(CANCER RESEARCH 50. 6600-6607. October 15. 1990]

Evaluation in Vitro of Adriamycin Immunoconjugates Synthesized Using anAcid-sensitive Hydrazone LinkerRobert S. Greenfield,' Takushi Kaneko, Ann Daues, Mary A. Edson, Kathleen A. Fitzgerald, Lee J. Olech,2 James

A. Grattan, George L. Spitalny, and Gary R. BraslawskyBristol-Myers Squibb Company, Pharmaceutical Research Institute, H'allingford, Connecticut 06492-7660

ABSTRACT

A novel method for linking Adriamycin (ADM) to monoclonal antibodies is described in which the 13-keto position of the anthracycline isused as the attachment site to the linker arm. A new ADM acylhydrazonederivative, Adriamycin 13-(3-(2-pyridyldithio)propionyl|hydrazone hy-drochloride, which contains a pyridyl-protected disulfide, was synthesizedand used for conjugation to monoclonal antibodies (MAbs) that »erethiolated with /V-succinimidyl 3-(pyridyldithiol)propionate or 2-imino-thiolane. This resulted in formation of a linker between M \l> and drugthat contained a disulfide bond. Conjugation conditions were optimizedto yield conjugates with high ADM:MAb molar ratios. The final inimu-noconjugate yields were found to decrease as the ADM:MAb molar ratioof the conjugates increased. Stability studies indicated that ADM wasreleased from the immunoconjugates at mildly acidic pHs ranging from4.5-6.5. Treatment of immunoconjugates with mild reducing agent dithi-othreitol resulted in release of an acylhydrazone derivative of ADM.Flow-cytometric studies showed that the binding activity of various MAbsfollowing conjugation to ADM was preserved at AI>\I:M \l> molar ratiosup to 10. Antibody-directed cytotoxicity was demonstrated under severalassay conditions using combinations of antigen-positive and antigen-negative cells and binding and nonbinding immunoconjugates. In severalexperiments, ADM immunoconjugates were more potent than equivalentamounts of unconjugated ADM.

INTRODUCTION

ADM1 is an important anti-tumor antibiotic used in the

treatment of leukemia, breast carcinoma, and other cancers.However, the efficacy of ADM therapy is limited by the dose-dependent toxic side effects, which include bone marrowsuppression and cardiotoxicity (1, 2). Recently, much interesthas focused on linking anticancer agents to antibodies directedagainst tumor-associated antigens in order to improve the therapeutic efficacy of the drugs.

A number of reports have shown that Adriamycin and theclosely related anthracycline Daunomycin can be linked topolyclonal and monoclonal antibodies using many differentchemical strategies (for reviews, see Refs. 3-5). The mostfrequently used approach for attachment has been through theamino sugar moiety of the anthracycline (4, 6). Alternatively,anthracyclines have been directly linked to antibodies throughcarbodiimide-mediated linkage of amino sugar to carboxylgroups on the antibody (4) and by cross-linking amino groupsof the drug and antibody with glutaraldehyde (4, 7). In other

Received 6/16/89; accepted 6/26/90.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1To whom requests for reprints should be addressed, at Bristol-Myers Company. Pharmaceutical Research and Development Division. 5 Research Parkway.P. O. Box 5100, Wallingford, CT 06492-7660.

2Present address: Bio-Rad. 1414 Harborway South. Richmond, CA 94804.3The abbreviations used are: ADM. Adriamycin; 2-IT, 2-iminothiolane; s,

singlet; d, doublet: t. triplet; m. multiplet; bs. broad singlet: DTT. dithiothreitol;ADM-HZN. Adriamycin 13-(3-(2-pyridyldithio)propionyl]hydrazone; HPLC.high-pressure liquid chromatography; MAb. monoclonal antibody; PBS. phosphate-buffered saline, pH 7.2; SPDP, A'-succinimidyl 3-(pyridyldithiol)-propionate.

studies, the amino sugar of anthracyclines has been linked tohigh-molecular-weight dextran carriers that were covalentlyattached to antibodies (8-11).

Modifications of the amino sugar of anthracyclines had beenshown to decrease the cytotoxic activity of the drug (12, 13).Drug conjugation strategies utilizing the amino sugar portionof the anthracycline molecule also resulted in a loss of thecytotoxic activity of the conjugated drug (4, 6). Dillman et al.(14) reported that mixtures of doxorubicin and MAb T101 weresignificantly more efficacious than TlOl-dextran-doxorubicinconjugates. More recently, an acid-sensitive c/s-aconityl linkerhas been used to attach MAbs to the amino sugar of doxorubicin(15) and daunorubicin (16). This conjugation scheme allows forrelease of unmodified drug under mild acidic conditions fromthe antibody. Other groups have reported that immunoconjugates in which anthracyclines were attached to the MAb at theC-14 position retained cytotoxic drug activity, and exhibitedselective (antibody-directed) cytotoxicity in vitro (17, 18). Thus,the activity of conjugated ADM appears to be best retained byusing linking strategies that preserve the amino sugar portionof the anthracycline.

In this study, we describe conjugation of ADM to MAbsusing a new methodology that has several advantages over otherlinker chemistries. ADM-HZN, a new derivative of ADM thatcontains a pyridyl-protected disulfide attached through the 13-keto position by an acylhydrazone bond, was synthesized. Immunoconjugates were prepared by reacting ADM-HZN withthiolated MAbs resulting in attachment of the drug to theantibody through the 13-keto position of ADM. Syntheticconditions are described that produced immunoconjugates withdrug/antibody molar ratios ranging between 2 and 10. Thestability of ADM-HZN immunoconjugates was studied following treatment with a mild reducing agent and low pH. Immunoconjugates prepared with a variety of MAbs were shown tohave retained MAb binding activity to a great extent. Retentionof cytotoxic drug activity of conjugated ADM was determinedin vitro on human lymphoma and carcinoma cells using colonyformation assays. Thus, synthesis of ADM immunoconjugatesby linking MAbs to the 13-keto position of ADM represents anew linker strategy in which high cytotoxic drug activity isretained and antibody-directed killing of antigen-bearing tumortarget cells can be demonstrated.

