**Radionuclide Molecular Imaging of EpCAM Expression in Triple-Negative Breast Cancer Using the Sca**ff**old Protein DARPin Ec1**

**Anzhelika Vorobyeva 1,2,\* , Ekaterina Bezverkhniaia 2,3 , Elena Konovalova <sup>4</sup> , Alexey Schulga 2,4 , Javad Garousi <sup>1</sup> , Olga Vorontsova <sup>1</sup> , Ayman Abouzayed <sup>5</sup> , Anna Orlova 2,5,6 , Sergey Deyev 2,4,7,8 and Vladimir Tolmachev 1,2**


Academic Editor: Krishan Kumar

Received: 29 July 2020; Accepted: 13 October 2020; Published: 14 October 2020

**Abstract:** Efficient treatment of disseminated triple-negative breast cancer (TNBC) remains an unmet clinical need. The epithelial cell adhesion molecule (EpCAM) is often overexpressed on the surface of TNBC cells, which makes EpCAM a potential therapeutic target. Radionuclide molecular imaging of EpCAM expression might permit selection of patients for EpCAM-targeting therapies. In this study, we evaluated a scaffold protein, designed ankyrin repeat protein (DARPin) Ec1, for imaging of EpCAM in TNBC. DARPin Ec1 was labeled with a non-residualizing [125I]I-*para*-iodobenzoate (PIB) label and a residualizing [99mTc]Tc(CO)<sup>3</sup> label. Both imaging probes retained high binding specificity and affinity to EpCAM-expressing MDA-MB-468 TNBC cells after labeling. Internalization studies showed that Ec1 was retained on the surface of MDA-MB-468 cells to a high degree up to 24 h. Biodistribution in Balb/c nu/nu mice bearing MDA-MB-468 xenografts demonstrated specific uptake of both [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 in TNBC tumors. [125I]I-PIB-Ec1 had appreciably lower uptake in normal organs compared with [99mTc]Tc(CO)3-Ec1, which resulted in significantly (*p* < 0.05) higher tumor-to-organ ratios. The biodistribution data were confirmed by micro-Single-Photon Emission Computed Tomography/Computed Tomography (microSPECT/CT) imaging. In conclusion, an indirectly radioiodinated Ec1 is the preferable probe for imaging of EpCAM in TNBC.

**Keywords:** EpCAM; radionuclide; molecular imaging; SPECT; iodine; PIB; breast; cancer

#### **1. Introduction**

Breast cancer is one of the most common types of cancer among women worldwide. It is a heterogeneous disease, which is categorized into four subtypes: Luminal A, luminal B, human epidermal growth factor receptor 2-positive (HER2-positive), and "basal-like" or "triple-negative" [1]. Triple-negative breast cancer (TNBC) does not express estrogen, progesterone, or human epidermal growth factor (HER2) receptors and is characterized by having an aggressive course, early metastatic spread, and poor prognosis [2]. Although initial response to chemotherapy in TNBC might be better compared to other breast cancer subtypes, early relapse is commonly observed [3–5]. Furthermore, the high rate of relapse among TNBC patients after surgery because of incomplete eradication of the tumor highlights the need for more effective therapies [2].

Targeted systemic treatment is well proven for estrogen/progesterone- or HER2-expressing breast cancer but is still not established to the same extent for TNBC [1]. Therefore, disseminated TNBC is particularly challenging to treat. The standard of care for TNBC patients is sequential chemotherapy. However, its indiscriminate toxicity to healthy tissues results in a narrow therapeutic window and limited efficacy. Targeted delivery of a cytotoxic payload (drug, toxin, or radionuclide) specifically to cancer cells would reduce the off-target toxicity and increase the therapeutic window and the efficiency of the treatment. In April 2020 the US Food and Drug Administration (FDA) approved the antibody-drug conjugate sacituzumab govitecan-hziy, which targets the tumor-associated calcium signal transducer 2 (TROP-2) antigen, for treatment of disseminated TNBC refractory to previous chemotherapies [6]. It showed a 33% response rate in patients with metastatic TNBC pretreated with chemotherapy [7]. Still, the development of therapeutics specific to other molecular targets could further increase the success rate in treatment of TNBC.

The epithelial cell adhesion molecule (EpCAM) is overexpressed in a large number of TNBC cases. The fraction of EpCAM overexpression in TNBC is 36–88%, depending on the scoring system [8–10], which makes it an attractive target for this malignancy. In patients with TNBC, EpCAM overexpression is associated with unfavorable prognosis [9] and correlates with poor survival, lymph node metastasis, and distant metastasis [11]. Several therapeutic strategies targeting EpCAM in TNBC are under preclinical and clinical development [12–14]. The anti-EpCAM monoclonal antibody adecatumumab has been evaluated in a phase II clinical study in patients with metastatic breast cancer [15]. Three of the 18 patients with high EpCAM expression and adecatumumab treatment developed new metastases up to week 6, compared with 14 of 29 patients with low EpCAM expression, indicating that response was related to EpCAM expression.

Due to heterogeneity of EpCAM expression in patients with TNBC, it is necessary to select the patients with high expression for EpCAM-targeted therapy. Radionuclide molecular imaging allows for non-invasive, whole-body evaluation of targeted protein expression. In the past years, a number of biomolecules against various targets have been investigated for radionuclide molecular imaging of breast cancer, such as somatostatin (SST) analogues targeting SST receptors, vasoactive intestinal peptide targeting affibody molecules against HER2, arginine-glycine-aspartate (RGD) peptides targeting integrin receptors, bombesin analogues targeting gastrin-releasing peptide receptors (GRPRs), and peptide analogues of alpha-melanocyte stimulating hormone-targeting melanocortin receptors [16].

Previously developed probes for radionuclide molecular imaging of EpCAM were mainly based on a monoclonal antibody (mAb) scaffold [17–19]. In comparison to mAbs, engineered scaffold proteins (ESP) have more favorable properties for imaging because of their small size, which enables rapid localization in tumors and fast decrease of blood-associated background activity due to renal clearance [20,21]. Additionally, ESPs can be genetically engineered to incorporate peptide-based chelators, e.g., histidine tags for labeling with technetium-99m tricarbonyl, and can generally tolerate harsh radiolabeling conditions, such as high temperature or changes in pH. A class of ESP, the designed ankyrin repeat proteins (DARPins), demonstrated excellent results for radionuclide molecular imaging of HER2 in preclinical studies, providing high tumor-to-nontumor tissue ratio shortly after injection [22–24]. The DARPin scaffold consists of four to six helix-turn-helix units and has molecular weight from 14 to 18 kDa, depending on the number of units. DARPins are currently the only class of ESPs with binders selected against EpCAM [25]. According to surface

plasmon resonance, DARPin Ec1 has affinity of 68 pM to EpCAM [25], which meets requirements for a high-affinity imaging probe.

High affinity of an imaging probe is an important precondition for successful imaging, but it is not sufficient. Selection of a radionuclide and chemistry for its conjugation to a targeting probe is essential. Experience with other targeting proteins (affibody molecules [26] and DARPins targeting HER2) [22–24,27] demonstrated that the selection of an optimal labeling approach can increase tumor-to-organ ratios by an order of magnitude. Modification of the protein surface by a radionuclide in combination with a chelator or linker for coupling results in the alteration of off-target interactions with blood vessels and normal tissues. This has an essential impact on unspecific uptake in normal organs and tissues. Furthermore, accumulation of activity in both tumors and normal tissues depends on physicochemical properties of radioactive metabolites, which are formed after internalization and intracellular proteolysis of a labeled protein. Labels having charged or bulky polar radiometabolites are trapped inside the cells after proteolysis. Such labels are called *residualizing* [28]. The residualizing labels provide a long retention of activity both in tumors and in normal tissues if the imaging probe is internalized. In the case when the radiometabolites are lipophilic, they are capable of diffusing through cellular membranes and leaving the cell. These so-called *non-residualizing* labels are associated with low cellular retention of activity after internalization. When the internalization of an imaging probe by malignant cells is rapid, the use of residualizing labels is the only option for sufficient accumulation of activity in a tumor. However, the use of non-residualizing labels might be appropriate when internalization of a targeting probe by the cancer cells is slow, and an unspecific uptake, first and foremost, in liver and kidneys results in rapid internalization in normal tissues. In this case, tumor retention of activity depends mainly on the high affinity of a probe to its molecular target. A successful use of non-residualizing radiohalogen labels was demonstrated earlier for ESPs such as HER2-binding Albumin-binding domain (ABD)-Derived Affinity ProTein 6 (ADAPT6) [29] and HER2-binding DARPins 9\_29 and G3 [22,23], as well as for Ec1 for imaging of EpCAM expression in pancreatic and ovarian cancer models [30,31]. Importantly, good retention in tumors was accompanied by a rapid clearance from normal tissues, which increased tumor-to-organ ratios. Apparently, slow internalization of Ec1 after binding to TNBC cells is critical for the use of a non-residualizing label for imaging in this cancer. Such a slow internalization of radiolabeled Ec1 was observed in the case of binding to pancreatic and ovarian cell lines [30,31], which was an indication that the cellular processing pattern would be similar for TNBC as well. However, our previous observations with ESP suggest that an internalization rate might depend on origin of cancer cells. For example, a rate of HER2-binding affibody [111In]In-DOTA-ZHER2:342-pep2 internalization by ovarian cancer cells was twice higher compared with a rate of internalization by breast cancer cells [32]. HER3-binding affibody-ABD fusion proteins were internalized by pancreatic cancer cells appreciably rapider than by prostate cancer cells [33]. Thus, evaluation of internalization of radiolabeled Ec1 by TNBC cells was necessary before in vivo experiments.

