Construction and Evaluation of the Tumor Imaging ... - ACS Publications

Mar 16, 2007 - Tumor-associated glycoprotein-72 (TAG-72) is overexpressed in a high percentage of epithelial cancers and has proven useful as a target...
0 downloads 0 Views 449KB Size
Download PDF
Bioconjugate Chem. 2007, 18, 677−684

677

Construction and Evaluation of the Tumor Imaging Properties of 123I-Labeled Recombinant and Enzymatically Generated Fab Fragments of the TAG-72 Monoclonal Antibody CC49 Ying Tang,†,‡,⊥ Shaoxian Yang,§ Jean Garie´py,§ Deborah A. Scollard,‡ and Raymond M. Reilly*,†,‡,| Department of Pharmaceutical Sciences, University of Toronto, Toronto, ON, Canada, Toronto General Research Institute, University Health Network, Toronto, ON, Canada, Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada, and Department of Medical Imaging, University of Toronto, Toronto, ON, Canada. Received August 21, 2006; Revised Manuscript Received December 7, 2006

Tumor-associated glycoprotein-72 (TAG-72) is overexpressed in a high percentage of epithelial cancers and has proven useful as a target for imaging and targeted radiotherapy. Our goal was to express a recombinant Fab (rFab) of the TAG-72 monoclonal antibody CC49 in Pichia pastoris and directly compare its tumor and normal tissue uptake and imaging properties with enzymatically generated Fab (eFab). In this study, the genes coding for CC49 Fab were cloned from hybridoma cells and expressed in P. pastoris. Fab was purified to homogeneity and its immunoreactivity toward bovine submaxillary mucin (TAG-72) confirmed by ELISA. The tumor and normal tissue localization of 123I-CC49 rFab and eFab were compared in athymic mice bearing s.c. LS174T colon cancer or TAG-72-negative A375 melanoma xenografts. Results showed that pure and immunoreactive rFab of CC49 was produced and labeled with 123I. At 24 h post i.v. injection (p.i.), tumor uptake for 123I-rFab in LS174T xenografts was 6.0% ID/g which was 18-fold higher than in A375 tumors. Tumor-to-normal tissue ratios increased between 2 and 24 h and exceeded 5:1 at 24 h p.i. of 123I-rFab. 123I-rFab exhibited significantly lower liver uptake at 12 h p.i. and lower kidney uptake at 2 h p.i. than 123I-eFab. LS174T tumors were imaged as early as 2 h after administration of 123I-rFab. We conclude that CC49 rFab can be produced in a P. pastoris host system and accumulated at comparable levels as eFab in LS174T colon cancer xenografts in mice. The lower liver uptake of 123I-rFab as compared with eFab suggests that it may be more useful for imaging liver lesions. No major effect, except for kidneys and liver, was observed on tumor and normal tissue uptake due to introduction of hexahistidine and FLAG affinity tags or peptide linkers in the scaffold of rFab.

INTRODUCTION Monoclonal antibodies (mAbs) directed against tumor-associated antigens overexpressed on the surface of tumor cells can deliver radionuclides to tumors for imaging. Small antibody fragments such as Fab or scFv are particularly suitable for this purpose because these fragments extravasate more efficiently, penetrate more deeply into tumor nodules, and clear more rapidly from the body compared with intact IgG, leading to higher tumor-to-normal tissue uptake ratios at early times (13). The utility of antibody fragments in tumor imaging is exemplified by anti-carcinoembryonic antigen (CEA) arcitumomab (CEA-Scan, Immunomedics Inc, Morris Plains NJ), which consists of Fab′ of murine anti-CEA IgG labeled with 99mTc (4). 99mTc-labeled arcitumomab has been used for imaging colorectal and breast cancers (5-7). Tumor-associated glycoprotein-72 (TAG-72) (8) is an epithelial mucin overexpressed in a high percentage of colorectal, ovarian, gastric, and breast cancers (9). CC49 is a murine high* Address correspondence and requests for reprints to Raymond M. Reilly, Ph.D., Leslie Dan Faculty of Pharmacy, University of Toronto, 144 College St., Toronto, Ontario M5S 3M2. Tel.: (416) 946-5522, fax: (416) 978-8511, e-mail: [email protected]. † Department of Pharmaceutical Sciences, University of Toronto. ‡ Toronto General Research Institute. § Department of Medical Biophysics, University of Toronto. | Department of Medical Imaging, University of Toronto. ⊥ Current address: Ying Tang, Ph.D., postdoctoral fellow, Radiation Oncology Branch, NIH/NCI, Bldg. 10, Rm. 3B42, 10 Center Dr., Bethesda, MD 20892. E-mail: [email protected].