MATERIALS AND METHODS

Chemical Reagents. SPDP and 2-IT were obtained from PierceChemical Co. Adriamycin-HCl was obtained from Sanraku, Inc. (Japan). All other reagents were of the highest research grade available.

Monoclonal Antibodies. Hybridomas secreting MAbs 5E9 (anti-trans-ferrin receptor) and T33A1 (anti-M, 40,000 human T-cell antigen) wereobtained from the American Type Culture Collection. Antibodies werepurified from ascitic fluid produced in BALB/c mice according to theprocedure of Brücket al. (19). Purified MAbs G28.5 (anti-A/r 50,000human B-cell antigen), G28.1 (anti-A/r 39,000 human B-cell antigen),and Id (anti-human non-small cell lung carcinoma glycolipid antigen)

6600

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

were provided by Drs. J. Ledbetter and I. Hellström(Oncogen, Seattle,WA).

Cell Lines. HSB2 (human T-cell leukemia), Daudi (Burkitt lym-phoma) and Namalwa (Burkitt's lymphoma) were obtained from the

American Type Culture Collection. HCT116 (human colon carcinoma)was a gift of Dr. M. Brattain. All cell lines were propagated in RPMI1640 containing 10% fetal bovine serum, penicillin, and streptomycin.Cells were grown at 37°Cin a humid atmosphere containing 5% CO2.

Thiolation of MAbs. Thiolation of MAbs with SPDP was carried outas follows (see Fig. 2). SPDP (50 HIM),dissolved in ethanol, was addedto MAb (5-10 mg/ml) in PBS to give a final concentration between 5and 10 mivi. The reaction mixture was incubated for 30 min at 30°C.

Unreacted SPDP was separated from SPDP-derivatized MAb by gelfiltration chromatography using a PD-10 column (Pharmacia). Thethiopyridyl protecting groups were removed by reduction with excessDTT. The reduced antibodies were passed through a PD-10 columnand the free thiol containing antibodies were used directly beforecondensation with the Adriamycin hydrazone derivative (describedbelow).

Reactive thiol groups were also introduced onto the MAb proteinusing 2-IT (see Fig. 2): MAbs (5-10 mg/ml in 50 mivitriethylamine-50mM NaCl-1 mM EDTA, pH 8.0) were mixed with 2-IT at a finalconcentration of 5-10 mM. The reaction was allowed to proceed for 90min at 4°C,and thiolated MAbs separated on a PD-10 column equili

brated with 2 M NaCl/PBS.The number of reactive thiol groups incorporated onto the MAbs

was determined using 5,5'-dithiobis-(2-nitrobenzoic acid) (<4i2 =

14,150), according to the procedure of Ellman (20).Synthesis of ADM-HZN. To a cooled solution of SPDP (70 mg,

0.22 mmol) in 3 ml of tetrahydrofuran was added 0.3 ml of IMhydrazine solution in isopropyl alcohol. After 20 min of stirring at 0°C,

the product was extracted with CH2C12 and washed with brine anddried over K2CO3.The residue obtained after evaporation of the solventswas chromatographed on neutral alumina (5% methanol-95% CH2C12)to give 21 mg (41%) of compound 2 (see Fig. 1). Compound 2 andAdriamycin-HCl (48 mg, 0.083 mmol) were dissolved in 5 ml ofmethanol and stirred in the dark at room temperature for 6 days. Thereaction was followed by a reversed-phase thin-layer chromatography(methanol:H2O, 2:1, containing 3% w/v NH4OAc). After this period,the solvent was evaporated and the residue was chromatographed on aCig column (methanol:H2O = 3:2, containing 3% w/v ammoniumacetate). The fractions were combined, lyophilized, and excess ammonium acetate was removed under reduced pressure. The residue wasdissolved in methanol and precipitated by addition of acetonitrile togive 45 mg (72%) of 4: m.p. > 125°Cdarkens its color and not well

defined; nuclear magnetic resonance (acetone-i4, 5) 1.25 (s, 3H, J = 6Hz), 1.77 (m, IH), 2.06 (m, IH), 2.30 (m, IH), 2.53 (d, IH, J = 15Hz), 2.89-3.18 (m, 6H), 3.71 (m, IH), 3.85 (m, IH), 3.97 (m, 1H),4.07 (s, 3H), 4.78 (s, 2H), 5.21 (m, IH), 5.58 (t, IH, J = 7 Hz), 7.12(m, IH), 7.64 (d, IH, J = 8 Hz), 7.75 (m, 2H), 7.90 (t, IH, J = 8 Hz),7.98 (d, IH, J = 8 Hz), 8.37 (d, IH, J = 4 Hz), 10.50 (s, 1H), 10.52 (s,IH), 14.19 (bs, IH); IR (KBr) 3438, 1674, 1618, 1579, 1419, 1286,1016, 988, 698 cm"1; fast atom bombardment mass spectrometry

(glycerol) m/e 755 (M + 1), 737, 645, 625, 609. The ADM-HZN usedin these studies was greater than 99% pure by HPLC analysis.

