Synthesis and Characterization of Oligomeric nido-Carboranyl

Nov 1, 1994 - Bioconjugate Chem. , 1994, 5 (6), pp 557–564 ... Ruthenium(0) Nanoparticle-Catalyzed Isotope Exchange betweenB andB Nuclei in Decabora...
2 downloads 0 Views 4MB Size
Bioconjugate Chem. 1994, 5, 557-564

557

Synthesis and Characterization of Oligomeric nido-Carboranyl Phosphate Diester Conjugates to Antibody and Antibody Fragments for Potential Use in Boron Neutron Capture Therapy of Solid Tumors Charng-Jui Chen,?Robert R. Kane,* F. James Primus,+ Gy6rgy Szalai,’ M. Frederick Hawthorne,$ and John E. Shively*,+ Division of Immunology, Beckman Research Institute of the City of Hope, Duarte, California 91010, and Department of Chemistry and Biochemistry, University of California, Los Angeles, California 90024. Received July 22, 1994@

Antibodies conjugated to oligomeric carboranyl compounds have a high potential as target species for boron neutron capture therapy (BNCT) of solid tumors. As a first step toward developing conjugates with BNCT capabilities, an oligomeric nido-carboranyl phosphate diester (Kane, R. R., Dreschel, K., and Hawthorne, M. F. (1993)J . Am. Chem. SOC.115,8853-88541, CBlO (10 nido-carboranes containing 90 boron atoms) with a pseudo-5’-terminal amino group, was conjugated to the anticarcinoembryonic antigen antibody T84.66 and its F(ab’) fragment. The homobifunctional linker disuccinimidyl suberate (DSS) was coupled to CBlO via its 5’-terminal amino group followed by removal of excess linker with organic solvent extraction and conjugation with intact antibody. Similarly, the heterobifunctional linker, m-maleimidobenzoyl-N-hydroxysuccinimide(MBS), was coupled to CBlO and conjugated to the hinge region sulfhydryl of the F(ab’) fragment of T84.66. The extent of reaction was monitored by the mobility shift of CB10-antibody conjugate on native polyacrylamide gels and the increased susceptibility of the CB10-antibody conjugate to staining with silver nitrate. CBlO was also labeled with radioiodine (I3lI) in a solid phase reaction with iodogen and used in double-label studies with lz5I-1abeledantibody. Although free CBlO bound very tightly to gel filtration media such as Sephadex G-25, the CB10-antibody conjugate passed through freely. After separation of CB10-antibody conjugate from free CBlO on Sephadex G-25, molar incorporations of CBlO were calculated. At a molar ratio of 1O:l (CBlO:T84.66), greater than 90% of T84.66 and 30% of its F(abY fragment were conjugated to CB10. The amount of CBlO covalently incorporated into mT84.66 ranged from 1.2 to 6.2 (moles per mole), with retention of immunoreactivity in the range of 80-90%. Biodistribution studies in Balb/C mice revealed high uptake of free CBlO or CB10-mT84.66 conjugate in the liver followed by rapid clearance presumably via a dehalogenation or biliary clearance mechanism. Tumor uptake at 48 h was 6.6% ID/g for CB10-mT84.66 conjugate compared to 33%ID/g in mT84.66 controls. These studies demonstrate reliable methods for the routine conjugation of oligomeric nido-carboranyl phosphate diesters to both antibody and antibody fragments, but suggest that the resulting conjugates are captured by the liver rendering them inefficient for tumor-targeting. Current chemical studies are being directed toward the synthesis of a variety of oligomeric carboranyl phosphate diester trailers expected to provide more acceptable biodistributions.

INTRODUCTION

Antibody-targeted boron neutron capture therapy is a binary approach to cancer therapy based on the concept that a tumor-specific antibody can deliver selectively large amounts of the stable isotope boron-10 to the targeted tumor. Boron-10 has a great propensity to capture thermal neutrons resulting in the emission of high energy, cytotoxic particles (log(n, a)7Li,2.3 MeV). Since the emitted helium and lithium nuclei have a translational path of about one cell diameter, high selectivity is expected following the deposition of significant amounts of boron in the tumor mass. It has been estimated that approximately 10-30 pg of boron-lO/g of tumor is needed to attain an acceptable therapeutic advantage (11. Therefore, a successful approach requires an efficient tumor-specific antibody carrying large numbers of boron-10 nuclei. We have chosen an antibody

(mT84.66)l directed to carcinoembryonic antigen (CEA), a well-characterized human tumor marker antigen which has been widely used for the in vitro diagnosis of human colon cancer (2). The mT84.66 has a high affinity constant for CEA, K,R = 2.6 x 1O1O M-I (3), and when radiolabeled with lllIn can provide up to 35% ID/g uptake in tumors in a human xenograft nude mouse model ( 4 ) and 48% sensitivity in presurgical imaging of human colon cancers (5). We have also generated a mouse/ human chimeric antibody, cT84.66, which bears the mT84.66 variable region joined to human constant regions for use in human cancer diagnosis and therapy (6). In previous studies we have demonstrated the conjugation of up to 600 boron atoms to mT84.66 using homogeneous nido-carborane-containingpeptides (Figure 1A). These conjugates exhibited high liver and low tumor uptake compared to unmodified antibody when tested in ~

Beckman Research Institute of the City of Hope. University of California. LS Abstract published in Advance ACS Abstracts, November 1, 1994. +

1043-1802/94/2905-0557$04.50/0

Abbreviations: BNCT, boron neutron capture therapy; CEA, carcinoembryonic antigen; DSS, disuccinimidylsuberate; MBS, maleimidobenzoyl-N-hydroxysuccinimideester; PBS, phosphatebuffered saline; cT84.44, chimeric anti-CEA antibody; mT84.66, murine anti-CEA antibody.

