`
`381
`
`Substituted 2-Iminothiolanes: Reagents for the Preparation of Disulfide
`Cross-Linked Conjugates with Increased Stability'
`
`Dane A. Goff and Stephen F. Carroll'
`Department of Biological Chemistry, XOMA Corporation, 2910 Seventh Street, Berkeley, California 94710.
`Received July 30, 1990
`
`Much attention has been focused recently on the stability of immunotoxin (antibody-toxin) conjugates
`linked by a disulfide bridge. Conflicting reports have appeared regarding the in vivo stability of such
`conjugates prepared with the two most commonly used cross-linking reagents, SPDP and 2-iminothi-
`olane. We have developed (i) a series of reagents based on 2-iminothiolane substituted at the 4- and/
`or 5-positions (X2ITs) which, based on model studies with simple amines, should show enhanced di-
`sulfide stability when conjugated with antibodies or other proteins and (ii) a real-time method for
`monitoring the rate and extent of conjugation of these reagents with amino groups. Depending upon
`the substituent, the stability of model-activated disulfides relative to unsubstituted 2-iminothiolane
`was increased from 5- to 4000-fold as measured by glutathione-induced release of thionitrobenzoic acid.
`This family of cross-linking reagents should allow the construction of disulfide cross-linked toxin, drug,
`or enzyme conjugates with enhanced stability in vivo.
`
`INTRODUCTION
`Immunotoxins are a class of therapeutic agents typically
`composed of an antibody, capable of binding to specific
`cell-surface antigens on target cells, covalently cross-
`linked to a cytotoxic protein. For most immunotoxins
`prepared to date, the cytotoxic protein is ricin A chain
`(RTA),2 which requires a reducible disulfide linkage with
`the targeting antibody for maximal expression of cyto-
`toxic activity ( I , 2). However, while some studies have
`indicated that the disulfide linkages prepared with RTA
`or other toxins and two of the most commonly used cross-
`linking reagents, 2-iminothiolane (2IT) and N-succinim-
`idyl 3-(2-pyridyldithio)propionate (SPDP), are stable in
`vivo (3,4), others (5,6) have suggested that such linkages
`may be unstable. In addition to decreasing the circulating
`levels of the antibody-toxin conjugate, such deconjuga-
`tion would liberate free antibody which can then compete
`with the conjugate for binding to target cells.
`Worrell et al. (7) and Thorpe et al. (8, 9) have shown
`that hindering access to the disulfide bond linking the
`antibody and toxin by introducing a methyl group adjacent
`to the sulfur atom of reagents based upon SPDP increases
`the stability of the resultant conjugate both in vitro and
`in vivo. Moreover, alterations in spacer length have been
`found to alter the stability of conjugates in vitro (10). When
`combined with the experimental discrepancies noted above,
`these results suggest that the in vivo stability of immu-
`notoxins may be influenced by the specific antibody-
`toxin pairing and can be enhanced by alterations in residues
`immediately adjacent to the disulfide bridge.
`* Author to whom correspondence should be addressed.
`A preliminary report of a portion of this work has appeared
`(Abstracts from the Fourth International Conference on Mon-
`oclonal Antibody Immunoconjugates for Cancer, San Diego, 1989).
`2 The abbreviations used are as follows: DTNB, 5,5'-dithiobis-
`(2-nitrobenzoic acid); PBS-EDTA, phosphate-buffered saline,
`pH 7.2, containing 0.5 mM EDTA, ethylenediaminetetraacetic
`acid; RTA, ricin toxin A chain; SPDP, N-succinimidyl 3-(2-
`pyridy1dithio)propionate; TNB, 5-mercapto-2-nitrobenzoic acid;
`2IT, 2-iminothiolane; XBIT, 4- and/or 5-substituted 2-iminothi-
`olanes.
`
`In an effort to more critically evaluate the requirements
`for improving disulfide bond stability in vivo, we have
`synthesized a series of sterically hindered reagents based
`on 21T (11). 21T offers several advantages over other
`possible linking reagents (Figure 1): (i) it reacts with
`primary amines to form stable amidinium derivatives,
`retaining the positive charge; (ii) inclusion of an aromatic
`disulfide such as 5,5'-dithiobis(2-nitrobenzoic acid) (DTNI3)
`in the reaction mixture both activates the newly exposed
`21T thiol and allows real-time spectrophotometric
`monitoring of the labeling reaction; and (iii) substitution
`on the 2-IT ring at the 4- and/or 5-positions can be used
`to introduce various groups into proximity with the di-
`sulfide bond, thus allowing alterations in the degree of
`steric hindrance. Here we have prepared a series of ster-
`ically hindered 21T molecules (X2ITs, Table I) and have
`evaluated their reactivity with amino groups and the di-
`sulfide stability of model conjugates.
`EXPERIMENTAL SECTION
`Materials. 21T and DTNB were obtained from Sigma.
`Anhydrous solvents, 10 M n-butyllithium in hexane, and
`all other reagents were from Aldrich Chemical Co., except
`as noted.