Conjugation of Thiolated MAbs with ADM-HZN. ADM-HZN wasdissolved in methanol and added to SPDP-thiolated MAbs in PBS orto 2-IT thiolated MAbs in 2 M NaCl/PBS (Fig. 2). In a typicalexperiment, 10 equivalents of ADM-HZN were added to MAbs containing 10-20 thiol groups. Conjugation reaction was allowed to incubate overnight at 4°C.The reaction mixture was centrifuged at 10,000x g and conjugated drug was separated from unreacted ADM-HZN bypassage through a PD-10 column. The amount of conjugated anthra-cycline bound to antibody was determined by absorbance at 495 nm((495= 8030). The amount of antibody protein was determined byabsorbance at 280 nm (1 mg/ml = 1.4 absorbance units). To correctfor the overlap of ADM absorbance at 280 nm, the following formulawas used:

Conjugates were analyzed for the presence of unconjugated ADM orADM derivatives using HPLC analysis. HPLC was done using aPhenomenex column packed with 5-^m IB-SIL C,8 beads. Unconjugated ADM-HC1 or ADM-HZN (0.1 Mmol) standards were applied tothe column and eluted with methanol and 10 mM ammonium phosphate, pH 4.5 (70:30), at 1.5 ml/min. Immunoconjugates containing0.5-5 /¿moldrug equivalents were mixed with an equal volume ofacetonitrile and the organic extract was analyzed for drug by HPLC asabove. All immunoconjugates contained no significant amount (<5%)of unconjugated drug by HPLC analysis.

Flow Cytometry. Immunoconjugates were incubated at 4°Cin 0.1 mlof growth media containing 1 x IO6tumor target cells. After 1 h, cells

were twice washed and incubated for an additional 30 min in 0.1 ml ofmedium containing 1:40 dilution of fluorescein isothiocyanate-labeledgoat anti-mouse IgG (Boehringer-Mannheim). Cells were analyzed ona Coulter Epics V fluorescence cell analyzer. Cell surface fluorescencewas compared with the fluorescence obtained using similarly dilutedunconjugated MAb. The percentage of control binding was determinedby dividing the fluorescence intensity (mean channel number) of theimmunoconjugate (at the dilution that unconjugated MAb saturated allantigenic sites on the tumor target cell) by the fluorescence intensity ofthe unconjugated MAb at the same dilution.

Binding Assay. Conjugate binding activity was determined by competition assay using I25l-labeled MAb. Cells (1 x IO6) were suspended

in 0.1 ml of RPMI 1640 containing 2% fetal bovine serum and mixedwith 0.1 ml of 2-fold serially diluted MAb or immunoconjugate atconcentrations starting at 50 Mg/"1'- The cell suspensions were incubated while mixing at 4°Cfor 1 h. Each point was done in duplicate.

Cells were washed twice and suspended in 0.1 ml containing 5 Mg/mlof 125I-labeledhomologous MAb. Samples were incubated at 4°Cfor 1

h and overlaid onto 0.15 ml 1:1 mixture of dibutyhdinonyl phthalatethat was cooled to 4°C.The samples were centrifuged at 10,000 x gfor 1 min at 4°Cand the cell-bound counts (pellet) determined using

an LKB gamma-counter.Soft Agar Colony Formation Assay. Lymphoma cells (1 x IO5) in

complete medium were exposed to serially diluted conjugated or unconjugated ADM. Triplicate determinations were done for each dilution. Controls consisted of similarly treated cells not exposed to drugs.Cells were incubated for various periods of time, washed, and suspendedin RPMI 1640 containing 15% fetal bovine serum and 0.3% agarose(Marine Colloid). One ml (5 x IO3 cells) of the cell suspension wasoverlaid onto a 0.4% agarose layer in 6-welI microtiter plates (Costar).Samples were incubated for 7-10 days at 37°Cand the resulting colonies

were stained with 0.5 ml of 1 mg/ml of p-iodonitrotetrazolium violet(Sigma) for 48 h. Colonies were counted using an Optimax 40-10 imageanalyzer and the inhibition of colony formation was determined bycomparing drug-, immunoconjugate-, or MAb-treated cells with theuntreated control.

Colony Formation on Plastic. Monolayer cultures were removed fromculture flasks with trypsin-EDTA (GIBCO), washed, and passedthrough a 22-gauge needle to obtain single cell suspensions. Conjugatedor unconjugated ADM was serially diluted in 0.2 ml of mediumcontaining 1 x IO5cells. Each dilution was done in triplicate. Controlsincluded untreated or MAb-treated cells. Cells were incubated for 24h, washed once in medium and 1 x IO3cells in 1 ml were plated in 12-well microtiter plates (Costar). Plates were incubated for 7-10 days at37°Cand fixed with absolute methanol (10 min). The colonies were

stained with crystal violet and counted on an Optimax 40-10 imageanalyzer.

Limiting Dilution Assay. Cytotoxicity of immunoconjugates was alsodetermined using a limiting dilution assay with Namalwa cells essentially as described by Colombatti et al. (21). Log cell kill was calculatedbased upon the plating efficiencies that were estimated by the portionof wells without growth at limiting cell concentrations.

RESULTS

MAb (mg/ml) =- (0.72 x

1.4Synthesis of ADM Immunoconjugates. Synthesis of the im

munoconjugates was accomplished by thiolating the MAbs with6601

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

heterobifunctional reagents SPDP or 2-IT and reacting thethiolated MAbs with the ADM-HZN derivative, which wasprepared as shown in Fig. 1. The addition of MAbs containingfree thiol groups to ADM-HZN, which contained a pyridyl-protected disulfide, led to the formation of a disulfide bond inthe linker arm joining ADM to the M Ab. The synthetic pathwayused in the preparation of the immunoconjugates is shown inFig. 2.