0 1994 American Chemical Society

558 Bioconjugate Chem., Vol. 5, No. 6, 1994

Chen et al.

\o/

-l



~

L J

10

OH

(M -2H) 2

C

100

-

90

-

184!.89

86

70

-

60

-

50

1

40

-

JO

-

7600.36

20b0

(1000

6000

8000

10000

Mass/Charge

Figure 1. Structure and mass spectrometric analysis of CB10.A. Structure of carboranyl peptide. B. Structure of carboranyl oligophosphate C. MALDI-TOF mass spectrometric analysis of carboranyl oligophosphate (CB10).The mass spectrum shows the doubly charged (MH2-),singly charged (MHl-1, and singly charged dimer (mlz = 7600) species. an animal model (7-9). The hydrophobicity of the boron cages present in the peptide reagent was believed to adversely affect the conjugates’biodistributions. In order to increase the hydrophilicity of the boron-rich conjugation reagents, nido-carborane cages were coupled using phosphate diesters as the linking groups (Figure 1B). Homogeneous oligomeric phosphate diesters containing up to 400 boron atoms have been synthesized, and these highly charged species exhibit extensive water solubility (20). The present report describes the preparation and characterization of the nido-carboranyl phosphate diester-whole antibody conjugates as well as related F(ab’)-conjugates. The biodistribution of a oligomeric nido-carboranyl oligophosphate diester-whole antibody conjugate in an animal model is also described. EXPERIMENTAL PROCEDURES

General. An oligomeric nido-carboranyl phosphate diester (Figure lB, hereafter called “CB 10”)containing a pseudo-5’-terminal amino group was synthesized by Midland Certified Reagent Company, Midland, TX, as described previously (10). The ammonium salt was converted to the sodium form by passage over Dowex 50W-A2 (sodium form) and dried in 200 nmol aliquots. Anti-CEA monoclonal antibody (mT84.66) was purified

from ascites by 40% ammonium sulfate precipitation and protein A affinity chromatography. Anti-CEA chimeric antibody, cT84.66 (61, was produced in a Unisyn bioreactor and purified by ion-exchange and protein G affinity chromatography. Disuccinimidyl suberate (DSS), maleimidobenzoic acid N-hydroxysuccinimide ester (MBS), Iodogen, and Ellman’s reagent were purchased from Pierce Chemical Co. Cysteine, silver nitrate, and N,”dimethylformamide (DMF) were obtained from Sigma. Ethyl acetate was supplied from Burdick & Jackson Labs. Mass Spectrometry. Matrix-assisted laser desorption-time of flight mass spectrometry (MALDI-TOF)was performed on a Kratos Kompact I11 mass spectrometer. The sample (dissolved in water, 10 pmol in 0.5 pL) was applied to the metal stage, mixed with 0.5 p L of matrix (a saturated solution of 3-hydroxypicolinic acid in 30% acetonitrile/70% water containing 0.1% TFA), and analyzed in the negative ion mode according to Wu et al. (11). The MH- and (MH#- peaks of an oligonucleotide were used to calibrate the instrument (mlz = 5735, 2867). Silver Nitrate Spot Test. Samples containing CBlO (2 pL, 1.0 mM) were spotted on Whatman 3 filter paper and air dried. Five pL of 10% silver nitrate solution prepared in 30%ammonium hydroxide were added to the same spot. In the presence of nido-carborane derivatives, brown color develops within 1 min.

Bioconjugate Chem., Vol. 5, No. 6,1994 559

Carboranyl Oligophosphate-Antibody Conjugates

Introduction of Linker Groups into CB10. The CBlO oligomer was dissolved in 50 mM sodium borate buffer, pH 7.0, to a final concentration of 1mM. Bifunctional linkers (DSS or MBS) were dissolved in ice cold DMF to a final concentration of 100 mM. Fifty pL of 1 mM CBlO was incubated with 25 pL of 100 mM bifunctional linker at room temperature for 2 h with occasional stirring. The excess unreacted bifunctional linker was removed by three extractions with 750 pL of ethyl acetate. The aqueous phase, containing linker-activated CB10, was used for conjugation with whole antibody or an antibody fragment. Preparation of CB10-mT84.66 Conjugate. Various amounts of DSS-CB10 were mixed with mT84.66 (1 mg/mL in PBS) a t room temperature for 2 h. The reaction was terminated by the addition of one-tenth volume 1M Tris buffer, pH 8.0. The control reaction was carried out under the same conditions except the DSSCBlO was quenched by reaction with Tris (10 pL of 1M Tris-HC1, pH 8) for 30 min prior to the addition of mT84.66. Preparation of F(ab)’and CB10-Fab’ Conjugate. Anti-CEA cT84.66 (30 mg in 10 mL of 0.1 M sodium acetate, pH 4.2 buffer) was digested with pepsin employing an antibody to pepsin ratio of 100:30 (w/w) at 37 “C for 4 h. The reaction was terminated with 2 mL of 2 M Tris base. The F(ab’)z fragment was purified on a Pharmacia Superose 12 column equilibrated in PBS and concentrated by ultrafiltration with an Amicon membrane to a final concentration of 10 mg/mL. The F(ab’) fragment was prepared by reduction of F(ab’)z with 10 mM cysteine in 40 mM ammonium carbonate, pH 8.0, at 37 “C for 2 h. The reduced F(ab’) was purified by a Sephadex G-25 column equilibrated in 50 mM ammonium citrate, pH 6.3, containing 2 mM EDTA and 100 mM NaC1. The free sulhydryl content was quantitated with Ellman’s reagent according to the manufacturer’s instructions (Pierce). Maleimide groups in samples of CBlO activated with MBS were determined by reacting an aliquot of the sample with a fixed amount of cysteine (in excess) and back-titrating the residual sulfhydryl groups with Ellman’s reagent. All buffers used for the preparation of F(ab’) were thoroughly degassed and saturated with nitrogen. The reduced F(ab’) was mixed with MBS-CB10 at a molar ratio of 1 : l O . The conjugation reaction was carried out under a nitrogen atmosphere for 2 h at room temperature and subsequently terminated with a 10- fold excess of iodoacetamide. Analysis of Conjugation Reactions. An aliquot from each reaction mixture was analyzed by gel electrophoresis on a Pharmacia Phastsystem using 7.5% or 1015% gradient native or SDS polyacrylamide gels. Gels were stained for protein with Coomassie Blue only or stained briefly with silver stain (Pharmacia kit) for carboranes followed by protein staining with Coomassie Blue. The staining procedures were carried out manually following the manufacturer’s instructions. Radioiodinationof CBlO and Antibodies. Radioiodination of CB10, mT84.66, and cT84.66 F(ab’) was performed as follows: The sample (10-20 pg of protein or oligomeric carboranyl phosphate diester in 10-20 pL of PBS) was added to a 1.2 mL polyethylene tube coated with 10-50 pg of Iodogen (10-50 pL of 1mg/mL reagent in chloroform, vacuum dried and rinsed with 10-50 pL of PBS). The tube was enclosed in a 4 mL Reacti-Vial (Pierce) and sealed with a silicone septum. The radioiodine (0.5-1.0 mCi in 5-10 ,uL of PBS; lZ5Ior 1311)was injected into the vial using a Hamilton syringe and allowed to react for 2 min at room temperature. The radiolabeled sample was removed with a syringe along

