Application of a new 68Ga-labeled cNGR dimer peptide to microPET imaging of ovarian cancer

Introduction: Peptides containing the asparagine-glycine-arginine (NGR) sequence have been found to specically bind to cluster of differentiation 13 (CD13) (aminopeptidase N), a tumor neovascular biomarker that is overexpressed on the surface of angiogenic blood vessels and various tumor cells and plays an important role in angiogenesis and tumor progression. The aim of this study was to evaluate the ecacy of a gallium-68 ( 68 Ga)-labeled dimeric cyclic NGR (cNGR) peptide as a new molecular probe that binds to CD13 in vitro and in vivo . Materials and Methods: A dimeric cNGR peptide conjugated with 1,4,7,10-tetraazacyclododecane-N,N',N'',N'''-tetraacetic acid (DOTA) and DOTA-c(NGR) 2 was synthesized and labeled with 68 Ga. In vitro uptake and binding analysis was performed in two ovarian tumor cell lines, ES2 and SKOV3, each of which have different expression levels of CD13. An in vivo biodistribution study was performed in normal mice, and micro positron emission tomography (PET) imaging was performed in nude mice xenografts with ES2 and SKOV3 tumors. Results : 68 Ga-DOTA-c(NGR) 2 with high radiochemical purity (>95%) was obtained and found to be stable at room temperature and when incubated with bovine serum at 37°C for 3 h. In vitro studies showed that uptake of 68 Ga-DOTA-c(NGR) 2 in ES2 cells increased over time, was higher than that in SKOV3 cells at all time points, and could be blocked by cold DOTA-c(NGR) 2 . Biodistribution studies demonstrated that 68 Ga-DOTA-c(NGR) 2 was mainly excreted from the kidney and rapidly cleared from blood. MicroPET imaging of ES2 tumor xenografts showed that focal uptake in tumors was distinctly observed from 1 to 1.5 h post-injection of 68 Ga-DOTA-c(NGR) 2 . Clear and high-contrast tumor visualization occurred at 1 h, which corresponded to the highest tumor/background ratio of 10.30±0.26. Moreover, accumulation of the probe in ES2 tumors apparently declined with pretreatment of unlabeled peptide, which further proved the specicity of 68 Ga-DOTA-c(NGR) 2 . In SKOV3 tumor models, the tumor was not obviously displayed under the same imaging protocols. Conclusion : We conclude that 68 Ga-DOTA-c(NGR) 2 might be a potential molecular probe for evaluating the expression levels of CD13 in different tumors, thereby providing a basis for targeting angiogenesis in cancer therapy.


Introduction
Angiogenesis, the formation of new vessels from the preexisting vascular system, is the basic process of tumor progression. Studies have shown that tumor angiogenesis is a complex multistep process that is mediated and controlled by multiple adhesion molecules and cell receptors. Thus, targeting tumor angiogenesis with anti-angiogenic agents represents a promising therapeutic strategy for cancer control and treatment. Aminopeptidase N, also known as cluster of differentiation 13 (CD13), a zinc-dependent membrane-bound exopeptidase, is usually upregulated on the endothelium of tumor neovasculature and in various solid cancers including melanoma, prostate, ovarian, lung, and breast [1][2][3][4]. CD13 is a key regulator implicated in angiogenesis and tumor progression, as well as a key binding receptor for peptides containing the asparagine-glycine-arginine (NGR sequence) [5][6][7]. In recent years, NGR peptidebased drug delivery and imaging studies have becoming an intriguing approach in cancer therapy and diagnosis. Molecular imaging of CD13 expression level is of great signi cance for guiding and monitoring the anti-angiogenesis therapy of NGR peptide-based drug delivery.
Ovarian cancer (OVCA), a serious threat to women's health, is characterized by high mortality and a high recurrence rate. It is the third most common gynecologic malignant tumor after endometrial carcinoma and cervical cancer in China, but its mortality rate is the highest of all gynecological malignancies [8].
Because CD13 is also expressed in OVCA and is involved in tumor progression, many studies have focused on treating OVCA by targeting CD13 [4,9,10]. However, due to the heterogeneity of the disease, CD13 is differentially expressed in OVCA tissues and cell lines such as ES2 and SKOV3 cells. ES2 cells are intensely positive for CD13, while SKOV3 cells express CD13 at a low levels [11].
In our previous study, gallium-68 ( 68 Ga)-labeled NGR linear monomer ( 68 Ga-1,4,7,10tetraazacyclododecane-N,N',N'',N'''-tetraacetic acid [DOTA]-NGR) was applied in A549 lung cancer xenografts [12], and was found to be concentrated in the tumor. However, the retention time of the probe in the tumor was short, so translation to clinic would be di cult. We expect to improve the biodistribution and pharmacokinetics of the probe through further modi cation. Some reports have shown that cyclic peptides are more stable than linear ones because cyclic peptides are not easily broken down by enzymes [13]. In addition, the dimer has higher a nity to targets than the monomer because the dimer can bind more sites in target cells [14][15].
In this study, we designed and synthesized a novel radioactive molecular probe, a dimer of cyclic NGR (cNGR) peptide radiolabeled with 68 Ga, and applied it to micro positron emission tomography (PET) imaging of OVCA xenografts to evaluate its ability to identify tumors with CD13 expression at different levels.