`General Procedures. Melting points are uncor-
`rected and were measured on a Fisher-Johns apparatus.
`'H NMR spectra were recorded at 60 MHz on a Varian
`EM-360 or at 400 MHz on a Bruker AM-400 spectrometer.
`Peaks are given as 6 values relative to tetramethylsilane
`as internal standard. Coupling constants (J) are given in
`hertz. IR spectra were obtained either as KBr disks or
`as neat films on NaCl plates on a Perkin-Elmer Model 1330
`spectrophotometer. The abbreviations used to describe
`the spectral peaks are v = very, s = strong, b = broad, sh
`= sharp, m = medium, w = weak. Wavelengths are given
`in inverse centimeters. Ultraviolet spectra were recorded
`on a Shimadzu Model 160 spectrophotometer. Flash
`chromatography was performed on silica gel 60 (Merck,
`230-400 mesh) with apparatus supplied by J.T. Baker.
`Elemental analyses were performed by Desert Analytics,
`Tuscon, AZ.
`
`1043- 1802/90/290 1-038 1 $02.50/0 0 1990 American Chemical Society
`
`IMMUNOGEN 2082, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`382 Bigconjugate Chem., Vol. 1, No. 6, 1990
`
`I
`
`A412
`
`Time
`
`H
`TNB
`Figure 1. Chemistry of the X2ITs. Compounds containing a
`primary amine (R-NHz) react with the X2ITs to create an ami-
`dinium linkage without neutralizing the positive charge. If the
`reaction is performed in the presence of DTNB, the newly exposed
`iminothiolane thiol cleaves DTNB in a disulfide-exchange
`reaction. This cleavage both activates the linker thiol (in
`preparation for conjugation) and releases free TNB, allowing real-
`time spectrophotometric monitoring of the rate and extent of
`reaction.
`Table I. Structures of Substituted X2ITs
`flJ=Nh*
`S
`
`compd
`ZIT
`
`4 3
`substitution
`-
`
`MZIT
`
`5-methyl
`
`TB2IT
`
`5-tert-butyl
`
`no."
`7a
`
`7b
`
`7c
`
`structure
`
`C H3!$2\
`ck3"
`
`N i 2
`
`Ph2IT
`
`5-phenyl
`
`DM2IT
`
`5,5-dimethyl
`
`7d
`
`@ - ( 3 = N ' H 2
`
`7e "0 NL
`
`CH3
`
`S2IT
`
`5-spiro
`
`7f
`
`0 Refers to the structures shown in Scheme I. * The fused ring
`structures are shown here with the numbering system common to
`the iminothiolane family.
`
`All synthetic reactions were stirred magnetically under
`an inert atmosphere; argon for reactions with n-butyl-
`lithium, otherwise nitrogen. The expression "dried" refers
`to passage of the organic layer through a pad of anhydrous
`sodium sulfate. Concentration of solutions in vacuo was
`performed under water aspiration with a Buchi RllO Ro-
`tavapor. Reactions were performed at room temperature
`unless otherwise noted. For each compound, the boldfaced
`designation in parentheses (e.g. 6b) denotes the cor-
`responding structure shown in Scheme I.
`4-Mercaptopentanenitrile (6b). To dry THF (100
`mL) cooled to -75 "C was added 10.0 M n-butyllithium
`(15.0 mL), followed carefully dropwise with a solution of
`CH3CN (7.84 mL, 0.15 mol) in THF (20 mL). After 15
`min 2-methylthiirane (11.75 mL, 0.15 mol) was added drop-
`wise. The white suspension stirred a further 5 min a t -75
`"C and then the cooling bath was removed. An exothermic
`reaction ensued to give a pale yellow homogeneous solution.
`After 2 h the reaction was quenched with 1/1 concentrated
`HCl/H2O (30 mL) and extracted (7 X 50 mL) with EtzO.
`
`Goff and Carroll
`Scheme 1.. Synthetic Scheme for the Preparation of
`X2ITs
`
`a
`Route B
`
`- 4 RmSAc
`6 R = S H
`
`a Details can be found in Experimental Procedures. Reagents:
`(a) LiCH&N/THF; (b) ClSOzCHs/benzene/Et3N; (c) CsSAc/
`DMF, (d) KSCN/EtOH/HZO or 3-methylbenzothiazole-2-thione/
`TFA/CHzC12; (e) HCl(g)/MeOH. The products of the reactions
`are indicated by underlined number (stage of synthesis), let-
`ter (R groups) combinations. For example, 7a refers to the final
`(2IT).
`X2IT hydrochloride where R1 = Rz = R3 =-
`The combined organic layers were dried and concentrated
`in vacuo to give a crude amber liquid (10.1 g, 59%).