In this scheme, the number of drug molecules convalentlybound to the antibody depended on the number of thiol groupsintroduced on the MAb. Up to 20 thiol groups could be attachedto various murine MAbs of the IgGl and IgG2 subclasses withhigh protein yields (60-100%). In addition, it was also foundthat substitution of up to 20 thiol groups onto MAbs did notlead to significant reduction in the binding activity of a numberof MAbs (data not shown).

The ADMrMAb molar ratio achieved depended upon thenumber of thiol groups linked to the MAb and the amount of

V!, V,

Fig. 1. Synthetic scheme representing the synthesis of ADM-HZN that wasused in the preparation of ADM immunoconjugates.

Av;-. N- •ss

SPDP

CONHMAb

DTT

MAb-NHCO^

ADM-HZN

-SH

MAb-NH,

2IT

NH,*cr ADM-HZN

0 OH

CH.0 I ••

OH'OH

2HIlNH

Fig. 2. Synthesis of ADM immunoconjugates by reaction of SPDP (A) or 2-IT (B) Ihiolaled MAbs with ADM-HZN. Condensation of the drug with eitherthiolated MAb leads to formation of disulfide bond in the linker arm and anacylhydrazonc bond at the 13-keto position of ADM.

ADM-HZN derivative added to the thiolated MAb. The scat-tergram in Fig. 3 shows that ADM:MAb ratios of 3-5 wereachieved when ADM-HZN was condensed with MAb containing approximately 7-15 thiol groups. Typically, a 10-fold molarexcess of ADM-HZN to protein were added in these reactions.ADMrMAb ratios could be increased by addition of ADM-HZN to MAbs containing 18-25 thiol groups. No significantdifferences in the ADM:MAb ratios were observed when SPDPor 2-IT was used to thiolate the MAb. However, final proteinyields following conjugation of drug to MAb appeared to besomewhat higher with SPDP as compared with 2-IT. Proteinyields of 50-80% were commonly obtained for SPDP-thiolatedMAbs, whereas yields of 20-50% were more common for 2-IT-thiolated MAbs.

Binding Activity of ADM Immunoconjugates. Table 1 showsthe retention of MAb binding activity for ADM-HZN immunoconjugates prepared with either the 5E9 or 3A1 MAbs. 5E9conjugates with molar ratios ranging between 3.5 and 8.5retained most of their binding activity as compared with uncon-jugated 5E9. 5E9 immunoconjugates prepared using 2-IT orSPDP had similar binding activities. The 3A1 conjugates prepared using SPDP showed some loss in MAb-binding activity.In general, conjugation of ADM to these and to other MAbs(not shown) resulted in loss of relatively small degrees of MAb-binding activity. We also observed that any significant losses inbinding activity and protein yields occurred following reactionof ADM-HZN with thiolated MAb rather than after the introduction of thiol groups onto the MAbs.

Stability of ADM Immunoconjugate under Acidic Conditions.Acylhydrazone bonds can undergo hydrolysis under acidic conditions. It should be possible to release unmodified ADMdirectly from the MAb protein by hydrolysis of the hydrazonebond at the 13-keto position. The stability of an L6 immuno-conjugate was determined at various pHs ranging from 4.0 to7.0 (Fig. 4). After incubation for 24 h, a single product, whichhad spectral characteristics and column retention time the sameas that of the ADM standard, was detected by HPLC. Theamount of ADM released from the immunoconjugate after 24h increased as the pH was lowered from 7 to 4. Kinetic studiesshowed that at pH 4.5, the release of ADM from MAb proteinis quantitative and rapid (tv, = 2.5 h, data not shown). Theimmunoconjugates were also found to be quite stable whenincubated for 24 h in mouse or human serum (data not shown).It thus appears that this chemical linkage strategy allows forrelease of unmodified ADM from the ADM-HZN immunoconjugate under mild acidic conditions typically found in endoso-mal and lysosomal vesicles.

Reduction of ADM Immunoconjugates with DTT. In the conjugation procedure, the linker arm is generated by formation ofa disulfide bond between the thiolated MAb and ADM-HZN.It should thus be possible to release an anthracycline moiety bytreatment of the immunoconjugate with a reducing agent suchas DTT. After treatment of an L6 conjugate with 10-fold excessDTT, a single anthracycline product, which had a columnretention time different from that of ADM-HC1 or ADM-HZN,appeared in the HPLC chromatogram (Fig. 5).

In similar studies, the same product was formed after thereduction of ADM-HZN with dithioerythritol (Fig. 6B). Theanthracycline derivative, Adriamycin 13-(3-thiopropionyl)hy-

drazone, released by reduction upon treatment with acid (pH2.8) quantitatively yielded unmodified ADM (Fig. 6C). Thus,free ADM can be obtained from ADM immunoconjugatesthrough 2 reaction pathways as depicted in Fig. 7.

6602

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

THIOLATING AGENT : SPDP THIOLATING AGENT : 2-IT

lOx

Fig. 3. A, Scattergram comparing theSH:MAb molar ratio with the finalADM:M Ab molar ratio achieved after condensation with ADM-HZN to SPDP-thiolatedMAbs. n, Conjugates made using MAb 5E9;O, those made using 3A1. B, Same as A exceptMAbs were thiolated with 2-IT.

