248
`
`Bioconjugate Chem. 1994, 5, 248-256
`
`Enhanced Stability in Vitro and in Vivo of Immunoconjugates Prepared
`with 5-Methyl-2-imino thiolane
`Stephen F. Carroll,' Susan L. Bernhard,t Dane A. Goff,+ Robert J. Bauer, Will Leach, and
`Ada H. C. Kung
`Departments of Biological Chemistry and Pharmacology/Toxicology, XOMA Corporation, 2910 Seventh Street,
`Berkeley, California 94710 . Received December 3, 1993"
`
`Substituted 2-iminothiolanes (X2ITs) are new heterobifunctional crosslinking agents designed for the
`preparation of disulfide-linked conjugates with enhanced resistance to reduction. Based upon 2-IT
`substituted at the 4 and/or 5 position, these reagents appear to function by sterically protecting the
`conjugate disulfide bond from attack by thiolate nucleophiles. Here, we have used the X2ITs to prepare
`and evaluate a series of immunoconjugates (antibody-cytotoxin conjugates) between the murine
`monoclonal antibody 791/T36, which recognizes a 72-kDa surface antigen present on many human
`tumor cells, and RTA30, the naturally occurring 30-kDa glycoform of ricin A chain. The X2IT-linked
`conjugates were also compared to immunoconjugates prepared with N-succinimidyl3-(2-pyridyldithio)-
`propionate (SPDP) and 4- [(succinimidyloxy)carbonyll-a-methyl-cu-(2-pyridyldithio)toluene (SMPT),
`as well as with methyl- and dimethyl-substituted structural analogs of SPDP. In vitro, 791-(X2IT)-
`TNB model compounds exhibited a 6000-fold range of stabilities. In contrast, the corresponding 791-
`(X2IT)-RTA30 immunoconjugates were up to 20-fold more stable than conjugates made with unhindered
`linkages. These improvements resulted in immunoconjugates with prolonged serum half-lives in animals.
`Our data indicate that one of the crosslinking agents, 5-methyl-2-iminothiolane (MZIT), has optimal
`properties for the preparation of disulfide crosslinked immunoconjugates intended for therapeutic use
`in that (i) it is highly water soluble and reacts rapidly with protein amino groups at neutral pH, preserving
`the positive charge, (ii) it forms conjugates with RTA30efficiently, and (iii) its conjugates exhibit enhanced
`disulfide bond stability in vitro and in vivo. The potential utility of M2IT and other X2ITs for the
`preparation of controlled release protein-drug conjugates is also discussed.
`
`INTRODUCTION
`Immunoconjugates (antibodies linked to cytotoxic pro-
`teins) represent a specialized class of protein-protein
`conjugates designed for therapeutic use (for a review, see
`refs 1 and 2). As such, they typically are prepared by
`covalently crosslinking an antibody molecule to a cytotoxin
`such as the A chain of ricin (RTA).' The antibody thus
`serves to target the action of the cytotoxic component to
`cells bearing the target antigen. Once internalized, the
`cytotoxin is released and then penetrates into the cytosol
`where it enzymically inactivates ribosomes, blocking
`protein synthesis and causing cell death. This approach
`for selective cellular elimination is currently being evalu-
`ated clinically for the treatment of autoimmune disorders
`and cancer (2-8).
`
`Most of the RTA immunoconjugates prepared to date
`have utilized one of two crosslinking reagents, SPDP or
`2IT, to generate a disulfide bond linking antibody to
`cytotoxin. That a reducible bond is required for maximal
`expression of cytotoxic activity has been demonstrated
`by numerous studies (1, 2, 9). However, many such
`conjugates are unstable in animals (10, II), where cleavage
`of the disulfide bond regenerates free antibody and
`cytotoxin. For immunoconjugate therapy this deconju-
`gation has two important consequences. First, it reduces
`the effective concentration of circulating immunoconju-
`gate, and as a result, larger clinical doses may be required.
`Second, the released antibody may remain in circulation
`much longer than does conjugate, where it can compete
`for antigen binding sites on target cells. Thus, the disulfide
`bond linking antibodies to cytotoxin such as RTA must
`be sufficiently labile to facilitate intracellular cytotoxicity,
`but it must also be sufficiently stable to survive admin-
`istration and delivery in vivo.
`To address these issues, several new crosslinking
`reagents have been prepared and tested for immunocon-
`jugate preparation, and the in vitro and in vivo properties
`of such conjugates have been studied. Each of these
`reagents contains one (12,131 or two (14) methyl groups
`adjacent to the disulfide bond, and each has generated
`conjugates with enhanced stability (12-15) and improved
`efficacy (15) in animals. Thus, hindering accesa of reducing
`agents to the antibody-cytotoxin linkage results in im-
`munoconjugates with improved in vivo stability and
`potency.
