Activatable fluorescent cys-diabody conjugated with indocyanine green derivative: consideration of fluorescent catabolite kinetics on molecular imaging

Abstract. Antibody fragments including diabodies have more desirable pharmacokinetic characteristics than whole antibodies. An activatable optical imaging probe based on a cys-diabody targeting prostate-specific membrane antigen conjugated with the near-infrared fluorophore, indocyanine green (ICG), was designed such that it can only be activated when bound to the tumor, leading to high signal-to-background ratios. We employed short polyethylene glycol (PEG) linkers between the ICG and the reactive functional group (Sulfo-OSu group), resulting in covalent conjugation of ICG to the cys-diabody, which led to lower dissociation of ICG from cys-diabody early after injection, reducing hepatic uptake. However, unexpectedly, high and long-term fluorescence was observed in the kidneys, liver, and blood pool more than 1 h after injection of the cys-diabody PEG-ICG conjugate. A biodistribution study using I125-labeled cys-diabody-ICG showed immediate uptake in the kidneys followed by a rapid decrease, while gastric activity increased due to released radioiodine during rapid cys-diabody-ICG catabolism in the kidneys. To avoid this catabolic pathway, it would be preferable to use antibody fragments large enough not to be filtered through glomerulus or to conjugate the fragments with fluorescent dyes that are readily excreted into urine when cleaved from the cys-diabody to achieve high tumor-specific detection.


Introduction
Monoclonal antibodies (mAbs) have been used for targeted therapy and imaging because of their high affinity for a diverse number of antigens overexpressed on cancer cells. 1,2 Molecular imaging based on mAbs labeled with radioisotopes can not only localize tumor within the whole body but also characterize it for drug targeting. 3,4 . However, the prolonged clearance time of whole mAbs reduces the target-to-background ratio, lowering sensitivity and specificity of labeled mAbs. 5 To overcome this problem, mAbs have been subjected to enzymatic cleavage or have been genetically modified to create fragments of various sizes, including variable region fragments (Fvs), diabody, fragments, antigen binding (Fabs), minibody, F(ab)'2 fragment, and CH 2 -deleted mAb. 6,7 If these fragments are too small, their renal clearance will be too rapid, compromising tumor uptake. Additionally, monovalent Fabs have lower binding affinities. Therefore, intermediate-size Fabs with mild prolongation of clearance but with bivalent binding to antigen, such as diabodies (50 to 60 kDa) or minibodies (75 to 80 kDa), would seem more likely to provide the highest tumor-to-background ratios. 7 Activatable imaging probes upon binding to the target are produced based on a relatively new concept that minimizes background signal, resulting in an increase in the tumor-to-background ratio. These activatable optical imaging probes using mAbs can be activated only when bound to a cell surface receptor followed by internalization. [8][9][10] . Whole immunoglobulin G (IgG)-based activatable probes with a long biological half-life have shown excellent target-specific tumor detection with minimal background signals. [11][12][13] In this study, we hypothesize that optimal pharmacokinetics of a cys-diabody combined with activatable labeling could improve the specific detection of cancer. Therefore, we have synthesized anti-prostate-specific membrane antigen (PSMA) cys-diabody activatably labeled with indocyanine green (ICG) derivatives 14 and tested them in PSMAexpressing tumor-bearing mice.

Synthesis of ICG Derivatives and IR700
Conjugated PSMA-Cys-Db ) at room temperature for 1 h, followed by purification with a size exclusion column (PD-10; GE Healthcare, Piscataway, New Jersey). The concentrations of ICG and IR700 were calculated by measuring the absorption with a UV-Vis system (8453 Value UV-Vis system; Agilent Technologies, Santa Clara, California) to confirm the number of fluorophore molecules conjugated to each antibody molecule. The protein concentration was also determined by measuring the absorption at 280 nm with a UV-Vis system. The number of ICG and IR700 per antibody was adjusted to 0.5 to 0.8 for each probe. We performed sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) as a quality control for each conjugate according to a previous report. 14 As a comparison, anti-human PSMA mAb (J591) conjugated with ICG was also synthesized as reported previously. 12