General
All chemicals (reagent grade), unless speci cally stated, were from commercial suppliers and no further puri cation was required. DOTA-c(NGR) 2 ( Fig. 1) was commissioned by Sangon-Peptide Biotech Co., Ltd.
(Ningbo, China) to synthesize using standard F-moc solid-phase chemistry. The yield was analyzed by high-performance liquid chromatography (HPLC) and mass spectrometry (MS) with a purity of more than 98%. 68 GaCl 3 was produced from 68 Ge-68 Ga radionuclide generator (ITG GmbH, Oberding, Germany) by elution with 5 mL of 0.05 M HCl.

Labeling and characterization of 68 Ga-DOTA-c(NGR) 2
A concentration of 10 mg DOTA-c(NGR) 2 was dissolved in 2 mL deionized water to make a 5 mg/mL solution. Then 80 µL of 1 mol/L HEPES (pH 5.0, Sigma-Aldrich Corporation, St. Louis, MO, USA) was added to 200 µL 68 GaCl 3 (37-74 MBq) eluent, and then incubated with 40 µL DOTA-c(NGR) 2 solution at 95 °C water bath for 10 min. Quality control was performed by radio-HPLC (Agilent Technologies, Santa Clara, CA, USA) with the VP-ODS C18 column (Shimadzu, Kyoto, Japan) and radioactivity detector (Zonkia Scienti c Instruments Co., Ltd., Anhui Province, China). Mobile phase A was water with 0.1% tri uoroacetic acid (TFA) and mobile phase B was acetonitrile with 0.1% TFA. The ow rate was 1 mL/min and peaks were detected at 220 nm, with the mobile phase starting from 80% solvent A and 20% solvent B to 30% solvent B at 20 min. To evaluate in vitro stability, ~ 3.7 MBq of 68 Ga-DOTA-c(NGR) 2 were incubated in phosphate-buffered saline (PBS) at room temperature and in bovine serum at 37 °C, and radiochemical purity was measured at 30-min and 1-, 2-, and 3-h time points.

Cell culture and animal model
The human ovarian cancer ES2 and SKOV3 cells were obtained from Nanjing Keygen Biotech Co., Ltd. (Nanjing, China). ES2 cells were grown as monolayer cultures in McCoy's medium 5A containing 10% fetal bovine serum (FBS), and SKOV3 cells were cultured in RPMI 1640 medium containing 10% FBS. Both cell lines were incubated at 37 °C in a humidi ed atmosphere containing 5% CO 2 . Cells in log phase were collected and prepared at a concentration of 5 × 10 7 /mL in PBS. A volume of 0.1 mL single-cell suspension was injected subcutaneously into the front ank of each female BALB/c-neu nude mouse (~ 4-6 weeks old, body weight of 18-25 g; Slac Laboratory Animal, Shanghai, China). Animal models were used for experiments when xenografts grew to 500-1000 mm 3 . All animal studies were performed according to a protocol approved by the Institutional Animal Care and Use Committee of Soochow University.