`Recently, we described the synthesis and preliminary
`characterization of a new family of crosslinking reagents,
`termed X2ITs (16), which are based upon 2-iminothiolane
`(17). The X2ITs offer several advantages over other
`crosslinking reagents: (i) they react with primary amines
`0 1994 American Chemical Society
`
`+ Current address: Chiron Corp., 4560 Horton Street, Em-
`eryville, CA 94608.
`* Current address: Gen Pharm Int., 297 N. Bernard0 Ave.,
`Mountain View, CA 94043.
`Abstract published in Advance A C S Abstracts, March 15,
`1994.
`Abbreviations: 2IT, 2-iminothiolane; 2-ME, 2-mercapto-
`ethanol; 2TP, 2-thiopyridine; DTNB, dithionitrobenzoic acid;
`DTPO, 2,2'dithiobis(pyridine N-oxide); DTDP, 2,2'dithiodipy-
`ridine; GSH, reduced glutathione; HPSEC, high-performance
`size-exclusion chromatography; MSPDP, the methyl-SPDP
`analog N-succinimidyl 3-(2-pyridyldithio)butyrate; RTA, ricin
`toxin A chain; RTABo, the 30-kDa glycoform of RTA; SAMBA,
`the dimethyl-SPDP structural analog N-hydroxysuccinimidyl
`3-methyl-3-(acetylthio)butanoate; SMCC, succinimidyl 4-(N-
`maleimidomethy1)cyclohexane-1-carboxylate; SMPT, 4-[(suc-
`cinimidyloxy)carbonyl] -a-methyl-a-(2-pyridyldithio)toluene;
`SPDP, N-succinimidyl 3-(2-pyridyldithio)propionate; TNB,
`thionitrobenzoic acid; TPO, 2-thiopyridine N-oxide; XBIT,
`2-iminothiolane substituted at the 4 and/or 5 position.
`
`1043- 18Q2/94/29Q5-Q248$Q4.5QlQ
`
`IMMUNOGEN 2076, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Immunoconjugates Made with 2-Iminothiolanes
`to form stable amidinium derivatives that retain the
`positive charge; (ii) inclusion of an aromatic disulfide (such
`as DTNB) in the reaction mixture both activates the newly
`exposed X2IT thiol and allows real-time spectrophoto-
`metric monitoring of the labeling reaction; and (iii)
`variation in the substituent at the 5-position (immediately
`adjacent to the linker thiol) alters the susceptibility of the
`resulting disulfide bond to reduction. For activated model
`compounds, these alterations in steric hindrance resulted
`in disulfide bonds that varied by over 4000-fold in their
`ability to be reduced by glutathione (16). The preparation
`and properties, both i n vitro and in vivo, of RTA
`immunoconjugates prepared with the X2IT reagents are
`the subject of this report.
`EXPERIMENTAL PROCEDURES
`Materials. Solutions of DTNB (Sigma Chemical Co.,
`St. Louis, MO) were prepared as described by Jocelyn
`(18). An ElmMlcm of 14.1 at 412 nm (19) was used to
`determine concentrations of the TNB anion. DTDP and
`DTPO were from Aldrich (Milwaukee, WI); the mM
`extinction coefficients (and wavelengths) used for 2TP
`(343 nm) and TPO (332 nm) were 7.06 and 4.16, respec-
`tively. Stock solutions of GSH (Sigma Chemical Co., St.
`Louis, MO) were prepared in PBS-EDTA (see below);
`prior to use, the concentration of free thiols was quantified
`by reaction with DTNB. DTT, 2-ME, Sephadex G25F,
`Phenyl-Sepharose (all from Sigma Chemical Co., St. Louis,
`MO), trisacryl GF-O5LS, and Ultrogel AcA44 (both from
`IBF Biotechnics, France) were purchased as indicated.
`All other reagents were of analytical grade.
`Crosslinking Reagents. 2IT, SMCC (both from Sigma
`Chemical Co., St. Louis, MO), and SPDP (Pierce Chemical,
`Rockford, IL) were obtained from the indicated sources.
