Activatable Fluorescent Molecular Imaging of Peritoneal Metastases following Pretargeting with a Biotinylated Monoclonal Antibody

Optical imaging has been proposed to improve the detectability of cancer lesions intraoperatively but requires high target-to-background ratios to identify small clusters of cancer cells. Activatable or “smart” optical probes are appealing because they only fluoresce when bound to their target site (1–6). Several strategies have emerged to create such smart optical probes. One approach, developed by Weissleder et al., is based on protease activation (ProSense 680 and 750, VisEn Medical, Inc., Woburn, MA) wherein fluorophores in the quenched state are released from their protein backbone by enzymes found in specific microenvironments, such as those overexpressed in neoplasms and atherosclerotic plaques (2–6). However, maximal fluorescence using extrinsic proteases often requires ≥24 h and uptake can be seen nonspecifically in organs that normally express the critical enzyme. Moreover, the protease is often found outside the cell and may not accurately reflect the location or margins of the tumor.

Here, we propose an alternative mechanism of target-specific optical activation for in vivo molecular imaging. This strategy is based on a modification of a known observation that the fluorescence of BODIPY conjugated to avidin or streptavidin is greatly (∼10-fold) increased after binding to biotin. This likely occurs because the photon-induced electron transfer from aromatic amino acids, such as tryptophan or tyrosine, to the excited state of BODIPY is inhibited by biotin-avidin binding, leading to a dequenching of the fluorophore (Fig. 1A). An in vitro assay based on this phenomenon is in use for measuring endosome fusion in live cells and studying membrane-interacting molecules (7–9). We have taken advantage of this BODIPY-biotin activation for in vivo imaging whereby a biotinylated monoclonal antibody is first given to pretarget peritoneal metastases and is then followed by a neutravidin-conjugated BODIPY-FL (nAv-BDPfl), which activates on binding. This method offers a highly specific and generally applicable paradigm for optical molecular imaging of tiny tumor clusters of peritoneal metastases.

Synthesis of biotinylated Erbitux and Zenapax. Amine-reactive biotin, sulfosuccinimidyl-6-(biotin-amido)hexanoate (Sulfo-NHS-LC-Biotin) was purchased from Pierce Chemical Co. (Rockford, IL). A recombinant, chimeric, IgG1 monoclonal antibody cetuximab (Erbitux) that binds specifically to the epidermal growth factor receptor (EGFR; HER1 or c-ErbB-1) was purchased from Bristol-Myers Squibb Co. (Princeton, NJ). As a control, humanized anti-human interleukin-2 receptor α subunit (IL-2Rα) antibody with a complimentary determination region against IL-2Rα grafted on a human IgG1 framework (daclizumab, Zenapax; ref. 10) was a generous gift of Dr. Thomas Waldman [Metabolism Branch, National Cancer Institute (NCI)/NIH, Bethesda, MD]. Zenapax is a good control antibody for Erbitux because it is isotype matched but also shows >98% protein sequence homology with Erbitux.

At room temperature, 12 μL of 10 mmol/L NHS-LC-Biotin dissolved in DMSO were added to 1 mg (6.7 nmol) of Erbitux or Zenapax in 388 μL of 0.1 mol/L Na₂HPO₄ and incubated for 15 min. Unreacted biotin was separated from the antibody by gel filtration using a Sephadex G50 (PD-10; GE Healthcare, Milwaukee, WI). Biotinylated Erbitux (b-Erb) and biotinylated Zenapax (b-Zen) samples were kept at 4°C.

The protein concentrations of b-Erb and b-Zen samples were determined by measuring the absorption at 280 nm and validated with Coomassie Plus protein assay kit (Pierce Chemical) by measuring the absorption at 595 nm with a UV-Vis system (8453 Value UV-Bis system, Agilent Technologies, Palo Alto, CA) using standard solutions of known concentrations of Erbitux and Zenapax (100, 200, and 400 μg/mL).