Flow cytometry analysis
ES2 and SKOV3 cells were harvested by trypsinization and washed twice with PBS. The cells were adjusted to a concentration of 2 × 10 5 /mL with PBS. The cells were labeled with mouse-anti-human CD13 monoclonal antibody and incubated at 4 °C for 30 min. The super uous antibodies were washed away by adding 1 mL cold PBS to each tube. The second antibody labeled with phycoerythrin was added and incubated at 4 °C for 30 min. The cells were washed twice and resuspended in 500 µL PBS. Then labeled cells were analyzed on a ow cytometer (FC-500; Beckman Coulter Inc., Sykesville, MD, USA) to quantify the expression level of CD13 on OVCA cells.

Immunohistochemical staining
Tumors were excised and xed in 4% buffered formalin (pH 7.0). Sections of para n-embedded tumor tissues were baked in an oven at 65 °C for 2 h, dewaxed by being soaked with dimethylbenzene twice, and dehydrated using deionized water and alcohol washes of increasing concentrations (80%, 95%, and 100%). Antigens on the sections were retrieved with antigen repair solution (0.01 M citric acid buffer, pH 6.0) and endogenous peroxidase activity was eliminated with 3% freshly prepared H 2 O 2 . After blocking in 3% BSA (SW3015; Solarbio, Beijing, China) at 37 °C for 30 min, sections were incubated with adequate diluted rabbit anti-CD13 (1:150; Nanjing Keygen Biotech) in a wet box at 4 °C for 12 h, followed by the addition of polymeric horseradish peroxidase-labeled rabbit IgG as secondary antibody. After incubation for 30 min at 37 °C and three washes with PBS, sections were stained with a substrate-chromogen solution containing 0.025% 3, 3'-diaminobenzidine for 10 min and counterstained with hematoxylin for 3 min before microscope observation at low (200×) and high (400×) magni cations.

In vitro cell binding assay
To study the in vitro binding a nity and speci city of 68 Ga-DOTA-c(NGR) 2 to CD13, ES2 and SKOV3 cells were seeded into 6-well plates at 1 × 10 6 cells/well and cultured overnight. Then 37 KBq 68 Ga-DOTAc(NGR) 2 was added and incubated at 37 °C for 5, 15, 30 min, 1, and 2 h after which the supernatant was suctioned, and cells were washed three times with PBS and harvested with 0.25% trypsin. Both the supernatant and cell suspensions were collected and counted using a gamma counter (CRC-55tR; Capintec, Florham Park, NJ, USA). Cell-based competitive binding assay was performed with ES2 cells.
The cells were incubated with 37 KBq 68 Ga-DOTA-c(NGR) 2 in the presence of increasing concentrations of unlabeled DOTA-c(NGR) 2 (0.2-3.2 µg/mL). After 2 h of incubation at 37 °C, cells in all phases were collected and measured by the same previous method. The data were tted with nonlinear regression using GraphPad Prism 7.0 (GraphPad Software, San Diego, CA, USA) to obtain the 50% inhibitory concentration (IC 50 ). Studies were performed in triplicate.

Biodistribution studies
Biodistribution studies were performed in 30 normal ICR (Institute of Cancer Research) mice (18-30 g; age, 4 weeks). At 10, 30, 60 min, and 2 h after injection of 3.7 MBq/0.1 mL 68 Ga-DOTA-c(NGR) 2 via the tail vein, mice (n = 6 per time point) were sacri ced by cervical dislocation, after which the blood, muscle and other organs of interest (brain, heart, liver, lungs, spleen, pancreas, kidneys, stomach, small intestine, and femur) were immediately harvested, weighed, and counted with a gamma counter. Data were normalized to the time of injection and expressed as percent injected dose per gram (% ID/g).

Statistical analysis
All quantitative data are presented as the mean ± standard deviation (SD). Statistical analysis was conducted by one-way analysis of variance and the Student's t-test. P < 0.05 was considered statistically signi cant.

Radiochemistry and stability
Under the condition of bathing at 95 °C for 10 min, 68 Ga-DOTA-c(NGR) 2 was easily prepared with a high labeling rate of 98.01% ± 1.44%, and further puri cation was not needed. The retention time of labeling yields on radio-HPLC was 4.86 ± 0.27 min (Fig. 2). For the stability study, the radiochemical purity was > 96% in saline and > 95% in bovine serum after 3 h of incubation (Fig. 3).