The goal of this study was to investigate whether our previous findings could be translated to triple-negative breast cancer and to evaluate the potential of DARPin Ec1 for imaging of EpCAM in a TNBC model in vivo. To evaluate the influence of residualizing properties of the radiolabel, technetium-99m tricarbonyl [99mTc]Tc(CO)<sup>3</sup> with residualizing properties was used as a comparator to the non-residualizing [125I]I-*para*-iodobenzoate ([125I]I-PIB) label. The 99mTc (T1/<sup>2</sup> = 6.01 h) is the most commonly used radionuclide in clinical single-photon emission computed tomography (SPECT). The <sup>125</sup>I (T1/<sup>2</sup> = 59.4 d) is a chemical analogue and a convenient preclinical surrogate for <sup>123</sup>I (T1/<sup>2</sup> = 13.27 h), which is used for SPECT, or <sup>124</sup>I (T1/<sup>2</sup> = 4.18 d), which is used for positron emission tomography (PET).

#### **2. Materials and Methods**

#### *2.1. General Procedures*

Sodium iodide [125I]NaI was purchased from PerkinElmer Sverige AB (Upplands Väsby, Sweden). Technetium-99m was obtained as pertechnetate by elution of Ultra-TechneKow generator (Mallinckrodt, Petten, The Netherlands) with sterile 0.9% sodium chloride (Mallinckrodt, Petten, The Netherlands). The CRS (Center for Radiopharmaceutical Sciences) kits for production of tricarbonyl technetium were purchased from the Center for Radiopharmaceutical Sciences (PSI, Villigen, Switzerland; contact e-mail: crs-kit@psi.ch). Instant thin-layer chromatography (iTLC) analysis was performed using iTLC silica gel strips (Varian, Lake Forest, CA, USA). The radioactivity distribution along iTLC strips was measured using a Cyclone storage phosphor system (Packard) and analyzed by OptiQuant image analysis software. Purification of radiolabeled proteins was performed using NAP-5 size-exclusion columns (GE Healthcare, Buckinghamshire, UK). Radioactivity was measured using an automated gamma-spectrometer with a NaI (TI) detector (1480 Wizard, Wallac, Finland). MDA-MB-468 breast cancer cells and Ramos lymphoma cells were purchased from the American Type Culture Collection (ATCC) and were cultured in Roswell Park Memorial Institute (RPMI) medium supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 100 IU/mL penicillin, and 100 µg/mL streptomycin, in a humidified incubator with 5% CO<sup>2</sup> at 37 ◦C, unless mentioned otherwise. Binding specificity and cellular processing experiments were performed using 35-mm Petri dishes (Nunclon Delta Surface, ThermoFisher Scientific, Roskilde, Denmark). Ligand Tracer experiments were performed using 89-mm Petri dishes (Nunclon Delta Surface, ThermoFisher Scientific, Roskilde, Denmark).

#### *2.2. Protein Production and Radiolabeling*

The EpCAM-targeting DARPin Ec1-H<sup>6</sup> (containing a hexahistidine tag at C-terminus) was produced based on sequences published previously [25]. Production and purification of DARPin Ec1 was performed as described previously [30].

Indirect radioiodination of Ec1 using N-succinimidyl-para-(trimethylstannyl)benzoate was performed as described earlier [28,31]. Acetic acid in water (0.1%, 10 µL) was added to radioiodine (5–15 µL, 17–42 MBq). Then, N-succinimidyl-p-(trimethylstannyl)benzoate (13 nmoles, 5 µg, 5 µL of 1 mg/mL in 5% acetic acid in methanol) and chloramine-T (40 µg, 10 µL, 4 mg/mL in water) were added. The reaction was stopped by addition of sodium metabisulfite (60 µg, 10 µL, 6 mg/mL in water) after 5 min of incubation at room temperature. Then, DARPin Ec1 (7.6 nmoles, 140 µg, 39 µL of 3.6 mg/mL in phosphate-buffered saline (PBS)) in 140 µL of 0.07 M borate buffer (pH 9.3) was added and incubated at room temperature for 30 min. The radiolabeled conjugate was purified on a NAP-5 column, pre-equilibrated with 1% bovine serum albumin (BSA) in PBS, and eluted with PBS. The labeling yield and purity were determined using radio-iTLC analysis in 4:1 acetone:water system.

Site-specific radiolabeling of DARPin Ec1-H<sup>6</sup> bearing a C-terminal His6-tag with tricarbonyl technetium-99m was performed as described earlier [31]. The solution eluted from the technetium generator (500 µL) containing 3–4 GBq of [99mTc]Tc pertechnetate was added to the CRS kit, and the mixture was incubated at 100 ◦C for 30 min. The obtained solution of [99mTc]Tc(CO)<sup>3</sup> (12 µL, 83–108 MBq) was mixed with a solution of DARPin Ec1 (40 µg, 2.18 nmol) in 33 µL of PBS and incubated at 60 ◦C for 60 min. The radiochemical yield and purity were determined using iTLC strips eluted with PBS. The radiolabeled DARPin Ec1 was purified using NAP-5 columns pre-equilibrated and eluted with PBS.

Radio high-performance liquid chromatography (HPLC) analysis was performed using a Hitachi Chromaster HPLC system with a radioactivity detector and Phenomenex Luna® C18 column (100 Å; 150 × 4.6 mm; 5 µm) at room temperature (20 ◦C). Solvent A was 0.1% trifluoroacetic acid (TFA) in H2O, solvent B was 0.1% TFA in acetonitrile, and the flow rate was 1 mL/min. For identity and purity analysis, the 20-min method with a gradient from 5 to 95% solvent B over 18 min and from 95% to 5% solvent B from 18 to 20 min was used.

The label stability under challenge conditions (excess of histidine for [9 9mTc]Tc(CO)3-Ec1, NaI, or in 30% ethanol for [125I]I-PIB-Ec1) was assessed by analysis with iTLC silica gel (SG) strips eluted by PBS or by 4:1 acetone:water. The stability in complete cell culture medium containing 10% fetal bovine serum after 24 h of incubation at 37 ◦C was assessed by passing the media through a NAP-5 size-exclusion column and collecting the high-molecular-weight fraction (containing molecules over 5 kDa) and the low-molecular-weight fraction (containing molecules below 5 kDa). The activity in the column, the high- and low-molecular-weight fractions, was measured using a gamma-spectrometer.

#### *2.3. Binding Specificity and Cellular Processing Assays*

In vitro studies were performed using EpCAM-expressing breast cancer cell line MDA-MB-468. One day before the experiment, cells were seeded in 3-cm Petri dishes (ca. 1 × 10<sup>6</sup> cells per dish) and three dishes per group were used.

Binding specificity to EpCAM was evaluated as described previously [30,31]. To saturate EpCAM receptors, 100-fold excess of nonlabeled Ec1 DARPin (200 nM) in cell culture medium was added to one group of cells and an equal volume of media only was added to the second group. After 30 min of incubation at room temperature, radiolabeled DARPins [125I]I-PIB-Ec1 or [99mTc]Tc(CO)3-Ec1 were added at 2 nM final concentration. After 6 h of incubation at room temperature, the medium was collected, cells were washed, and trypsin was added to detach the cells. The cell suspension was collected, and the radioactivity of cells and medium was measured to calculate the percent of cell-bound radioactivity. The data were analyzed using unpaired two-tailed *t*-test.

Cellular retention and processing were studied during continuous incubation using an acid-wash method [32]. To study cellular processing during continuous incubation, radiolabeled [ 99mTc]Tc(CO)3-Ec1 or [125I]I-PIB-Ec1 (1 nM) were added to cells, which were incubated at 37 ◦C in a humidified incubator for 1, 2, 4, 6, and 24 h. At these time points, the media were collected from one group and cells were washed once with serum-free media. To collect the membrane-bound fraction, the cells were treated with 0.2 M glycine buffer containing 4 M urea (pH 2.0) on ice for 5 min causing dissociation of membrane-bound protein. The buffer was collected, and the cells were washed once with the same buffer. Then the cells were treated with 1 M NaOH for 30 min to lyse the cells containing internalized fraction, and the solution was collected. The activity in every fraction was measured. The maximum value of cell-associated activity for each dataset was taken as 100% and the other dataset values were normalized to it. To study cellular retention and processing after interrupted incubation, the cells were incubated with [99mTc]Tc(CO)3-Ec1 or [125I]I-PIB-Ec1 (10 nM) for 1 h at 4 ◦C. Then the media were removed, the cells were washed, fresh medium was added, and the cells were placed in a humidified incubator at 37 ◦C. At 1, 4, and 24 h, the medium was collected and cells were washed and treated as described above to evaluate the membrane-bound and internalized fractions.

For analysis of radiocatabolites in the supernatant after interrupted incubation, a part of it (500 µL) was separated using a NAP-5 size-exclusion column, pre-equilibrated with 1% BSA in PBS. Fractions containing activity associated with the high-molecular-weight compounds (first 900 µL) and low-molecular-weight compounds (3.6 mL) were collected. The activity in each fraction and the column were measured using a gamma-spectrometer. The residual activity left on the columns after separation was below 4% from the total activity. As a control for stability of the label, [99mTc]Tc(CO)3-Ec1 and [ <sup>125</sup>I]I-PIB-Ec1 were incubated in complete media at 37 ◦C in a humidified incubator for 24 h and analyzed as described above.