`Distillation in vacuo gave 6b as a colorless liquid (6.1 g,
`35%): 1H NMR (60 MHz, CDCl3) 2.72-3.28 (m, 1 H, CH),
`2.53 ft, J = 7, 2-CH2), 1.62-2.18 (m, 2 H, 3-CH2), 1.48 (d,
`1 H, J = 7, exchanges with D20, SH), 1.40 (d, 3 H, J = 6,
`5-CH3); IR (film) 2550 (SH), 2245 (CN).
`4-Mercapto-4-methylpntanenitrile (6e). To a 20 mL
`of dry T H F cooled a t -75 "C was added 10.9 M n-BuLi
`(3.0 mL), followed dropwise by a solution of CH3CN (1.60
`mL, 30 mmol) in T H F (5 mL). After addition was
`complete the mixture, containing a white precipitate of
`LiCHzCN, was stirred for 10 min. Then 2,2-dimethylthi-
`irane (19) (2.7 g, 30 mmol) was added dropwise. The
`cooling bath was removed and an exothermic reaction
`ensued to give a pale yellow homogeneous solution. After
`1.5 h the reaction was quenched by the addition of 5 mL
`of 1/1 concentrated HCl/H20. The mixture was extracted
`(3 X 10 mL) with EtOAc. The combined organic layers
`were dried and concentrated in vacuo to give a pale yellow
`liquid (3.66 g, 92% ) with a strong thiol stench, which was
`used without further purification: 'H NMR (60 MHz,
`CDC13) 2.88 (t, 2 H, J = 7, 2-CH2), 2.05 (t, 2 H, J = 7,
`3-CHz), 1.53 (s, 6 H, 2 CH3); IR (film) 2250 (w), 2190 (m).
`4-Hydroxy-4-phenylbutanenitrile (2d) (13). Sodium
`amide (10.61 g, 0.271 mol) was transferred to a tared 500-
`mL three-neck round-bottom flask in a glove bag under
`Ar. An addition funnel and thermometer were added, then
`150 mL of dry THF was cannulated in. The suspension
`was cooled to -30 "C on a CC&/dry ice bath and then CH3-
`CN (13.5 mL, 0.259 mol) was added dropwise, while the
`temperature was kept <-20 "C. This was followed by
`addition of styrene oxide (26.0 mL, 0.228 mol) over 30 min.
`The mixture was allowed to warm to 0 "C over 80 min,
`and then to room temperature over 75 min. The mixture
`
`IMMUNOGEN 2082, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`Substituted 2-Iminothiolanes
`was then recooled to -35 "C and quenched with saturated
`aqueous NH&l (35 mL). The mixture was diluted with
`EtOAc (200 mL), and the layers were separated. The
`aqueous layer was extracted (2 X 50 mL) with EtOAc. The
`combined organic layers were dried and concentrated in
`vacuo. TLC indicated two major products. Repeated flash
`chromatography on silica (hexanes/EtOAc) gave 2d as a
`pale yellow oil (3.48 g, 10%): TLC (70/30 hexanes/
`EtOAc) R, = 0.52; 'H NMR (60 MHz, CDC13) 7.30 (s, 5
`H, ar), 4.73 (m, 1 H, benzylic), 2.16-2.80 (m, 2 H), 1.70-
`2.16 (m, 2 H); IR (film) 3420 (vs, b, OH), 2240 (CN).
`4-Phenyl-4-(thioacetyl)butanenitrile (4d). Hy-
`droxy nitrile 2d (3.48 g, 21.6 mmol) in benzene (20 mL)
`with Et3N (3.13 mL, 22.5 mmol) was cooled to 0 "C and
`treated dropwise with methanesulfonyl chloride (2.54 g,
`22.2 mmol) in benzene (15 mL). The ice bath was removed
`and stirring continued for 30 min. The Et3NHCl was
`filtered off and the filtrate concentrated in vacuo to give
`cyanomesylate 3d: 'H NMR (60 MHz, CDC13) 7.37 (s, 5
`H, ar), 5.58 (m, 1 H, benzylic), 2.72 (s, 3 H, SOZCH~), 2.03-
`2.67 (m, 4 H). Crude 3d was dissolved in dry DMF (5 mL)
`and treated with a solution of CsSAc (22 mmol, prepared
`according to the method of Kellog and Strijtveen, ref 15)
`in DMF (10 mL). After overnight stirring the mixture was
`diluted with Et20 (150 mL) and rinsed (5 X 20 mL) with
`HzO. The organic layer was dried and concentrated in
`vacuo to give a brown liquid which was purified by flash
`chromatography (hexanes/EtOAc). Thioacetate 4d was
`obtained as an orange oily liquid (3.37 g, 72%): TLC (90/
`10 hexanes/EtOAc) Rf = 0.21; 'H NMR (60 MHz, CDC13)
`7.27 (s, 5 H, ar), 4.58 (m, 1 H, benzylic), 2.30 (s, 3 H, SAC),
`2.10-2.30 (m, 4 H); IR (film) 2250 (CN), 1688 (C=O).