Fluorescence-activated cell sorting and immunohistochemical staining
As shown in Fig. 2, uorescence-activated cell sorting (FACS) revealed that expression rates of CD13 on ES2 cells and SKOV3 cells were 87.2% and 27.6% (Fig. 4), respectively. In accordance with the results of FACS, immunohistochemical staining showed that CD13 was signi cantly expressed on the membranes of the tumor and endothelial cells of the new vasculature in ES2 tumor tissue, at signi cantly higher levels than in SKOV3 tumor tissue (Fig. 5).

In vitro cell binding assay
Cell-binding studies were performed with ES2 cells and SKOV3 cells, and blocking studies were performed with ES2 cells. Signi cant binding differences were observed between ES2 cells and SKOV3 cells (Fig. 4).
The results showed that the binding of 68 Ga-DOTA-c(NGR) 2 to ES2 cells increased over time with highest binding of 3.45% ± 0.51% achieved at 2 h of incubation, which was signi cantly higher than binding to SKOV3 cells at all time points with a highest binding of 1.79% ± 0.34% (P < 0.05) (Fig. 6). In blocking assays with excess cold peptide. 68 Ga-DOTA-c(NGR) 2 binding to ES2 cells signi cantly decreased at all time points with highest binding of 2.52% ± 0.15% (P = 0.038). Cell-based competitive binding assay in ES2 cells demonstrated that 68 Ga-DOTA-c(NGR) 2 binding was blocked by DOTA-c(NGR) 2 in a dosedependent relationship, and the IC 50 value was 160.1 nM (Fig. 7).

Biodistribution studies
The dynamic biodistribution results of 68 Ga-DOTA-c(NGR) 2 in healthy mice are shown in Fig. 5.
Radiouptake in the kidney peaked at 5 min (16.7 ± 5.80% ID/g) and then rapidly decreased by more than 50% at 30 min and 85% at 120 min post-injection, followed by the liver and lung. Quick clearance of 68 Ga-DOTA-c(NGR) 2 from blood was observed, and the blood uptake was 7.44 ± 1.64% ID/g at 5 min and 0.31 ± 0.05% ID/g at 60 min post-injection. Activity in brain, heart, stomach, intestines, spleen, pancreas, and bone was markedly lower (Fig. 8).

MicroPET imaging and blocking experiments
After injection with 68 Ga-DOTA-c(NGR) 2 , a series of microPET images from ES2 tumor-bearing mice were collected at 30 min, 1, and 1.5 h and were represented by coronal and transverse views (Fig. 6). High focal accumulation in tumor was visualized at 1 and 1.5 h and was then reduced at 2 h. Marked accumulation of radioactivity in the bladder was observed at all time points. With the exception of 30 min, distribution of radioactivity in liver was barely observed in subsequent images. Quantitative analysis by ROI technology revealed that the uptake of 68 Ga-DOTA-c(NGR) 2 in ES2 tumors was 0.62 ± 0.09% ID/g at 1 h and 0.53 ± 0.08% ID/g at 1.5 h, whereas in SKOV3 tumors it was 0.32 ± 0.03% ID/g and 0.24 ± 0.05% ID/g, respectively. The T/B ratio in ES2 tumors was 10.30 ± 0.26 at 1 h and 8.04 ± 1.75 at 1.5 h, but in SKOV3 tumors was 3.99 ± 0.18 and 4.24 ± 0.73, respectively. In ES2 tumors blocked by unlabeled peptide, it was only 0.16 ± 0.03% ID/g at 1 h and 0.14 ± 0.02% ID/g at 1.5 h. (Fig. 9)