#### *2.4. A*ffi*nity Measurements Using LigandTracer*

The kinetics of [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 binding to living MDA-MB-468 and Ramos cells were measured using LigandTracer and evaluated using the TraceDrawer Software (both from Ridgeview Instruments, Vänge, Sweden) as described earlier [34]. Briefly, 2 × 10<sup>6</sup> MDA-MB-468 cells were seeded to a local area of an 89-mm Petri dish one day before the experiment. Ramos cells growing in suspension were attached to a Petri dish following a method developed by Bondza et al. [35]. Biomolecular anchor molecule (BAM) (SUNBRIGHT® OE-040CS, NOF Corporation) was dissolved in water to a concentration of 2 mg/mL. An area of a 98-mm Petri dish (Nunc, Cat No 263991) about 1.5 cm in diameter and 5 mm from the rim of the dish was covered with 400 µL of BAM solution (0.8 mg) and was incubated under sterile conditions at room temperature for 1 h. The BAM solution was aspirated and 400 µL of Ramos cell suspension (5 × 10<sup>6</sup> cells/mL, 2 × 10<sup>6</sup> ) were added dropwise to the BAM-coated area. Cells were allowed to attach to the dish for 40 min. Then the dish was tilted to remove the remaining cell suspension. The dish was covered with complete cell culture medium (10 mL) and placed into the incubator overnight. Cell attachment was confirmed the next day by observing the cells under a microscope.

To measure the binding during association phase, three concentrations of [125I]I-PIB-Ec1 (1.8, 5.4, and 14.5 nM) or [99mTc]Tc(CO)3-Ec1 (0.2, 0.6, and 1.8 nM) were added to cells, followed by exchange of media and measurement of retention in the dissociation phase. Binding kinetics were recorded at room temperature and dissociation constants were calculated based on association and dissociation rates.

#### *2.5. Ethical Statement*

The described procedures were reviewed and approved by the Animal Research Committee at Uppsala University (ethical permission 4/16 from 26 February 2016) and were performed in accordance with the Swedish national legislation on protection of laboratory animals.

#### *2.6. Animal Studies*

To select an optimal label, a dual-label biodistribution study was performed. To establish MDA-MB-468 xenografts, 10<sup>7</sup> cells were implanted subcutaneously in 8-week-old Balb/c nu/nu mice. For specificity control, 10<sup>7</sup> EpCAM-negative Ramos cells were implanted. At the time of experimentation (two to three weeks after implantation), the weights of the animals were 17 ± 2 g in the MDA-MB-468 group and 17 ± 0 g in the Ramos group. Average tumor weights were 0.05 ± 0.04 g for MDA-MB-468 and 0.05 ± 0.04 g for Ramos. Groups of four animals per data point were used.

A well-established, dual-label approach [36,37] was selected for animal studies. In this methodology, a mixture of compounds labeled with different nuclides is co-injected into animals, and the distribution of each labeled compound is determined by gamma-spectrometry of tissue samples. A precondition for this approach is that the gamma-spectra of nuclides can be resolved. This is the case for <sup>125</sup>I and 99mTc (Figure S7). An advantage of the dual-label methodology is that the factors related to a host animal (e.g., individual features of metabolic rate and blood circulation) and xenografts (e.g., vascularization or presence of necrotic areas) act in the same way on both tracers. This method enables the use of a paired t-test, which provides high statistic power with a small number of animals.

Mice were injected with a mixture of both [125I]I-PIB-Ec1 (non-residualizing label) and [ 99mTc]Tc(CO)3-Ec1 (residualizing label) and the biodistribution was measured 6 h and 24 h post injection (pi). The injected activity was 40 kBq/mouse for technetium-99m and 20 kBq/mouse for iodine-125. The injected protein dose was adjusted to 4 µg/mouse using unlabeled protein. The labeled proteins were injected into the tail vein. Before dissection, mice were anesthetized by an intraperitoneal (i.p.) injection of ketamine and xylazine solution and sacrificed by heart puncture. The dose of ketamine was 250 mg/kg and the dose of xylazine was 25 mg/kg. The organs and tissues were collected and weighed and the activity was measured using an automated gamma-spectrometer. Whole submandibular salivary gland, lung, liver, spleen, pancreas, stomach, and kidneys were sampled for measurements. A small section of small intestines was emptied of content to measure the uptake in intestinal walls. Activity in the rest of the intestinal tract was measured to estimate hepatobiliary excretion. The rest of the body was also collected and its activity was measured. The energy ranges for measurements of <sup>125</sup>I and 99mTc were 18–85 keV and 110–160 keV, respectively. Correction for counts' spillover was performed automatically by the software of the gamma-spectrometer. Activity in a sample was considered as nonmeasurable (NM) if a count rate for a sample plus background

was less than two-fold higher than for background (approximately 0.005% of the injected activity). The percentage of injected dose per gram of sample (%ID/g) was calculated.

In addition, an in vivo saturation experiment was performed. EpCAM receptors in MDA-MB-468 xenografts were saturated by co-injection of unlabeled protein at 0.5 mg/mouse (125-fold molar excess to 4 µg Ec1 dose used for biodistribution) together with the injection of [125I]I-PIB-Ec1 and [ 99mTc]Tc(CO)3-Ec1, and the biodistribution measurement was performed 6 h pi. To confirm specificity, the uptake of [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 was measured in EpCAM-negative Ramos lymphoma xenografts 6 h pi.

Whole body SPECT/CT scans of mice bearing MDA-MB-468 xenografts were performed using nanoScan SPECT/CT (Mediso Medical Imaging Systems, Budapest, Hungary). Mice were injected with [125I]I-PIB-Ec1 (4 µg, 0.3 MBq) and [99mTc]Tc(CO)3-Ec1 (4 µg, 9.4 MBq). Imaging at 6 h pi was performed after mice were sacrificed by CO2. The acquisition time was 20 min. CT scans were acquired using X-ray energy peak of 50 keV, 670 µA, 480 projections, and 5.26-min acquisition time. SPECT raw data were reconstructed using Tera-Tomo™ 3D SPECT reconstruction technology (version 3.00.020.000; Mediso Medical Imaging Systems Ltd.): High dynamic range, 30 iterations, one subset. CT data were reconstructed using Filter Back Projection and fused with SPECT files using Nucline 2.03 Software (Mediso Medical Imaging Systems Ltd.). Images are presented as maximum-intensity projections in the red, green, and blue (RGB) color scale.

To analyze the radioactive species in urine after injection of [125I]I-PIB-Ec1, it was injected intravenously (i.v.) into two healthy Naval Medical Research Institute (NMRI) mice (Scanbur AS, Karlslunde, Denmark) (10 µg, 1.9 MBq per mouse). Mice were kept in separate cages covered with absorbent paper, which was later checked for the activity indicating any release of urine. One hour after the injection, mice were anesthetized by i.p. injection of ketamine and xylazine solution and sacrificed by cervical dislocation. Urinary bladders containing urine were excised, cut through, the urine was collected in Eppendorf tubes, and the activity was measured (0.58 MBq in 50 µL for mouse 1; 0.68 MBq in 100 µL for mouse 2). An equal volume of ice-cold acetonitrile was added to each tube and the tubes were centrifuged at 15,000 rpm at 4 ◦C for 15 min. The solution was filtered through a 0.45-µm filter and analyzed by radio-HPLC. The same procedure with addition of acetonitrile and centrifugation was performed for the intact [125I]I-PIB-Ec1 as a control to check that it would be detectable during HPLC analysis.

Radio-HPLC analysis was performed using a Hitachi Chromaster HPLC system with a radioactivity detector and Phenomenex Luna® C18 column (100 Å; 150 × 4.6 mm; 5 µm) at room temperature (20 ◦C). Solvent A was 0.1% trifluoroacetic acid (TFA) in H2O, solvent B was 0.1% TFA in acetonitrile, and the flow rate was 1 mL/min. The 30-min method with a gradient from 5 to 95% solvent B over 28 min and from 95% to 5% solvent B from 28 to 30 min was used.

#### *2.7. Statistical Analysis*

The in vitro specificity and cellular processing data are presented as the mean ± standard deviation (SD) of three samples. Statistical analysis was performed using GraphPad Prism (version 7.02; GraphPad Software, Inc., La Jolla, CA, USA). The *p* < 0.05 was considered a statistically significant difference. The data were analyzed using an unpaired two-tailed *t*-test. The biodistribution data for dual-label experiments at 6- or 24-h time points were analyzed using a paired two-tailed *t*-test.

#### **3. Results**

#### *3.1. Radiolabeling*

DARPin Ec1 was labeled site-specifically with 99mTc(CO)<sup>3</sup> using a hexahistidine tag at C-terminus to provide a residualizing label. Labeling of DARPin Ec1 with [125I]I-*para*-iodobenzoate was performed by attaching the N-hydroxysuccinimide ester derivative of [125I]I-PIB to amino groups of lysines. Data concerning radiolabeling of DARPin Ec1 with [125I]I-PIB and technetium-99m tricarbonyl are

presented in Table 1. Size-exclusion chromatography provided radiochemical purities over 99%. Both labeling methods provided stable labels (Tables 2 and 3).


**Table 1.** Labeling and characterization of radiolabeled Ec1 variants.

**Table 2.** In vitro stability of [99mTc]Tc(CO)<sup>3</sup> -Ec1.


Samples were incubated in PBS or with 1000-fold molar excess of histidine at 37 ◦C. Analysis was performed in duplicates.