with two to three rinses of 20 pL each of PBS and, in the case of antibody or its fragments, further purified by gel filtration on a Pharmacia PDlO column in PBS. Radiolabeled CBlO was not further purified because of its high binding to Sephadex G25 and other gel permeation media. Purification of CB10-mT84.66 Conjugate. In a typical experiment, 5 pL of the 1251-CB10-antibody conjugate or control reaction were loaded onto 100 pL of Sephadex G-25 gel packed in a small pipette tip. The sample was eluted with 250 pL of PBS, followed by another 200 pL of PBS containing 0.05% Triton X-100 in PBS. Fractions (50 pL) were collected and counted. For the preparation of larger amounts of conjugate, a 1 mL Sephadex G-25 column was used to remove the free CB10. Determination of CBlO Content in CB10-mT84.66 Conjugates. Variable amounts of DSS-1311-CB10 were mixFd with a fixed amount of lZ5I-mT84.66(1mg/mL) at room temperature for 2 h. The reaction mixture was terminated by adding 1/10 volume of 1M Tris buffer, pH 8.0. The control reaction was carried out under the same conditions except the DSS-CB10 was inactivated with Tris-buffer, pH 8, for 30 min before mixing with antibody. An aliquot from each control and conjugation reaction mixture was analyzed on a Phastsystem 7.5% native or SDS polyacrylamide gel. The CBlO content of the CB10antibody conjugate was determined by differential monitoring of radioactivity of the conjugated band cut from the SDS gel. The remainder of the reaction mixture was loaded onto a Sephadex G-25 column and eluted with PBS. The radioactivity of the eluate was measured and calculated as the percentage of total input. Immunoreactivity of CB10-mT84.66 Conjugates. CEA coupled to Sepharose 4B (0.5 mL) was equilibrated in PBS in a spin column. The double-labeled conjugate (1311-CB10,lZ5I-mT84.66,or lz5I-cT84.66F(ab’)) or the control samples purified from Sephadex G-25 were loaded to each column, incubated for 15 min at 37 “C, and washed three times with PBS containing 1%BSA. The percent of the reactivity was calculated as the percentage of the counts bound to the column to the total number of counts added. Biodistribution Studies. Biodistribution studies were carried out in female athymic nude mice (NCr-nu, Simonsen, Gilroy, CAI under approval of the Institutional Research Animal Care Committee. Animals were injected sc with 1 x lo6 LS-174T colon carcinoma cells (ATCC, Rockville, MD). Seven days after tumor inoculation, approximately 4 pCi of dual 1311- and lz5I-labeled control or conjugate preparations were administered by tail vein injection. Animals were euthanized at 24 and 48 h after injection of the radioiodinated preparations and blood, tumor, and normal tissues were removed. Radioactivity was determined in a multichannel y counter with counts appearing in the lZ5Ichannel corrected for I3lI cross-over. RESULTS

Characterization of CB10. CBlO is an oligomeric nido-carboranyl phosphate diester with an amino group a t the pseudo-%-terminus and a thymidine at the pseudo3’-terminus (Figure 1B). It was synthesized by adding 10 carboranyl monomers, followed by a hexylamine linker residue, to a thymidine-derivatized resin using standard phosphoramidate chemistry on a DNA synthesizer (10). The oligomer was removed from the resin and deprotected by treatment with concentrated ammonium hydroxide (30 min at 80 “C). The ammonium hydroxide treatment also converted the closo-carborane structure

Chen et al.