ìte4

m<

B--

6- •

B ••

•

10 15 20 25

io-0MKÅ’

6-_1

oz3

4-Zz

o2-o-C'Ban

*n•n

•ai

i i « 4:5

10 15 20 25 30

SH/MAB MOLAR RATIO SH/MAB MOLAR RATIO

Table Relative binding affinity estimates after ADM conjugation to MAbproteins

InhibitorSPDP

linker5E9-ADM3A1-ADM2-IT

linker5E9-ADM3A1-ADMMolar

ratio3.54.04.96.88.52.63.34.26.75.66.76.0[Ai-Trace X IO"9M5E9-'"I3A1-'"15E9-'"I

4.26.74.24.20.13A1-125I

4.21.38.37.35E9-'"1

4.14.73A1-'"I

4.0mX

10"'M3.42.14.02.12.11.31.31.31.34.04.01.3/(conjx IO7

liters/mol3.25.12.61.03.01.60.71.55.10.80.93.12.71.6

" [Ii], Molar concentration of antibody conjugate giving 50% inhibition oftracer antibody; [7",], molar concentration of antibody giving 50% inhibition oftracer antibody; A'conJ,relative affinities calculated as

K conjIT,]KA,

[I,]

where AuAis the equilibrium constant of the unconjugated MAb as determined byScatchard analysis.

Cytotoxic Activity of ADM Immunoconjugates on HumanLymphoid Tumor Cells. ADM-HZN immunoconjugates weretested in vitro for cytotoxicity using colony formation in softagar. Fig. 8 compares the cytotoxic activity of 5E9-ADM-7.5and 3A1-ADM-7.0 immunoconjugates (thiolated with 2-IT)after 1.5-h incubation with the 5E9 antigen-positive and 3A1antigen-negative Burkitt's lymphoma cell line, Daudi. Compar

ison of the dose-response curves shows that the 5E9-ADM-7.5conjugate, which retained 93% of the original binding activityfor antigen-bearing target cells, was significantly more potentthan 3A1-ADM-7.0, the nonbinding control conjugate.

The limiting dilution assay, which provides a measure of thelog cell kill, was used to test for cytotoxic drug activity of thesame two ADM-HZN-immunoconjugates towards Namalwacells (5E9+, 3A1—)using a longer exposure format (24 h). Asdescribed in Fig. 9, 5E9-ADM-7.5 conjugate produced 1-2 logsgreater cell kill over a wide concentration range as comparedwith the nonbinding 3A1-ADM-7.0 conjugate. Up to 5 logs of

TIME IMIN.I

Fig. 4. Stability of L6-ADM-9.0 immunoconjugates over the pH range between 4-7. The conjugate was incubated in phosphate buffers at the indicatedpHs for 24 h at 37°C.The amount of free ADM released was determined by

HPLC. Untreated control represented the chromatogram of conjugate in pH 7.4phosphate buffer.

TIME tMIHi

Fig. 5. Release of anthracycline from an ADM immunoconjugate by reductionwith DTT. L6-ADM-9.0 was treated with excess DTT at room temperature for15 min and applied to HPLC column. ADM-HCI and ADM-HZN standardswere run at the same time and the peaks from the chromatograms are indicated.L6-ADM-9.0 had no detectable peak of unconjugated drug prior to the additionof DTT.

cell kill were measured at the highest dose. The activity of the5E9 conjugate was equal to and at several of the lower concentrations greater than an equivalent dose of free ADM. Whilecytotoxic activity for the nonbinding 3A1 immunoconjugate

6603

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

A. ADM-HZN

=4:B. ADM-HZN + DIE

C. PRODUCT B tpH 2.8

Fig. 6. Generation of ADM from ADM-HZN following reduction and acidhydrolysis. A. HPLC chromatogram of ADM-HZN; B, HPLC chromatogramfollowing treatment of ADM-HZN with 1.6 M excessdithioerythritol. C. HPLCchromatogram following hydrolysis of the compound in B with pH 2.8 buffer.Arrows, position of ADM-HCI standard eluted under identical Chromatographieconditions.

was detected, the level of cytotoxicity was less than an equivalent amount of unconjugated ADM.

The 5E9-ADM-7.5 and 3A1-ADM-7.0 conjugates describedabove were synthesized using 2-iminothiolane as the protein-thiolating agent. Immunospecific cytotoxicity was also observedwith immunoconjugates that were prepared using SPDP as thethiolating agent. Fig. 10 shows selective cytotoxic activity onDaudi cells of 5E9 and 3A1 immunoconjugates prepared usingSPDP as the thiolating agent. Selective cytotoxicity was alsoto be demonstrated with immunoconjugates prepared withMAbs other than 5E9. As shown in Fig. 11, G28.1-ADM-9.0conjugate prepared using SPDP was cytotoxic toward 2 antigen-positive cell lines (Daudi, Namalwa) but not the antigen-

negative cell line (HSB2). HSB2 cells had comparable sensitivity to ADM (50% inhibitory concentration = 4.5 x IO"8 M) asDaudi (7 x 10~8M).

Cytotoxic Activity: Carcinoma Cells. Preferential killing ofantigen-positive cells by 5E9-ADM-7.5 was also observed usingan anchorage-dependent carcinoma cell line. As shown in Fig.12, greater cytotoxicity was observed when the HCT 116 cells(5E9+, 3A1-) were reacted with 5E9-ADM-7.5 as comparedwith 3A1-ADM-7.0.