**Table 3.** In vitro stability of [125I]I-PIB-Ec1.


Samples were incubated in PBS, 30% ethanol or with 1000-fold molar excess of NaI at 37 ◦C. Analysis was performed in duplicates.

#### *3.2. Characterization of Radiolabeled DARPins In Vitro*

In vitro evaluation was performed using EpCAM-expressing MDA-MB-468 breast cancer cells. To demonstrate binding specificity of [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 to EpCAM, the EpCAM receptors were saturated with 100-fold molar excess of nonlabeled Ec1 before addition of the radiolabeled compound. Blocking the EpCAM receptors resulted in a significant (*p* < 0.001) decrease of both [ <sup>125</sup>I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 uptake (Figure 1). This demonstrated a saturable character of radiolabeled Ec1 binding to MDA-MB-468 cells.

**Figure 1.** In vitro specificity of epithelial cell adhesion molecule (EpCAM) targeting using [ 99mTc]Tc(CO)<sup>3</sup> -Ec1 (**A**) and [125I]I-PIB-Ec1 (**B**) in EpCAM-expressing MDA-MB-468 cells. Uptake by cells was significantly (*p* < 0.001) reduced when 100-fold molar excess of nonlabeled Ec1 designed ankyrin repeat protein (DARPin) was added to the blocked groups. Final concentration of radiolabeled compound was 2 nM. Data are presented as mean from three samples ± SD.

The binding kinetics of [99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 to MDA-MB-468 cells were measured using LigandTracer (Figure 2, Figures S1 and S2). A rapid binding and slow dissociation were observed. The equilibrium dissociation constant (KD) values for both probes were in the picomolar range (Table 1). As a control for nonspecific interactions with cells, binding of [ 99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 to EpCAM-negative Ramos cells was also measured using LigandTracer (Figures S3 and S4). The signal detected from Ramos cells was comparable to the background and it was much lower than the signal detected from MDA-MB-468 cells.

**Figure 2.** LigandTracer sensorgrams of [99mTc]Tc(CO)<sup>3</sup> -Ec1 (**A**) and [125I]I-PIB-Ec1 (**B**) binding to MDA-MB-468 cells. The association was measured at 0.2, 0.6, and 1.8 nM concentrations for [ 99mTc]Tc(CO)<sup>3</sup> -Ec1 and at 1.8, 5.4, and 14.5 nM concentrations [125I]I-PIB-Ec1.

The processing of [99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 by MDA-MB-468 breast cancer cells during continuous incubation is shown in Figure 3. For [99mTc]Tc(CO)3-Ec1, the total cell-associated activity increased continuously over 24 h incubation and the internalized fraction had also a tendency to a slow increase. For [125I]I-PIB-Ec1, the maximum of total cell-associated activity was reached at 6 h and it slowly decreased by 24 h. This could be explained by the non-residualizing properties of the [125I]I-PIB label and the diffusion of iodine catabolites from the cells after internalization. A characteristic feature of radiolabeled Ec1 was the quite low internalization. The internalized fraction for [99mTc]Tc(CO)3-Ec1 was approximately 15% of the total cell-associated activity at 24 h.

**Figure 3.** Cellular processing of [99mTc]Tc(CO)<sup>3</sup> -Ec1 (**A**) and [125I]I-PIB-Ec1 (**B**) by MDA-MB-468 cells during continuous incubation. Cells were incubated with the DARPins (1 nM) at 37 ◦C. Data are presented as the mean of three samples ± standard deviation (SD). Error bars might not be seen when they are smaller than data point symbols.

Cellular processing and retention of [99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 by MDA-MB-468 cells after interrupted incubation is shown in Figure 4. After initial release of both probes from the membranes due to the shift in equilibrium, the cell-associated activity for [99mTc]Tc(CO)3-Ec1 remained constant until 24 h (Figure 4A). On the opposite, the cell-associated activity continued to decrease in the case of [125I]I-PIB-Ec1 (Figure 4B). This decrease was associated with the gradual build-up of low-molecular-weight (<5 KDa) radioactive compounds in the media (25 ± 1% at 4 h and 70 ± 1% at 24 h) (Figure 4D). For [99mTc]Tc(CO)3-Ec1, the low-molecular-weight activity fraction was only 18 ± 3% at 24 h. Most likely, these low-molecular-weight compounds are the products of intracellular degradation released from cells, since only less than 5% of activity was in the low-molecular-weight fraction after 24-h incubation of both compounds in cell-free complete media. The difference in the amount of low-molecular-weight-associated activity reflects, most likely, the difference in intracellular retention of radiometabolites of iodine and technetium labels. Identity and purity of the radiolabeled [ 99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 was confirmed by radio-HPLC analysis (Figure S5).

**Figure 4.** Cellular retention of activity after interrupted incubation of [99mTc]Tc(CO)<sup>3</sup> -Ec1 (**A**) and [125I]I-PIB-Ec1 (**B**) with MDA-MB-468 cells. Cells were first incubated with the DARPin variants (10 nM) at 4 ◦C for 1 h and then the media were exchanged and the cells were incubated at 37 ◦C for 1, 4, or 24 h. A fraction of the supernatant at every time point was analyzed using NAP-5 size-exclusion columns and compared to the control when the radiolabeled DARPins were incubated in complete media for 24 h (**C**,**D**). Numbers in panels (**C**,**D**) show a percentage of activity associated with low-molecular-weight compounds at each time point. Data for retention are presented as the mean of three samples ± SD and data for NAP-5 analysis are presented as the mean of two samples ± SD. LMW = low molecular weight, HMW = high molecular weight. Error bars might not be seen when they are smaller than data point symbols.

#### *3.3. In Vivo Studies*

The results of the specificity test (Figure 5) demonstrated that the uptake of both [99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 in EpCAM-negative Ramos xenografts was much lower (*p* < 0.001, unpaired t-test) than in EpCAM-positive MDA-MB-468 xenografts. In addition, saturation of EpCAM by co-injecting a

large excess of unlabeled Ec1 resulted in a significant (*p* < 0.01, unpaired *t*-test) reduction in tumor uptake of both variants.

**Figure 5.** In vivo specificity of EpCAM targeting using [99mTc]Tc(CO)<sup>3</sup> -Ec1 (**A**) and [125I]I-PIB-Ec1 (**B**). Uptake of both DARPin variants was significantly (*p* < 0.01, unpaired *t*-test) higher in EpCAM-positive MDA-MB-468 xenografts than in EpCAM-negative Ramos xenografts 6 h post injection (pi). EpCAM blocking in MDA-MB-468 xenografts by co-injecting a large excess of unlabeled Ec1 also resulted in a significant decrease of tracer uptake. Data are presented as mean ± SD for four mice.

The tumor uptake and retention of [99mTc]Tc(CO)3-Ec1 was significantly (*p* < 0.05, paired *t*-test) higher than tumor uptake and retention of [125I]I-PIB-Ec1 (2.6 ± 0.2 vs. 1.7 ± 0.2%ID/g 6 h pi). Distribution of activity in normal organs and tissues was quite different for 99mTc- and <sup>125</sup>I-labeled Ec1 (Table 4). The renal uptake of [99mTc]Tc(CO)3-Ec1 was more than 100-fold higher compared to [125I]I-PIB-Ec1 even 24 h pi. A combination of a high kidney retention of activity and low activity accumulation in gastrointestinal tract indicated a renal excretion pathway of DARPin Ec1. The renal and hepatic uptake of [99mTc]Tc(CO)3-Ec1 was higher than the tumor uptake. [125I]I-PIB-Ec1 had appreciably lower hepatic and renal uptake. Overall, [125I]I-PIB-Ec1 had the lowest uptake in normal tissues. Already by 6 h after injection of [125I]I-PIB-Ec1 only ca. 4% of ID was left in mice and over 95% of ID was excreted in comparison with [99mTc]Tc(CO)3-Ec1, when about 30% of ID was excreted by 6 h (Table S1). Accordingly, [125I]I-PIB-Ec1 provided significantly higher tumor-to-organ ratios compared with [99mTc]Tc(CO)3-Ec1 (Table 5). At 6 h pi, [125I]I-PIB-Ec1 had two-fold higher tumor-to-blood, ca. 100-fold higher tumor-to-liver, 20-fold higher tumor-to-spleen and tumor-to-pancreas, and eight-fold higher tumor-to-muscle ratios than [99mTc]Tc(CO)3-Ec1. By 24 h pi, the tumor-to-blood and tumor-to-kidney ratios increased further for the [125I]I-PIB label, while no improvement was observed for the technetium-99m label.


**Table 4.** Biodistribution of [99mTc]Tc(CO)<sup>3</sup> -Ec1 and [125I]I-PIB-Ec1 in Balb/c nu/nu mice bearing MDA-MB-468 xenografts 6 and 24 h pi.

Data are presented as mean percent of injected dose (%ID)/g ± SD for four mice. Data for the rest of the intestines with contents and rest of the body are presented as %ID per whole sample. *<sup>a</sup>* Significant difference between [ 99mTc]Tc and [125I]I at the same time point (paired t-test). *<sup>b</sup>* Significant difference between values for [99mTc]Tc at 6 and 24-h time point (unpaired t-test). *<sup>c</sup>* Significant difference between values for [125I]I at 6- and 24-h time point (unpaired *t*-test). NM = nonmeasurable.


**Table 5.** Tumor-to-organ ratios of [99mTc]Tc(CO)<sup>3</sup> -Ec1 and [125I]I-PIB-Ec1 in Balb/C nu/nu mice bearing MDA-MB-468 xenografts at 6 and 24 h pi.