The biotin labeling ratio was determined by the HABA assay (Pierce Chemical). HABA/avidin solution (900 μL) was put into a 1 mL cuvette, and the absorbance of this solution at 500 nm was measured and recorded as A₅₀₀ HABA/avidin (11). This measurement procedure was repeated twice with 250 and 500 μg/mL solutions, and the average number of biotin molecules per Erbitux or Zenapax molecule was determined. The number of biotin molecules conjugated to Erbitux and Zenapax was 10 (b-Erb) and 9 (b-Zen), respectively. By changing the concentration of NHS-LC-Biotin, the number of biotin molecules per Erbitux antibody was adjusted to either 2, 12, or 23.

To validate that biotinylation did not compromise the binding ability of antibody, unconjugated Erbitux, 12b-Erb, and 23b-Erb were radiolabeled with ¹²⁵I with modified chloramine-T method. The binding assay to A431 cells was done. In brief, 40 μg of antibodies were dissolved in PBS (pH 7.4) and 400 μCi of Na¹²⁵I were added (GE Healthcare). Then, 6 μg of chloramine-T hydride were added in the reaction mixture and incubated for 5 min. The solution was applied to the Sephadex G50 column (PD-10), and the antibody fractions were collected. The labeling yields were 81%, 72%, and 76% for Erbitux, 12b-Erb, and 23b-Erb, respectively. For the binding assay, 4 ng of each antibody were incubated with 2 × 10⁶ A431 cells in 300 μL of PBS/0.025% bovine serum albumin for 2 h on the ice. To determine the nonspecific binding, two other reaction mixtures containing 40 μg Erbitux were added to block the specific binding. The cells were washed twice and counted with ¹²⁵I to determine the binding fraction.

Synthesis of BODIPY-FL–conjugated Erbitux. Amido-reactive BODIPY-FL was purchased from Molecular Probes, Inc. (Eugene, OR). At room temperature, 500 μg (3.3 nmol) of Erbitux in Na₂HPO₄ were incubated with 10 to 100 nmol (1–10 μL/10 mmol/L) of BODIPY-FL-succinimidyl ester in DMSO at pH 8.5 for 15 min. The mixture was purified with a Sephadex G50 column (PD-10). Erbitux-conjugated BODIPY-FL (Erb-BDPfl) was kept at 4°C as stock solutions.

The protein concentration of Erb-BDPfl samples was determined with Coomassie Plus protein assay kit by measuring the absorption at 595 nm with a UV-Vis system (8453 Value UV-Bis system) using standard solutions of known concentrations of Erbitux (100, 200, and 400 μg/mL). Then, the concentration of BODIPY-FL was measured by the absorption at 508 nm, respectively, with the UV-Vis system to confirm the number of fluorophore molecules conjugated with each Erbitux molecule. By changing the concentration of the Erbitux solution, the number of BODIPY-FL molecules per Erbitux was adjusted to 2.2.

Synthesis of nAv-BDPfl. Neutravidin was purchased from Pierce Biochemical, Inc. (Milwaukee, WI). At room temperature, 400 μg (5.9 nmol) of neutravidin in 198 μL of Na₂HPO₄ were incubated with 12 nmol (2 μL/6 mmol/L) of BODIPY-FL-succinimidyl ester in DMSO for 15 min. The mixture was purified with Sephadex G50 (PD-10). nAv-BDPfl was kept at 4°C.

The protein concentration was determined with Coomassie Plus protein assay kit by measuring the absorption at 595 nm with a UV-Vis system (8453 Value UV-Bis system). Then, the BODIPY-FL concentration was measured by the absorption at 504 nm with a UV-Vis system to confirm the number of BODIPY-FL molecules conjugated with each neutravidin molecule. The number of BODIPY-FL molecules per avidin was 1.3.

Measurement of fluorescence enhancement of nAv-BDPfl. To investigate the fluorescence enhancement of nAv-BDPfl on binding to biotinylated antibody, fluorescence intensity and emission spectra of nAv-BDPfl were measured in the presence of Erbitux or b-Erb by the Maestro In-Vivo Imaging System (CRi, Inc., Woburn, MA). Erbitux (5 μg) or b-Erb (5 μg) in 390 μL PBS was placed in a nonfluorescent 96-well plate (Costar, Corning, Inc., Corning, NY) and 5 μg nAv-BDPfl was added to each well. To investigate the activation potential of nAv-BDPfl, high-dose biotin (5 μg) was also added to each of the mixed solutions consisting of nAv-BDPfl and Erbitux and nAv-BDPfl and b-Erb. For the target-specific activation, 5 μg nAv-BDPfl and 5 μg b-Erb in 390 μL PBS and 5 μg nAv-BDPfl and 5 μg b-Zen in 390 μL PBS were placed in a nonfluorescent 96-well plate. A band pass filter from 445 to 490 nm and a long pass filter over 515 nm were used for emission and excitation light, respectively. The tunable filter was automatically stepped in 10-nm increments from 500 to 800 nm, while the camera captured images at each wavelength interval with constant exposure. A region of interest (ROI) as large as each well was drawn to determine the emission spectra using commercial software (Maestro software, CRi). The mean fluorescence intensity in arbitrary unit (a.u.) as well as the SD of each well were measured using ImageJ software.³

