.2bsa to a retronecine-presenting immunogen. As a consequence, anti-retronecine antibodies were present in all antisera obtained in this study. Attempts at creating nonhomologous systems using the serum from system A with the coating from system B to detect alkaloids 1-5 were largely unsuccessful. These results are presented in Table 4. Biological samples tested using the assays reported here would no doubt contain non-PA alkaloids (such as indolizidine and quinolizidine alkaloids) which could potentially cross-react with the PA antiserum to give a false “PA positive”. To probe this issue, swainsonine (10) and lupinine (11) were selected as representative alkaloids for cross-reactivity studies. Lupine seeds which contain lupine alkaloids, including lupinine, were also tested for cross-reactivity. Reassuringly, relatively large concentrations (2500 ppb) of swainsonine and lupinine had no measurable effect on the I50 of system A or system B assays (Table 5) and hence do not impede their ability to monitor biological samples for PAs. The utility of our competitive ELISA for detection of PAs in actual plant samples was demonstrated by extracting S. vulgaris and C. retusa samples with refluxing methanol and assaying (Table 6). C. retusa seeds are

Bioconjugate Chem., Vol. 7, No. 2, 1996 195

ELISA Detection of Pyrrolizidine Alkaloids

rich in monocrotaline 2 and hence yield a positive ELISA result. Although the PAs in Senecio occur largely as N-oxides, these too can be detected by this assay since Senecio contains senecionine. In contrast, lupine extracts contain lupinine and lupinidine alkaloids and do not result in a positive assay. CONCLUSIONS

Sensitive, compound-specific ELISA methods have been developed for the detection of the pyrrolizidine alkaloids retrorsine and monocrotaline. These assays are comparable in sensitivity to GC methods [reported 0.1 ppm sensitivity: see Deinzer et al. (1982) and Wiedenfeld (1981)] and have the potential advantages of low cost, fast and easy performance, and portability. The linear ranges of inhibition curves using system A for the detection of retrorsine is 0.5-100 ppb with an I50 of 0.9 ( 0.2 ppb. Using system A, good cross-reactivity was observed toward monocrotaline (I50 ) 760 ppb), senecionine (I50 ) 100 ppb), isatidine (retrorsine N-oxide, I50 ) 1 ppb), and retronecine (I50 ) 420 ppb). The linear ranges in inhibition curves using system B for the detection of monocrotaline is 5-5000 ppb with an I50 of 36 ( 9 ppb. Retronecine is detected at 10.9 ppb, and little crossreactivity is observed toward other PAs. Indeed, system B holds great potential as a compound-specific detection method. System C detected retronecine at 600-10000 ppb with an I50 of 3000 ( 600 ppb [less sensitive than that previously reported by our laboratory (Roseman et al., 1992); little cross-reactivity was observed toward other PAs]. All three systems described were capable of detecting retronecine at some level, and none of these systems show cross-reactivity with either swainsonine or lupinine. ACKNOWLEDGMENT

We are pleased to acknowledge support of this work by the National Institutes of Health (ES04274 and ES04699) and the National Science Foundation (CHE9406891). We thank R. Bryan Miller (Department of Chemistry, University of California, Davis) for useful discussions. Supporting Information Available: 1H and 13C NMR data for compounds 7.9, 8.9, and 9 (6 pages). Ordering information is given on any current masthead page. LITERATURE CITED Allen, J. R., Hsu, I. C., and Carsten, L. A. (1975) Dehydroretronecine-induced rhabdomyosarcomas in rats. Cancer Res. 35, 997-1002. Ames, B. N., Durston, W. E., Yamasaki, E., and Lee, F. D. (1973) Carcinogens are mutagens. Simple test system combining liver homogenates for activation and bacteria for detection. Proc. Natl. Acad. Sci. U.S.A. 70, 2281-2285. Anderson, G. W., Zimmerman, J. E., and Callahan, F. M. (1964) J. Am. Chem. Soc. 86, 1839-1842. Bober, M. A., Kurth, M. J., Milco, L. A., Roseman, D. M., Miller, R. B., and Segal, H. J. (1990) A pyrrolizidine alkaloid enzymelinked immunosorbent assay detection strategy. ACS Symp. Ser. 451, 176-183. Bookser, B. C., and Bruice, T. C. (1991) Syntheses of Quadruply Two- and Three-Atom, Aza-Bridged, Cofacial Bis(5,10,15,20tetraphenylporphyrins). J. Am. Chem. Soc. 113, 4208-4218. Bull, L. B., Culvenor, C. C. J., and Dick, A. T. (1968) in The Pyrrolizidine Alkaloids, p 133, North-Holland, Amsterdam. Chesney, C. F., Allen, J. R., and Hsu, I. C. (1974) Right ventricular hypertrophy in monocrotaline pyrrole treated rats. Exp. Mol. Pathol. 20, 257-268. Culvenor, C. C. J. (1980) in Toxicology in the Tropics (R. L. Smith and E. A. Bababunmi, Eds.) Taylor and Francis, London. Deinzer, M. L., Arbogast, B. L., and Buhler, D. R. (1982) Gas Chromatographic Determination of Pyrrolizidine Alkaloids in Goat’s Milk. Anal. Chem. 54, 1811-1814.