Data are presented as mean ± SD for four mice. *<sup>a</sup>* Significant difference between [125I]I and [99mTc]Tc at the same time point (paired t-test). *<sup>b</sup>* Significant difference between values for [99mTc]Tc at 6- and 24-h (unpaired t-test). *<sup>c</sup>* Significant difference between values for [125I]I at 6- and 24-h (unpaired *t*-test). NM= nonmeasurable.

To study if the lower activity retention in kidneys in case of [125I]I-PIB-Ec1 is caused by the excretion of radiometabolites, we performed radio-HPLC analysis of urine collected 1 h after [125I]I-PIB-Ec1 injection in healthy NMRI mice (Figure S6). It was observed that the majority of activity excreted with urine was in the form of radiocatabolites already 1 h after the injection of [125I]I-PIB-Ec1.

MicroSPECT/CT imaging using [99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec in Balb/c nu/nu mice bearing MDA-MB-468 xenografts at 6 h pi confirmed the results of the biodistribution studies (Figure 6). The tumor was visualized using both radiolabeled variants of DARPin Ec1. The use of non-residualizing label [125I]I-PIB-Ec1 provided lower retention of activity in normal organs and kidneys and higher imaging contrast compared to [99mTc]Tc(CO)3-Ec1.

**Figure 6.** Micro-Single-Photon Emission Computed Tomography/Computed Tomography (microSPECT/CT) imaging of EpCAM expression in BALB/C nu/nu mice bearing EpCAM-positive MDA-MB-468 xenografts at 6 h pi using [99mTc]Tc(CO)<sup>3</sup> -Ec1 (**A**) and [125I]I-PIB-Ec (**B**). Arrows indicate: T—tumor, K—kidneys, L—liver, LN—lymph node. The scale in panel A was adjusted to the first red pixel in the tumor.

#### **4. Discussion**

Efficacy of targeted therapies is critically dependent on expression level of the molecular target in tumors. In the case of absence or too low expression level of the target, there will be no therapeutic effect. Unfortunately, an unspecific toxicity to normal organs and tissues will be preserved. This is a particularly apparent risk for a targeted delivery of cytotoxic payloads, such as drugs, toxins, or alpha or beta particle-emitting radionuclides. Thus, personalizing treatment by stratification of patients according to expression of a molecular target in tumors is an essential precondition for successful targeted therapy. Although the percentage of EpCAM-overexpressing TNBC tumors is high, identification of patients eligible for therapy is necessary to avoid overtreatment of patients having tumors with low expression level. Radionuclide molecular imaging is a promising approach, as it enables visualization of multiple metastases addressing heterogeneity of expression.

Imaging data can be used to select patients for targeted therapy in different ways. The most common method is the determination of tumor-to-reference organ ratio. A typical example is the so-called Krenning score, which is applied for selection of patients with neuroendocrine tumors for somatostatin receptor-targeted radionuclide therapy using [177Lu-DOTA<sup>0</sup> ,Tyr<sup>3</sup> ]octreotate [38,39]. The scale grades the tumor uptake of [<sup>111</sup> In-DTPA<sup>0</sup> ]octreotide in planar gamma camera images: Grade 1, no tumor uptake; grade 2, tumor uptake is equal to normal liver tissue; grade 3, tumor uptake is greater than normal liver tissue; and grade 4, tumor uptake is higher than normal spleen or

kidney uptake. It has been demonstrated that the response strongly correlates with the grade [38]. It has been shown that tumor-to-spleen uptake ratio for <sup>111</sup>In and <sup>68</sup>Ga-labeled affibody molecules [40] and tumor-to-contralateral breast ratio for 99mTc-labeled ADAPT6 [41] correlates strongly with the level of HER2 expression in breast cancer. Alternatively, the uptake of an imaging probe in tumor might be quantified using PET and correlated with expression level [42].

An essential factor for successful radionuclide imaging is high imaging contrast because it determines the diagnostic sensitivity. Thus, imaging agents providing high tumor-to-organ ratios (ratios of activity concentration in tumors to the concentration in normal organs), especially to organs that are frequent metastatic sites, are required. TNBC has a predominant metastasis to visceral organs, first and foremost, liver and lungs [5,43]. Thus, sufficiently high tumor-to-liver and tumor-to-lung ratios are the preconditions for successful translation to clinics. In addition, a high tumor-to-blood ratio is an essential parameter for evaluation of an imaging agent as a blood-borne activity and might contribute to the background signal.

Selection of a fitting label is important for obtaining of a high contrast. One strategy to achieve a high imaging contrast is based on using non-residualizing labels. After binding of radiolabeled proteins to cell-surface receptors on cancer cells or to scavenger receptors on cells in excretory organs (e.g., kidneys and liver), a protein-receptor complex is internalized and the radiolabeled protein is degraded in lysosomes with formation of radiocatabolites. Radiocatabolites of residualizing labels (typically, radiometals) are retained inside the cells as they are not able to diffuse through the lipophilic membranes. However, the radiocatabolites of non-residualizing labels rapidly diffuse from the cells, return to blood circulation, and are excreted with urine. When the internalization of radiolabeled proteins is slow in tumors and rapid in normal organs and tissues, the use of non-residualizing labels might enable fast clearance of activity from normal organs and tissues and provide higher imaging contrast than the use of residualizing labels.

Three different variants of non-residualizing radioiodine labels were evaluated for DARPins: A product of so-called direct electrophilic radioiodination, when the radioiodine is incorporated into the phenolic ring of tyrosine [22,23,30], a conjugate with [125I]-PIB [30,31], and a site-specific conjugate of [125I]I-iodo-[(4-hydroxyphenyl)ethyl]-maleimide (HPEM) to a unique cysteine engineered to a C-terminus of DARPin G3 [24]. It turned out that the use of the HPEM label results in a high level of hepatobiliary excretion and accumulation of activity in the content of gastrointestinal tract. This is undesirable, as it creates an unacceptable high background for imaging of visceral TNBC metastases. The use of direct iodination resulted in accumulation of radiocatabolites in Na/I-symporter-expressing organs (first and foremost, in salivary gland and stomach) but also in in visceral organs such as pancreas and intestines [30]. Overall, previous studies suggest that indirect radioiodination using PIB would be optimal for labeling of Ec1.

A potential disadvantage of the use of [125I]I-PIB is a random attachment of N-succinimidyl-*para*-iodobenzoate to amino groups of lysines in DARPin Ec1. As DARPin Ec1 contains eight lysines, this labeling might result in a mixture of labeled proteins with a different number and positions of [125I]I-PIB label. However, previous studies with proteins of similar size showed that the specificity and affinity is usually preserved with this type of labeling [27,44–46].

This study showed that the binding of both [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 to TNBC cells in vitro (Figure 1) can be blocked with an excess of unlabeled Ec1, which confirmed its specificity.

Previous studies have demonstrated that kinetics and affinity of binding of radiolabeled proteins to their molecular targets depend on cell line origin [47,48]. This might be associated with the molecular context of cellular membranes, including such factors as co-expression of other receptors or cell-surface proteins, glycosylation patterns, and homo- and heterodimerization. The results of LigandTracer measurements (Figure 2) demonstrated that binding of both tracers to TNBC cells was characterized by rapid association and very slow dissociation rates. The equilibrium dissociation constants were 121 ± 21 pM and 58 ± 5 pM for [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1, respectively. Thus, affinity of [125I]I-PIB-Ec1 binding to TNBC cells was somewhat lower compared to binding to

ovarian cancer OvCAR-3 cells (35 ± 1 pM) [31] or pancreatic cancer BxPC-3 cells (58 ± 13 pM) [30]. According to stoichiometric calculations, the average number of pendant groups conjugated per Ec1 molecule should be 0.3, which should have a minimal impact on binding properties. However, it might happen that one of the pendant groups is conjugated to a lysine close to the binding site and might create a hindrance. This might explain the differences in affinities between [99mTc]Tc(CO)3-Ec1, which was labeled site specifically, and [125I]I-PIB-Ec1. Still, the picomolar affinity is a good precondition for a strong retention of an imaging probe by malignant cells in vivo.

Assessment of the internalization rate of Ec1 during continuous incubation of cells with the tracer was initially performed using [99mTc]Tc(CO)3-Ec1 (Figure 3). A residualizing label is the most suitable for such kind of test because the data obtained by a non-residualizing label might be deceptive due to the "leakage" of radiometabolites from cells leading to underestimation of internalized radioactivity. The cellular processing study showed a slow internalization (approximately 15% of cell-bound activity at 24 h after incubation start) of Ec1 by TNBC cells. The difference between cellular processing of [ 99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 was further elucidated by studying cellular retention after interrupted incubation (Figure 4). The use of radioiodine label was associated with lower retention of activity by cells. In addition, a decrease of cell-associated radioiodine activity was accompanied with an increase of low-molecular-weight radioiodinated compounds in the incubation media. Most likely, this was caused by diffusion of radiometabolites of a non-residualizing label from cells. We considered the combination of the slow internalization and high affinity as a good rationale to proceed with in vivo studies.

Animal studies demonstrated clearly specific accumulation of both [125I]I-PIB-Ec1 and [ 99mTc]Tc(CO)3-Ec1 in MDA-MB-468 xenografts (Figure 5). Saturation of EpCAM by co-injection of a large excess of unlabeled Ec1 resulted in significantly lower uptake of the tracers in TNBC xenografts. In addition, the activity uptake in EpCAM-negative Ramos xenografts was much lower than in EpCAM-positive MDA-MB-468 xenografts.