`The X2IT crosslinkingreagents, which have the structures
`shown in Table 1, were synthesized as described previously
`(16). Stock solutions were prepared in water; concentra-
`tions were determined spectrophotometrically by using
`appropriate extinction coefficients at 248 nm (16). In
`addition, three other crosslinkers were prepared for these
`studies. MSPDP and SMPT were synthesized according
`to Worrell et al. (12) and Thorpe et al. (13), respectively,
`with minor modifications. The structure of each linker
`was confirmed by lH NMR. N-Hydroxysuccinimidyl
`3-methyl-3-(acetylthio)butanoate (SAMBA), a dimethyl-
`substituted structural analog of SPDP, was prepared as
`follows: 3-Methyl-3-(acetylthio)butanoic acid (ref 10,2.17
`g, 12.3 mmol) in CHzClz (20 mL) was treated with
`N-hydroxysuccinimide (1.86 g, 16.2 mmol) and dicyclo-
`hexylcarbodiimide (3.34 g, 16.2 mmol) at room temperature
`for 66 h under N2. The reaction mixture was filtered,
`concentrated i n vacuo, and then subjected to flash
`chromatography on Si02 with elution in hexane/EtOAc
`(80/20, v/v, then 50/50). The desired ester was obtained
`as a pale yellow oil (2.78 g, 82% yield) that gave a white
`solid upon standing at room temperature: mp 63 "C; TLC
`(hexane/EtOAc, 80/20, v/v) Rf = 0.27; lH NMR (60 MHz,
`CDC13) 3.23 (s, 2H), 2.80 (8, 4H, NHS ester), 2.27 ( 8 , 3H,
`SAC), 1.60 (s, 6H).
`Preparation of Linker-Modified Antibody. The
`murine IgG2b monoclonal antibody 791/T36 (791, M, ca.
`150000) was produced in an Accusyst hollow fiber
`bioreactor (Endotronics, Minneapolis, MN) and purified
`as described (20). The purified antibody was derivatized
`with each crosslinker so as to incorporate an average of
`1-1.5 linkers per mol of antibody. For modification with
`SPDP, MSPDP, SAMBA, or SMCC, 791 antibody at 2
`mg/mL in reaction buffer (0.1 M NaP04, 0.1 M NaC1, pH
`7.5) was first reacted with a 5- to 10-fold molar excess of
`
`Bioconjugate Chem., Vol. 5, No. 3, 1994 249
`crosslinker (previously dissolved in absolute ethanol).
`Following a 20-min incubation at 20 "C, excess reagent
`and reaction byproducts were removed by size-exclusion
`chromatography on a GF-05LS column equilibrated in
`reaction buffer at 4 "C. The number of crosslinkers
`introduced into the antibody was determined by spec-
`trophotometric analysis following DTT-induced release
`of the 2TP leaving group (21). For some experiments, the
`2TP leaving groups were replaced with TNB by mild
`reduction of the linker-modified antibody (0.1 mM DTT,
`30 min, 25 "C), followed by reaction with 2 mM DTNB
`(30 min, 25 "C). The TNB-activated antibody was then
`purified by size-exclusion chromatography on a column of
`G5-05LS equilibrated in phosphate buffered saline con-
`taining 1 mM EDTA, pH 7.4 (PBS-EDTA) and stored at
`4 "C.
`The reaction of 21T and the X2ITs with 791 antibody
`was monitored spectrophotometrically as follows (16): 791
`antibody (3 mg/mL; 20 pM) and DTNB (2.5 mM) in
`reaction buffer were equilibrated at 25 "C in a l-cm
`disposable cuvette and placed in a dual-beam spectro-
`photometer. An identical solution prepared without
`antibody was placed in the reference position. To initiate
`the reaction, X2IT (freshly dissolved in water) was rapidly
`added to each cuvette with mixing to a final concentration
`of 0.5 mM, and the absorbance at 412 nm was monitored.
`When the A412 reached a value of 0.28 (20 pM TNB, or 1
`mol of TNB per mol of 7911, the reaction mixture was
`rapidly desalted on a l-cm X 20-cm column of Sephadex
`G25F equilibrated at 4 "C in PBS-EDTA. The excluded
`protein peak was pooled and stored at 4 "C. The number
`of linkers introduced per mole of antibody was determined
`spectrophotometrically as follows: The A2, of the linker-
`activated protein was first measured. Then, following
`reaction with 2 mM DTT, released TNB was quantified
`at 412 nm. The corrected proteinA2, was calculated from
`the equation
`A,,(protein) = A,,(nonreduced) - (0.33A4,,(reduced))
`
`and the concentration of 791 antibody was determined by
`using anElmMlCm of 179 at 280 nm (E1 mg/mllcm = 1.2). The
`linkedantibody ratio was then calculated from the molar
`values of TNB and protein.