Determination of Quenching Capacity In Vitro
The quenching capacity of each conjugate was investigated by denaturation with 1% SDS as described previously. 14 Briefly, the conjugates were incubated with 1% SDS in phosphate-buffered saline (PBS) for 15 min at room temperature. As a control, the samples were incubated in PBS. The change in fluorescence signal intensity of IR700 and ICG was investigated with an in vivo imaging system (Maestro, CRi Inc., Woburn, Massachusetts). The red filter set for IR700 uses a bandpass filter, which ranges between 615 and 665 nm (excitation), and a long-pass filter over 700 nm (emission); the near-infrared filter set for ICG uses a bandpass filter from 710 to 760 nm (excitation) and a long-pass filter over 800 nm (emission). Regions of interest (ROIs) were placed on ICG spectrum images with reference to white light images to measure fluorescence intensities of solutions. The Maestro software was used for calculating ROI signal data.

Stability in Mouse Serum
Each probe [5 to 6 μg in PBS (40 μL)] was incubated in mouse serum (40 μL) for 0, 0.5, and 1 h at 37°C, followed by imaging with a Maestro in vivo imaging system. 1% SDS was added to each probe to dequench. Fluorescence recovery in mouse serum was calculated by the following equation: (Fluorescence signal in mouse serum-Fluorescence signal in PBS)/(Fluorescence signal in SDS-Fluorescence signal in PBS) ×100.

Cell Culture
A PSMA transfected PC3 cell line PC3-PSMA+(PC3pip) and a control blank-vector transfected PC3 cell line PC3-PSMA −(PC3flu) were used for PSMA targeting studies with ICG derivatives conjugated PSMA-Cys-Db. Both cell lines were established at the Cleveland Clinic Foundation. All cell lines were grown in RPMI 1640 medium (Life Technologies, Gaithersburg, Maryland) containing 10% fetal bovine serum (Life Technologies) and 1% Pen-Strep (Biofluids, Camarillo, California). All cell cultures were maintained in 5% carbon dioxide at 37°C in a humidified incubator.

Fluorescence Microscopy Studies
PC3-PSMA+ or PC3-PSMA− cells (1 × 10 4 ) were plated on a covered glass-bottomed culture well and incubated for 16 h. PSMA-Cys-Db-ICG, PSMA-Cys-Db-PEG4-ICG, and PSMA-Cys-Db-PEG8-ICG (10 μg∕mL) were then added to cells and incubated for either 1 or 8 h, followed by washing once with PBS, and fluorescence microscopy was performed using an Olympus BX81 microscope (Olympus America Inc., Melville, New York) equipped with the following filters: excitation wavelength 672.5 to 747.5 nm, emission wavelength 765 to 855 nm for ICG. Transmitted light differential interference contrast images were also acquired.

Tumor Model
All in vivo procedures were conducted in compliance with the Guide for the Care and Use of Laboratory Animal Resources (1996), U.S. National Research Council, and approved by the local Animal Care and Use Committee. Six-to-eight-week-old female homozygous athymic nude mice were purchased from Charles River (NCI-Frederick, Frederick, Maryland). During the procedure, mice were anesthetized with isoflurane. PC3flu cells (2 × 10 6 ) were injected subcutaneously into the left dorsum of each mouse, and three days later (because they grow faster), PC3-pip cells (2 × 10 6 ) were injected into the right dorsum of the same mouse. Mice were imaged when tumors grew to about 7 to 8 mm, as reported previously. 12

In Vivo Imaging Studies
Each probe (PSMA-Cys-Db-ICG, PSMA-Cys-Db-PEG4-ICG, PSMA-Cys-Db-PEG8-ICG, PSMA-Cys-Db-IR700, and J591-ICG) (25 μg∕100 μL in PBS/mouse) was administered via tail vein into tumor-bearing mice (PC3-PSMA+ and PC3-PSMA−). The mice were anesthetized with isoflurane, and fluorescence imaging data were obtained at 24 h with a Pearl Imager (LI-COR) using the 700 and 800 nm fluorescence channel for IR700 and ICG, respectively. ROIs were placed on ICG spectrum images with reference to white light images to measure fluorescence intensities of PC3-PSMA+, PC3-PSMA− tumor, and liver at every time point up to 24 h. The software Pearl Cam (LI-COR Biosciences) was used for calculating ROI signal data of each point. After acquisition of images at 24 h, mice were sacrificed with carbon dioxide. Ex vivo images of resected tumors, liver, and kidneys were obtained.