560 Bioconjugate Chem., Vol. 5, No. 6,1994

to the corresponding anionic nzdo-cage. The initially isolated ammonium salt of CBlO was converted to its corresponding sodium salt on Dowex 50W-A2 resin (sodium form) and was used directly for coupling to crosslinkers. Both the sodium and ammonium salts of CBlO were extremely water soluble. The mass of CBlO as determined by negative ion MALDI-TOF MS was 3669 (Figure IC). The predicted mass of fully protonated CBlO is 3514.9 while that of its sodium salt is 3956.5. The intermediate value of 3669 probably corresponds to a mixture of species differing in their relative contents of sodium and proton neutralized structures. The unusual peak width may be due to this phenomenon as well as the abundant (20%) loB isotope. Coupling of Linkers to CB10. Because of difficulties in purifying CBlO or CB10-linker conjugates (CB10 bound strongly t o dextran and polyacrylamide-based gel permeation media as well as silica-based C18 media), a simple organic solvent extraction method was developed to separate excess linker from free or linker-coupled CB10. The efficiency of the extraction of the MBS bifunctional linker with ethyl acetate was tested by back titration of underivatized cysteine with Ellman’s reagent after a fured amount of cysteine was allowed to react with MBS in the aqueous phase (1:3 molar ratio of MBS to cysteine). After three extractions with ethyl acetate (10 vol), MBS was completely removed. The recovery of CBlO during extraction was monitored either by a silver nitrate spot test or by counting of radiolabeled CB10. The recovery of CBlO was ’95%. The bifunctional linker DSS was also removed eficiently by ethyl acetate extraction. The aqueous phase which contained CB 10-coupled linker was used for whole antibody or antibody fragment conjugation. Preparation and Analysis of CBlO-F(ab’) Conjugate. The F(ab’12 fragments were generated by pepsin treatment of cT84.66 and reduced with 10 mM cysteine for 2 h at 37 “C. Excess reducing reagent was removed by gel filtration. Greater than 90% of F(ab’)2fragments were reduced to F(ab‘) when analyzed by nonreducing SDS gel electrophoresis. The reduced F(ab’) fragment contained 1.8 equiv of the sulhydryl function per mole of F(ab’) fragment as measured by Ellman’s reagent. The sulfhydryl content decreased significantly upon storage (50%in 2 days), but remained in a monomeric F(ab’) form when analyzed by SDS polyacrylamide gel electrophoresis. The CBlO-F(ab’) conjugate was prepared by the addition of MBS-CB10 at a molar ratio of 1O:l to F(ab’) (2 mg/mL) immediately after the removal of excess reducing agent and analyzed by SDS gel electrophoresis. The CBlO-F(ab)’ conjugate migrated more slowly than the unconjugated F(ab)’ during SDS gel electrophoresis (Figure 2A). When the gel was stained briefly with silver nitrate followed by Coomassie Blue, the CBlO-F(ab)’ stained intensively as a broad brown band immediately above the blue band corresponding to the unconjugated F(ab)’ fragment. Unconjugated CBlO migrated at the dye front as one or two bands. The best yield for the conjugation of CBlO t o F(ab)’ was about 30% as judged by the Coomassie Blue staining (Figure 2, Panel A, lane 2). Although a wide variety of conditions were explored to increase the conjugation efficiency, including higher concentrations of F(ab‘) and higher molar ratios of CBlO to F(ab’), the maximum yield obtained was about 30%. The low conjugation efficiency might be a result of the tendency of the F(ab’) fragment to form intramolecular disulfide bonds, as judged by its rapid loss of free sulfhydryl content without F(ab’)z fragment formation. Preparation and Analysis of CB10-mT84.66 Conjugate. A 5-15-fold molar excess of DSS-CB10 was

130 75-

50 39 -

..I+-

Figure 2. Dual stain analysis of the CBlO-F(ab’) conjugate on SDS gel electrophoresis. F(ab’j(2 mg/mLj prepared from antiCEA cT84.66 was mixed with MBS-CB10 a t a molar ratio of 1:lO. The conjugation reaction was carried out in the presence of 50 mM ammonium acetate, pH 6.3, containing 2 mM EDTA, 100 mM NaCl for 2 h at room temperature. Lane 1: control reaction, F(ab’) was alkylated with iodoacetamide prior to mixing with MBS-CB10. Lane 2: conjugation reaction. Panel A: staining with Coomassie Blue. Panel B: the gel was stained briefly with silver nitrate and followed by Coomassie Blue staining. The upper arrow indicates the CBlO-F(ab‘) conjugate, which was detected as a broad brown band on the top of the unconjugated F(ab’) band. The bottom arrow indicates free CB10.

incubated with mT84.66 (1 mg/mL) in PBS at room temperature for 2 h. When an aliquot of the conjugation mixture was analyzed by SDS polyacrylamide gel under nonreducing conditions, no significant difference in mobility was observed between conjugated and unconjugated antibody (data not shown). However, when the CB10-mT84.66 conjugation mixtures were analyzed on a native gel, the mobility of the CB10-mT84.66 conjugate was markedly retarded (Figure 3, lanes 2-4). Unconjugated CBlO migrated at the dye front. At higher molar ratios (15:l) of DSS-CB10 t o mT84.66 the conjugate exhibited less retardation compared to free mT84.66 (Figure 3, lane 2). The reason for the lessened mobility shift a t higher CBlO incorporation levels is not clear. When the gel was stained briefly with silver nitrate prior to Coomassie Blue, the CB10-mT84.66 conjugate stained brown and unconjugated antibody stained blue, similar t o the pattern observed for the CBlO-F(ab)’ conjugate. This staining pattern is due to the facile reduction of silver ion by incorporated CB 10 (under these conditions proteins are not stained with silver nitrate). The mobility shift together with the unique silver staining pattern provided a convenient method with which to monitor the conjugation reaction. As shown in Figure 3, lane 2, when the DSS-CB10 to mT84.66 ratio increased t o 15:1, more than 90% of the mT84.66 was conjugated to CB10. Partial Purification of CB10-mT84.66 Conjugate. In preliminary experiments, we investigated the use of gel filtration media to remove excess unconjugated CBlO from CB10-antibody conjugate. Since we already knew that low concentrations of CBlO bind tightly t o Sephadex G-25, we reasoned that excess CBlO would be bound to