DISCUSSION

In this study, we describe a novel method for linking ADMto MAbs via an acylhydrazone bond at the 13-keto position ofADM. Immunoconjugates were constructed by condensation ofthiolated MAbs with a new 13-acylhydrazone ADM derivative,ADM-HZN, which contains a reactive pyridyl disulfídebond.This results in formation of a linker arm containing a disulfídebond and more importantly, an acid-sensitive acylhydrazonebond at the 13-keto position of ADM. One of the uniquefeatures of this conjugation strategy is that there are 2 cleavablesites in the linker arm that allow for the anthracycline to bereleased from the MAb protein. Unmodified ADM was shownto be rapidly released by hydrolysis under mild acidic pHsranging from 4-5.5, which are typically found in endosomaland lysosomal vesicles. The disulfídebond in the middle of thelinker arm was shown to be cleaved by mild reducing agents.This leads to release of an ADM derivative, which upon treatment with mild acid generates unmodified ADM. Thus, thehigh cytotoxic drug activity observed with the ADM immunoconjugates most likely is a direct consequence of the ability ofunmodified ADM to be readily released from the MAb protein

Fig. 7. Chemical pathways for generatingunmodified ADM from ADM-immunoconju-gates by direct hydrolysis or following reduction of the disulfide bond.

~S--nflB

0 OH

6604

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

„-5

im]

Fig. 8. Cytotoxicity of ADM (•),5E9-ADM-7.5 (•),and 3A1-ADM-7.0 (O)immunoconjugates on the Dandi cell line (phenotype: 5E9+, 3A1-). Immuno-conjugates and ADM were incubated with Daudi cells for 1.5 h. washed, andCytotoxicity determined using soft agar colony formation assay as described. Bars,SD of triplicates.

Fig. 9. Cytotoxic effects of 5E9-ADM-7.5 (•),3A1-ADM-7.0 (A), and ADM(•)towards Namalwa cells (phenotype 5E9+, 3A1 -) as determined by the limitingdilution assay. Cells were incubated with immunoconjugates for 22 h, washed,and log cell kill was determined as described in "Materials and Methods."

in the intracellular environment of tumor cells (described below).

In other reports, immunoconjugates prepared by linkingMAbs to the amino sugar of anthracyclines through an acid-sensitive c/s-aconityl linker were described to have retained highcytolytic activity (15, 16). However, the m-aconityl linkerchemistry can lead to formation of heterogeneous chemicallinkages between MAb and drug. One of the advantages of theADM-HZN chemistry is that uniform chemical linkages between drug and MAb are formed. Ogden et al. (22) reportedthat 30% of the ADM in the immunoconjugates synthesizedusing the m-aconityl linker was noncovalently bound to theMAb. ADM-HZN immunoconjugates have been placed inbuffer containing 6 M urea or 1% sodium dodecyl sulfate andfree drug was extracted with acetonitrile. Under these denaturing conditions, less than 5% of total drug in the conjugate wasfound by HPLC in the organic extract (data not shown). Thesefindings, along with the demonstration of quantitative releaseof drug under acidic and reducing conditions, indicate that

almost all of the ADM in the conjugate preparations wascovalently bound to MAb.

Experimental evidence presented indicated that the cytotoxicactivity of the ADM-HZN immunoconjugates could be attributed to delivery of drug to target cells by the MAb. The cytotoxicactivity of the ADM immunoconjugates was shown to be greaterwhen tested on antigen-bearing target cells than on non-antigen-bearing cells. Immunospecific cytotoxic effects were observedwith conjugates prepared with either SPDP or 2-IT. Two 5E9

log ADM (M/L)Fig. 10. Preferential Cytotoxicity of ADM immunoconjugates prepared using

SPDP as the thiolating agent. 5E9-ADM (•)and 3A1-ADM (A) conjugates wereincubated with Daudi cells for 1.5 h and Cytotoxicity was determined using thesoft agar colony formation assay. Bars, SD of triplicates.

lOOr

Zo

trou.

oCJ

mMI

LOG ADM (M/L)Fig. 11. G28. l-ADM-9.0 immunoconjugate is selectively cytotoxic towards

two G28.1 antigen-positive cells Daudi (•)and Namalwa (•),but not G28.I-antigen-negative cell line HSB-2 (A). Assay protocol same as in Fig. 11. Bars, SDof triplicates.

6605

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

«o-

••-,r.-fiiì

0.1 5.0CONCENTRATIONlug/mil

HECHA

Fig. 12. Preferential cytotoxic effects of ADM-immunoconjugates on anchorage-dependent HCT116 (5E9+. 3A1-) cells. HCT116 cells were exposed to 5E9-ADM-7.0 (H), 3AI-ADM-7.0 (D). or medium alone (R) for 24 h. and cytotoxicitywas determined by colony formation in liquid medium. Bars. SD of quadruplicates.

antigen-positive lymphomas (Daudi, Namalwa) and one 5E9-positive carcinoma (HCT116) were preferentially killed by the5E9 immunoconjugate but not by the nonbinding control 3A1immunoconjugate. Immunospecific cytotoxicity was also demonstrated with a G28.1 immunoconjugate, which was cytotoxictoward two antigen-bearing target cells but was not cytotoxictoward the antigen-negative target cells. Binding and cytotoxicity properties of the immunoconjugates reported here appearto represent an improvement over immunoconjugates describedin the literature in which ADM or daunomycin were directlylinked to antibodies through the amino sugar moiety (3-5). Inother reports, amino sugar-linked immunoconjugates containedlow ADM:MAb ratios, had impaired antibody binding properties, and exhibited reduced drug activity (4, 6).