Cell culture. EGFR (HER1)-overexpressing A431 human epidermoid carcinoma cells were purchased from the American Type Culture Collection (Manassas, VA). A431 cells were grown in DMEM (Life Technologies, Gaithersburg, MD) containing 10% fetal bovine serum (Life Technologies), 0.03% l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin in 5% CO₂ at 37°C.

Transfection of green fluorescent protein to the A431 cell. Green fluorescent protein (GFP)-expressing plasmid was purchased from Clontech Laboratories, Inc. (Mountain View, CA). The plasmid was transfected into the A431 cells to validate the results with targeted fluorophores (see below). The transfection of GFP was done with an electroporation method using Gene Plus II (Bio-Rad Laboratories, Hercules, CA). Briefly, 3 μg of GFP-expressing plasmid were mixed with 2 × 10⁶ A431 cells in 400 μL of the cell culture medium (DMEM with 10% FCS). The cell suspension was then placed in a pulse cuvette (Bio-Rad Laboratories), and 250 V pulses were delivered after 950 cycles.

Flow cytometry. One-color flow cytometry was done to assess the antibody pretargeting by b-Erb and the fluorescence enhancement of nAv-BDPfl on its binding to biotin on b-Erb. A431 cells (5 × 10⁵) were plated on a 12-chamber culture well and incubated for 16 h. Erbitux (10 μg/mL) or b-Erb (10 μg/mL) was added to the medium, and the cells were incubated for 2 h. For the investigation of target-specific activation, 10 μg/mL b-Erb or 10 μg/mL b-Zen was added to the medium and the cells were incubated for 2 h. For signal amplification study, 10 μg/mL 2b-Erb or 10 μg/mL 12b-Erb was added to the medium and the cells were incubated for 2 h. Then, cells were washed twice with PBS and 10 μg/mL nAv-BDPfl was added to each well and incubated for another 1 h. Cells were washed twice with PBS, and flow cytometry was done using a FACScan cytometer (Becton Dickinson, Franklin Lakes, NJ). The argon ion 488-nm laser was used for excitation, and signals from cells were collected using a 530/30-nm band pass filter. All data were analyzed using CellQuest software (Becton Dickinson). The fluorescing capability of cells in each condition was referred to as the mean fluorescence index (MFI).

Fluorescence microscopy. A431 cells (2 × 10⁴) were plated on a cover glass bottom culture well and incubated for 16 h. Erbitux (10 μg/mL) or b-Erb (10 μg/mL) was added to the medium, and the cells were incubated for 2 h. Cells were washed twice with PBS and 10 μg/mL nAv-BDPfl was added to each well and incubated for another 1 h. Cells were washed twice with PBS, and fluorescence microscopy was done using an Olympus BX51 microscope (Olympus America, Inc., Melville, NY) equipped with the following filters: excitation wavelength of 470 to 490 nm and emission wavelength of 515 nm long pass. Transmitted light differential interference contrast (DIC) images were also acquired.

Tumor model. All procedures were carried out in compliance with the Guide for the Care and Use of Laboratory Animal Resources (1996), National Research Council, and approved by the local Animal Care and Use Committee. The i.p. tumor implants were established by i.p. injection of 2 × 10⁶ A431 cells or GFP-positive A431 cancer cells (GFP-A431) suspended in 200 μL PBS in female athymic mice (NCI Animal Production Facility, Frederick, MD). Experiments with tumor-bearing mice were done at 10 days after injection of the A431 or GFP-A431 cells.