affinity, second-generation TAG-72 mAb which has been conjugated to radionuclides for radioimmunodetection and radioimmunotherapy in patients with primary and metastatic TAG-72-positive cancers (10-16). Engineered scFv-based antibody fragments of CC49 have been reported (17-19). The very fast clearance of these CC49 scFvs from the blood suggested that small antibody fragments are attractive agents for imaging TAG-72-positive tumors. Recombinant scFv-based antibody fragments are technically favored over fragments containing interchain disulfide bonds such as Fab in part because the heterologous expression of scFv is more convenient and the yield is higher. In comparison with scFvs however, Fab is thermodynamically more stable because of the interchain disulfide bond and mutual stabilization between the VH/VL and CH1/CL interface. In addition, Fab is kinetically more stable than scFv as manifested by a slow unfolding rate, which renders Fab more resistant to denaturation (20, 21). Although murine Fab has the potential to be immunogenic in humans, the incidence of human anti-mouse antibody (HAMA) response reported from several studies in practice is very low (less than 2%) (2, 22). In this study, we report for the first time the synthesis of recombinant Fab (rFab) of monoclonal antibody CC49 in a heterologous Pichia pastoris expression system and their tumor targeting properties compared to enzymatically generated Fab (eFab) produced by proteolytic digestion of CC49 IgG. Our rationale for producing the rFab was several fold. First, we were interested in evaluating the ability of the P. pastoris expression system to produce rFab of antitumor antibodies. Second, we were interested in evaluating the effects of affinity tags that are often incorporated into recombinant antibodies (including the

10.1021/bc060260r CCC: $37.00 © 2007 American Chemical Society Published on Web 03/16/2007

678 Bioconjugate Chem., Vol. 18, No. 3, 2007

Tang et al.

Figure 1. (A) Linear representation of coexpression plasmids pPIC-Fd-L and pPIC-L-Fd. (B) Schematic representation of rFab. A FLAG tag (amino acid sequence DYKDDDDK) is fused to the N-terminus of L through a linker sequence (GGGGS); A His6 tag (amino acid sequence HHHHHH) was fused to the C-terminus of Fd through a GGGGS linker. L and Fd are covalently linked by a disulfide bond.

rFab) on their tumor and normal tissue localization. Third, we were interested in designing a method of producing Fab fragments of CC49 that would be more reproducible than that of enzymatic digestion of IgG, which is fraught with risk for heterogeneous generation of Fab with low molecular weight contaminating peptides as well as smaller fragments and residual intact IgG. Finally, although in the current study the rFab and eFab were radiolabeled with 123I, we speculated that the generation of rFab of CC49 with an incorporated His6 tag may allow radiolabeling in the future with 99mTc(I)carbonyl complexes as described by Waibel et al. (23).

EXPERIMENTAL PROCEDURES Materials. The CC49 hybridoma cell line was provided by Dr. J. Schlom at the U.S. National Cancer Institute (Bethesda MD). The hybridoma secretes murine CC49 IgG1,κ. E. coli TOP10F′ (Invitrogen, San Diego, CA) was used for cloning, subcloning, and propagation of plasmids. KM71H (aox1::ARG4, arg4) strain of P. pastoris was used as the host for protein expression. The pCR 2.1 vector (Invitrogen) was used for cloning the genes encoding the light chain (L) and Fd of the heavy chain from the hybridoma cells; the pPICZRA (Invitrogen) vector containing an R-factor secretion-signal peptide was used to construct coexpression plasmids for rFab expression. Taq and Pfx DNA polymerases (Invitrogen) were used for polymerase chain reaction (PCR). Restriction enzymes and ligases were from New England Biolabs Ltd. (Beverly, MA). Nickel-nitrilotriacetic acid (Ni-NTA) agarose was from Qiagen Inc. (Valencia, CA). Immobilized Protein-A was from Pierce