For drug targeting with MAbs, one method to increase theamount of drug delivered to the tumor cell surface is to attachas many drug molecules to MAb as possible. We were able tooptimize the chemistry to yield conjugates that attainADM:MAb ratios between 6 and 10 using several MAbs ofdifferent isotypes. The amount of protein recovered after condensation with the ADM-HZN derivative dropped off dramatically when molar ratios greater than 10 were attempted. Itappeared that the major limitation in obtaining conjugates withmolar ratios greater than 10 was due to solubility problemsassociated with ADM-protein molecular interactions. Attachment of drugs to intermediate high-molecular-weight carriershas been used in the past to increase the drug:MAb molarratios. Up to 50 ADM molecules have been reported per antibody when dextran was used as a carrier for anthracyclines (3).Additionally, daunomycin has been linked to polyglutamic acidcarriers to produce conjugates with 10-20 drugs/MAb (23). Weare currently studying methods to link ADM to high-molecular-weight carriers using the hydrazone linker.

Unconjugated ADM-HZN, which has the same 13-acylhy-drazone bond as present in ADM-immunoconjugates, wasfound to be 10 to 50 times less potent than ADM in vitro (datanot shown) and in vivo (see accompanying article). However,ADM-HZN conjugates were consistently found to be morepotent than free ADM-HZN and in some experiments morepotent than parent drug ADM (for example, see Fig. 9). It canbe speculated that the high cytolytic activity of ADM immunoconjugates may be due to the mechanism by which antibody-bound drug is delivered to the intracellular environment of thetarget cell. MAbs 5E9, 3A1, and G28.1 are rapidly internalizedfollowing binding to tumor cell surface (data not shown). ADMconjugated to modulating MAbs probably enters the cellthrough the same endocytic pathway that leads to internaliza-

tion of membrane-bound antibodies and ligands (for review, seeRef. 24). Internalization of conjugates by this pathway deliversthe conjugated ADM-HZN to intracellular acidic endosomalvesicles and lysosomes. The acid-sensitive nature of the acylhy-drazone linker in the conjugates would allow for the release ofunmodified ADM from the MAb protein in intracellular acidicvesicles. These differences in uptake mechanisms and intracellular trafficking between free and antibody-bound ADM mayaccount for the differences in potency observed in vitro. Themechanism of action of ADM-immunoconjugates and quantitative aspects of ADM delivery are the subject of other studies.4

Delivery of ADM to cell surface membrane components mayalso be a contributing factor to the increased potency observedbecause it has been postulated that the cytotoxicity of ADM ismediated by a membrane effect (25).

Hydrolysis of the acid-sensitive 13-acylhydrazone bond depends upon time, temperature, and pH. The stability of theimmunoconjugate in vivo must be long enough to allow fordelivery of intact MAb-drug to the site of the tumor. In studiesusing 2 human tumor xenograft models, we have found thathighly effective targeting of ADM can be achieved using ADM-HZN immunoconjugates (26). These studies strongly indicatethat conjugates prepared using this new linker strategy haveappropriate stability in vivo that allows for MAb-targeting ofADM to tumors. Structure-activity studies with anthracyclineshave also shown that the sugar residue is an important contributor to cytotoxic activity. Although attachment of ADM withthe hydrazone chemistry may leave the amino sugar residueexposed for possible reaction with certain enzymes, pharma-cokinetic studies in mice have shown no evidence of the cleavageof the sugar residue leading to the formation of the aglycone.5

In conclusion, we have described a new chemical linkerstrategy for construction of acid-sensitive ADM immunoconjugates and demonstrated the efficacy and chemical propertiesof these conjugates in vitro. This methodology for the preparation of acid-sensitive immunoconjugates can be utilized toconjugate more potent 13-keto anthracyclines that containblocked amino sugar residues [i.e., 3'-deamino-3'-(3-cyano-4-

morpholinyl)adriamycin] that cannot be attached using the cis-aconityl linker chemistry (15, 16). The efficacy of ADM-HZNconjugates has been extensively evaluated in two human tumorxenograft models. The demonstration of immunospecific anti-tumor activity and reduced toxicity of ADM conjugates isreported in Ref. 26.

ACKNOWLEDGMENTS

The authors wish to acknowledge Susan Gawlak for performingseveral of the cytotoxicity assays. Flow cytometry studies were assistedby Dr. R. Mittler and Kurt Edinger.

REFERENCES

1. Mathe, G.. Maral, R., and DeJager. R. Anthracyclines. Current Status andFuture Developments. New York: Masson Publishing USA. 1982.

2. Casazza, A. M., Isetta. A. M., Giuliani. F.. and DiMarco. A. Immunode-pressive activity of daunomycin and Adriamycin. In: M. Staquet, H. Tagnon.Y. Kenis. G. Bonadorna, S. K. Carter, G. Sokal. G. Trouet, M. Ghione, C.Praga, L. Lenaz. and O. S. Karin (eds.), Adriamycin Review, pp. 123-131.Medikon, Belgium: European Press, 1975.

3. Chose. T. I.. Blair, A. H., and Kulkarni. P. N. Preparation of antibody-linkedcytotoxic agents. Methods Enzymol.. 93: 280-333. 1983.

4. Arnon. R.. and Sela, M. In vitro and In vivo efficacy of conjugates ofdaunomycin with anti-tumor antibodies. Immunol. Rev.. 62: 5-27, 1982.

4G. R. Braslawsky and R. S. Greenfield, manuscript in preparation.5 Unpublished data.

6606

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

ADRIAMYCIN HYDRAZONE IMMUNOCONJUGATES

5. Upcslacis. J., and Hinman. L. Chemical modification of antibodies for cancerchemotherapy. Annu. Rep. Med. Chem., 23: 151-160, 1988.

6. Hurwitz. E., Marón, R., Arnon, R., Wilchek, M., and Sela, M. The covalentbinding of daunomycin and adriamycin to antibodies, with retention of bothdrug and antibody activities. Cancer Res., 35: 1175-1181. 1975.