`2-(Cyanomethyl)-l-mercaptocyclohexane (6g). To
`dry THF (30 mL) at -75 "C was added 10.0 M n-BuLi (3.59
`mL), followed dropwise by CH3CN (1.88 mL, 35.9 "01).
`Ten minutes after addition was complete, cyclohexene
`sulfide (17) (3.90 g, 34.2 mmol) in THF (10 mL) was added
`dropwise. When addition was complete the cooling bath
`was removed. After 2 h the mixture was recooled to 0 "C
`and quenched by addition of a mixture of concentrated
`HCl(6 mL) and HzO (15 mL). The layers were separated,
`and the aqueous layer was extracted (2 X 25 mL) with
`EtOAc. The combined organic layers were dried and
`concentrated in vacuo, followed by vacuum distillation to
`give 6g as a colorless liquid (2.26 g, 43%): bp 85-89 "C
`(0.3 mm); lH NMR (60 MHz, CDC13) 2.68 (d, 2 H, CH2-
`CN), 2.32 (s, 1 H), 1.00-2.30 (m, 10 H); IR (film) 2250 (CN).
`l-(a-Cyanoet hy1)- 1-mercaptocyclohexane (6f). To
`a solution of 10.0 M n-BuLi (0.67 mL) in THF (10 mL)
`at -75 "C was added CH&N (0.35 mL, 0.67 mmol) drop-
`wise. After 10 min, 7-thiaspiro[5.2]octane [0.86 g, 0.67
`mmol, prepared from 7-oxaspiro[5.2]octane (20) according
`to the method of ref 161 in 2 mL of dry THF was added.
`The cooling bath was removed and after 2 h the reaction
`was quenched with a mixture of concentrated HCl(l.5 mL)
`and H2O (3.5 mL). The layers were separated, and the
`aqueous layer was extracted twice with EtOAc (25 mL).
`The combined organic layers were dried and concentrated
`in vacuo to give crude 6f as a foul-smelling oil (0.48 g),
`which was used without further purification.
`5,5-Dimethyl-4-mercaptohexanenitrile (6c). To a
`solution of 10.0 M n-BuLi (5.6 mL) in dry THF (50 mL)
`at -75 "C was added CH3CN (2.9 mL, 56 mmol) drop-
`wise. After 10 min, a solution of 2-tert-butylthiirane (16)
`(5.94 g, 51 mmol) in dry THF (5 mL) was added drop-
`wise. After 10 min the cooling bath was removed and the
`mixture stirred for 2 h, whereupon it was recooled to 0 "C
`and quenched with 1/1 concentrated HCl/HzO (12 mL).
`
`Bioconjugate Chem., Vol. 1, No. 6, 1990 383
`
`The mixture was diluted with CH2C12, and the layers were
`separated. The organic layer was rinsed four times with
`H20, dried, and Concentrated in vacuo to give a pale yellow
`liquid (4.96 gj. Kugelrohr distillation [%lo0 "C (0.6 mm)]
`gave 2.94 g (34%) of colorless 6c: 'H NMR (60 MHz,
`CDC13) 3.70 (dd, 1 H, J = 6, 10, CH), 2.36-3.36 (m, 3 H),
`1.70-2.36 (m, 2 H), 1.03 (s, 9 H, t-Bu); IR (film) 3200-
`3300 (8, b), 2960 (vs, b), 2250 and 2210 (both sh, m, CN),
`1620,1475,1375.
`General Procedure for the Synthesis of Substituted
`2-Iminothiolanes (7b-g). The appropriate 4-mercap-
`tonitrile (6) or 4-(thioacetyl)nitrile (4) was dissolved in dry
`MeOH (ca. 10 mL/g) and bubbled vigorously with HCl-
`(g) for 3-5 min. The mixture was stirred overnight or until
`IR spectroscopy showed complete loss of the nitrile ab-
`sorbance at 2200 cm-'. The solvent was removed in vacuo.
`The residue was slurried in EtOAc and then warm EtOH
`was added until a solution was obtained. The mixture was
`cooled to -20 "C. If crystallization did not occur within
`several hours, Et20 was added. Further cooling gave the
`following hydrochlorides.
`5-Methyl-2-iminothiolane hydrochloride (M2ITsHC1,
`7b): yield 6.67 g (83%); mp 114.5-115.5 "C (EtOAc/
`EtOH); 'H NMR (400 MHz, DMSO-&) 7.46 (t, 1 H, J =
`50.7, NH),4.13-4.22 (m, 1 H, 5-H), 3.31-3.39 (m, 1 H), 3.19-
`3.28 (m, 1 H), 2.39-2.47 (m, 1 H), 1.88-1.97 (m, 1 H), 1.45
`(d, 3 H, J = 6.7,5-CH3); IR (KBr) 3410,2810 (vs, vb), 1615
`(s, C=N), 1535,1450,1415,1240 1025,990,885,690. Anal.
`Calcd for C5HloClNS: C, 39.60; H, 6.65; N, 9.24; S, 21.14.
`Found: C, 39.63; H, 6.62; N, 9.02; S, 21.34.
