¹²⁴I-Labeled Engineered Anti-CEA Minibodies and Diabodies Allow High-Contrast, Antigen-Specific Small-Animal PET Imaging of Xenografts in Athymic Mice

Production and Purification of T84.66 Minibody and Diabody

The anti-CEA scFv-C_H3 minibody and diabody (shown schematically in Fig. 1) used in these studies were produced and purified as previously described in detail (16). Briefly, the T84.66/GS18 minibody and diabody were expressed in NS0 murine myeloma cells using the pEE12 expression vector and a cytomegalovirus promoter and glutamine synthetase as the selectable marker (Lonza Biologics) (17). Protein was purified by hydrophobic interaction chromatography (Source ISO; Amersham Pharmacia Biotech) followed by anion exchange chromatography (HQ50; PE Biosystems) using a BioCAD 700E perfusion chromatography system (PE Biosystems). Activity was determined by enzyme-linked immunosorbent assay using microtiter wells coated with a recombinant CEA protein (N and A3 domains) (18) and antihuman Fc (minibody) or antimurine Fab (diabody) (Jackson ImmunoResearch) for detection.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Schematic drawing shows domain structure of parental intact T84.66 antibody (A), T84.66 diabody (B), and T84.66 minibody (C). Anti-CEA minibody (80 kDa) and diabody (55 kDa) were derived from T84.66, a high-affinity and highly specific antibody that recognizes an epitope on A3 domain of CEA. Minibody (C) has glycine-serine-rich 18-amino-acid linker (GS18) between variable light (V_L) and variable heavy chains (V_H) and human IgG1 hinge (H) joining them to CH3 domain, also derived from human IgG1. Gene encoding diabody (B) encodes glycine-serine-rich 8-amino-acid linker (GS8) between V_L and V_H regions. Amino acid sequence is GSTSGGGSGGGSGGGGSS for GS18 and GGGSGGGG for GS8.

Radioiodination of T84.66 Minibody and Diabody

Radioiodination of the engineered antibody fragments was accomplished by the IODO-GEN method (Pierce Biotechnology) as previously described (12). Carrier-free ¹²⁴I was produced at the Cyclotron Facility of Memorial Sloan-Kettering Cancer Center as previously described (19). Labeling reactions (0.1–0.2 mL) typically contained 0.2–0.3 mg of purified protein and 65–98 MBq (1.75–2.65 mCi) of ¹²⁴I. Radioiodinated proteins were purified by size-exclusion high-performance liquid chromatography (HPLC) on either a TSK2000 (Tosoh Biosep) or Superdex 75 (Amersham Biosciences) column, with phosphate-buffered saline (PBS)/1% human serum albumin (HSA) as the running buffer. The radiolabeling efficiency was determined by integrating areas on the HPLC trace and determining the radioactivity associated with the 80-kDa protein peak for minibody or 55-kDa peak for diabody as a percentage of total radioactivity eluted. Immunoreactivity was determined by incubation of the labeled protein with a 20-fold excess (w/w) of either purified CEA or the recombinant N-A3 fragment of CEA in 0.15 mL of PBS/1% HSA as previously described. Samples were analyzed by size-exclusion HPLC on tandem Superose 6 HR 10/30 columns (Amersham Biosciences) to assess formation of Ab:Ag complexes.

Cell Lines and Xenografts

The LS174T human colorectal carcinoma cell line (CL 188) and C6 rat glioma cell line (CCL 107) were obtained from the American Type Culture Collection and maintained under standard conditions. Xenografts were established in 8-wk-old CD1 female nude mice (Charles River Laboratories) by subcutaneous inoculation of 1–2 × 10⁶ cells 10–14 d before the imaging studies. All animal handling and studies were performed in accordance with the guidelines of the University of California, Los Angeles (UCLA), Animal Research Committee.

PET Scanning

Mice were imaged using the prototype small-animal PET scanner developed at the Crump Institute for Molecular Imaging, UCLA (20). Thyroid uptake of radioiodine was blocked by pretreatment using 10 drops of saturated KI per 100 mL of drinking water for 24 h before injection of radioiodinated protein. Stomach uptake was blocked by administration of 1.5 mg of potassium perchlorate in 0.2 mL of PBS by gastric lavage 30 min before injection. The mice were injected in the tail vein either with 1.9–3.1 MBq (53–85 μCi) of ¹²⁴I-minibody or with 3.1 MBq (85 μCi) of ¹²⁴I-diabody in saline/1% HSA. After the appropriate uptake time had elapsed, the mice were anesthetized by intraperitoneal injection of a mixture of ketamine (80 mg/kg final dose) and xylazine (10 mg/kg final dose), placed prone, and imaged using the PET scanner with the long axis of the mouse parallel to the long axis of the scanner. Acquisition time was 28 min (4 min per bed position; 7 bed positions). Images were reconstructed using a 3-dimensional filtered backprojection reconstruction algorithm (21) for quantitation (see below) or using a maximum a posteriori reconstruction protocol (22) for presentation of images. Some mice were also imaged using ¹⁸F-FDG 2 d before the ¹²⁴I-minibody scan. These were administered 5.6–7.4 MBq (150–200 μCi) of ¹⁸F-FDG via the tail vein, and scans were acquired as described above, beginning 1 h after tracer injection. In some experiments, after scanning, the animals were euthanized; the tumors were excised, weighed, and counted in a well counter (Cobra II Auto-Gamma; Packard); and the %ID/g was calculated.