(Rockford, IL). All oligonucleotide primers for PCR were synthesized by ACGT Corporation (Toronto, ON). Expression of rFab in P. pastoris and Purification. The pPIC-Fd-L or pPIC-L-Fd plasmids (supplemental experimental procedures and Figure S1 in Supporting Information) were linearized at the unique Pme I site in the first 5′AOX1 promoter region before electrotransforming P. pastoris KM71H cells using a MicroPulser electroporator (Bio-Rad, Hercules, CA). The transformed KM71H cells were selected for resistance to Zeocin (100 µg/mL) on YPD agar plates (1% yeast extract, 2% peptone, 2% dextrose, and 2% agar). Transformed clones were grown at 30 °C in buffered glycerol-complex medium to an optical density of OD600 ) 10 and then were transferred to buffered methanol-complex medium at a cell density of 15% (w/w, wet cell weight) to induce protein expression. The methanol concentration and induction time were optimized. At the time point of maximum rFab yield determined by ELISA (see Supporting Information), the culture was centrifuged at 3000g for 15 min at 4 °C to isolate the supernatant containing secreted rFab. rFab was precipitated from the culture supernatant by adding ammonium sulfate to 50% saturation at 25 °C (31.3 g ammonium sulfate per 100 mL of supernatant). rFab was purified by affinity chromatography on a Ni-NTA agarose (Qiagen) column. Protein bound to the column was eluted with 100 mM imidazole in 20 mM Tris-HCl and 0.5 M NaCl, pH 8.0. The eluate was buffer-exchanged into PBS, pH 7.0, using Centricon YM-30 ultrafiltration devices (Amicon, Billerica, MA; Mr cutoff 30 kDa).

123I-Labeled

Recombinant Fab Fragments

Bioconjugate Chem., Vol. 18, No. 3, 2007 679

Figure 2. (A) Coomassie-stained SDS-PAGE gel of the crude culture supernatant collected at 48 h after methanol induction. The unpurified sample was concentrated five times and electrophoresed under nonreducing (lane 1) and reducing (β-mercaptoethanol, lane 2) conditions on a 4-20% Tris-HCl gradient gel. (B) SDS-PAGE comparison of purified rFab (lanes 1 and 3) and papain-produced Fab (lanes 2 and 4) of CC49 under nonreducing (lanes 1 and 2) and reducing (lanes 3 and 4) conditions. (C) Western blot of purified rFab (lanes 1 and 3) and papain-produced Fab (lanes 2 and 4) of CC49 under nonreducing (lanes 1 and 2) and reducing (lanes 3 and 4) conditions. The proteins were probed with a goat-anti mouse Fab. (D) Coomassie-stained SDS-PAGE gel of rFab under nonreducing (lane 1) and reducing (lane 2) conditions. (E) Western blot of purified rFab under nonreducing (lane 1) and reducing (lane 2) conditions. Proteins were probed with tetra-histidine antibody. (F) Western blot of purified rFab under nonreducing (lane 1) and reducing (lane 2) conditions. Proteins were probed with FLAG M2 antibody.

Enzymatic Preparation of CC49 Fab. Fab fragments of CC49 were prepared by proteolysis of CC49 IgG using immobilized papain (Pierce). After digestion, the supernatant was purified on immobilized Protein-A (Pierce). Purified enzymatically produced Fab (eFab) was buffer-exchanged into PBS, pH 7.0, using Centricon YM-30 ultrafiltration devices. Analysis of the Purity of rFab. The identity and purity of CC49 rFab was assessed by SDS-PAGE, Western blot, and sizeexclusion high performance liquid chromatography (HPLC). For SDS-PAGE, rFab (1-2 µg) was electrophoresed under reducing (β-mercaptoethanol) and nonreducing conditions on a 4-20% Tris-HCl gradient mini-gel (Bio-Rad, Mississauga, ON) stained with Coomassie brilliant blue R-250 (Bio-Rad). Western blot was performed using a panel of horseradish peroxidase (HRP)conjugated antibodies as probes, including goat anti-mouse Fab (Sigma), anti-tetra-histidine (Qiagen), and anti-FLAG M2 (Sigma). Each antibody was diluted at 1:2000 in PBS/0.5% Tween-20. rFab was also analyzed by size-exclusion HPLC on a BioSep SEC-S2000 column (Phenomenex Inc, Torrance, CA) eluted with 100 mM NaH2PO4 buffer, pH 7.0, at a flow rate of