`5-( l,l-Dimethylethyl)-2-iminothiolane hydro-
`chloride (TB21T*HCI,7c): yield 1.22 g (41%); mp 224-
`226 "C; 1H NMR (400 MHz, DMSO-de) 7.43 and 7.40 (2
`t, J = 50.7, major t centered at 7.40, combined integral
`1.7 H, NH2+), 4.12 (dd, 1 H, J = 5.2, 10.9, H-5), 3.17-
`3.33 (m, 2 H, 3-CHz), 2.28-2.32 (m, 1 H, H-4), 1.96-2.06
`(m, 1 H, H-4), 1.01 (s, 9 H, t-Bu); IR (KBr) 2800-3000 (vs,
`vb), 1630 (s, C=N), 1535, 1475, 1410, 1375, 1260, 1185,
`1000,895,695. Anal. Calcd for C8H16ClNS: C, 49.60; H,
`8.32; N, 7.23; S, 16.55. Found: C, 49.52; H, 8.37; N, 7.37;
`S, 16.34.
`5-Phenyl-2-iminothiolane hydrochloride (Ph-
`2IT*HCl,7d): yield 1.22 g (44%); mp 172 "C dec; 'H NMR
`(400 MHz, DMSO-&) 7.52 (d, 2 H, J = 6.9), 7.36-7.45 (m,
`3 H), 5.35 (dd, 1 H, J = 5.5,10.4, 5-H), 3.30-3.44 (m, 2 H,
`3-CH2), 2.64-2.71 (m, 1 H), 2.39-2.48 (m, 1 H); IR (KBr)
`3360,2870 (vs, vb), 1615 (s, br, C=N), 1515, 1495, 1450,
`1405,1325,1020,985,830,770,760,700,675. Anal. Calcd
`for CloH12ClNS: C, 56.20; H, 5.66; N, 6.55; S, 15.08.
`Found: C, 55.94; H, 5.54; N, 6.60; S, 15.05.
`5,5-Dimethyl-2-iminothiolane hydrochloride
`(DM2IT.HCI,7e): yield 3.0 g (65%); mp 156-159 "C; lH
`NMR (400 MHz, DMSO-&) 7.43 (t, 1 H, J = 51.3, NH),
`3.43 (t, 2 H, J = 7,1,3-CH2), 2.18 (t, 2 H, J = 7.1, 4-CHz),
`1.59 (s, 6 H, 2 CH3); IR (KBr) 2900 (vs, vb), 1610 (s, b,
`C=N), 1510, 1325, 1240, 1215, 1120, 990, 875, 845, 690.
`Anal. Calcd for CsH12ClNS: C, 43.50; H, 7.30; N, 8.45; S,
`19.35. Found: C, 43.08; H, 7.45; N, 8.63; S, 18.88.
`8-Imino-7-thiaspiro[5.4]decane hydrochloride
`(S2IT-HCl, 7f): yield 0.17 g (35%); mp 130-135 "C; 'H
`NMR (400 MHz, DMSO-&) 7.37 (t, J = 50.8) and 7.34 (t,
`J = 50.7) (the major t is at 7.34, the combined integral of
`both 1.8 H, NH2+, 3.39 (t, 2 H, J = 7.1, 3-CHz), 2.20 (t, 2
`H, J = 7.1, 4-CHz), 2.00 (m, 2 H), 1.69-1.79 (m, 4 H), 1.58
`(m, 1 H), 1.33-1.38 (m, 3 H); IR (KBr) 2500-3600 (vs, vb),
`1610 (s, C=N), 1525, 1445, 1405, 1130, 1100, 1020, 990,
`875, 690.
`t-8-Imino-7-thiabicyclo[4.3.0]nonane hydrochloride
`
`IMMUNOGEN 2082, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`Sioconjugate Chem., Vol. 1, No. 6, 1990
`384
`(RZITsHCl, 7g): yield 3.13 (59%); mp 254-255 "C dec;
`'H NMR (400 MHz, DMSO-&) 7.34 (t, 1 H, J = 50.7), 3.57
`(td, 1 H, J = 11.5, 3.4, H-9), 3.24 (dd, 1 H, J = 17.0, 5.7,
`H-3,,), 2.91 (dd, 1 H, J = 17.0, 13.1, H-3,,), 2.22 (m, 1 H),
`1.94-2.03 (m, 2 H), 1.75-1.84 (m, 2 H), 1.45-1.54 (m, 1 H),
`1.28-1.39 (m, 3 H); IR (KBr) 2840 (vs, vb), 1624 (s, C=N),
`1528,998,895, 880,694. Anal. Calcd for CsH&lNS: C,
`50.12; H, 7.36; N, 7.31; S, 16.72. Found: C, 50.02; H, 7.29;
`N, 7.21; S, 16.79.
`Preparation of Thiol-Activated Model Compounds.