Goodrow, M. H., Harrison, R. O., and Hammock, B. D. (1990) Hapten Synthesis, Antibody Development, and Competitive Inhibition Enzyme Immunoassay for s-Triazine Herbicides. J. Agric. Food Chem. 38, 990-996. Harrison, R. O., Goodrow, M. H., Gee, S. J., and Hammock, B. D. (1990) Hapten synthesis for pesticide immunoassay development ACS Symp. Ser. 451, 14-27. Herscovic, J., Egron, M.-J., and Antonakis, K. (1982) New oxidative systems for alcohols: molecular sieves with chromium(VI) reagents. J. Chem. Soc., Perkin Trans. 1, 19671973. Hooper, P. T. (1974) Pathology of Senecio jacobaea poisoning of mice. J. Pathol. 113, 227-230. Jones, A. A., Culvenor, C. C. J., and Smith, L. W. (1982) Pyrrolizidine alkaloids - a carbon-13 NMR study. Aust. J. Chem. 35, 1173-1184. Keeler, R. F. (1983) Plant metabolites that are teratogenic in offspring and toxic in the dam. Toxicon (Suppl. 3), 221-225. Kuhara, K., Takanashi, J., Hirono, I., Furuya, T., and Asada, Y. (1980) Carcinogenic activity of clivorine, a pyrrolizidine alkaloid isolated from Ligularia dentata. Cancer Lett. 10, 117-122. Kurth, M. J., Milco, L. A., Bober, M. A., Miller, R. B., Mount, M. E., and Wicks, B. (1989) A competitive enzyme-linked immunosorbent assay (ELISA) to detect retronecine and monocrotaline in vitro. Toxicon 27, 1059-1064. Kurth, M. J., Milco, L. A., and Miller, R. B. (1992) Trifunctional reagents for substrate-protein conjugation: application to pyrrolizidine alkaloid analogs. Tetrahedron 48, 1407-1416. Laurent, P., Magne, L., De Palmas, J., Bignon, J., and Jaurand, M.-C. (1988) Quantitation of elastin in human urine and rat pleural mesothelial cell matrix by a sensitive avidin-biotin ELISA for desmosine. J. Immunol. Methods 107, 1-11. Leonard, N. J. (1960) in The Alkaloids (R. H. F. Manske, Ed.) Vol. 6, p 35, Academic Press, New York. Li, Q. X., Zhao, M. S., Gee, S. J., Kurth, M. J., Seiber, J. N., and Hammock, B. D. (1991) Development of Enzyme-Linked Immunosorbent Assays for 4-Nitrophenol and Substituted 4-Nitrophenols. J. Agric. Food Chem. 39, 1685-1692. Mattocks, A. R. (1986) in Chemistry and Toxicology of Pyrrolizidine Alkaloids pp 191-219, 147-150, 322-323, pp 213-217, Academic Press, London. Mattocks, A. R., and Bird, I. (1983) Pyrrolic and N-oxide metabolites formed from pyrrolizidine alkaloids by hepatic microsomes in vitro: relevance to in vivo hepatotoxicity. Chem.-Biol. Interact. 43, 209-222. McLean, E. K. (1970) Toxic actions of pyrrolizidine (Senecio) alkalois. Pharmacol. Rev. 22, 429-483. Peterson, J. E., and Culvenor, C. C. J. (1983) in Handbook of Natural Toxins (R. F. Keeler, and A. T. Tu, Eds.) Vol. 1, p 637, Dekker, New York. Robins, D. J. (1982) The pyrrolizidine alkaloids. Prog. Chem. Org. Nat. Prod. 41, 115-203. Roeder, E. (1990) Carbon-13 NMR spectroscopy of pyrrolizidine alkaloids. Phytochemistry 29, 11-29. Roseman, D. M., Wu, X., Milco, L. A., Bober, M., Miller, R. B., and Kurth, M. J. (1992) Development of a Class-Specific Competitive Enzyme-Linked Immunosorbent Assay for the Detection of Pyrrolizidine Alkaloids in Vitro. J. Agric. Food Chem. 40, 1008-1014. Segall, H. J., and Dallas, J. L. (1983) Proton NMR spectroscopy of pyrrolizidine alkaloids. Phytochemistry 22, 1271-1273. Segall, H. J., Wilson, D. W., Dallas, J. L., and Haddon, W. F. (1985) trans-4-Hydroxy-2-hexenal: a reactive metabolite from the macrocyclic pyrrolizidine alkaloid senecionine. Science 229, 472-475. Smith, L. W., and Culvenor, C. C. J. (1981) Plant sources of hepatotoxic pyrrolizidine alkaloids. J. Nat. Prod. 44, 129152. Svoboda, D. J., and Reddy, J. K. (1972) Malignant tumors in rats given lasiocarpine. Cancer Res. 32, 908-913. Wiedenfeld, H., Pastewka, U., Stengl, P., and Ro¨der, E. (1981) On the gas-chromatographic determination of the pyrrolizidine alkaloids of some Senecio-species. Planta Med. 41, 124-128.

BC950084E