Bioconjugate Chem., Vol. 5, No. 6,1994 561

Carboranyl Oligophosphate-Antibody Conjugates

Sephadex G-25 Elution Profile h

5)

u

-

v)

fl I

I v

80 -

ICB 10 alone

70 60

W

-

CB 1O+IgG Conjugate

.-

u

.U

2

'

50 -

,h 40 -

e$

CBlO+IgG Control

30 -

2010

-

0

3

1 _ 1

1

1

2

3

4

5

6

7

8

9

B

Fraction Number

Figure 4. Analysis of 1251-CB10-mT84.66 conjugate by Sephadex G-25 chromatography. DSS-1251-CB10 was mixed with mT84.66 (1mg/mL in PBS) at a molar ratio of 1 O : l . After 2 h of incubation, 5 pL of the conjugation mixture was loaded onto a 100 pL Sephadex G-25 column and was eluted with 250 pL of PBS, followed by 200 pL of PBS containing 0.05% Triton X 100. Tris buffer inactivated DSS- 1251-CB10was used in the control reaction. Counts for each fraction (50pL) and counts irreversible bound to the column (B) were measured and plotted a s the percentage of the total input. 1-125 IgG

Figure 3. Analysis of the CB10-mT84.66 conjugate by native polyacrylamide gel electrophoresis. The CB10-mT84.66 conjugate and the control reaction mixtures were analyzed by electrophoresis on a Phast-system 7.5% native polyacrylamide gel. The gel was stained briefly with silver stain and followed by Coomassie Blue stain. Lane 1: Control reaction. Lane 2-4: the conjugation reactions were carried out at a DSS-CBlOI antibody molar ratio of 15 (lane 2), 10 (lane 3), and 5 (lane 4). The top left arrow indicates the CB10-antibody conjugates. The bottom left arrow indicates free CB10. The right arrow indicates unconjugated antibody.

the column, while conjugated CBlO would be eluted with the antibody. To test this possibility, five ,uL of the la51labeled CB10-antibody conjugate reaction mixture was loaded onto a 100 ,uL bed volume of Sephadex G-25 in a micropipette tip and was eluted with PBS followed by PBS containing 0.05% Triton X-100. Two control reactions were carried out under the same conditions, one with 1251-CB10only, and another with a mixture of inactivated 1251-CB10 and antibody. When CBlO alone was loaded to the column, about 15%of the radioactivity was eluted from the column, while 85% was bound to the resin. The addition of Triton X-100 did not improve the dissociation of CBlO from the Sephadex G-25 once the CBlO had bound (Figure 4). When the conjugation reaction mixture was loaded onto the Sephadex G-25 column, approximately 40% of the labeled CBlO eluted from the column, while 53% bound t o the column. An additional 7% was eluted by 0.05% Triton X-100 in PBS. The control mixture containing unconjugated CBlO and antibody gave 23% of the counts in the eluate, an 8% increase over the CBlO only control. These results indicate that CBlO can noncovalently bind t o mT84.66, thus complicating the analysis of the reaction mixture by Sephadex gel filtration. Furthermore, it is possible that a small percentage of the CB10-mT84.66 conjugate was irreversibly bound to the column. In order to investigate this possibility, double labeled CB10-mT84.66 conjugate was prepared in which mT84.66 was la51 labeled, and the CBlO was 1311labeled. As shown in Figure 5 , greater than 98% of the doubled labeled 1311CB10-1251-mT84.66conjugate passed through the column. The remaining counts bound to the column were solely due to free l3lI-CBIO. This was further confirmed by analysis of the conjugation reaction mixture before and

50

T

Conjugate

r

1

2

3

4

5

6

B

F'raction Number 1-131 CBlO

T

n

1

2

3

4

5

6

B

Fraction Number

Figure 5. Profile of double-labeled CB 10-mT84.66 conjugate on Sephadex G-25. The experimental procedures were the same as described in Figure 4 (CB10 to antibody molar ratio of lO:l), and CBlO with 1311. except that mT84.66 was labeled with 1251 The control and conjugation reaction mixture was loaded onto a Sephadex column and eluted with PBS. The counts for each fraction (50 pL)and the counts irreversibly bound to the column (B) were measured and plotted as the percentage of the total input. Upper panel: 1251 counts (mT84.66). Lower panel: 1311 counts (CB10).

after passing through Sephadex G-25 resin by SDS gel electrophoresis followed by the autoradiography (Figure 6). These results show that the gel filtration purified sample contains no unconjugated CBlO and confirm that gel filtration is an appropriate method for the removal of excess CBlO from CB-mT84.66 conjugate. Determination of CBlO Content in Conjugates. Double-labeled CB10-mT84.66 conjugate was used for the determination of the nonspecific binding of CBlO and quantitation of the incorporation of CBlO into mT84.66.

Chen et al.