`TNB-activated X2ITs were prepared by aminolysis in the
`presence of the aromatic disulfide DTNB. Each X2IT
`(final concentration 0.5 mM) was added to a solution of
`1.0 mM DTNB in 30 mM NH4HC03, pH 8.0. After 1 h
`at 25 "C, the reactions were terminated, and solvent was
`removed, by lyophilization. The activated model com-
`pounds (NH2-X2IT-TNB) were resuspended in 0.5 mL of
`25 mM sodium citrate, pH 3.2, and applied to separate
`1-mL columns of S-Sepharose Fast Flow (Pharmacia)
`equilibrated in the same buffer. Each column was washed
`with citrate buffer until the absorbance a t 412 nm
`approached zero, and the model disulfides were then eluted
`with PBS-EDTA.
`Preparation of INDl-M2IT-RTA. A mixture of the
`murine IgG2a INDl antibody (20 pM, 3 mg/mL, 2 mL),
`DTNB (5.0mM),and MBIT(3.4mM)inPBS-EDTAwas
`incubated at 25 OC until the absorbance of the reaction
`mixture at 412 nm reached 0.54. Excess reagents and
`reaction byproducts were then removed by desalting on
`a 1 cm X 20 cm column Sephadex G25F equilibrated at
`4 "C in PBS-EDTA. The thiolated INDl-M2IT-TNB
`antibody so prepared (6.4 pM, 5 mL) was mixed with a
`5-fold molar excess of RTA, and the mixture was incubated
`at 4 "C for 16 h. The IND1-M2IT-RTA
`immunotoxin
`was subsequently purified by size-exclusion chroma-
`tography on a 1 cm X 32 cm column of AcA44 equilibrated
`in PBS at 4 "C.
`
`RESULTS
`Synthesis and Characterization of Substituted
`2-Iminothiolanes, The XBIT-HCl analogues (Table I)
`were prepared by cyclization of y-mercapto or y-thio-
`acetyl nitriles in methanolic HC1 (Scheme I). Both the
`parent unsubstituted 21T (7a) and M2IT (7b) have been
`previously prepared in this way (11,12). The nitriles were
`prepared either by ring opening of the appropriate ep-
`oxide with sodium or lithium acetonitrile (13,14), followed
`by mesylation of the y-hydroxy nitrile and displacement
`with cesium thioacetate (15) (route A), or by ring opening
`of a thiirane with lithium acetonitrile (route B). The thi-
`iranes were prepared from the corresponding epoxides
`either with potassium thiocyanate in EtOH (or MeOH)/
`H2O (16) or with 3-methylbenzothiazole-2-thione (1 7).
`Route B was useful when nucleophilic substitution of a
`mesylate would be difficult [TBBIT (7c) and DMBIT (7e)l.
`Similarly, the cis-fused bicyclic R2IT (cis-7g) could not
`be prepared by substitution of mesylate 3g or the
`corresponding triflate with cesium thioacetate, but the
`trans-fused (trans-7g) compound was readily prepared via
`cyclohexene sulfide (route B). The reaction of lithioac-
`etonitrile with 2-phenylthiirane proceeded poorly, so in
`this case Ph2IT (7d) was prepared via route A. The
`preparation of nonracemic 5-substituted X2ITs was not
`investigated, but should be feasible from commercially
`available chiral epoxides such as (R)-(+)-styrene oxide and
`(S)-(-)-propylene oxide or the derived thiiranes (16,18).
`The XBIT-HCl compounds (7a-g) were characterized
`by lH NMR, IR, and UV spectroscopy and elemental
`
`Goff and Carroll
`
`Table 11. Properties of Substituted 2-Iminothiolanes
`reaction rates (25 "C); k, s-*
`hydrolysisn
`glycineb
`linker
`(XIOb)
`t, 1 mM, 248 nm
`A,,
`(xio-3)
`1.4
`5.3
`8.72
`247
`21T
`1.7
`3.4
`9.07
`250
`R2IT
`0.7
`4.8
`9.27
`247
`M2IT
`0.2
`5.2
`9.00
`246
`Ph2IT
`0.8
`4.4
`11.7
`248
`TB2IT
`0.9
`3.5
`10.8
`DM2IT 247
`0.7
`3.0
`10.7
`248
`SPIT
`The X2ITs (100 pM) were incubated at 25 O C in PBS-EDTA
`and changes in the absorbance at 248 nm were monitored. Plots of
`log [XBIT] vs time were linear, and first-order rate constants were
`calculated from the slopes. The rates of reaction of the X2ITs (100
`pM) with glycine (160 mM) in PBS-EDTA were examined by coupling
`the reaction with 0.5 mM DTNB and monitoring changes at 412 nm.
`First-order rate constants were determined from the linear slopes of
`log [XZIT] vs time plots.