0.8 mL/min using a System Gold Model 125 HPLC interfaced with a Model 166 UV detector (Beckman-Coulter, Mississauga, ON) set at 280 nm. Protein Concentration. The molar extinction coefficient for rFab at 280 nm [ ) 75 560 cm-1 M-1, equivalent to E ) 1.49 cm-1 (mg/mL)-1] was calculated (24) based on the amino acid sequence for CC49 Fab (25-28). The protein concentration was measured spectrophotometrically at 280 nm. Immunoreactivity of CC49 rFab with TAG-72. The ability of rFab or eFab to displace the binding of 111In-labeled CC49 IgG to bovine submaxillary mucin (BSM), a source of TAG72 (29), was measured. CC49 IgG was labeled with 111In (MDSNordion, Kanata, ON) to a specific activity of 237 kBq/µg (6.4 µCi/µg) following derivatization with diethylenetriaminepentaacetic acid (DTPA) as previously reported (30). Each well of Immulon2 HB strips (Dynex Technologies Inc., Chantilly, VA) was coated with 20 ng of BSM in 100 µL of PBS, pH 7.0, at 4 °C overnight. Nonspecific binding was blocked by 5% BSA in PBS, pH 7.0. rFab or eFab (1.0 or 2.5 µM) in 40 µL of PBS containing 1% BSA was added in duplicate, then 111In-CC49

680 Bioconjugate Chem., Vol. 18, No. 3, 2007

Figure 3. Size-exclusion HPLC of purified rFab of CC49. The solid line represents UV absorbance at 280 nm. rFab was detected as a single peak with a retention time of tR 9.2 min.

IgG (50 000 CPM) at a final concentration of 5 nM in 40 µL of PBS was added in each well. After incubation at room temperature for 2 h, the wells were rinsed with PBS and the radioactivity bound measured in a γ-counter. Tissue Distribution and Imaging of TAG-72-Positive Xenografts in Mice. The tumor and normal tissue distribution of 123I-labeled rFab or eFab were evaluated in female athymic mice bearing s.c. LS174T human colon cancer (TAG-72 positive) and A375 human melanoma (TAG-72 negative) xenografts (17). Four-week-old BALB/c nu/nu athymic mice (Charles River, Wilmington, MA) were injected s.c. with 1 × 106 LS174T cells in the right flank and with 2 × 106 A375 cells in the left flank. rFab and eFab were labeled with 123I using the Iodogen method (31) and then purified on a protein desalting spin column (Pierce). The radiochemical purity (RCP) of 123IrFab or eFab was >98% by paper chromatography (Whatman No. 1, Maidstone, England) developed in 85% methanol (Rf 123I-Fab: 0.0; R 123I- iodide). The specific activity of 123I-rFab f and eFab were 37 kBq/µg. At 10-14 days when the implanted tumors reached 5-8 mm in diameter, the mice were injected i.v. (tail vein) with 15 µg of 123I-rFab or eFab (0.5 MBq). The mice received Lugol’s iodine solution in their drinking water 2 days prior to radiopharmaceutical injection to attempt to block thyroid accumulation of radioiodine. Groups of three mice were sacrificed at 2, 6, 12, and 24 h postinjection (p.i.). Tumors and samples of normal tissues were collected and weighed, and their radioactivity content was measured in a γ-counter. Radioactivity uptake was expressed as percent injected dose per gram (% ID/ g) and as tumor-to-normal tissue (T/NT) ratios. In a separate study, tumor-bearing mice were injected i.v. with 3 MBq (15 µg) of 123I-rFab. Posterior whole body images were obtained up to 24 h p.i. using a small field-of-view gamma-camera fitted with a 4-mm pinhole collimator (ADAC Model TransCam, ADAC Laboratories Inc., Milpitas, CA). The principles of Laboratory Animal Care (NIH Publication No. 86-23, revised 1985) were followed, and animal studies were conducted under a protocol approved by the Animal Care Committee at the University Health Network (No. TG: 00-026) following Canadian Council on Animal Care (CCAC) guidelines. Statistical Comparisons. Statistical significance was assessed using an unpaired, two-tailed Student’s t-test or one-way ANOVA (P < 0.05).

RESULTS Construction of Coexpression Plasmids To Express rFab of CC49. The genes coding for the L chain and the Fd chain

Tang et al.

Figure 4. Comparison of the displacement of the binding of 111Inlabeled CC49 IgG to bovine submaxillary mucin (a source of TAG-72 antigen) coated onto wells in a microELISA plate by 1.0 or 2.5 × 10-6 M rFab or eFab. B/Bo represents the ratio of radioactivity bound in the presence compared to absence of rFab or eFab