Low accumulation in normal tissue is an important precondition for a high-contrast imaging. The results of the biodistribution experiments demonstrated advantages in the use of non-residualizing labels for slowly internalizing, high-affinity imaging probes (Table 4). The difference in the renal uptake of [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 was spectacular already at 6 h after injection. A high re-absorption in the proximal tubuli of the kidneys is a common feature of imaging probes based on short peptides [49] and ESPs, such as affibody molecules, ADAPTs, or DARPins [20,50,51]. In the case of short peptides, the re-absorption could be suppressed by blocking scavenger receptors with cationic amino acids or succinylated bovine gelatin (Gelofusine) [49]. However, the use of these and a number of other substances potentially blocking or reducing renal reabsorption was unsuccessful in the case of DARPins [51]. Thus, there is no conventional way to reduce the renal uptake of DARPins labeled with a residualizing label. However, the use of [125I]I-PIB-Ec1 resulted in more than a 70-fold lower renal uptake compared with [99mTc]Tc(CO)3-Ec1. Radio-HPLC analysis of the urine 1 h after injection of [ <sup>125</sup>I]I-PIB-Ec1 demonstrated that the activity was excreted predominantly as radiometabolites (S6). This suggests that non-residualizing properties of the radioiodine label were critical for the reduction of renal retention. Even more impressive (and more relevant to imaging of EpCAM expression in TNBC metastases) was the reduction of hepatic uptake, which was 160-fold lower for [125I]I-PIB-Ec1. While the uptake of [99mTc]Tc(CO)3-Ec1 in tumor was nearly 7-fold lower than in liver, the tumor uptake of [125I]I-PIB-Ec1was 15-fold higher than the hepatic uptake.

It has to be noted that the tumor uptake of [99mTc]Tc(CO)3-Ec1 was significantly (*p* < 0.05, paired *t*-test) higher than tumor uptake and retention of [125I]I-PIB-Ec1 (2.6 ± 0.2 vs. 1.7 ± 0.2%ID/g 6 h pi). This can be explained by the lower affinity of [125I]I-PIB-Ec1 compared with its 99mTc-labeled counterpart. In addition, we can expect that some internalization would take place during six hours and leakage of catabolites would affect the retention of activity in tumors. However, the magnitude of the uptake reduction was appreciably smaller compared with the uptake in normal tissues. For this reason, tumor-to-organ ratios were several folds higher for [125I]I-PIB-Ec1

than for [99mTc]Tc(CO)3-Ec1 (Table 3). Particularly, tumor-to-blood, tumor-to-lung, tumor-to-liver, and tumor-to-muscle ratios were 19 ± 3, 8 ± 2, 15 ± 2, and 42 ± 10 at 6 h after injection, respectively. Accordingly, an experimental microSPECT imaging using [125I]I-PIB-Ec1 permitted clear visualization of EpCAM expression in TNBC xenografts (Figure 5). It has to be noted that the tumor-associated activity decreased with time after injection of [125I]I-PIB-Ec1. At 24 h pi, the tumor uptake of [125I]I-PIB-Ec1was only 0.27 ± 0.05%ID/g. This was more than five-fold lower compared with the tumor uptake of [ 99mTc]Tc(CO)3-Ec1. Such low uptake makes this tracer unsuitable for imaging the next day after injection. Thus, non-residualizing labels are suitable for imaging only a few hours after injection even in the cases of slow internalization by malignant cells.

It should also be noted that the observed differences in the biodistribution between [ 99mTc]Tc(CO)3-Ec1 and [125I]I-PIB-Ec1 might not only be due to residualizing properties of labels but also be influenced by other label properties (polarity, site specificity of labeling), as well as by differences in pharmacokinetics.

In this study, <sup>125</sup>I was used as a label because of its convenient half-life (60 days). However, its low-energy electromagnetic radiation (max. 35.5 keV) permits its use only in small rodents and is unsuitable for clinical translation. Two iodine radioisotopes, <sup>123</sup>I (T1/<sup>2</sup> = 13.3 h, Eγ = 159 keV) and positron-emitting <sup>124</sup>I (T1/<sup>2</sup> = 100 h, β+ 23%), are suitable for radionuclide imaging using single-photon computed tomography (SPECT) and positron emission tomography (PET), respectively. Since the chemical properties of isotopes are identical, only a minor re-optimization is required to change the labeling chemistry from <sup>125</sup>I to these nuclides [24,52].

Summarizing, this study demonstrated that [125I]I-PIB-Ec1 has high (subnanomolar) affinity to EpCAM. It binds in a specific manner to EpCAM-expressing TNBC cell line in vitro and accumulates specifically in EpCAM-expressing TNBC xenografts in mice. The tumor-to-organ ratios, which are a measure of imaging contrast and sensitivity, are appreciably higher for [125I]I-PIB-Ec1 than for [ 99mTc]Tc(CO)3-Ec1. The [125I]I-PIB-Ec1 is capable of visualization of EpCAM-expressing TNBC xenograft in a mouse. This creates a rationale for further clinical studies concerning imaging of EpCAM in TNBC and correlation of imaging and biopsy data.

#### **5. Conclusions**

Both [125I]I-PIB-Ec1 and [99mTc]Tc(CO)3-Ec1 demonstrated specific uptake in EpCAM-positive TNBC xenografts. Radioiodine provided better tumor-to-organ ratios compared to [99mTc]Tc(CO)<sup>3</sup> label. Radioiodinated DARPin Ec1 is a promising agent for same-day imaging of EpCAM expression in triple-negative breast cancer using SPECT.

**Supplementary Materials:** The following are available online, Figure S1: representative curves of the LigandTracer measurement of [99mTc]Tc(CO)<sup>3</sup> -Ec1 binding to MDA-MB-468 cells, Figure S2: representative curves of the LigandTracer measurement of [125I]I-PIB-Ec1 binding to MDA-MB-468 cells, Figure S3: representative curves of the LigandTracer measurement of [99mTc]Tc(CO)<sup>3</sup> -Ec1 binding to Ramos cells, Figure S4: representative curves of the LigandTracer measurement of [125I]I-PIB-Ec1 binding to Ramos cells; Figure S5: radio-HPLC analysis of [125I]I-PIB-Ec1 and [99mTc]Tc(CO)<sup>3</sup> -Ec1 in comparison to the non-labeled Ec1, Figure S6: radio-HPLC analysis of mouse urine 1 h after injection of [125I]I-PIB-Ec1 in comparison to the intact [125I]I-PIB-Ec1, Figure S7: resolution of gamma-spectra of iodine-125 and technetium-99m, Table S1: biodistribution of [99mTc]Tc(CO)<sup>3</sup> -Ec1 and [125I]I-PIB-Ec1 in Balb/c nu/nu mice bearing MDA-MB-468 xenografts at 6 and 24 h.

**Author Contributions:** A.V., S.D., and V.T. contributed to the concept and study design. E.K. and A.S. performed the production and purification of proteins. S.D. supervised the production and purification of proteins. J.G., O.V., A.A., A.O., and V.T. participated in planning and performing the in vivo experiments, data treatment, and interpretation. A.V. performed the labeling chemistry, in vitro and in vivo studies, data treatment, and interpretation. E.B. participated in data treatment and interpretation. A.V. and E.B. wrote the first draft of the manuscript. All authors revised the manuscript critically, read, and approved the final manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the grant of Ministry of Science and Higher Education of the Russian Federation (075-15-2019-1925). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

**Acknowledgments:** The authors thank Sara S. Rinne for proofreading the paper.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **References**


52. Orlova, A.; Wållberg, H.; Stone-Elander, S.; Tolmachev, V. On the selection of a tracer for PET imaging of HER2-expressing tumors: Direct comparison of a 124I-labeled affibody molecule and trastuzumab in a murine xenograft model. *J. Nucl. Med.* **2009**, *50*, 417–425. [CrossRef] [PubMed]

**Sample Availability:** Samples of the compound DARPin Ec1 are available from the authors.

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Initial In Vitro and In Vivo Evaluation of a Novel CCK2R Targeting Peptide Analog Labeled with Lutetium-177**

**Anton Amadeus Hörmann <sup>1</sup> , Maximilian Klingler <sup>1</sup> , Maliheh Rezaeianpour 1,2 , Nikolas Hörmann <sup>3</sup> , Ronald Gust <sup>3</sup> , Soraya Shahhosseini <sup>2</sup> and Elisabeth von Guggenberg 1,\***


Academic Editor: Krishan Kumar

Received: 27 August 2020; Accepted: 1 October 2020; Published: 8 October 2020

**Abstract:** Targeting of cholecystokinin-2 receptor (CCK2R) expressing tumors using radiolabeled minigastrin (MG) analogs is hampered by rapid digestion of the linear peptide in vivo. In this study, a new MG analog stabilized against enzymatic degradation was investigated in preclinical studies to characterize the metabolites formed in vivo. The new MG analog DOTA-DGlu-Pro -Tyr-Gly-Trp-(*N*-Me)Nle-Asp-1Nal-NH<sup>2</sup> comprising site-specific amino acid substitutions in position 2, 6 and 8 and different possible metabolites thereof were synthesized. The receptor interaction of the peptide and selected metabolites was evaluated in a CCK2R-expressing cell line. The enzymatic stability of the <sup>177</sup>Lu-labeled peptide analog was evaluated in vitro in different media as well as in BALB/c mice up to 1 h after injection and the metabolites were identified based on radio-HPLC analysis. The new radiopeptide showed a highly increased stability in vivo with >56% intact radiopeptide in the blood of BALB/c mice 1 h after injection. High CCK2R affinity and cell uptake was confirmed only for the intact peptide, whereas enzymatic cleavage within the receptor specific C-terminal amino acid sequence resulted in complete loss of affinity and cell uptake. A favorable biodistribution profile was observed in BALB/c mice with low background activity, preferential renal excretion and prolonged uptake in CCK2R-expressing tissues. The novel stabilized MG analog shows high potential for diagnostic and therapeutic use. The radiometabolites characterized give new insights into the enzymatic degradation in vivo.