In vivo spectral fluorescence imaging. Erbitux and b-Erb (100 μg each) were diluted in 300 μL PBS and injected into the peritoneal cavities of mice with peritoneally disseminated A431 cancer implants. For the investigation of target-specific activation of nAv-BDPfl, 100 μg each of b-Erb and b-Zen in 300 μL PBS were injected into the peritoneal cavities of the tumor-bearing mice. At 21 h after injection of each antibody, i.p. injection of 100 μg nAv-BDPfl in 300 μL PBS was done (Fig. 2). For comparison study between Erbitux and b-Erb, mice were sacrificed with carbon dioxide immediately and 3 h after nAv-BDPfl injection, whereas for comparison study between b-Zen and b-Erb, the mice were sacrificed only 3 h after nAv-BDPfl injection. For the in vivo control study using Erb-BDPfl, the mice were sacrificed 1 and 3 h after i.p. injection of 100 μg Erb-BDPfl.

The abdominal cavity was surgically exposed, and spectral fluorescence images of the peritoneal cavities as well as close-ups of the peritoneal membranes were obtained using the Maestro In-Vivo Imaging System. A band pass filter from 445 to 490 nm and a long pass filter over 515 nm were used for emission and excitation light, respectively. The tunable filter was automatically stepped in 10-nm increments from 500 to 800 nm, while the camera captured images at each wavelength interval with constant exposure. All experiments were done in triplicate.

Semiquantitative comparison of fluorescence intensities of tumors. Aggregated A431 tumors were removed from each of the mice and placed on a nonfluorescent board side by side in each experiment, and spectral fluorescence imaging was done. Semiquantitative side-by-side comparison of fluorescence intensity was done between two tumors pretargeted with Erbitux and b-Erb or b-Zen and b-Erb. A ROI as large as each tumor was drawn to determine the fluorescence intensity as well as the histogram using ImageJ software. The dynamic range of the fluorescent intensity in a.u. was split into equal-sized 256 bins (1–256). All experiments were done in triplicate.

Assessment of the sensitivity and specificity for the detection of peritoneal cancer foci. Side-by-side spectral fluorescence imaging of A431 cancer-bearing mice and normal athymic mice without tumors was done 3 h after i.p. injection of 100 μg nAv-BDPfl. Pretargeting with 100 μg b-Erb was done in both mice 21 h before nAv-BDPfl administration. The abdominal cavity was surgically exposed, and spectral fluorescence images of the peritoneal cavities as well as close-ups of the peritoneal membranes were obtained. A band pass filter from 445 to 490 nm and a long pass filter over 515 nm were used for emission and excitation light, respectively. The tunable filter was automatically stepped in 10-nm increments from 500 to 800 nm, while the camera captured images at each wavelength interval with constant exposure. All experiments were done in triplicate. Sixty fluorescent bumps on the peritoneal membranes of three tumor-bearing mice were randomly selected, and histologic examination (H&E stain) was done to validate the accuracy of b-Erb pretargeting image.