up to the first cysteine residue in the hinge region of the heavy chain that forms the disulfide bond with the L chain were cloned (supplemental experimental procedures in Supporting Information). A coexpression plasmid based on the P. pastoris secretion vector pPICZRA was constructed to coexpress the L and Fd chains of CC49 (Figure 1). In the plasmid (pPIC-L-Fd or pPICFd-L, Figure 1A), the expression of the L and Fd chains was driven individually by a methanol-inducible 5′AOX1 promoter. Expression of CC49 rFab in P. pastoris in a Shake Flask. CC49 rFab was expressed in P. pastoris strain KM71H transformed with plasmid pPIC-L-Fd or pPIC-Fd-L. L and Fd chains were expressed as separate polypeptides, assembled into rFab in the cells through the formation of a disulfide bond between L and Fd peptide chains and then secreted into the culture medium as folded rFab (Figure 1B). Two transformants Clone #24-5 (pPIC-Fd-L) and Clone #5-3 (pPIC-L-Fd) resulted in the greatest rFab expression as measured by ELISA, and a large-scale expression was carried out in a 500 mL Erlenmeyer flask over 48 h induced with 1% methanol added to the culture medium every 12 h. The yield of purified rFab in shake-flask expression was 1-2 mg/L determined by measuring the OD280. Purity of CC49 rFab. The crude culture supernatant containing expressed rFab was collected after methanol induction. The supernatant was concentrated five times before being electrophoresed under nonreducing (Figure 2A, lane 1) and reducing (β-mercaptoethanol, Figure 2A, lane 2) conditions on a 4-20% Tris-HCl gradient gel. After purification, the SDS-PAGE gel revealed a single band at Mr 51 kDa (Figure 2B, lane 1). This apparent molecular weight (Mr) was consistent with the calculated Mr based on the amino acid sequence of CC49 rFab (2528). This band at Mr 51 kDa dissociated into a single band at 30 kDa under reducing conditions (Figure 2B, lane 3). Western blot revealed that both the 51 kDa and 30 kDa bands were reactive with goat anti-mouse Fab antibodies (Figure 2C, lanes 1 and 3). The 51 kDa and 30 kDa bands (Figure 2D) were also reactive with antibodies directed against tetra-histidine (Figure 2E) identifying the His6 tag at the C-terminus of the Fd chain and were reactive with anti-FLAG antibodies (Figure 2F) identifying the FLAG tag at the N-terminus of the L chain. These results confirmed that the 51 kDa protein was rFab and contained an intermolecular disulfide bond covalently linking the L and Fd chains. For comparison, SDS-PAGE analysis (Figure 2B, lanes 2 and 4) and Western blot (Figure 2C, lanes 2 and 4) of eFab prepared by proteolytic digestion of CC49 IgG using immobilized papain are shown. The slightly greater Mr for rFab compared with eFab (Mr 51 kDa vs 48 kDa) was

123I-Labeled

Bioconjugate Chem., Vol. 18, No. 3, 2007 681

Recombinant Fab Fragments

Table 1. Biodistribution of 123I-rFab in Athymic Mice Bearing s.c. TAG-72-Positive LS174T Human Colon Cancer Xenografts and TAG-72-Negative A375 Human Melanoma Xenograftsa 2 h, p.i.

6 h, p.i.

12 h, p.i.

24 h, p.i.

tissue

%ID/g

T/NT

%ID/g

T/NT

%ID/g

T/NT

%ID/g

T/NT

blood heart lung liver kidneys spleen stomach intestines muscle A375 LS174T

14.7 ( 1.73 6.5 ( 1.06 6.6 ( 0.70 3.8 ( 0.29 32.3 ( 4.96 3.5 ( 0.24 13.3 ( 1.99 2.8 ( 0.20 1.9 ( 0.32 4.8 ( 0.97 12.9 ( 2.21

0.9 ( 0.14 2.1 ( 0.44 1.9 ( 0.20 3.4 ( 0.34 0.4 ( 0.05 3.6 ( 0.40 1.0 ( 0.06 4.5 ( 0.43 6.9 ( 0.26 2.7 ( 0.21

5.3 ( 0.24 2.1 ( 0.13 2.8 ( 0.32 1.5 ( 0.30 7.9 ( 0.40 1.5 ( 0.23 6.5 ( 2.74 1.4 ( 0.30 0.9 ( 0.07 1.8 ( 0.22 10.1 ( 2.14

1.9 ( 0.41 4.8 ( 0.82 3.4 ( 0.40 6.7 ( 1.26 1.3 ( 0.28 6.8 ( 0.81 2.0 ( 0.76 7.4 ( 1.31 11.6 ( 2.33 5.7 ( 1.07