Digital Whole-Body Autoradiography (DWBA)

The mice were sacrificed and frozen in carboxymethyl cellulose (Aldrich) in preparation for sectioning using a Cryostat (PMV). Coronal cross-sections were obtained with a thickness of 45 μm. DWBA was performed using a Fujifilm BAS 5000 PhosphorImager (Fujifilm Medical Systems U.S.A., Inc.) and digital plates with a final resolution of ∼100 μm. DWBA data were analyzed using Mac BAS software, version 2.4 (Fujifilm Medical Systems U.S.A., Inc.).

Data Analysis

Preliminary quantitation of the small-animal PET images was performed using the Crump Institute Integrated Imaging Software Package, or CRIIISP (Crump Institute for Molecular Imaging, UCLA). From the 3-dimensional filtered backprojection reconstruction, several planes (of a total of 64) encompassing the tumors were selected in the coronal orientation and averaged. Regions of interest (ROIs) were drawn for both the control and LS174T CEA-positive tumors centered on the peak of the activity profile. A soft-tissue region in the neck, as well as the area of maximum activity in the abdominal region, was also included for ROI analysis. Approximately equal-sized ROIs were drawn. ROI counts per pixel per minute were converted to counts per cubic centimeter per minute using a calibration factor obtained from scanning a cylinder containing a known amount of ¹²⁴I activity. After decay correction, these data were converted to %ID/g by dividing by the known amount of injected activity. Target-to-background ratios were then determined for individual mice and averaged. The effects of ROI positioning were determined by averaging at least 3 ROIs and assessing the variability across regions. Statistical analysis was performed using the Excel 2000 (Microsoft) software package.

¹²⁴I-Labeling of Anti-CEA Minibody and Diabody

As shown in Figure 2, ¹²⁴I was readily conjugated to the T84.66 minibody and T84.66 diabody after short incubations (3–5 min). Specific activities after labeling (n = 3) ranged from 81 to 133 kBq/μg (2.2–3.6 μCi/μg). The immunoreactivity of the labeled protein, as determined by size-exclusion HPLC analysis after incubation with excess antigen, was >90% in the case of minibodies and 42% for diabodies.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Size-exclusion HPLC analysis of ¹²⁴I-radiolabeled anti-CEA minibody (A) and diabody (B). ¹²⁴I was conjugated to T84.66 minibody and T84.66 diabody as described in Materials and Methods. Radiolabeling efficiency was determined by integrating areas on HPLC trace and determining radioactivity associated with 80-kDa protein peak for minibody or 55-kDa peak for diabody as percentage of total radioactivity eluted. Labeling efficiencies were 33% (not shown) or 46% for minibody (A) and 88% for diabody (B). Peak fractions (based on protein absorbance) were pooled for animal studies. Smaller peaks represent unincorporated label and low-molecular-weight components.

PET Imaging of Xenografts Using ¹²⁴I-Anti-CEA Minibody

To evaluate tumor targeting of the ¹²⁴I-minibody, antigen-positive (LS174T colorectal carcinoma) and antigen-negative control (C6 glioma) xenografts were established by subcutaneous inoculation in athymic mice. PET imaging studies were conducted on animals bearing CEA-positive tumors averaging 142 mg (range, 11–487 mg). In one of the study phases, each of 4 mice was injected in the tail vein with ∼2 MBq (53 μCi) of ¹²⁴I-minibody, and whole-body PET scans were obtained beginning 4 and 18 h after administration. The image in Figure 3C clearly demonstrates high uptake into the positive (LS174T) tumor and low activity in the control (C6) tumor. The positive tumor-to-control tumor uptake ratio for the ¹²⁴I-anti-CEA minibody was 3.46:1 at 4 h and 10.85:1 at 18 h, demonstrating high specificity (n = 8; Table 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

(A) Subcutaneous LS174T (left shoulder) and C6 glioma (right shoulder) xenografts were established in nude mouse. (B) PET imaging with ¹⁸F-FDG was done 2 d before injection of ¹²⁴I (10/64 planes shown). (C) Mouse was injected with 3.1 MBq (85 μCi) ¹²⁴I-minibody and imaged at 18 h by PET (10/64 planes shown). Both these images (B and C) are on same color scale. (D) After 18-h scan, mouse was euthanized and frozen, and whole-body coronal sections were cut in cryostat and processed for DWBA. (E) DWBA confirms specific localization of ¹²⁴I minibody to CEA-positive tumor and low levels of background activity.