Biodistribution Study
PC3-pip (PSMA+) and PC3-flu (PSMA−) bearing mice were divided into two groups (n ¼ 3 to 4) with approximately equal distributions of tumor sizes on the day of study. 125 I-PSMA-Cys-Db-ICG was prepared using the Iodo-Gen procedure and purified with a PD-10 size exclusion column. 125 I-PSMA-Cys-Db was also prepared as control to confirm the alteration of biodistribution of PSMA-Cys-Db after conjugation with ICG. The specific activities of the radiolabeled Dbs were 9.4 mCi∕mg for PSMA-Cys-Db and 6.9 mCi∕mg for PSMA-Cys-Db-ICG. 125 I-PSMA-Cys-Db or 125 I-PSMA-Cys-Db-ICG (37 kBq∕25 μg∕100 μL in PBS/mouse) was injected via tail vein, and the biodistribution was determined at 1, 6, and 24 h postinjection. Organs of interest were excised, weighed, and the radioactivity counts were determined with an NaI well-type scintillation counter (ARC-370M, Aloka, Tokyo, Japan), using the injected dose as a standard. Data were calculated as the percentage injected dose per gram of tissue (%ID/g).

Statistical Analysis
Quantitative data were expressed as mean AE s.e:m: Means were compared using two-way repeated measures analysis of variance with the Bonferroni correction of multiple comparisons. P value of <0.05 was considered statistically significant.

Fluorescence Microscopy Studies
Microscopy studies (Fig. 2)    , and J591-ICG using the 700 and 800 nm fluorescence channel of the fluorescence camera for IR700 and ICG, respectively. Covalent conjugation of ICG with short PEG linkers successfully reduced the nonspecific uptake in the liver at early time intervals after injection unlike PSMA-Cys-Db labeled with the original ICG-Sulfo-OSu, which had high liver uptake. However, unexpectedly, there was prolonged high background activity especially in the kidneys, liver, and circulation at 1, 3, and 6 h postinjection with PEG linkers. This resulted in low fluorescence ratios (1.0 to 1.7) of PSMA+ to PSMA− cells for PSMA-Cys-Db-ICG, PSMA-Cys-Db-PEG4-ICG, and PSMA-Cys-Db-PEG8-ICG, although there was a slight improvement in target-to-background ratios after introduction of short PEG linkers to ICG. On the other hand, PSMA-Cys-Db-IR700 (always-on probe) mainly accumulated in the kidneys, but there was low uptake in the liver, which was similar to the biodistribution of intact Cys-Db reported previously. 15 The fluorescence ratios of PSMA+ to PSMA− tumors were increased on PSMA-Cys-Db-IR700 and improved over time. As a comparison, J591-ICG selectively accumulated in PC3-PSMA+ tumors within 24 h with low background signal, except in the liver and bowel. The fluorescence intensity in PC3-PSMA+ tumors was maintained on day 3, as reported previously 12 (data not shown).
125 I-PSMA-Cys-Db-ICG showed a biodistribution similar to 125 I-PSMA-Cys-Db. 125 I-PSMA-Cys-Db-ICG was rapidly taken up by the kidneys within 1 h postinjection and showed a fast blood clearance and significantly higher accumulation in PSMA positive tumors than in PSMA negative tumors at 6 and 24 h after probe administration. Notably, the radioactivity observed in the kidneys was sharply reduced within 6 h, while that in the stomach was dramatically increased. These results suggest that the cys-diabody is dehalogenated and possibly catabolized in the kidneys, and the fragment of cys-diabody including the iodine ion can be slowly released into the circulation followed by gradual uptake in the stomach and thyroid.