3.0 ( 0.24 1.0 ( 0.07 1.6 ( 0.09 0.8 ( 0.18 3.8 ( 0.58 0.8 ( 0.15 3.6 ( 1.62 0.7 ( 0.16 0.5 ( 0.05 0.9 ( 0.07 7.4 ( 0.93

2.4 ( 0.19 7.6 ( 0.42 4.8 ( 0.34 9.5 ( 0.83 2.1 ( 0.36 9.0 ( 0.68 2.7 ( 0.82 11.1 ( 1.42 15.4 ( 1.39 8.2 ( 0.64

1.1 ( 0.07 0.4 ( 0.07 0.6 ( 0.05 0.4 ( 0.06 1.2 ( 0.14 0.4 ( 0.03 1.0 ( 0.12 0.2 ( 0.02 0.2 ( 0.06 0.3 ( 0.02 6.0 ( 0.27

5.2 ( 0.10 14.1 ( 1.99 9.4 ( 0.38 14.7 ( 1.55 5.0 ( 0.39 15.9 ( 0.82 6.4 ( 0.56 25.1 ( 1.09 32.0 ( 9.31 18.0 ( 1.24

a

Tissue distribution was expressed as % ID /g. T/NT: Tumor (LS174T)-to-normal tissue ratios. Values are mean ( SE (n ) 3 mice/time point).

Table 2. Biodistribution of 123I-eFab in Athymic Mice Bearing s.c. TAG-72-Positive LS174T Human Colon Cancer Xenografts and TAG-72-Negative A375 Human Melanoma Xenograftsa 2h

a

6h

12 h

tissue

%ID/g

T/NT

%ID/g

T/NT

%ID/g

T/NT

blood heart lung liver kidneys spleen stomach intestines muscle A375 LS174T

10.0 ( 1.28 4.4 ( 0.71 6.8 ( 0.65 5.4 ( 0.64 89.1 ( 4.77 4.9 ( 1.01 30.1 ( 9.79 5.1 ( 0.97 2.8 ( 0.23 3.9 ( 0.30 11.1 ( 1.19

1.2 ( 0.22 2.7 ( 0.59 1.7 ( 0.29 2.1 ( 0.40 0.1 ( 0.02 2.5 ( 0.66 0.5 ( 0.19 2.4 ( 0.55 4.0 ( 0.61 3.3 ( 0.62

7.3 ( 1.60 2.9 ( 0.94 4.0 ( 1.22 3.2 ( 0.96 9.9 ( 1.11 2.9 ( 0.97 15.8 ( 4.97 2.6 ( 0.78 1.3 ( 0.21 2.1 ( 0.67 8.4 ( 0.20

1.2 ( 0.23 3.6 ( 1.07 2.5 ( 0.61 3.4 ( 1.37 0.9 ( 0.12 3.6 ( 0.94 0.6 ( 0.20 3.9 ( 1.16 6.9 ( 1.02 4.9 ( 1.47

2.1 ( 0.17 0.8 ( 0.03 1.1 ( 0.10 2.1 ( 0.26 3.6 ( 0.48 1.0 ( 0.07 2.4 ( 0.36 0.6 ( 0.06 0.2 ( 0.03 0.5 ( 0.03 5.7 ( 0.78

2.8 ( 0.50 6.9 ( 1.20 5.1 ( 0.41 2.7 ( 0.10 1.6 ( 0.12 5.8 ( 0.52 2.4 ( 0.21 10.1 ( 0.46 23.0 ( 3.38 10.4 ( 1.60

Tissue distribution was expressed as % ID /g. T/NT: Tumor (LS174T)-to-normal tissue ratios. Values are mean ( SE (n ) 3 mice/time point).