`analysis. The recrystallized X2ITs are easily handled and
`stable for at least 6 months when stored dry at 4 "C. The
`compounds are readily soluble in H2O or polar solvents
`such as DMSO. The lH NMR spectra of the X2ITs are
`straightforward, showing signals for the 4-CHz from 1.9
`to 2.7 ppm, for the 3-CH2 from 2.9 to 3.4 ppm, and for the
`5-CH (when present) from 3.5 to 5.4 ppm. The trans ring
`fusion of R2IT (7g) was assigned on the basis of the
`assumed anti opening of cyclohexene sulfide with lithio-
`acetonitrile and by the 11.5 Hz coupling observed between
`the bridgehead protons.
`The UV spectra of the X2ITs have a maximum at 246-
`250 nm with mM extinction coefficients of ca. 10 (Table
`11). The UV absorbance of the X2ITs is lost upon ring
`opening and this provides a convenient method of
`monitoring hydrolysis or aminolysis rates. The rates of
`aqueous hydrolysis of the X2ITs (7a-g) in PBS could
`therefore be followed by (i) the loss of absorbance at 248
`nm or (ii) coupling the reaction with DTNB. As shown
`in Table 11, aqueous hydrolysis of the X2ITs was relatively
`slow, followed first-order kinetics, and showed little
`dependence upon ring substitution. The rates shown are
`for the direct optical assay; similar values were obtained
`when the released thiol was trapped with DTNB and the
`reaction monitored at 412 nm (data not shown). The rates
`of X2IT hydrolysis were thus unaffected by the presence
`of DTNB.
`Reactivity with Glycine. Reaction of the X2ITs with
`amino groups was analyzed by incubating each cross-
`linker with glycine at pH 7.2. In order to better mimic
`the reaction with proteins, these reactions were performed
`in the presence of DTNB, thus allowing both real-time
`monitoring of the reaction, as well as activation of the newly
`exposed iminothiolane thiol. Each X2IT was therefore
`incubated with glycine and DTNB, and changes in the ab-
`sorbance a t 412 nm were monitored. Under these
`conditions, the reaction rates were first-order and were
`again unaffected by the ring substituent (Table 11). Similar
`rates were also obtained when the reactions were monitored
`optically at 248 nm in the absence of DTNB (data not
`shown). Under the experimental conditions employed, the
`rates of reaction of the X2ITs with glycine were typically
`5000-fold greater than the rates of aqueous hydrolysis.
`Preparation and Stability of Model Disulfides. In
`order to assess the effect of the various ring substituents
`on subsequent disulfide bond stability, TNB-activated
`model conjugates were prepared by reacting each X2IT
`with ammonium bicarbonate in the presence of DTNB.
`Aminolysis with ammonium bicarbonate was chosen for
`these experiments because excess NH3 and buffer could
`
`IMMUNOGEN 2082, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`> Bioconjugate Chem., Vol. 1, No. 6, 1990 385
`
`Substkuted 2-Iminothiolanes
`
`IZ0
`
`100 -
`80 -
`60 -
`
`m z c
`
`A A21T
`0 Ph2IT
`A M21T
`TB2lT
`
`40-
`
`--
`
`0
`
`I 0 0
`
`Z O O
`
`300
`
`400
`
`500
`
`600
`
`T h e (seconds)
`Figure 2. Glutathione-induced release of T N B from NHz-
`X2IT-TNB model conjugates. Samples of the activated model
`conjugates (20 pM) in PBS-EDTA were placed in a cuvette ther-
`mostated to 25 O C and, at T = 0, reduced glutathione was added
`to a final concentration of 200 pM. The release of T N B was
`monitored optically at 412 nm for 500 s, and 2-mercaptoetha-
`no1 was then added to a final concentration of 200 mM to
`determine maximal release of TNB. The plots for the different
`linkers were normalized by determining percent maximal T N B
`release, as calculated by dividing the amount of T N B released
`at any timepoint by the amount released with 2-mercaptoeth-
`anol.
`Table 111. Relative Stability of TNB-Activated
`NHZ-Iminothiolanes
`TNB release rate (25 “C),” stability increase
`k (X104), s-l
`relative to 21Tb
`318
`1.0
`NHz-ZIT-TNB
`113
`2.8
`NHz-R2IT-TNB
`4.5
`NHz-Ph2IT-TNB
`70.0
`32.5
`9.8
`NHz-M2IT-TNB
`3.22
`99
`NHz-TB2IT-TNB
`4184
`NHz-DMPIT-TNB
`0.076
`a Reactions contained 20 pM NHz-X2IT-TNB in PBS-EDTA at
`25 OC and were initiated by the addition of reduced glutathione as
`described in the legend to Figure 2. The concentration of added glu-
`tathione was varied between 50 pM and 200 mM in order to achieve
`linear plots of log [NHz-X2IT-TNB] vs time, and first order rate
`constants were then determined. * Calculated by dividing the rate
`constant for TNB release for each model conjugate by that obtained
`for NHz-BIT-TNB.