Discussion
The optimization of pharmacokinetics and stability of activatable optical probes is important for successful implementation.
We have shown that ICG with short PEG linkers can be covalently conjugated to IgG with superior efficacy compared with ICG-Sulfo-OSu without a PEG linker. 14 When ICG-Sulfo-OSu was conjugated directly without a PEG linker, there was less covalent binding, resulting in higher liver uptake. By inserting PEG linkers, the activation capacity was slightly reduced due to decreasing quenching efficacy probably because of increased distances between ICG and aromatic residues on mAb molecules. 11,14 However, an activation capacity of at least 10-fold was retained for both PSMA-Cys-Db-PEG4-ICG and PSMA-Cys-Db-PEG8-ICG. In in vivo fluorescence imaging studies, PSMA-Cys-Db covalently conjugated with ICG with short PEG linkers showed little nonspecific uptake in the liver immediately (5 min) after injection compared with PSMA-Cys-Db conjugated with ICG-Sulfo-OSu. However, an unanticipated increase in background signal was observed especially in the kidneys, liver, and circulation from 1 h to 1 day after administration. The biodistribution study using 125 I-labeled PSMA-Cys-Db and PSMA-Cys-Db-ICG were similar with relatively rapid clearance through the kidneys that is consistent with previous reports. The biodistribution study suggested rapid renal catabolism and deiodination of probes followed by intense accumulation of free radioiodine in the stomach by 6 h after injection. Cys-diabody fragments particularly were subjected to filtration through glomerulus, reabsorption by the proximal convoluted tubule (PCT), and dehalogenated upon internalization in the PCT cells according to the previous reports. 16,17 ICG could be expected to undergo similar catabolism, albeit with a slower release than free iodine. ICG would be released after catabolism of the conjugate within the lysosome. Released ICG would bind to plasma proteins, resulting in increased fluorescence as shown in Fig. 1(c), would then circulate in the blood bound to protein, and would then gradually accumulate into the liver and finally be excreted via the hepatobiliary system as nonconjugate ICG does. Therefore, these fluorescently activated catabolites of ICG, which are formed within a day after injection, persistently increase the background signal in fluorescent images.
PSMA-Cys-Db-IR700 (always-on probe) immediately accumulated in the kidneys by glomerular filtration after injection and then was reabsorbed by PCT similar to radiolabeled diabodies. The difference in the pathway after metabolism in the kidneys between PSMA-Cys-Db-ICG and PSMA-Cys-Db-IR700 might be attributed to the difference in the chemical characteristics of fluorophores. After degradation and release into the circulation, ICG can bind to plasma protein and be excreted into the bile through the liver, whereas IR700 is rapidly excreted into the urine. 18 Therefore, although PSMA-Cys-Db-IR700 is an always-on probe, background activity of IR700 decreases more rapidly than that of ICG. ICG has been approved for human use for 50 years with a good safety profile and is a practical fluorophore for clinical application. However, when conjugated with small proteins such as diabodies, which are catabolized in the kidneys, fluorescently activated ICG and derivatives can be recirculated and excreted through the liver, leading to a prolonged increased background in abdomen on fluorescence images. Therefore, J591(whole IgG)-ICG is apparently a better activatable probe for the PSMA-specific cancer detection than diabody-ICG probes.
In conclusion, optimal design of activatable optical probes should consider the excretion pathway of fluorescent catabolites. A large proportion of such agents can be catabolized in the kidney and, therefore, not reach the target in large amounts. From this point of view, activatable labeling of imaging probes  with ICG is better for somewhat larger proteins such as IgG, F (ab)'2, and minibody, which are not filtered through the glomerulus, however, it is relatively inferior for small proteins including cys-diabody, Fab, and Fv fragments, which are filtered rapidly through the glomerulus. When small proteins are used as targeting moieties, we might have to consider more hydrophilic and, therefore, more readily urinary excretable fluorophores such as IR700 for activatable labeling.