expected because the rFab included a His6 tag, a FLAG tag, and the GGGGS linker (Figure 1B), which contributed 2.8 kDa to the molecular mass. Size-exclusion HPLC (Figure 3) showed a single peak with a retention time (tR) of 9.2 min corresponding to rFab. Immunoreactivity of CC49 rFab with TAG-72. rFab competed with 111In-CC49 IgG for binding to BSM, a source of TAG-72 antigen (Figure 4). The specific binding of 111InCC49 IgG to TAG-72 was reduced to 76 ( 5.1% by 1.0 × 10-6 M rFab and to 57 ( 5.2% by 2.5 × 10-6 M rFab. A 2.5fold lower concentration of eFab (1.0 × 10-6 M) was required to reduce the binding of 111In-CC49 IgG to 57 ( 1.4% (Figure 4), indicating that the affinity of rFab for TAG-72 was 2- to 3-fold lower than that of rFab. Tissue Distribution and Imaging of TAG-72-Positive Xenografts in Mice. 123I-rFab accumulated in LS174T xenografts reaching a level of 12.9 ( 2.2% ID/g at 2 h p.i. (Table 1). Tumor retention of 123I-rFab decreased from 10.1 ( 2.1% ID/g at 6 h to 7.4 ( 0.9% ID/g at 12 h p.i. (P > 0.05 compared with that at 2 h) and to 6.0 ( 0.3% ID/g at 24 h p.i. (P ) 0.04 compared with that at 2 h). In contrast, the retention of 123IrFab in control TAG-72-negative A375 tumors was significantly lower at all time points (P < 0.05) and decreased more than 16 fold between 2 and 24 h p.i. 123I-rFab was rapidly eliminated from the blood. By 24 h p.i., the blood concentration decreased 13 fold compared to that at 2 h p.i. (1.1 ( 0.07 vs 14.7 ( 1.73, P < 0.01). The greatest tissue uptake occurred in the kidneys at 2 h p.i. (32.3 ( 4.9% ID/g). The specific accumulation of 123I-rFab in LS174T tumors was reflected in the T/NT ratios (Table 1). The ratios of tumor (LS174T) to all tissues including the TAG-72-negative A375 tumors increased from 2 to 24 h p.i. For example, the tumor/blood (T/B) ratio increased from 0.9 ( 0.1 at 2 h p.i. to 5.2 ( 0.1 at 24 h (P < 0.0001). At 24 h p.i. the ratio of radioactivity in LS174T to that in A375 tumors was 18.0 ( 1.2. Uptake of 123I-rFab in LS174T tumors (Table

1) was not significantly different than that of 123I-eFab (Table 2; P > 0.05). Interestingly, the uptake of 123I-eFab in the kidneys was significantly greater at 2 h than that of 123I-rFab (89.1 ( 4.8 vs 32.3 ( 4.9 %ID/g, P < 0.01) but was similar to that of 123I-rFab at later times (P > 0.05). T/NT ratios were similar for 123I-rFab and 123I-eFab except for a much higher tumor/ kidney ratio at 2 h (P < 0.001) as well as greater tumor/liver and tumor/spleen ratios at 12 h after injection (P < 0.05) for 123I-rFab. LS174T xenografts were clearly imaged as early as 2 h after injection of 123I-rFab, whereas similarly sized A375 tumor could not be differentiated from the body background (Figure 5).

DISCUSSION The second-generation, high-affinity TAG-72 monoclonal antibody CC49 has proven to be one of the most clinically useful antitumor antibodies for radioimmunoimaging and radioimmunotherapy of solid tumors (10-17). In this study, we generated rFab of CC49 using a P. pastoris expression system and directly compared its tumor and normal tissue localization properties with those of enzymatically generated CC49 Fab (eFab) in athymic mice implanted s.c. with LS174T human colon cancer xenografts. To our knowledge, such a direct comparison of eFab and rFab of an antitumor monoclonal antibody has never been reported, despite the potential that genetically engineered antibodies may demonstrate altered binding affinities and tissue distribution due to their incorporated affinity tags (e.g., polyhistidine) or other modifications. The CC49 rFab differs from eFab in that rFab contains a His6 tag, a FLAG tag, and the GGGGS peptide linkers. Despite a 2-to-3-fold lower TAG-72binding affinity for rFab in Vitro, both rFab and eFab exhibited a similar pattern of tumor retention and tissue distribution in tumor-bearing mice except in the liver and kidneys. Kidney retention of 123I-rFab was significantly lower than that of 123I-eFab (32.3 ( 4.9 vs 89.1 ( 4.8% ID/g, P < 0.01)

682 Bioconjugate Chem., Vol. 18, No. 3, 2007

Tang et al.

Figure 5. Posterior whole body image of a representative athymic mouse bearing a s.c. TAG-72-overexpressing LS174T human colon cancer xenograft on the right flank (white arrow). A s.c. TAG-72-negative A375 human melanoma xenograft on the left flank (arrowhead only) served as the negative control tumor. Each mouse was injected i.v. with 15 µg of 123I-rFab (3 MBq) and imaged at 2, 6, 12, and 24 h p.i. The series of images were obtained from the same mouse. 250 000, 100 000, 100 000, and 50 000 counts were collected at 2, 6, 12, and 24 h p.i., respectively.