**Keywords:** cholecystokinin-2 receptor; minigastrin; molecular imaging; targeted radiotherapy; lutetium-177

#### **1. Introduction**

Radiolabeled peptide analogs for theranostic use in the diagnosis and treatment of cancer need to fulfill important prerequisites, such as high receptor affinity, appropriate metabolic stability, high and persistent tumor uptake, as well as low uptake in non-target tissue and fast blood clearance [1]. Up to date, targeting G-protein coupled receptors overexpressed on the surface of tumor cells for nuclear medicine applications is mainly limited to radiolabeled somatostatin analogs. The cyclic somatostatin analog octreotide with high affinity to the somatostatin receptor subtype 2 is used for symptomatic and biochemical control in the treatment of neuroendocrine tumors [2].

Radiolabeled octreotide derivatives have been successfully introduced in routine nuclear medicine applications for diagnosis and treatment of neuroendocrine tumors [3]. A major milestone in this respect is the recent approval of Lutathera® for peptide receptor radionuclide therapy by the European Medicines Agency (EMA) and by the Food and Drug Administration (FDA) [4,5]. So far, this success could not be translated to radiolabeled peptide analogs targeting other receptors. The reason often lies in the rapid metabolism in vivo of the linear peptide sequences derived from natural peptide hormones leading to cleavage of amino acids by enzymatic degradation and subsequent loss of affinity to the target receptor [1]. Radiolabeled peptide analogs targeting the cholecystokinin-2 receptor (CCK2R), overexpressed in different tumors, such as small cell lung cancer, stromal ovarian cancers, gastrointestinal stromal tumors, astrocytoma and especially medullary thyroid carcinoma (MTC), have shown to be very promising for application in diagnosis and therapy [6,7]. The reported clinical use has mainly focused on the diagnosis and treatment of patients with advanced MTC [8]. In the last decades, several attempts have been made to develop a radiolabeled CCK2R targeting peptide analog with suitable pharmacological properties for theranostic applications. First prove of principle studies with a radioiodinated gastrin analog confirmed the feasibility of CCK2R targeting [9]. Different CCK2R-targeting peptide analogs conjugated to the bifunctional chelators diethylenetriaminepentaacetic acid (DPTA) and 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) radiolabeled with trivalent radiometals have been developed and evaluated in clinical studies [8]. The MG analog DTPA-DGlu-Glu-Glu-Glu-Glu-Glu-Ala-Tyr-Gly-Trp-Met-Asp-Phe-NH<sup>2</sup> (DTPA-MG0), derived from human MG, displayed a high renal uptake hindering the therapeutic use [10]. With the removal of the penta-Glu sequence in the truncated MG analog DOTA-DGlu-Ala-Tyr-Gly-Trp-Met-Asp-Phe-NH<sup>2</sup> (DOTA-MG11), the renal uptake was efficiently reduced, however, so was the stability in vivo [11,12]. Thus, the clinical applicability of the peptide analogs developed so far is limited.

CCK2R-targeting peptide analogs are potential substrates of various enzymes such as the angiotensin converting enzyme (ACE), neutral endopeptidase (NEP), aminopeptidase A (APA) and cathepsins [13–17]. Extensive preclinical research was undertaken to improve the stability in vivo as well as the targeting properties. Different modifications were introduced in the linear peptide sequence of different CCK2R targeting peptide analogs, such as the incorporation of unnatural amino acids, inversion of the configuration of amino acids, cyclisation of the linear peptide and dimerization [18–21]. Most of these developments have not led to the required improvements necessary for successful clinical application. Two MG analogs, DOTA-(DGlu)6-Ala-Tyr-Gly-Trp-Met-Asp-Phe-NH<sup>2</sup> (PP-F11) labeled with indium-111 as well as DOTA-(DGlu)6-Ala-Tyr-Gly-Trp-Nle-Asp-Phe-NH<sup>2</sup> (PP-F11N) labeled with lutetium-177, which are derived from MG0 by inversion of the configuration of the penta-Glu motif, are currently examined in clinical studies (ClinicalTrials.gov Identifier: NCT03246659 and NCT02088645) [22–24]. Besides chemical modification of the peptide, in situ stabilization by co-injection of enzyme inhibitors was investigated. For <sup>111</sup>In-labeled DOTA-MG0 and DOTA-MG11, the use of phosphoramidone improved the tumor uptake, whereas for DOTA-MG0, a concomitant increase of renal retention occurred [25]. Sauter et al. studied the effect of two NEP inhibitors, phosphoramidone and thiorphan, on the targeting properties of <sup>177</sup>Lu-labeled DOTA-MG11, PP-F11 and PP-F11N. Only for DOTA-MG11 an improved tumor uptake could be achieved, whereas no improvement was found for PP-F11 and PP-F11N [14]. The results suggest that in situ stabilization is highly dependent on the individual radiopeptide and cannot be generalized. Besides that, the long-term use of protease inhibitors, especially NEP inhibitors, can potentially cause side effects that are not yet well understood [26].

In our recent studies we could introduce new modifications within the C-terminal sequence Trp-Met-Asp-Phe-NH2, known to be essential for CCK2R binding [27–29]. Most favorable properties were found for the new MG analog with the sequence DOTA-DGlu-Ala-Tyr-Gly-Trp-(*N*-Me)Nle-Asp-1Nal-NH<sup>2</sup> (DOTA-MGS5), in which methionine is replaced by *N*-methylated norleucine ((*N*-Me)Nle) and phenylalanine by 1-naphtylalanine (1Nal) [28]. Besides leading to highly improved stability in vivo, the introduced modifications also led to an enhanced receptor-specific cell uptake, as well as to a highly increased tumor uptake, when radiolabeled with different radiometals. In nude BALB/c mice bearing CCK2R-expressing tumor xenografts, a tumor uptake of more than 20% of the injected activity per gram (IA/g) was observed. This corresponds to a three-fold improvement compared to PP-F11 and PP-F11N (~6.7% and 6.9% IA/g) [14,28]. Nevertheless, some degradation products were still detected by radio-HPLC analysis of blood obtained from mice injected with <sup>177</sup>Lu-labeled DOTA-MGS5 [28].

With the aim of further improving the in vivo stability of DOTA-MGS5, we have explored additional modification of the peptide sequence. The amino acid proline (Pro) is a promising candidate for amino acid exchange and forms a tertiary amide bond which similarly to *N*-methylated peptide bonds and triazoles may improve the stability in vivo [27,30]. In this study, a preliminary preclinical characterization of the new MG analog DOTA-DGlu-Pro-Tyr-Gly-Trp-(*N*-Me)Nle-Asp-1Nal-NH<sup>2</sup> (**1**) was carried out focusing on the enzymatic stability of the <sup>177</sup>Lu-labeled radiopeptide in vivo. For this purpose, besides characterizing the stability in vitro in different media, metabolic stability studies were carried out in BALB/c mice giving first insights into the enzymatic degradation and biodistribution profile of this new radiolabeled MG analog. To characterize the <sup>177</sup>Lu-labeled metabolites formed in vivo, different possible metabolites of **1** were synthesized. Furthermore, the receptor affinity of the new MG analog, as well as selected metabolites, was studied in A431 human epidermoid carcinoma cells stably transfected with human CCK2R (A431-CCK2R). The same cell line was used to investigate the cell uptake of the <sup>177</sup>Lu-labeled peptide analog and selected radiometabolites. Mock-transfected A431 cells (A431-mock) were used as negative control.

#### **2. Results**

#### *2.1. Peptide Synthesis and Radiolabeling*

The amino acid sequence and chemical structure of **1** is displayed in Figure 1. **M1**–**M6** were synthesized by standard solid phase peptide synthesis starting from 100 mg of resin following the synthesis protocol described below. For **M7** and **M8,** which were synthesized by different strategies, conjugation with DOTA (DOTA-tris(tert-butyl) ester or DOTA mono-*N*-hydroxysuccinimide ester) was carried out in solution. After purification by reversed phase HPLC (RP-HPLC), characterization by mass spectrometry and lyophilization, the metabolites **M1**–**M8** were obtained with a purity >95% (with the exception of **M1** and **M4,** for which a purity of 93–94% was achieved). The amino acid sequences and analytical data for **1** and the different metabolites **M1**–**M8** are shown in Table 1.

## DOTA-DGlu-Pro-Tyr-Gly-Trp-(N-Me)Nle-Asp-1Nal-NH2 (**1**)

**Figure 1.** Amino acid sequence and chemical structure of **1**.


**Table 1.** Summary of the analytical data of **1** and the metabolites **M1**–**M8** (radio t<sup>R</sup> obtained after labeling with lutetium-177).

For experiments in vitro, labeling with lutetium-177 was carried out at a low molar activity of 10–20 MBq/nmol, yielding nearly quantitative labeling and allowing the use of the radiolabeled conjugates without further purification. The radiolabeled conjugates used in animal studies were radiolabeled at a higher molar activity of ~40 MBq/nmol. Hydrophilic impurities were removed by solid phase extraction (SPE) to obtain the radiolabeled peptides with a radiochemical purity of >99%. The radio-HPLC chromatograms of the radiolabeled compounds are shown in Figure 2.