`Preparation and Purification of Immunoconju-
`gates. Immunoconjugates containing SPDP-, SMPT-,
`and SMCC-activated 791 antibody and RTA30 (the
`naturally occurring 30-kDa glycoform of RTA) were
`prepared essentially as described (16,20). Briefly, linker-
`activated antibody (1-2 mg/mL) in PBS-EDTA was
`reacted with a 5-fold molar excess of freshly reduced
`RTA~o. The disulfide exchange (SPDP, SMPT) or ma-
`leimide-based (SMCC) conjugation reactions proceeded
`for 16 h a t 4 "C. For SAMBA-activated 791 antibody, the
`conjugation reaction proceeded differently. The free thiol
`on RTA30 (5 mg/mL in reaction buffer) was first activated
`by reaction with 2 mM DTNB, and the RTA30-SS-TNB
`was purified by size-exclusion chromatography on Sepha-
`dex G25F. SAMBA-modified 791 antibody was then
`treated with 50 mM hydroxylamine (pH 7.5) for 30 min
`at 25 "C to remove the S-acetyl protecting group, and
`conjugation was initiated by the addition of RTA30-SS-
`TNB (3-fold molar excess).
`Immunoconjugates were separated from excess RTAso
`and reaction byproducts by chromatography on a 1- X
`50-cm column of Ultrogel AcA44 equilibrated at 4 "C in
`reaction buffer. The number of RTA30 molecules
`crosslinked to antibody was determined by densitometric
`analysis of samples following sodium dodecyl sulfate
`
`IMMUNOGEN 2076, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`250 Bioconjugate Chem., Vol. 5, No. 3, 1994
`polyacrylamide gel electrophoresis in 5 % gels under
`nonreducing conditions (22) and Coomassie blue staining.
`The monoconjugate species (1 RTA per 791 antibody) of
`selected immunoconjugates was purified by hydrophobic
`interaction chromatography on Phenyl-Sepharose (201,
`so as to remove residual free antibody and immunocon-
`jugates containing multiple RTA30 moieties.
`Disulfide Bond Stability Assay. The susceptibility
`of 791-RTA30 immunoconjugates to reduction in vitro was
`evaluated in a high-performance size-exclusion chromato-
`graphic assay (HPSEC) which quantifies physical disso-
`ciation of the antibody-RTA3o conjugate2 (23). Immu-
`noconjugates (0.23 mg/mL in PBS-EDTA) were incubated
`at 37 "C for 30 min with increasing concentrations of
`reduced glutathione (0-10 mM). Upon completion, free
`thiols were quenched by the addition of excess iodoacetic
`acid (pH 7.5; final concentration, 50 mM), and aliquots
`were chromatographed on a BioSil TSK-250 column
`(BioRad Labs, Richmond, CA) equilibrated at 25 "C in 50
`mM NaP04,lOO mM Na2S04, pH 6.8. The flow rate was
`1.0 mL/min, the column effluent was monitored at 280
`nm, and the amount of RTASO released was quantified by
`area integration. By comparison to samples incubated
`with 50 mM 2-ME (which resulted in 100% deconjugation),
`plots were constructed correlating percent RTAso release
`with the concentration of glutathione in the incubation
`mixture. That concentration of glutathione which released
`50% of the conjugated RTAN was termed the RC50.
`Cytotoxicity Assay. The cytotoxicities of 791-RTA30
`immunoconjugates were determined using the 791T/M
`osteosarcoma cell line, which expresses the antigen rec-
`ognized by 791 antibody (20). Cells (4 X 105/mL) were
`incubated in a humidified 5% COz incubator with in-
`creasing concentrations of immunoconjugates at 37 "C.
`After 42 h, 3H-thymidine (1 pCi/well) was added, and
`incubation was continued for an additional 18 h. Upon
`completion, cell-associated radioactivity was determined
`by liquid scintillation counting. The IC50 was calculated
`as the concentration of immunoconjugate necessary to
`inhibit incorporation of radioactivity by 50% relative to
`untreated controls. These IC50 values were corrected for
`the number of RTA~o molecules conjugated to antibody
`for each preparation by multiplying the conjugate IC50 by
`the RTA/Ab ratio. These normalized values compensate
`for slight variations in the number of cytotoxins per
`antibody in the different preparations and are expressed
`in terms of pM RTA30.
`Pharmacokinetic Studies. Pharmacokinetic studies
`of selected immunoconjugates were performed in 5-week-
`old male Sprague-Dawley rats (Simonsen Laboratories,
`Gilroy, CA) weighing an average of 149 g (range 122-175
`g) at the initiation of dosing. All animals were delivered
`healthy to the XOMA animal care facility, where they
`were acclimated for at least 5 days prior to dosing, and
`were housed using standard NIH guidelines for husbandry
`procedures.
`Only purified monoconjugates were used in these studies.