The sensitivity and specificity of spectral imaging for the detection of peritoneal disseminated cancer foci were studied using four tumor-bearing mice. The i.p. tumor xenografts were established 14 days after i.p. injection of 2 × 10⁶ GFP-A431 cancer cells suspended in 200 μL PBS in female athymic mice (NCI Animal Production Facility). Twenty-one hours after i.p. injection of 100 μg b-Erb in 300 μL PBS, 100 μg nAv-BDPfl in 300 μL PBS was injected into the peritoneal cavity of GFP-A431 tumor-bearing mice. Spectral fluorescence images of the peritoneal membranes were obtained. A band pass filter from 445 to 490 nm and a long pass filter over 515 nm were used for emission and excitation light, respectively. The tunable filter was automatically stepped up in 5-nm increments from 500 to 800 nm, while the camera captured images at each wavelength interval with a constant exposure. Standard emission spectra were obtained from the GSA-A431 cell pellets and the nAv-BDPfl solution under the same conditions as this sensitivity and specificity study. The unmixed images of GFP, nAv-BDPfl, and autofluorescence were created using the standard emission spectra. For each mouse, two to three different parts of the peritoneal membranes were randomly selected and spread out on a nonfluorescent plate before close-up spectral fluorescence imaging was done. ROIs were drawn within the nodules depicted by unmixed GFP images, unmixed nAv-BDPfl images, or both. Additional ROIs were drawn in the surrounding nontumorous areas on the unmixed GFP images (standard reference for noncancerous foci). The average fluorescence intensity of each ROI was calculated on the GFP and the nAv-BDPfl spectral unmixed images using commercial software (Maestro software version 2). The minimum possible diameter for the ROI was 0.8 mm; thus, all visible nodules with short axis diameters of ≥0.8 mm on either image were included for analysis. nAv-BDPfl–positive nodules were defined as having an average fluorescence intensity of ≥10 a.u. on the unmixed nAv-BDPfl images, whereas nAv-BDPfl–negative nodules were defined as having an average fluorescence intensity of <10 a.u.. The number of foci positive for both nAv-BDPfl and GFP and the number positive only for nAv-BDPfl or GFP were counted. Sensitivity was defined as the number of peritoneal foci positive for both nAv-BDPfl and GFP divided by the number of peritoneal foci positive for GFP. Specificity of nAv-BDPfl was defined as the number of peritoneal foci negative for both GFP and nAv-BDPfl divided by the number of peritoneal foci negative for GFP.

Quality control study. With the same chemical reaction, a single Erbitux molecule could be conjugated with up to 23 biotins. 12b-Erb and 23b-Erb showed no loss of binding ability (72% and 68%) compared with unconjugated Erbitux (70%). In addition, the biotinylation did not increase the nonspecific binding of 12b-Erb (2.9%) and 23b-Erb (2.3%) compared with unconjugated Erbitux (3.1%). Therefore, the conjugation of 12 or 23 biotins did not compromise the binding of antibody to the specific antigen.

nAv-BDPfl activation after binding to a biotinylated antibody. To investigate the fluorescence enhancement of nAv-BDPfl on binding to a biotinylated antibody, a solution of 5 μg nAv-BDPfl was combined with 5 μg b-Erb in 390 μL PBS. As a control, 5 μg nAv-BDPfl was combined with 5 μg Erbitux in 390 μL PBS. All the solutions containing nAv-BDPfl have the same emission peak at a wavelength of 540 nm (Fig. 1B). The fluorescence of nAv-BDPfl increases strongly (∼10-fold) on binding to the biotinylated antibody b-Erb. The spectral peak of nAv-BDPfl bound to b-Erb was substantially higher than that of nAv-BDPfl in the presence of non-b-Erb. When 5 μg biotin was added to both solutions (nAv-BDPfl + Erbitux and nAv-BDPfl + b-Erb), the spectral peaks became almost identical. The signal intensity of nAv-BDPfl + Erbitux and nAv-BDPfl + b-Erb was 19.43 ± 1.5 a.u. and 173.1 ± 5.1 a.u. (mean ± SD), respectively. However, the coadministration of 5 μg biotin to both nAv-BDPfl + Erbitux and nAv-BDPfl + b-Erb resulted in signal intensities of 220.9 ± 16.3 a.u. and 223.8 ± 12.0 a.u. (mean ± SD), respectively.

Flow cytometry shows significant optical activation of nAv-BDPfl after pretargeting with biotinylated antibody. To investigate whether nAv-BDPfl is activated on cancer cells pretargeted with a biotinylated monoclonal antibody, single-color flow cytometry of A431 cancer cells was done after pretargeting with either b-Erb or Erbitux followed by nAv-BDPfl instillation. The percentage of fluorescence-gated A431 cells that corresponds to activated nAv-BDPfl is shown in Fig. 3A. The percentages of positive cells were 0.5% for A431 cells alone and 1.6% for A431 cells incubated with nAv-BDPfl without prior pretreatment with antibody. The MFI values were 4.8 a.u. for A431 cells alone and 5.1 a.u. for A431 cells incubated with nAv-BDPfl. The percentages of positive cells were 1.1% for cells pretargeted with Erbitux followed by nAv-BDPfl instillation but 99.9% for cells pretargeted with b-Erb followed by nAv-BDPfl instillation. The MFI values of cells pretargeted with Erbitux and b-Erb followed by nAv-BDPfl instillation were 6.8 and 680.0 a.u., respectively. These results indicate that, once a biotinylated antibody pretargets a cell, nAv-BDPfl will be activated on binding to the biotinylated antibody, but in the absence of a biotinylated antibody, nAv-BDPfl will not be activated.