562 Bioconjugafe Chem., Vol. 5, No. 6,1994

11

conjugate ranged from 1.2 to 6.2 (difference between conjugate and control). In addition, the control and conjugate were analyzed by SDS polyacrylamide gel electrophoresis by excising and counting the gel bands corresponding to conjugated and unconjugated mT84.66 (Table 2). The calculated CBlO incorporation for both the nonspecific binding control and specific binding conjugate were lower but comparable with the CBlO content measured from the Sephadex G-25 gel filtrate. Immunoreactivity of CB10-mT84.66 Conjugate. After removal of excess CBlO from the conjugate by Sephadex G-25 chromatography, the immunoreactivity of the modified antibodies was tested by binding to a CEA-Sepharose4B spin column. The double labeled (1311CB10, 1251-mT84.66,or 1251-F(ab>’)conjugates and the control samples (mixture of inactivated CBlO and mT84.66 purified from Sephadex G-25) were loaded onto the CEA affinity column, and washed three times with PBS containing 1%BSA. The percent immunoreactivity was calculated as the percentage of the counts bound to the column over input counts. As shown in Table 3, the immunoreactivity of the control sample was 92%. This suggests that the nonspecific association of CBlO with antibody did not alter its immunoreactivity. The CB10mT84.66 conjugate retained greater than 80% of immunoreactivity even for the highest levels of CBlO incorporation. The CBlO-F(ab’) conjugate gave essentially the same immunoreactivity (ca. 60%) as control F(ab’). In this case, the loss of immunoreactivity was due to radioiodination conditions. Biodistribution of CB10-mT84.55 Conjugates. Because the yields of Fab’ conjugates were low, biodistribution studies with double-labeled whole antibody conjugates were carried out in tumor-bearing nude mice. Radioiodinated whole mT84.66-CB10 conjugate was prepared as described above using a 151 ratio of 1251labeled CBlO to l3l1-mT84.66. A control preparation consisting of a mixture of radioiodinated CBlO and mT84.66 was also produced as described above. The CEA binding properties of the control and conjugate preparations were very similar to that shown in Table 3. The distribution of mT84.66 and CBlO in blood and selected tissues a t 24 h after injection is shown in Figure 7. The

7

L

Figure 6. SDS-PAGE analysis of 1251-CB10-mT84.66 conjugate before and after Sephadex G-25 filtration. The unfractionated conjugation mixture and the second fraction of eluate from the Sephadex G-25 column (Figure 4) were analyzed on a Phastsystem 10-15% gradient SDS gel, followed by autoradiography. Lane 1: the conjugation reaction mixture of 1251-CB10 and antibody. Lane 2: the peak fraction eluted from the Sephadex G-25 column. Upper arrow indicates the radioactive band which corresponds to the CBlO conjugated antibody. The lower arrow indicates free CB10.

Increasing amounts of DSS-CB10 were incubated with a fixed amount of mT84.66 a t room temperature for 2 h. An aliquot from each control and conjugation reaction mixture was analyzed by electrophoresis on a 7.5%native and SDS polyacrylamide gel. The rest of the reaction mixture was partially purified by passing through a 1 mL Sephadex G-25 column. The calculated CBlO incorporation of control and CB10-mT84.66 conjugates after removal of free CBlO is shown in Table 1. The moles of CBlO incorporated per mole into mT84.66 was dependent on the molar ratio of CBlO to mT84.66 used in the conjugation reaction. When the molar ratio of CB10/ mT84.66 in the conjugation reaction was increased from 5 to 25, the amount of CBlO incorporated into mT84.66 was increased from 2.4 to 11.3 mol/mol of mT84.66. However, the nonspecific binding also increased from 1.2 to 5.1 mol of CBlO/mol of mT84.66. Thus, the moles of CBlO covalently incorporated into the CB10-mT84.66

Table 1. CBlO Incorporationinto the CB10-mT84.66 Conjugate after Separation by Sephadex G-25 Filtrationu input molar ratio 5 10 15 20 25

l3lI-CB10 eluatehnput (%) control conjugate

1251-mT84.66eluatehnput (%) control conjugate ~

23 19 19 20 20

48 45 46 46 44

~~

~

97 95 97 98 96

99 96 96 96 97

CBlO/mT84.66 output molar ratio control conjugate 2.4 1.2 4.5 2.0 7.2 2.3 10.0 4.1 11.3 5.1

a The antibody was 1251 labeled and CBlO was 1311labeled and reacted with DSS as described in the Experimental Procedures. m e r mT84.66 and DSS-CB10 were reacted a t the indicated input molar ratios (CBlO/mT84.66; column l), conjugate was separated from unconjugated material on a 1mL Sephadex column and output molar ratios were calculated (columns 6 and 7). Controls are treated in the same way, except that the DSS-CB10 was treated with Tris base to prevent covalent coupling to mT84.66.

Table 2. CBlO Incorporation into CB10-mT84.66 Conjugate after Separation by SDS PolyacrylamideGel Electrophoresis” 1311-CB lO-mT84.66Anput (%)

1251-mT84.66/input (%)

input molar ratio

control

conjugate

control

5 10 15 20 25

21 20 16 14 15

27 24 23 28 33

81 80 79 82 81

conjugate

CBlO/mT84.66 output molar ratio control

conjugate

~

~

82 81 82 76 81

1.2 2.0 2.7 3.7 5.2

1.5 3.1 5.1 6.3 10.0

a The double-labeled sample was prepared as described in Table 1and analyzed by SDS gel electrophoresis under nonreducing conditions. The gel bands correspondingto antibody were excised and the radioactivity for each band was measured. The percentage of the radioactivity compared to the total input was calculated. CBlO incorporation into the antibody was calculated based on the measurement of the ratios to 1311in the excised conjugate band (columns 6 and 7). of 1251

Carboranyl Oligophosphate-Antibody Conjugates

Bioconjugate Chem., Vol. 5, No. 6, 1994 563

Table 3. I2%mT84.66 or F(ab) Binding (Immunoreactivity) and 1311-CB10Binding to a CEA Affinity Column" immunoreactivity (%I control conjugate

input molar ratio CB10-mT84.66

~

5 10 15 20 25

93 104 92 100 102

82 85 90 86 84

a The conjugates were synthesized with 1251-mT84.66or F(ab') and 1311-CB10. The immunoreactivity was calculated as the percentage of the 1251counts bound to a CEA-Sepharose 4B spin column. The control was 1251-labeled mT84.66 or the F(ab') fragment. The input ratio for F(ab') was 10 with 61% immunoreactivity for the control and 64% for the conjugate.

.