`Each monoconjugate was radiolabeled with 1251 by the
`Iodogen method (24) to a specific activity of 0.3-2 mCi/
`mg and was injected intravenously (33-50 pg/kg) into 42
`rats per study (three rats per timepoint). At selected
`timepoints (0.05, 0.25, 0.5, 0.75, 2, 4, 8, 12, 18, 24, 36, 48,
`72, and 96 h), blood samples were collected via the orbital
`sinus, and serum aliquots were counted in an LKB y
`
`Carroll, S. F., Goff, D., Reardan, D., and Trown, P. W. (1989)
`Abstracts from the fourth international conference on monoclonal
`antibody immunoconjugates for cancer, San Diego, CA, p 161.
`
`Table 1. Reaction of Substituted 2-Iminothiolanes with 791
`Antibody.
`
`Carroll et al.
`
`linker
`21T
`
`M2IT
`
`structure
`
`substitution
`(none)
`
`(") +NH2
`5-methyl "0 hi2
`
`+NH2
`
`cF
`C% -7 -cs) fNH2
`
`cH3
`
`Ph2IT
`
`5-phenyl
`
`TB2IT
`
`5-tert-butyl
`
`DMPIT
`
`5-dimethyl
`
`SPIT
`
`5-spiro
`
`R2IT
`
`4,5-ring
`
`reaction rate4
`k X lo6
`5.0
`
`4.0
`
`8.8
`
`7.4
`
`4.6
`
`4.6
`
`5.6
`
`a Rates of reaction of X2ITs (0.5 mM) with 791 antibody (20 pM)
`as monitored by coupling the reaction with 2.5 mM DTNB in 0.1 M
`NaP04,O.l M NaCI, pH 7.5, and monitoring the change in absorbance
`at 412 nm. First-order rate constants were determined from the
`linear slopes of plots for log [XBIT] against time.
`
`counter. Serum samples from each timepoint were also
`analyzed by SDS-PAGE and autoradiography to deter-
`mine the fraction of intact monoconjugate prior to
`pharmacokinetic analysis (13). Pharmacokinetic param-
`eters were determined from a two compartmental analysis
`using the program PCNONLIN (Statistical Consultants,
`Inc., Lexington, KY).
`
`RESULTS
`Reaction of X2ITs with Proteins. The structures of
`the X2IT crosslinking reagents are summarized in Table
`1. Prior studies had shown that the reactivity of the X2ITs
`with the amino group of glycine was relatively unaffected
`by the X2IT ring substituent (16). We therefore examined
`the reactivity of the X2ITs with protein amino groups, in
`preparation for conjugate production. Each X2IT (0.5
`mM) was incubated at pH 7.5 with the murine IgG2b
`monoclonal antibody 791 (20pM) in the presence of DTNB
`(2.5 mM), and changes in the absorbance at 412 nm were
`recorded. Following reaction of the X2ITs with the protein
`amino groups, DTNB undergoes disulfide exchange with
`the newly exposed X2IT thiol to yield a free TNB group
`(monitored at 412 nm) and an activated 791-(X2IT)-SS-
`TNB molecule. This coupling of the reactions between
`protein modification and TNB production simplifies the
`analysis of the rate and extent of the reaction. As was
`found for the reaction with glycine (16), reaction rates for
`the X2ITs with 791 antibody followed first-order kinetics
`and varied less than 2-fold for the entire series of
`crosslinkers (Table 1). Similar reaction conditions were
`therefore employed for the preparation of protein con-
`jugates utilizing each X2IT linker.
`Stability of Model Disulfides. The relative stability
`of X2IT model protein disulfides was assessed by mea-
`suring the rates of release of the TNB leaving group from
`791-(X2IT)-SS-TNB molecules following incubation with
`200 pM reduced glutathione (GSH). For comparison, two
`additional control analogs were examined. 791-(SPDP)-
`SS-TNB was prepared by derivatizing 791 antibody with
`the heterobifunctional crosslinking reagent SPDP and
`replacing the 2TP leaving group with TNB. A similar
`
`IMMUNOGEN 2076, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Immunoconjugates Made with 2-Iminothiolanes
`
`Bioconjugate Chem., Vol. 5, No. 3, 1994 251
`
`100
`
`90
`
`80
`
`2 70
`a
`$ 60
`m
`f 50
`
`c.
`2
`n.
`a,
`
`40
`
`30
`
`20
`
`10
`
`0
`0
`
`21T
`SPDP
`R21T
`M21T
`SMPT
`Ph21T
`TB2lT
`DM21T
`S21T
`
`1 0 0
`
`2 0 0
`
`300
`
`4 0 0
`
`500
`
`600
`
`Time (seconds)
`Figure 1. Glutathione-induced release of TNB from 791-TNB
`analogs. Samples of activated conjugates (791-(X)-SS-TNB, 10
`pM) in PBS-EDTA were placed in a cuvette at 25 "C, and at T
`= 0 "C, reduced glutathione was added to a final concentration
`of 200 pM. The release of TNB was monitored spectrophoto-
`metrically at 412 nm for 500 s, and 2-ME was then added to a
`final concentration of 200 mM to determine maximal release of
`TNB. Results were normalized by quantifying percent maximal
`release, as determined by dividing the absorbance at any
`timepoint by that obtained with 200 mM 2-ME and then
`multiplying the product by 100.