7. Levy, R., Hurwitz. E., Marón, R., Arnon. R., and Sela, M. The specificcytotoxic effects of daunomycin conjugated to antitumor antibodies. CancerRes., 35: 1182-1186. 1975.

8. Belles-Isles, M., and Page M. In vitro activity of daunomycin-anti-alphafe-toprotein conjugate on mouse hepatoma cells. Br. J. Cancer, 41: 841-842,

1980.9. Hurwitz. E.. Marón, R., Arnon, R., Wilchek. M., and Sela. M. Daunomycin-

immunoglobulin conjugates, uptake and activity in vitro. Eur. J. Cancer. 14:1213-1220, 1978.

10. Hurwitz, E.. Kashi. R.. Burowsky. J. D., Arnon, R.. and Haimovich, J. Site-directed chemotherapy with drug bound to anti-idiotypic antibody to alymphoma cell-surface IgM. Int. J. Cancer. 31: 745-748. 1983.

11. Pimm, M. V.. Jones, J. A.. Price. M. R., Middle. J. G.. Embleton, M. J..and Baldwin, R. W. Tumor localization of monoclonal antibody against ratmammary carcinoma and suppression of tumor growth with adriamycinantibody conjugates. Cancer Immunol. Immunother., 12: 125-134. 1982.

12. Vaniamolo, K., Acton, E. M., and Henry, P- W. Antitumor activity of somederivatives of daunomycin at the amino and methylkctone functions. J. Med.Chem., 15: 872-875, 1972.

13. DiMarco. A. Adriamycin (NSC-123I27): mode and mechanism of action.Cancer Chemother. Rep., 6:91-106, 1975.

14. Dillman, R. O.. Shawler, D. L.. Johnson. D. E.. Meyer. D. L., Koziol, J. A.,and Frincke, J. M. Preclinical trials with combinations and conjugates ofT101 monoclonal antibody and doxorubicin. Cancer Res., 46: 4886-4891,

1986.15. Yang, H. M., and Reisfeld, R. A. Doxorubicin conjugated to a monoclonal

antibody directed against a melanoma-associated proteoglycan suppressesgrowth of established tumor xenografts in nude mice. Proc. Nati. Acad. Sci.USA, «5:1189-1193, 1988.

16. Dillman, R. O., Johnson, D. E.. Shawler. D. L., and Koziol, J. A. Superiority

of an acid-labile daunorubicin-monoclonal antibody immunoconjugate compared to free drug. Cancer Res.. 48: 6097-6102. 1988.

17. Gallego, J., Price, M. R., and Baldwin, R. W. Preparation of four dauno-mycin-monoclonal antibody 791T/36 conjugates with anti-tumor activity.Int. J. Cancer. 33: 737-744, 1984.

18. Pietersz, G. A.. Smyth. M. J.. and McKenzie, I. F. C. Immunochemotherapyof a murine thymoma with the use of idarubicin monoclonal antibodyconjugates. Cancer Res., 48: 926-931. 1988.

19. Brück.C.. Porteteli!. D., Glineur, C., and Bollen, H. One-step purificationof mouse monoclonal antibodies from ascitic fluid by DEAE-affigel bluechromatography. J. Immunol. Methods, 53: 313-319, 1982.

20. Ellman, G. L. Tissue sulfhydryl groups. Arch. Biochem. Biophys.. 82: 70-77, 1959.

21. Colombatti. M.. Nabholz. M.. Gros. O., and Brown, C. Selective killing oftarget cells by antibody-ricin A chain or antibody-gelonin hybrid molecules:

comparison of cytotoxic potency and use in immunoselcction procedures. J.Immunol.. 131: 3091-3095. 1983.

22. Ogden. J. R.. Leung. K.. K nuli S. A., Telander. M. W., Avner. B. P.. Liao.S. K., Thurman. G. B., and Oldham. R. K. Immunoconjugates of doxorubicinand murine antihuman breast carcinoma monoclonal antibodies prepared viaan /V-hydroxysuccinimide active ester intermediate of r/s-aconityl-doxorubi-cin: preparation and in vitro cytotoxicity. Mol. Biother., /: 170-174, 1989.

23. Tsukada, Y., Kato, Y., Umemoto, N., Takeda, Y., Hará, T., and Hirai, H.An anti-«-fetoprotein antibody-daunomycin conjugate with a novel poly-t.-glutamic acid derivative as intermediate drug carrier. J. Nati. Cancer Inst.,73: 721-729, 1984.

24. Pastan, I., and Willingham. M. C. (eds.). Pathway of endocytosis. In: Endo-cytosis. pp. 1-44. New York: Plenum Publishing Corp., 1985.

25. Tritton. T. R.. and Yee. G. The anticancer agent Adriamycin can be activelycytotoxic without entering cells. Science (Washington DC), 217: 248-250,1982.

26. Braslawsky, G. R.. Edson, M. A.. Pearce. W., Kaneko, T.. and Greenfield.R. S. Antitumor activity of Adriamycin (Hydrazone-linked) immunoconju-gates compared with free Adriamycin and specificity of tumor cell killing.Cancer Res.. 50:6600-6607. 1990.

6607

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

1990;50:6600-6607. Cancer Res   Robert S. Greenfield, Takushi Kaneko, Ann Daues, et al.   Synthesized Using an Acid-sensitive Hydrazone Linker

of Adriamycin Immunoconjugatesin VitroEvaluation

  Updated version

  http://cancerres.aacrjournals.org/content/50/20/6600

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/50/20/6600To request permission to re-use all or part of this article, use this link

on April 9, 2020. © 1990 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from