Comparison of the PET image encompassing the C6 and LS174T xenografts, and a corresponding DWBA section (Fig. 3E), clearly demonstrates that the region of high uptake observed in the scan correlates with autoradiographically detected tumor-associated radioactivity. Furthermore, an ¹⁸F-FDG PET scan of the same animal, performed before the ¹²⁴I-minibody scan, allowed visualization of both tumors (Fig. 3B). The accumulation of ¹⁸F-FDG activity in the control and CEA-positive tumors was not significantly different. These studies confirm that the observed high localization of the ¹²⁴I-minibody was antigen specific.

Visualization of CEA-Positive Xenografts Using ¹²⁴I-Anti-CEA Diabody

Experiments were conducted similarly using the ¹²⁴I-anti-CEA diabody in animals (n = 4) bearing CEA-positive tumors with an average weight of 183 mg (range, 40–420 mg). PET imaging performed 4 h after administration of ¹²⁴I-anti-CEA diabody revealed rapid and specific uptake in the CEA-positive tumors (Fig. 4 C), with a tumor-to-background uptake ratio of 3.99:1 (Table 1). The contrast between the CEA-positive LS174T and CEA-negative C6 xenografts was even more striking at 18 h (Figs. 4D and 4E); a ratio could not be calculated because the C6 xenografts were not distinguishable.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Comparison of ¹²⁴I minibody and ¹²⁴I diabody by serial PET imaging at 4 and 18 h. Mice bearing LS174T (LS) and C6 rat glioma (C6) xenografts were injected via tail vein with 1.9–3.1 MBq (65–85 μCi) ¹²⁴I minibody (A and B) or diabody (C–E) and imaged at 4 and 18 h. At 18 h, background activity in central region was minimal in ¹²⁴I diabody (D and E)–injected animal, resulting in high contrast, which can be seen in these maximum a posteriori reconstructed images (10/64 planes shown). A–D are on common scale. D was rescaled to E to illustrate excellent contrast achieved with ¹²⁴I diabody at 18 h.

Tumor-to-Background Ratios from Serial Imaging

Imaging by PET allowed evaluation of tracer localization and distribution over time in the same animal. All animals were imaged 4 and 18 h after injection of minibody or diabody. In animals that were injected with ¹²⁴I-anti-CEA minibody, there was substantial activity in the blood pool at the 4-h time point and hence the tumor-to-background ratio was only about 0.8:1 (Fig. 4A; Table 1). Subsequently, this ratio improved to 3.5:1 at the 18-h time point as most of the activity in the nonspecific areas cleared, whereas the activity in the CEA-positive tumors remained high (Fig. 4B). Background ratios of tumor to soft tissue averaged 3.05:1 at 4 h and improved to 11.03:1 at 18 h (Table 1).

Higher tumor-to-background ratios were achieved in animals administered ¹²⁴I-labeled anti-CEA diabody. In these animals (n = 4), the tumor-to-background (background in abdominal region) ratios improved from 1.8:1 at 4 h after injection (Fig. 4C) to 6.1:1 at 18 h after injection (Figs. 4D and 4E). Background ratios of tumor to soft tissue were even higher, increasing from 4:1 at 4 h to 11:1 by the 18-h scans. This improvement in the tumor-to-background ratios achieved with ¹²⁴I-labeled anti-CEA diabody, especially regarding the central activity in the abdominal region, provides greater contrast and renders a clear image of the CEA-positive tumor (Fig. 4E).

Tissue Counting

After the 18-h scans, animals not used in DWBA studies were sacrificed, and activity in various tissues was quantitated using a γ-counter. The results in Table 2 confirm high tumor uptake in the LS174T model, with a mean uptake of 20.55 ± 2.74 %ID/g for the ¹²⁴I-T84.66 minibody (n = 5), consistent with previous work using ¹²³I/¹³¹I-radiolabeled T84.66 minibody (5,13). Uptake of ¹²⁴I-minibody by the control C6 xenografts and activities in heart, liver, and kidney were significantly lower than uptake in the positive tumors, and the tumor-to-control ratios determined by counting exceeded 13:1 for the indicated tissues and were consistent with ratios determined from the PET scans.

Similarly, biodistribution data for the anti-CEA diabody (Table 2) yielded activity levels of 4.53 ± 0.50 %ID/g in the LS174T xenografts (n = 4) at 18 h, comparable to the previously published 4.56 ± 0.5 %ID/g at 24 h using ¹²³I-T84.66 diabody (12). Control C6 tumor activity was only 0.23 ± 0.03 %ID/g, resulting in a tumor-to-control tumor activity ratio of 21.3:1 (Table 2). The highest observed tumor-to-normal ratio was 61.6:1 for LS174T to heart. Again, these results validate the high contrast observed in the PET images using ¹²⁴I-diabody as the tracer.