`Table 2. Relative Stabilities of TNB-Activated 791
`Antibody Analogs
`
`stability increase
`relative to
`TNB release ratea
`(k X 104)
`analog
`21Tb
`SPDP"
`791-(2IT)-SS-TNB
`235
`1.0
`0.4
`791-(R2IT)-SS-TNB
`21.7
`10.8
`4.7
`791-(M2IT)-SS-TNB
`18.2
`12.9
`5.6
`791-(Ph2IT)-SS-TNB
`11.5
`20.4
`8.9
`52.2
`4.5
`79 1-( TB2 IT)-S S-TNB
`22.7
`791-(DM2IT)-SS-TNB
`6030
`2620
`0.039
`79 1-( S 2 IT)-SS-TNB
`6530
`2830
`0.036
`791-(SPDP)-SS-TNB
`2.3
`1.0
`102
`16.0
`6.9
`14.7
`791-(SMPT)-SS-TNB
`(20 pM) and
`a Reaction mixtures contained 791-X-SS-TNB
`reduced glutathione (40-103 mM) and were incubated at 25 "C and
`monitored at 412 nm. Plots of log [791-X-SS-TNB] vs time were
`linear for all analogs except 791-(SMPT)-SS-TNB. Pseudo-first-
`order reaction rates were calculated by computerized nonlinear curve
`fitting (GraF'it, version 2.0, Erithacus Software Ltd., Staines, U.K.).
`Relative increase in disulfide stability compared to the 21T analog.
`c Relative increase in disulfide stability compared to the SPDP analog.
`procedure was used to prepare 791-(SMPT)-SS-TNB,
`which incorporates the methyl-hindered crosslinking
`reagent developed by Thorpe et al. (13). Figure 1 indicates
`that the X2IT reagents create model protein disulfides
`which vary greatly in their susceptibility to reduction by
`GSH. However, each of the substituted X2ITs produced
`linkages that were significantly more stable than those
`produced by SPDP or 21T. At appropriate concentrations
`of reductant, pseudo-first-order rate constants for TNB
`release were calculated (Table 2). Relative to 2IT, the
`most stable linkages (DM2IT and SBIT) were more than
`6000-fold more resistant to reduction by GSH. The order
`from least to most stable was as follows: 21T < R2IT <
`M2IT < Ph2IT < TB2IT < DMBIT < SBIT. For the most
`stable analogs (those made with DMBIT and SBIT),
`prolonged incubation (30-60 min) with 200 mM 2-ME
`was required for complete release of TNB. In this assay,
`the stability of the SMPT analog was intermediate between
`those of M2IT and Ph2IT.
`
`v
`
`a
`Q 4 -
`E '- 3 - b
`2 -
`1 -
`mAb -
`
`0
`
`- 200
`
`(Ir- 90
`Figure 2. Conjugate formation by 791-TNB analogs. Each 791-
`(X2IT)-SS-TNB analog (1 TNB per 791 antibody) was incubated
`with a &fold molar excess of freshly reduced RTA30 in PBS-
`EDTA. The final concentration for both 791-(X2IT)-TNB and
`RTA30 was 1.6 mg/mL. After 16 h at 4 "C, aliquots (10 pg) were
`in a 5 % gel under nonreducing
`analyzed by SDS-PAGE
`conditions. Upon completion, the gel was stained with Coomassie
`blue.
`
`Preparation of RTA30 Immunoconjugates. Anti-
`body-RTA immunoconjugates are typically prepared by
`performing a disulfide-exchange reaction between the free
`-SH group of RTA and an activated linker disulfide present
`on the antibody. Because this exchange reaction (like the
`stability assay described above) is essentially a reductive
`cleavage of the activated linker-SS-TNB bond, variations
`might be expected in the efficiency with which the linker-
`activated antibody is converted to immunoconjugate.
`Activated 791-(X2IT)-SS-TNB antibodies (1.0-1.3 link-
`ers/Ab) were therefore individually reacted with a 5-fold
`molar excess of RTA30 for 16 h at 4 "C and then aliquots
`were analyzed by SDS-PAGE. The results (Figure 2)
`suggest an inverse correlation between disulfide bond
`stability and the efficiency of conjugation, as determined
`by either the disappearance of the free antibody band or
`by the appearance of higher molecular weight conjugate
`bands. Utilizing densitometry to quantify the conversion
`of antibody into immunoconjugate, we found that the
`efficiency of conjugation followed the order 21T > MBIT
`> R2IT = Ph2IT > TB2IT. Under identical conditions,
`SMPT-activated antibody was converted to immunocon-
`jugate roughly as efficiently as was TBBIT-activated
`antibody (data not shown).