`be removed by lyophilization, thereby minimizing ionic
`effects. Conditions were chosen such that each linker was
`quantitatively converted to the corresponding activated
`model conjugate (NHr X2IT-TNB). Following lyophiliza-
`tion, the model conjugates were purified by cation-
`exchange chromatography to remove unreacted DTNB and
`TNB. The relative stabilities of these mixed disulfides
`were then examined by reacting the model conjugates with
`reduced glutathione. The release of TNB was monitored
`at 412 nm, providing a direct measure of glutathione-
`induced disulfide bond cleavage.
`A comparative rate plot for the activated X2IT model
`conjugates (20 pM) incubated with 200 pM glutathione is
`shown in Figure 2. By varying the concentration of glu-
`tathione (0.05 to 250 mM) for the different compounds,
`first-order rate constants for TNB release were determined
`(Table 111). The results indicate that the relative stability
`of the disulfide bonds formed by the X2ITs vary by a factor
`of roughly 4000, with DMBIT being the most stable,
`followed by TB2IT and M2IT.
`Preparation of Protein-Protein Conjugates. To
`confirm the utility of the XBIT reagents for the preparation
`of antibody-toxin conjugates, one of the linkers (M2IT)
`was reacted with the murine IgG2a antibody INDl (21)
`
`in the presence of DTNB. The reaction was monitored
`optically at 412 nm and was terminated when the absor-
`bance indicated 1.9 activated thiols/mol of protein.
`Following purification, an aliquot was removed and treated
`with 0.1 mM DTT; the activated IND1-M2IT-TNB
`antibody contained 2.0 TNB/mol. Subsequent conjugation
`with RTA gave IND1-MBIT-RTA in high yield.
`
`DISCUSSION
`The X2ITs described here represent a new family of
`cross-linking reagents that should prove useful in the
`preparation of stabilized protein conjugates linked by a
`disulfide bridge. Like 21T (II), the X2ITs are highly water
`soluble, they react with amino groups to produce a stable
`amidinium linkage (thereby preserving the positive charge),
`and the reaction rate and extent can be monitored in real
`time by including an aromatic disulfide in the reaction
`mixture. In addition, the relative stability of the conjugate
`disulfides can be controlled by appropriate substitution
`on the 21T ring, particularly at the 5-position. With the
`model conjugates prepared here, disulfide bond stability
`was increased from 6- to 4000-fold relative to unsubsti-
`tuted 21T and was well-correlated with the degree of steric
`hindrance (see below). Importantly, these increases were
`achieved without adversely affecting either the rate of
`reaction with glycine or the aqueous lability of the linker.
`In designing cross-linking reagents for the preparation
`of disulfide-linked immunoconjugates, several features
`critical to actual therapeutic use must be evaluated (1).
`Some considerations (and consequences) include (i) the
`effect of linker derivatization on protein function (reduced
`binding or enzymatic activity, altered charge, etc.), (ii)
`variations in subsequent conjugation efficiency (lower
`efficiency necessitates higher linkerlprotein ratios and
`reagent needs), and (iii) relative stability in vivo (rapid
`deconjugation reduces the effective serum concentration
`and liberates competitive ligand). Whereas some of these
`concerns are empirical (i.e., the effect of linker derivati-
`zation), the remainder can be readily controlled by
`appropriate linker chemistry and selection.
`Each of the above concerns can be minimized by the use
`of a linking reagent that allows both efficient conjugation
`with thiol-containing compounds, as well as the ability to
`control disulfide bond stability. Thus, the absolute number
`of linkers/molecule can be minimized, reducing the
`probability that protein function will be affected.
`Preliminary results suggest that the efficiency of con-
`jugation (i.e., the efficiency with which the aromatic leaving
`groups are replaced by the protein thiol) for the X2ITs
`ranges between 60 and 10076, depending upon the di-
`aryl disulfide present in the reaction mixture (unpublished
`data). These results, together with the preservation of
`positive charge and the controlled bond stability, support
`the use of the X2ITs in the preparation of human
`therapeutics.
`Mechanistically, the observed increases in the stability
`of model X2IT disulfides conferred by a-alkylation can
`be rationalized by assuming that the incoming thiolate
`nucleophile (glutathione in this case) must attack the
`sulfur atom derived from the XBIT (see Figure 1). It has
`been postulated that this thiol-exchange process is a nu-
`cleophilic substitution with the attacking thiolate
`approaching along the extension of the S-S bond (22).
`According to this model, the a*(S-S) orbital interacts with
`the approaching nucleophile. Molecular modeling indicates
`that a-methyl groups do provide significant shielding of
`the adjacent S and that this shielding increases with
`increased steric bulk (H < methyl < tert-butyl < dime-
`
`IMMUNOGEN 2082, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`
`
`388 Bioconjugate Chem., Vol. 1, No. 6, 1990
`thyl). That this increased shielding closely parallels the
`measured increases in disulfide bond stability (1-, lo-, loo-,
`and 4000-fold, respectively) strongly argues in favor of steric
`protection.
`The model reactions and conjugations described here
`have been successfully extended to the construction of
`other conjugates between monoclonal antibodies and RTA
`(manuscript in preparation). Preliminary data suggest tha