at 2 h p.i., but the difference was not significant at later times, i.e., 6 and 12 h (P > 0.05). The reason for the lower initial kidney uptake of 123I-rFab may be related to the net negative charges introduced by the FLAG tag. It was previously reported that introduction of three negatively charged amino acids reduced the kidney retention of radioiodinated CC49 scFv fragments at early time points after administration to mice bearing s.c. LS174T xenografts (32). It was proposed that changes in the isoelectric point (pI) could affect kidney uptake. However, the difference in kidney uptake among three CC49 scFv variants, i.e., scFv[pI 8.1], scFv[pI 5.8], and scFv[pI 5.2], was not statistically significant except at 0.5 h p.i. At this time, the scFv with the lowest pI [pI 5.2] showed the highest kidney uptake, whereas scFv with intermediate pI [pI 5.8] had the lowest uptake (32). It appears, therefore, that pI did not correlate directly with kidney uptake among the three scFv variants that were studied. The pI values of eFab and rFab were not measured in our study, but the theoretical pI values were calculated using the Swiss-Prot database (33) based on the amino acid sequences. These pI values were 6.3 for eFab and 7.6 for rFab. His6 was also reported to affect the blood clearance and kidney retention of a radioiodinated (scFv)2 of CC49. The blood clearance for the His6-tagged (scFv)2 was slightly slower while the kidney retention was lower at early time points (20.0 and 14.8 %ID/g at 0.5 and 1 h p.i., respectively) than that for (scFv)2 (41.2 and 24.7 %ID/g at 0.5 and 1 h p.i., respectively) (18). In our study,

the liver retention of 123I-rFab was lower than that of 123I-eFab and the tumor-to-liver ratio was higher for 123I-rFab at all times examined. The reason for the lower liver uptake of 123IrFab is not known, but this finding suggests that it could be more useful than 123I-eFab for scintigraphic detection of hepatic lesions. The blood concentration of 123I-rFab at 12 h was higher than that of 123I-eFab (3.0 ( 0.2 vs 2.1 ( 0.2% ID/g, P < 0.05). However, this did not significantly lower the T/B ratio for 123IrFab compared to 123I-eFab (2.4 ( 0.2 vs 2.8 ( 0.5, respectively, P > 0.05) because the tumor uptake for 123I-rFab was higher. In short, the 3-fold lower TAG-72 binding affinity for rFab than eFab in Vitro failed to translate into any significant difference in tumor accumulation. Consistent with this finding, it was previously noted that a 320-fold decrease in Kd of an anti-HER2/neu scFv resulted in only a 7-fold increase in tumor accumulation in mice (34). These and other observations indicate that in Vitro receptor-binding affinity is not the sole determinant of the accumulation of antibody fragments in tumors (35, 36). Nevertheless, it has been reported that increased avidity by means of engineering bivalent, dimeric (scFv)2 (Mr 54 000) improved tumor retention 2 to 36-fold at 24 h p.i. compared to monovalent scFv (Mr 27 000) in mice (37, 38). T/B ratios between divalent (scFv)2 and monovalent scFv, however, were similar (37, 38). The enhanced tumor retention of divalent

123I-Labeled

Recombinant Fab Fragments

(scFv)2 was likely a combined result of higher avidity, and prolonged residence time as a result of the larger molecular size (37, 38). Schlom’s group and others studied the tumor retention and tissue distribution in mice bearing LS174T xenografts of radioiodinated Fab′ produced by pepsin digestion of CC49 IgG. The tumor retention of radioiodinated Fab′ was 3.9 to 6.2% ID/g at 6 h p.i.; it was 2.9 to 3.7% ID/g at 24 h (17, 19). This level of tumor retention is lower than that observed in our study for 123I-rFab (10.1 ( 2.1% ID/g at 6 h; 6.0 ( 0.3% ID/g at 24 h) or 123I-eFab (8.4 ( 0.2% ID/g at 6 h). It was noted that the size of LS174T tumor xenografts in our study was similar to that previously used by Schlom’s group (0.5-0.8 cm diameter) to evaluate the tumor uptake of CC49 Fab′. The blood concentration was also much lower for radioiodinated Fab′ (0.9-1.68 and 0.1 %ID/g at 6 and 24 h, respectively) than observed in our study for 123I-rFab (5.3 and 1.1 %ID/g at 6 and 24 h, respectively) (17, 19). It is unclear whether the slight differences in amino acid sequences between Fab′ and rFab, or between Fab′ and eFab, may be responsible for these disparities in biodistribution and blood clearance, because in our study, mice were injected with a higher dose of rFab or eFab (15 µg/mouse) than previously examined for Fab′ (