To further investigate the number of biotin per antibody that is required to amplify the signal of nAv-BDPfl, flow cytometry of A431 cells was done after pretargeting either with 2b-Erb or 12b-Erb. The MFI value of A431 cells pretargeted with 2b-Erb followed by nAv-BDPfl instillation was 97.6 a.u., whereas the MFI of A431 cells pretargeted with 12b-Erb followed by nAv-BDPfl instillation was 1,110.6 a.u. (Fig. 3B). Based on this comparison study, as few as 2 biotin molecules per antibody can activate the nAv-BDPfl and result in a significant rightward shift (>1 order shift) on flow cytometry.

nAv-BDPfl is activated on the surface of pretargeted A431 cells. To investigate the location where nAv-BDPfl is activated, fluorescence microscopy and DIC images were obtained. Fluorescence microscopy of A431 cells pretargeted with b-Erb followed by nAv-BDPfl administration showed strong fluorescence on the surface of the cells (Fig. 3C). Intracellular fluorescence was not observed after 2 h of pretargeting with b-Erb and 1 h after incubation with nAv-BDPfl, whereas fluorescence microscopy of A431 cells incubated with b-Erb alone or nAv-BDPfl alone or A431 cells pretargeted with non-b-Erb followed by nAv-BDPfl instillation showed minimal fluorescence (Fig. 3C). These results indicate that nAv-BDPfl is activated on the surface of A431 cancer cells on binding to the pretargeting biotinylated antibody b-Erb; however, internalization of the b-Erb-nAv-BDPfl complex is minimal within this incubation period.

Peritoneal cancer foci are clearly visualized in vivo using the b-Erb pretargeting strategy. To show that nAv-BDPfl can be activated at target cancer foci in an i.p. cancer model using A431 cancer cells, the spectral fluorescence imaging was done of a surgically exposed mouse after pretargeting with Erbitux or b-Erb. Twenty-one hours after i.p. injection with 100 μg Erbitux or 100 μg b-Erb in tumor-bearing mice, 100 μg nAv-BDPfl was injected into the peritoneal cavity of each mouse. Immediately after injection with nAv-BDPfl, the aggregated tumors were clearly visualized in a mouse pretargeted with b-Erb, whereas the tumors were not depicted in a mouse pretargeted with non-b-Erb (Fig. 4A). Some background signal contamination was due to the unbound b-Erb. At 3 h after injection of nAv-BDPfl, the tumor remained bright but the background signal decreased markedly by the transperitoneal clearance of noncellular b-Erb and nAv-BDPfl complexes (Fig. 4A) resultantly in a high signal-to-background image. Close-up images of the peritoneal membranes showed submillimeter cancer foci with optical enhancement (Fig. 4A). As a control, Erbitux directly conjugated to BODIPY-FL (Erb-BDPfl) was injected into A431 tumor-bearing mice and spectral fluorescence imaging of the peritoneal cavity as well as the peritoneal membrane was done 1 or 3 h after i.p. injection. Unlike the pretargeting with b-Erb followed by nAv-BDPfl instillation, the Erb-BDPfl failed to visualize the peritoneal cancer foci due to the high background signals both 1 and 3 h after injection with Erb-BDPfl (data not shown).

To avoid the autofluorescence from the intestine, aggregated tumor nodules were extracted and placed side by side on a nonfluorescent plate and spectral fluorescence imaging was done. Ex vivo spectral fluorescence imaging showed that the fluorescence intensity of the tumors pretargeted with b-Erb was higher than that with Erbitux both immediately and 3 h after incubation with nAv-BDPfl (Fig. 4B). The histogram of the tumors pretargeted with b-Erb had a larger proportion of counts for pixels in the higher fluorescent intensity range than Erbitux both immediately and 3 h after injection with nAv-BDPfl (Fig. 4B). These results indicate that nAv-BDPfl is activated in vivo at the target tumor on binding to biotinylated antibody b-Erb, but nAv-BDPfl cannot be activated with nonbiotinylated antibody.