`No immunoconjugates were detected in reaction mix-
`tures containing DM2IT- or S2IT-modified antibody,
`suggesting that RTA30 (like GSH and 2-ME) could not
`easily displace the TNB leaving group from these two
`linkers disubstituted at the 5 position (immediately
`adjacent to the disulfide bond). Similarly, little or no
`conjugation was detected with DM2IT- or S2IT-modified
`antibody even following prolonged incubation with RTA3o
`for several months at 4 O C or after increasing the incubation
`temperature to 25 or 37 "C. Antibody activated by reaction
`with a dimethyl-substituted analog of SPDP (synthesized
`according to Worrell et al. (12)) was also incapable of
`making conjugates when incubated with an excess of RTA30
`(data not shown).
`Conjugation Efficiency Is Influenced by the Leav-
`ing Group. Because the reactions of RTA30 with DM2IT-
`, TNB- or S2IT-TNB-activated antibodies were not pro-
`ductive, the effect of alternate leaving groups on con-
`jugation efficiency was investigated. Initially, 791-(SP-
`DP)-SS-2TP was studied, together with SPDP linkages
`activated by two additional diary1 disulfides. As before,
`the 2TP leaving group,of SPDP was first removed by mild
`reduction, and the newly exposed linker thiol was then
`reacted with either DTNBor DTPO, thus generating 791-
`(SPDP)-SS-TNB and 791-(SPDP)-SS-TPO,
`re-
`
`IMMUNOGEN 2076, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`252 Bioconjugate Chem., Vol. 5, No. 3, 1994
`
`Carroll et al.
`
`0
`
`CH,
`
`0
`
`200
`
`4 0 0
`
`6 0 0
`
`Time (seconds)
`Figure 3. Glutathione-induced release of leaving groups from
`activated 791 antibody. Aliquota of 791-(SPDP)-SS-2TP were
`converted to the corresponding 791-(SPDP)-SS-TNB and 791-
`(SPDP)-SS-TPO analogs by mild reduction and subsequent
`reaction with the corresponding diary1 disulfide (DTNB and
`DTPO, respectively). The activated 791 antibodies were isolated
`by size-exclusion chromatography, and glutathione-induced
`release of the leaving groups was monitored as described in the
`legend for Figure 1. The final concentration of GSH in these
`assays was 40 pM.
`
`spectively. These compounds were then evaluated for their
`susceptibility to reduction by GSH. As shown in Figure
`3, the TPO derivative was most easily reduced, followed
`by TNB and then 2TP. On the basis of first-order rate
`constants, the release of TPO was &fold faster than TNB
`and 15-fold faster than 2TP. When these activated
`antibodies were reacted with RTAm, conjugation efficiency
`was also highest for the TPO analog, followed again by
`TNB and then 2TP (data not shown). Essentially identical
`results were obtained with the M2IT-activated 791 an-
`tibody (791 reacted with M2IT in the presence of DTDP,
`DTNB, or DTPO); i.e., the TPO derivative was most easily
`reduced and was most efficiently conjugated with R T A ~ o
`(data not shown). In fact, the conjugation efficiency of
`791-(M2IT)-SS-TPO exceeded 95%, even when only a
`3-fold molar excess of RTABo was used for conjugation. As
`before, however, no immunoconjugates were detected
`following reaction of 791-(DM2IT)-SS-TPO or 791-
`(S2IT)-SS-TPO with RTA~o under any of the reaction
`conditions tested.
`Disulfide Bond Stability and Cytotoxicity of 791-
`RTA30 Immunoconjugates in Vitro. The in vitro
`stabilities of 791-RTA30 immunoconjugates were analyzed
`directly by monitoring thiol-dependent release of RTA30.
`In addition to the conjugates described above, three
`additional immunoconjugates were also prepared, purified,
`and tested. 791-(MSPDP)-SS-RTA30
`(which incorpo-
`rates a methyl-substituted analog of SPDP) and the
`thioether-linked conjugate 791-(SMCC)-CS-RTAm (which
`is not reducible) were prepared by standard reactions with
`linker-modified antibodies. The third conjugate was
`prepared in a effort to evaluate an immunoconjugate
`disubstituted at the carbon atom adjacent to the disulfide
`bond and, since conjugations with DM2IT- and S2IT-
`linked antibodies were unsuccessful, required alternate
`chemistries. An analog of SPDP was therefore prepared
`(SAMBA) which incorporated two methyl groups adjacent
`to the linker thiol and an S-acetyl protecting group instead
`of the usual 2TP moiety. Following reaction of the
`
`SAMBA @-NH
`
`0
`y 3
`I1
`- c - c H , - - C - - S S a
`I
`CH3
`Figure 4. Linkage structures formed by the different croaslinking
`agents. For the X2ITs, only M2IT is shown as an example.