Discussion
Although to date, no antiangiogenic agents can eradicate tumors alone, effective inhibition of tumor angiogenesis might arrest tumor progression and has been shown to synergize with other cancer treatments [16 -17]. As one targeted tumor therapy, only a few patients can actually bene t from antiangiogenic therapy due to the biological heterogeneity of tumors [18][19]. Molecular imaging using radioisotopes labeled with targeted drugs or its analogs makes it possible to select patients, predict e cacy, and dynamically assess response to therapy [20]. Detection of speci c molecular targets is an important application of molecular imaging in the management of malignancies.
In nuclear medicine, PET has become a very important imaging technique for the diagnosis and treatment e cacy evaluation of cancer. Currently, 18 F-FDG is still the most widely used radiotracer for PET imaging. The acquisition of 18 F-FDG depends on the availability of a nearby cyclotron. Due to the high price and operating cost of the cyclotron, only a small number of medical institutions has both a PET and cyclotron. Moreover, 18 F-FDG traces tumor cell glucose metabolism and is used for the diagnosis and evaluation of therapeutic effects. It is a non-speci c tumor imaging agent, which limits its application to guide targeted therapy, especially selecting the right patients. 68 Ga, a generator-based positron radioisotope, is relatively easy to obtain by 68 Ge/ 68 Ga radionuclide generators and shows equally high-quality images on PET. 68 Ga 3+ can form stable complexes with many ligands containing oxygen and nitrogen as donor atoms by aid of chelators, which makes it an appealing alternative to 18 F [21].
Radioisotope-labeled RGD peptides have been widely studied as molecular probes for tumor angiogenesis imaging [22][23][24][25]. However, RGD targets integrin αvβ3, which works well for tumors with positive expression of integrin αvβ3, but cannot be used for tumors that do not express or have low expression of αvβ3. CD13 is another important target of tumor and angiogenesis, and NGR can be speci cally combined with CD13 to form a new probe, which can be an important complement to the RGD probe. To improve the targeting and biological metabolic characteristics of the NGR probe, we refer to the successful experience of RGD modi cation. The cyclic dimer peptide has a more stable structure and higher a nity than the monomer, and PEG modi cation can increase the hydrophobicity of peptides and thus improve the pharmacokinetics of the probe [13,26]. In this study, we designed a cyclic NGR dimer linked by PEG-4 and labeled it with 68 Ga to form a new radiolabeled probe targeting CD13. 68 Ga-DOTA-c(NGR) 2 with high radiochemical purity was obtained after relatively simple labeling steps without further puri cation, and the yield maintained stability in bovine serum beyond 3 h. To verify the speci city and cellular binding kinetics of the probe, two ovarian cancer cell lines, ES2 cells and SKOV3 cells, expressing CD13 at signi cantly different levels were utilized to mimic heterogeneity from the same tumor. An in vitro binding assay demonstrated that the binding of 68 Ga-DOTA-c(NGR) 2 to ES2 cells was signi cantly higher than that in SKOV3 cells and could be inhibited with unlabeled DOTA-c(NGR) 2  Biodistribution studies demonstrated that a large proportion of 68 Ga-DOTA-c(NGR) 2 was metabolized through the urinary system and rapidly cleared from blood, whereas its early uptake in other organs such as liver, lungs, and spleen was probably correlative to the blood supply of these organs. The distribution characteristics were also con rmed by subsequent imaging studies of tumor xenografts. Series of microPET images of ES2 tumor model showed that focal uptake in the tumor was distinctly observed In our previous study, 68 Ga labeling linear NGR conjugated with DOTA, 68 Ga-DOTA-NGR, was synthesized, which showed high selectivity and a nity for CD13 and in signi cant uptake in CD13-positive lung tumors [12]. Compared with this study, a cNGR dimer was used and its higher a nity to CD13 and higher uptake in tumors were observed, consistent with our prediction. However, we noted that the uptake decreased in the tumor at 2 h post-injection. Further optimization of the structure of the probe may improve its pharmacokinetic characteristics and increase a nity to CD13-positive tumors.
It is worth noting that our work is only a preliminary study on the physicochemical and biological characteristics of a newly designed cyclic dimer probe based on the NGR sequence. It was not compared with other probes of NGR derivatives in previous studies under the same experimental conditions. Conclusion 68 Ga-DOTA-c(NGR) 2 are easily synthesized with high radiochemical purity and stability. Due to its high selectivity and a nity for CD13, 68 Ga-DOTA-c(NGR) 2 might be a potential molecular probe for evaluating CD13 expression levels in different tumors, thereby providing a basis for targeted therapy.       Uptake of 68Ga-DOTA-c(NGR)2 in ES2 and SKOV3 cells.

Figure 7
Competitive binding assay. The uptake in ES2 cells could be inhibited competitively by unlabeled DOTAc(NGR)2 in dose-dependent manner, and the IC50 value was 160.1 nM.