A. Antibody (1-131)

25-

8. CBlO (1-125)

BLOOD

LIVER

0 SPLEEN

M-

KIDNEYS

Table 4. Tumor Localization of CBlO Following Injection of Whole CB10-mTS4.66 Conjugate" time (h) 24 48

% iniected doselg mT84.66 (1311) CBlO (1251)tumor wt (g) 37.23 f 3.86b 1.16 f 0.09 0.16 f 0.03 control conjugate 5.32 f 0.31 0.89 f 0.06 0.22 f 0.07 33.71 f 3.47 0.83 f 0.06 0.42 f 0.07 control 6.61 f 0.54 0.88 f 0.06 0.23 f 0.04 conjugate

group

a Athymic nude mice bearing LS-174T colon carcinoma xenografis were injected iv with a mixture (control) of 1311-labeled antibody 1251-labeledCBlO or the double labeled CB10-mT84.66 conjugate. Mean f SEM.

+

clearance of antibody from the blood. Mice injected with antibody alone had similar levels of circulating antibody as compared to the control group (data not shown). The tumor localization of antibody and CBlO in the control and conjugate groups is depicted in Table 4. At both time points, accumulation of antibody in control group tumors was five to seven times higher than that in tumors from animals injected with conjugate. The reduced tumor targeting of the conjugate reflects its more rapid elimination from the blood (Figure 7A). Tumor uptake of CBlO was low and did not differ between the two groups a t both time points. DISCUSSION

Control

Conjugate

Control

Conjugate

Figure 7. Biodistribution of CB10-mT84.66 conjugate in athymic nude mice bearing LS-174T colon carcinoma xenografts. The control group consisted of animals injected with a mixture of I3lI-labeled antibody plus 1251-labeledCB10. The conjugate was prepared with 1251-labeledCBlO and 1311-labeled antibody, and purified by gel filtration. Measurement of (A) antibody and (B) CBlO distribution 24 h after injection.

level of circulating mT84.66, as indicated by 1311activity, was markedly reduced in animals receiving the conjugate as compared to the control preparation (Figure 7A). Depressed levels of mT84.66 in normal tissues was also evident in animals injected with the conjugate. By contrast, animals injected with either the control or conjugate had similar elevated levels of CB10, as indiactivity, in the liver followed by spleen and cated by 1251 large intestine (Figure 7B). Blood levels of CBlO were very low in both animal groups. Uptake of either mT84.66 or CBlO in other normal tissues such as heart, muscle, stomach, and small bowel was unremarkable and is not shown for clarity. An identical distribution pattern was observed at 48 h (data not shown). The elevated level of CBlO in the livers of mice injected with the conjugate suggests that the low level of circulating antibody observed in this group (Figure 7A) was due to the liver-mediated clearance of the conjugate. While it might be expected that the 1311labeled antibody in the conjugate should follow the 1251 -labeled CBlO levels in the liver, this was not observed, and was probably due to dehalogenation of radioiodine from the antibody in the liver. Dehalogenation of 1251 labeled CBlO did not occur in the liver, suggesting that radioiodine attached to CBlO is metabolically stable. Since CBlO was not covalently attached to mT84.66 in the control preparations, the liver uptake of CBlO in the control group did not affect the

A method has been developed to rapidly synthesize oligomers of carboranyl phosphate diesters which mimic nucleotides and contain multiple nido-carborane cages (1-20 or more) (IO). Starting with appropriately protected carboranyl monomers, the oligophosphate synthesis was performed on a standard automated DNA synthesis instrument with no modifications to the standard reagents or procedures. For convenience, the oligomers begin with a 3'4hymidine (which also contributes U V absorbance at 260nm), while the 5' terminal group is a hexylamine moiety to allow crosslinking to proteins. During deprotection with ammonium hydroxide the protecting groups are removed and each closo-CB cage is converted to the corresponding nido-structure. Thus, the CB oligomers are more negatively charged (two negative charges per CB) than their oligonucleotide counterparts (one negative charge per nucleotide). This property has made it difficult to analyze the CB oligomers by conventional positive ion mass spectrometry. However, we were able to confirm the mass of a CB2Omer (IO) and the CBlO used in this work by negative ion electrospray mass spectrometry and MALDI-TOF, respectively. Although the CB oligomers are very water soluble they exhibit some unusual properties, such as high binding to gel filtration media and anion exchangers (they are not bound to cation exchangers). In this work, attempts to purify CBlO by reversed phase HPLC in 0.1% TFA-acetonitrile were unsuccessful because the sample became hydrophobic after elution from the C18 column. We attribute this property to the expected increase in hydrophobicity of the fully protonated oligomer. Due to this problem, we decided to couple CBlO to antibody and then purify the conjugate by gel filtration. Two cross-linking agents were studied, MBS for conjugation of the amino-CB10 to the sulfhydryl group of an F(ab') antibody fragment and DSS for conjugation of amino-CB10 to whole antibody (via the +amino group of lysine). An organic extraction procedure was shown to successfully remove excess crosslinker from CB 10 with no loss of the carborane oligomer. Conjugation of CBlO to either whole antibody or F(ab') fragment resulted in a product which was retarded relative to the protein on