`Table 3. Stability and Cytotoxicity of 791-RTAao
`Immunotoxins in Vitro
`
`stability increase
`relative to
`ICE4
`(pMRTAm)
`RCaa 21Tb SPDP'
`immunotoxin
`1.9
`791-(21T)-SS-RTAso
`70.6
`0.6
`1.0
`11.8
`6.2
`791-(R2IT)-SS-RTAm
`65.8
`3.7
`26.2
`13.8
`791-(M2IT)-SS-RTAm
`69.0
`8.3
`16.5
`8.7
`791-(Ph2IT)-SS-RTA30
`94.6
`5.2
`791-(TB2IT)-SS-RTA30
`93.6
`9.5
`5.0
`3.0
`79l-(SPDP)-SS-RTAso
`89.1
`3.2
`1.7
`1.0
`72.2
`6.2
`3.7
`791-(MSPDP)-SS-RTA30
`11.8
`89.6
`32.9
`17.3
`10.4
`791-(SAMBA)-SS-RTA30
`5.7
`3.0
`791-(SMPT)-SS-RTAso
`76.1
`1.8
`9949.5
`nde
`791-(SMCC)-CS-RTA30
`The concentration of reduced glutathione, in mM, that releases
`50% of the RTA3o from the immunotoxin. b Obtained by dividing
`the RCm for each conjugate by 1.9, the value for 791-(2IT)-SS-
`RTA~o. Obtained by dividing the RCw value for each conjugate by
`3.2, the value for 791-(SPDP)-SS-RTAm. The concentration of
`immunotoxin that inhibited protein synthesis in 791T/M cells by
`50%. The data are expressed in terms of RTAso equivalents. The
`RTA/Ab ratios varied between 1.1 and 1.5. Not determined. The
`amount of RTA~o released by reducing agents did not exceed 10%.
`
`SAMBA NHS ester with antibody amino groups, a free
`-SH group was exposed on the linker by treatment with
`hydroxylamine. Thiol-activated RTA30-SS-TNB was
`then added, and conjugation occurred via disulfide ex-
`change, thus producing 79l-(SAMBA)-SS-RTA30. The
`linkage structures of these and other immunoconjugates
`are shown in Figure 4 (note that the linkage made by
`SAMBA is identical to that which would have been made
`by dimethyl-SPDP).
`Following preparation and purification, the stability of
`the disulfide bond linking antibody and RTA30 to reduction
`with GSH i n vitro was then examined. The calculated
`RCmvalues (the concentration of GSH causing 50% release
`of RTA30) for all immunoconjugates are shown in Table
`3. On the basis of these analyses, the dimethyl-substituted
`SAMBA conjugate was the most stable, followed closely
`by conjugates made with M2IT and Ph2IT. However,
`unlike the 791-TNB protein-leaving group compounds
`(which exhibited stabilities over an 6000-fold range), the
`
`IMMUNOGEN 2076, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Immunoconjugates Made with 2-Iminothiolanes
`
`5
`X
`c
`
`c
`
`c
`
`c 1
`- E
`3
`-
`5
`e
`2
`
`1 oc
`
`80
`
`60
`
`40
`
`20
`
`0
`
`1 0
`
`20
`
`30
`
`40
`
`50
`
`Time (hours)
`Figure 5. Stability of 791-RTA30 immunoconjugates in rats.
`1251-labeled immunoconjugate monoconjugates were injected iu
`into rats as described in the Experimental Procedures. As a
`function of time thereafter, serum samples were collected and
`analyzed by SDS-PAGE and autoradiography. The percentage
`of radioactivity in serum associated with intact immunoconjugate
`at each timepoint was then quantified by densitometry.
`
`increases in stability for the series of 791-RTA30 protein-
`protein conjugates varied by less than a factor of 20. In
`all cases, the 2IT-linked forms were the least stable.
`The cytotoxic activity of each 791 immunoconjugate
`was accessed against 791T/M cells in our standard 60-h
`assay (20). The IC50 values calculated for all conjugates
`are presented in Table 4. Despite the 20-fold range of
`stabilities in the in vitro disulfide stability assay, each of
`the conjugates was highly cytotoxic and of similar potency
`for antigen-bearin