Target-specific activation of nAv-BDPfl was shown by in vitro and in vivo. To show that the optical activation of nAv-BDPfl was target specific, in vitro spectral fluorescence imaging was done using b-Erb and b-Zen as a control. Two cell-free mixed solutions consisting of 5 μg nAv-BDPfl and 5 μg b-Zen (b-Zen + nAv-BDPfl) in 390 μL PBS and 5 μg nAv-BDPfl and 5 μg b-Erb (b-Erb + nAv-BDPfl) in 390 μL PBS were placed on a nonfluorescent 96-well plate, and spectral fluorescence imaging was done. Unmixed images were generated (Fig. 5A) and a ROI was placed in each well. The signal intensities of b-Zen + nAv-BDPfl and b-Erb + nAv-BDPfl were 155.2 ± 9.4 a.u. and 180.8 ± 16.6 a.u. (mean ± SD), respectively. Based on the phantom study, the activation potential of b-Zen and b-Erb was comparable (<15% difference in fluorescence intensity of the b-Erb + nAv-BDPfl solution). Flow cytometry showed a significant rightward shift when A431 cells were pretargeted with 10 μg/mL b-Erb for 2 h followed by 1-h incubation with 10 μg/mL nAv-BDPfl (Fig. 5B); however, only a minimal shift was observed in A431 cells pretargeted with 10 μg/mL b-Zen for 2 h followed by 1-h incubation with 10 μg/mL nAv-BDPfl (Fig. 5B).

To confirm the target-specific activation of nAv-BDPfl in living animals, in vivo spectral fluorescence imaging was done. At 21 h after i.p. injection of 100 μg b-Zen or 100 μg b-Erb in A431 tumor-bearing mice, 100 μg nAv-BDPfl was injected into the peritoneal cavity of each mouse. At 3 h after injection with nAv-BDPfl, the mouse pretargeted with b-Zen and the mouse pretargeted with b-Erb were placed side by side on a nonfluorescent plate and the abdominal cavities were surgically exposed and in vivo spectral fluorescence imaging was done. The spectral fluorescence images clearly visualized the aggregated tumors as well as small peritoneal implants in a mouse pretargeted with b-Erb, whereas the tumors were not successfully visualized in a mouse pretargeted with b-Zen due to the insufficient signal-to-background ratio (Fig. 5C). Then, the aggregated tumor nodules were extracted and placed side by side on a nonfluorescent plate and spectral fluorescence imaging was done to compare the fluorescence intensities between the tumors. Ex vivo spectral fluorescence image showed that the fluorescence intensity of the tumors pretargeted with b-Erb was higher than that with b-Zen 3 h after injection with nAv-BDPfl (Fig. 5D). The histogram of the tumors pretargeted with b-Erb had a larger proportion of counts for pixels in the higher fluorescent intensity range than b-Zen (Fig. 5D). These results indicate that nAv-BDPfl will be activated specifically at the target lesions by binding to the biotinylated antibody previously bound to target cells both in vitro and in vivo.

Sensitivity and specificity of b-Erb pretargeting of peritoneal metastases are 96% and 98%, respectively. To validate that there is no nAv-BDPfl signal from normal tissues, side-by-side spectral fluorescence imaging of A431 cancer-bearing mice and normal mice without tumors was done 3 h after i.p. injection of 100 μg nAv-BDPfl. Pretargeting with 100 μg b-Erb was done 21 h before nAv-BDPfl injection. Spectrally unmixed nAv-BDPfl images showed that there were no nodular foci or bumps on the peritoneal membranes in normal mice, whereas there were several fluorescent foci on the peritoneal membranes of A431 tumor-bearing mice (Fig. 6A). The histology of 60 fluorescent bumps randomly selected from the tumor-bearing mice (n = 3) confirmed that all 60 bumps were A431 cancer foci. The sensitivity and specificity of spectrally unmixed nAv-BDPfl imaging for the detection of GFP-A431 peritoneal cancer foci were studied using dual-color in vivo spectral fluorescence imaging of GFP and nAv-BDPfl (Fig. 6B). A total of 188 peritoneal tumor foci in four mice was identified by the unmixed nAv-BDPfl images, the unmixed GFP images, or both. Additionally, 188 ROIs were created in the nontumorous areas (i.e., where no tumors were visible on the GFP images). One hundred and seventy-eight foci showed nAv-BDPfl fluorescence intensities of ≥10 a.u. among the 185 GFP-positive foci (Fig. 6C). One hundred and eighty-eight foci showed nAv-BDPfl fluorescence intensities of <10 a.u. among the 191 GFP-negative foci (i.e., fluorescence intensities of <10 a.u. on unmixed GFP images). Thus, the spectral unmixed nAv-BDPfl imaging had a sensitivity of 96% (178 of 185) and a specificity of 98% (188 of 191).