564 Bioconjugafe Chem., Vol. 5, No. 6,1994

polyacrylamide gel electrophoresis in the presence or absence of SDS. Free CBlO migrates near the dye front on gel electrophoresis, consistent with its highly negatively charged nature. Thus, it appears that the anionic nature of CBlO was not sufficient to increase the electrophoretic mobility of the protein but, in fact, actually retarded its migration. The conjugates were stained with a silver nitrate stain, specific for the carborane, and with Coomassie Blue, specific for the protein. The double staining pattern clearly revealed that the conjugate was retarded relative to the unmodified protein. The whole antibody conjugate could be purified by gel filtration with irreversible binding of free CBlO to the gel filtration media. In this case, the strong binding characteristic of the CBlO to gel filtration media was overcome by its conjugation to antibody. Analysis of the double-labeled purified conjugate by SDS gel electrophoresis revealed that the conjugate contains no free CB10. Analysis of the doubly labeled conjugate revealed a specific incorporation of 1-10 CBlOmers per antibody molecule, corresponding to 90-900 boron atoms per antibody. In addition, the purified conjugate retained over 80% of its immunoreactivity as measured against a radiolabeled antibody control. When the purified conjugate was injected into animals bearing a CEA positive tumor xenograft, a large portion of the Conjugate (20-25% ID/g at 24 h) biodistributed to the liver. The tumor uptake was reduced from 30-40% ID/g in antibody controls to 5 7 % ID/g in conjugate injected animals. These results are almost identical to those we obtained for carboranyl peptides attached to antibody (8). In those studies up to 600 boron atoms were attached per antibody, with retention of 80-90% immunoreactivity, and in biodistribution studies gave 15-25% ID/g in liver and 4-5% ID/g in tumor. Recently, Barth et al. (12)conjugated highly boronated “starburst dendrimers” to the antimelanoma antibody IB16-6 using MBS and SPDP crosslinkers. Preparations with up to 2200 boron atoms per antibody molecule retained up to 82% immunoreactivity. However, in biodistributions studies, 11%ID/g was localized in the liver and only 0.4% ID/g in tumor xenografts. These results are similar to ours for liver uptake, but 10 times lower than ours for tumor accretion. Thus, three very different approaches to generate boron rich oliogmers have been devised, as well as conjugation methods leading to antibodies which retain high immunoreactivities. In spite of this exciting progress, all of the boron rich oligomers have dramatically altered the antibodytargeting characteristics by increasing liver uptake at the expense of tumor targeting. It is possible that these compounds bind to an as yet unidentified receptor in the liver. Further studies are needed to understand the nature of the liver uptake so that a rational approach can be used to overcome this problem. We continue to be optimistic about this approach, since the formidable problems of chemical synthesis and antibody conjugation have been solved. It is reasonable to assume that further modifications of the carboranyl oligomers will reduce liver uptake.

Chen et al. ACKNOWLEDGMENT

This research was supported by grants CA53870 and CA31753 from the NIH. R.R.K. was supported in part by a Tumor Cell Biology Fellowship (NIH NRSA CA09056). We gratefully acknowledge the technical assistance of Toby ONeil. LITERATURE CITED (1) Hawthorne, M. F. (1993) The role of chemistry in the development of boron neutron capture therapy of cancer. Angew. Chem., Znt. Ed. Engl. 32, 950-984. (2) Shively, J. E., and Beatty, J. D. (1985) CEA-related antigens: molecular biology and clinical significance. Crit. Rev. Oncol. Hematol. 2 , 355-399. (3) Wagener, C., Yang, Y. H. J., Crawford, F. G., and Shively, J . E. (1983) Monoclonal antibodies for carcinoembryonic antigen and related antigens as a model system: a systematic approach for the determination of epitope specificities of monoclonal antibodies. J . Zmmunol. 130, 2308-2315. (4) Jakowatz, J. G., Beatty, B. G., Vlahos, W. G., Porduminski, D., Philben, V. J., Williams, L. E., Paxton, R. J., Shively, J . E., and Beatty, J. D. (1985) High specific activity indium111-labeled anti-carcinoembryonic antigen monoclonal antibody: biodistribution and imaging in nude mice bearing human colon cancer xenografts. Cancer Res. 45,5700-5706. (5) Beatty, J. D., Williams, L. E., Yamauchi, D., Morton, B. A., Hill, L. R., Beatty, B. G., Paxton, R. J., Merchant, B., and Shively, J . E. (1990)Presurgical imaging with indium-labeled anti-carcinoembryonic antigen for colon cancer staging. Cancer Res. 50S, 922s-926s. (6) Neumaier, M., Shively, L., Chen, F.-S., Gaida, F.-J., Ilgen, C., Paxton, J., Shively, J. E., and Riggs, A. D. (1990) Cloning of the genes for T84.66, an antibody that has a high specificity and affinity for carcinoembryonic, and expression of chimeric humadmouse T84.66 genes in myeloma and Chinese hamster ovary cells. Cancer Res. 50, 2128-2134. (7) Hawthorne, M. F. (1991) Biochemical application of boron cluster chemistry. Pure and Appl. Chem. 24, 327-334. (8) Varadarajan, A., and Hawthorne, M. F. (1991) Novel carboranyl amino acids and peptides: reagents for antibody modification and subsequent neutron-capture studies. Bioconjugate Chem. 2, 242-253. (9) Paxton, R. J., Beatty, B. G., Varadarajan, A., and Hawthorne, M. F. (1992)Carboranyl peptide-antibody conjugates for neutron-capture therapy: preparation, characterization, and in vivo evaluation. Bioconjugate Chem. 3, 241-247. (10) Kane, R. R., Drechsel, K., and Hawthorne, M. F. (1993) Automated synthesis of carborane-derived homogeneous oligophosphates: reagents for use in the immunoprotein mediated boron neutron capture therapy of cancer. J . A m . Chem. SOC.115,8853-8854. (11) Wu, K. J., Steding, A., and Becker, C. H. (1993) Matrixassisted laser desorption time of flight mass spectrometry of oligonucleotides using 3-hydroxypicolinic acid as an ultraviolet matrix. Rapid Commun. Mass Spectrom. 7, 142-146. (12) Barth, R. F., Adams, D. M., Soloway, A. H., Alam, F., and Darby, M. V. (1994) Boronated starburst dendrimer-monoclonal antibody immunoconjugates: evaluation as a potential delivery system for neutron capture therapy. Bioconjugate Chem. 5, 58-66.