Optical imaging probes are unique in that they can be activated and deactivated. Activatable probes permit very high target-to-background imaging, which, in turn, improves the sensitivity of the probe. Several activation mechanisms exist, including quenching-dequenching, photo-induced electron transfer, and fluorescence resonance energy transfer. Here, we exploit a known property of BODIPY, which shows signal amplification after biotin binding. By biotinylating a monoclonal antibody, pretargeting can occur. When a nAv-BDPfl complex is subsequently given, dramatic increases in fluorescence signal are observed corresponding only to the pretargeted cells (Figs. 2 and 4). Our results show that submillimeter cancer implants can be visualized in vivo with this two-step activation method. This is not the first use of avidin in targeting fluorophores to peritoneal implants (12, 13). Previous studies relied on the binding of avidin to lectin binding proteins (synonyms of asialo receptors, β-d-galactose receptors), which are glycoproteins found on the cell surface of some ovarian cancer cells. Once bound, the avidin-fluorophore complex internalizes into the cancer cells, thus allowing optical imaging of tiny peritoneal implants (12, 13). However, there are two limitations with lectin-targeted optical imaging. First, the unbound optical agent must be cleared from the peritoneal cavity rapidly to avoid high background signals (13). Second, the lectin-targeted optical probes can only image β-d-galactose receptor-positive cancer cells (12–14). The current report focuses on establishing a generic two-step fluorescent activation mechanism whereby many different targeting antibodies can be used to activate a single neutravidin-BODIPY conjugate. In the example shown here, pretargeting with biotinylated antibody (b-Erb) and subsequent administration of the activatable optical probe nAv-BDPfl led to the activation of fluorescence only in cancer cells in the peritoneal cancer model. Because biotin-neutravidin binding-induced activation of nAv-BDPfl is independent of antigen-antibody binding, virtually any cancer-specific antigen can be targeted with its specific antibody. One caveat is that the known immunogenicity of avidin in humans is mitigated substantially by the use of neutravidin; however, alternatives may be needed for clinical use. Nevertheless, we have shown that it is possible to develop a family of imaging probes based on a two-step activation schema, in which the first step is the administration of a targeted biotinylated antibody and the second step is the administration of an activatable fluorophore, which remains the same regardless of the targeting antibody.

This method benefits from a target amplification effect because ∼10 biotin molecules are conjugated to each antibody. Thus, a single antibody, such as anti-HER1, can bind up to 10 nAv-BDPfl molecules, greatly amplifying the net fluorescence. Taken together with the 10-fold activation of fluorescence by biotin binding, this method has the potential to achieve ∼100-fold higher signal-to-background ratio compared with the use of a nonbiotinylated antibody-fluorophore conjugate. In addition, this method does not require the enzymatic activation or biological clearance of unbound reagents or internalization of nAv-BDPfl complex (12, 13). Thus, this method has the potential to be highly specific and highly sensitive for the detection of tiny cancer deposits.

In conclusion, a two-step pretargeting and activation technique can be used to visualize submillimeter peritoneal cancer implants. This method can be applied for targeting any cancer-specific receptors or antigens by use of their corresponding specific antibody while using a general activatable fluorophore, which remains the same. The combination of the target amplification due to a 10-fold ratio of biotin to pretargeting antibody, the 10-fold activation of nAv-BDPfl after binding to biotin, and the gain in contrast with the use of spectrally unmixed imaging allow the detection of tiny foci of peritoneal implants. We hope this method may be useful in guiding physicians to detect and treat disease that would otherwise escape detection.

Grant support: Intramural Research Program of the NIH, NCI, Center for Cancer Research.

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 in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.