**Figure 2.** Radio-HPLC chromatograms of [ <sup>177</sup>Lu]Lu-**1** and its <sup>177</sup>Lu-labeled metabolites **M1**–**M8** (\* indicating radiolabeled with lutetium-177).

#### *2.2. Characterization In Vitro*

The stability of <sup>177</sup>Lu-labeled **1** and **M1**–**4** in fresh human serum, as well as the stability of [177Lu]Lu-**1** in liver and kidney homogenates, was analyzed for up to 24 h after incubation. The percentage of intact radiopeptide found over time is presented in Figure 3. [ <sup>177</sup>Lu]Lu-**1** showed a very high stability in human serum with values of 96.2 ± 1.3% intact peptide after 24 h incubation. A higher degree of degradation was found in liver homogenate with >90% intact radiopeptide up to 20 min after incubation and decreasing to 83.6 ± 0.2, 52.9 ± 2.0 and 49.3 ± 1.7%, at 30 min, 2 h and 24 h, respectively. In kidney homogenate, a faster metabolic breakdown was observed. However, still 65.6 ± 2.8 and 43.9 ± 1.0% intact radiopeptide were present 20 min and 30 min after incubation. At later time points, the radiolabeled conjugate was completely degraded (<5 and <1% at 2 and 24 h after incubation, respectively). In human serum, a high stability was also found for [ <sup>177</sup>Lu]Lu-**M1** (89.1 ± 1.3%), [ <sup>177</sup>Lu]Lu-**M2** (98.1 ± 0.3%) and [ <sup>177</sup>Lu]Lu-**M4** (98.4 ± 0.01%) after 24 h incubation in human serum, whereas [ <sup>177</sup>Lu]Lu-**M3** showed a different behavior. A fast enzymatic degradation occurred in serum with only 29.7 ± 0.8% of intact [ <sup>177</sup>Lu]Lu-**M3** 24 h after incubation.

The logD values calculated from the octanol/PBS distribution of the different <sup>177</sup>Lu-labeled peptides resulted in a hydrophilicity profile in the order of [ <sup>177</sup>Lu]Lu-**M4** (−4.19 ± 0.30) > [ <sup>177</sup>Lu]Lu-**M2** (−4.18 ± 0.17) > [ <sup>177</sup>Lu]Lu-**M3** (−4.13 ± 0.11) > [ <sup>177</sup>Lu]Lu-**M1** (−4.07 ± 0.35) > [ <sup>177</sup>Lu]Lu-**1** (−1.96 ± 0.07).

Protein binding in human serum as analyzed by size exclusion chromatography was found to be lowest for [ <sup>177</sup>Lu]Lu-**M3** (23.4 ± 1.9%), followed by [ <sup>177</sup>Lu]Lu-**1** (37.8 ± 2.9%), [ <sup>177</sup>Lu]Lu-**M4** (41.7 ± 0.2%), [ <sup>177</sup>Lu]Lu-**M1** (45.4 ± 1.8%) and [ <sup>177</sup>Lu]Lu-**M2** (48.1 ± 2.3%) for the time point of 24 h after incubation.

**Figure 3.** Stability of (**a**) [ <sup>177</sup>Lu]Lu-**1** after incubation in human serum as well as in rat liver and rat kidney homogenates, and of (**b**) selected radiolabeled metabolites **M1**–**M4** in human serum, as analyzed up to 24 h after incubation.

#### *2.3. Cell Internalization and Receptor Binding Studies*

For [ <sup>177</sup>Lu]Lu-**1** incubated with A431-CCK2R cells, a high internalization with uptake values of 44.4 ± 2.7% after 1 h incubation (Figure 4) was observed. As expected, the <sup>177</sup>Lu-labeled metabolites showed no internalization into A431-CCK2R cells ([ <sup>177</sup>Lu]Lu-**M1**: 0.18 ± 0.03%, [ <sup>177</sup>Lu]Lu-**M2**: 0.06 ± 0.02%, [ <sup>177</sup>Lu]Lu-**M3**: 0.05 ± 0.03% and [ <sup>177</sup>Lu]Lu-**M4**: 0.05 ± 0.02%). Additionally, after 2 h incubation a very high receptor-specific uptake could be confirmed for [ <sup>177</sup>Lu]Lu-**1**, with values further increasing to 66.6 ± 0.3%, whereas no internalization occurred for the <sup>177</sup>Lu-labeled metabolites. Specificity of the cell uptake was proven by contemporaneous incubation in A431-mock cells finding a cell uptake of <0.2% and <0.4% for [ <sup>177</sup>Lu]Lu-**1** at 1 and 2 h, respectively.

**Figure 4.** Cell uptake of (**a**) [ <sup>177</sup>Lu]Lu-**1** in A431-CCK2R and A431-mock cells and (**b**) <sup>177</sup>Lu-labeled **M1**–**M4** in A431-CCK2R cells, after 1 h and 2 h incubation.

In competition assays against [Leu 15 ]gastrin-I substituted with iodine-125, a high binding affinity to CCK2R was confirmed for **1** (IC50: 0.69 ± 0.09 nM) on A431-CCK2R cells, comparable to pentagastrin used as a reference (IC50: 0.76 ± 0.11 nM) (Figure 5). For **M1**–**M4** no binding affinity could be observed, confirming the complete loss of receptor binding after removal of the C-terminal amide function and the C-terminal amino acids (*N*-Me)Nle, Asp and 1Nal. Based on these findings, the cell uptake and receptor binding of the remaining metabolites were not tested.

**Figure 5.** Competitive binding curves against [ 125 I][3-iodo-Tyr 12 ,Leu 15 ]gastrin-I for (**a**) non-labeled **1** in comparison with pentagastrin, as well as (**b**) non-labeled **M1**–**M4.**

#### *2.4. Stability In Vivo and Biodistribution Studies*

The metabolic stability in vivo, as shown in Figure 6, was monitored after intravenous injection of [ <sup>177</sup>Lu]Lu-**1** in BALB/c mice. A high resistance against enzymatic degradation was found at 10 min post injection (p.i.) with 80.5% intact radiopeptide present in blood. A high percentage of intact radiopeptide in blood was also observable after 30 and 60 min p.i. with values of 64.1% and 56.9%, respectively. Analysis of the urine of the mice showed that only 29.1%, 19.4% and 18.0% intact [ <sup>177</sup>Lu]Lu-**1** was excreted at the different time points studied. In the soluble phase extracted from liver homogenate, 71.9%, 53.9% and 47.7% intact peptide were found after 10, 30 and 60 min p.i., respectively, reflecting the stability of the radiolabeled conjugate during circulation. Metabolism during excretion was confirmed by the analysis of the soluble phase extracted from kidney homogenate with values of 29.7%, 18.4% and 7.0% intact radiopeptide after 10, 30 and 60 min p.i., respectively.

**Figure 6.** Intact [ <sup>177</sup>Lu]Lu-**1** detectable in blood and urine, as well as liver and kidney homogenates obtained from BALB/c mice, as analyzed up to 60 min p.i.

The preliminary evaluation of the biodistribution profile of [ <sup>177</sup>Lu]Lu-**1** showed a fast clearance from the blood with low non-specific uptake in most tissues. The observed uptake values are summarized in Table 2. The whole body activity ranged from 61.78% IA at 10 min p.i. to 12.27% IA at 1 h after injection. As shown in Figure 7a, the blood pool activity rapidly declined from 17.79% to 2.05% IA/g at 10 and 60 min, respectively. Consequently, a very low background activity was found in muscle with values ranging from 2.56% to 0.36% IA/g at 10 min and 60 min, respectively. Low non-specific uptake and rapid washout was observed also for lung (17.45% to 1.95% IA/g), heart (7.13% to 0.99% IA/g), femur (5.97% to 0.52% IA/g) and spleen (3.70% to 0.79% IA/g). Excretion occurred mainly through the kidneys with activity values decreasing from 14.49% to 7.01% IA/g from 10 to 60 min p.i. (Figure 7b). The activity values observed in liver (5.37% and 1.26% IA/g) and intestine (2.63% and 1.02% IA/g) for the same time points were much lower and comparable to the non-specific uptake in other organs. When looking at the washout of radioactivity from different tissues, a higher retention of radioactivity was observed for CCK2R-expressing stomach and pancreas, which was more prominent for stomach. The percentage decrease of radioactivity at 1 h versus 10 min p.i. was lower for stomach (59%) and pancreas (75%) in comparison to the washout of 79–91% from non-excretory organs (blood, lung, heart, femur, spleen and muscle), indicating a receptor-specific uptake in stomach and pancreas (Figure 7c, Table 2). No further blocking studies were performed to investigate the specificity of the uptake in more detail.

**Figure 7.** Uptake values in selected tissues as obtained from metabolic biodistribution studies with [ <sup>177</sup>Lu]Lu-**1** in BALB/c mice at 10, 30 and 60 min p.i.: (**a**) background activity in blood and muscle, (**b**) excretory organs kidney and liver as well as (**c**) CCK2R-expressing pancreas and stomach. Values are expressed as % IA/g.

e

s s


**Table 2.** Biodistribution profile of [177Lu]Lu-**1** in BALB/c mice after 10 min, 30 min and 1 h p.i. (30–40 MBq, 0.8 nmol). Values are expressed as % IA/g.

#### *2.5. Identification of the Radiometabolites Formed In Vivo*

For the evaluation of the radiometabolites formed in vivo, blood and urine collected from BALB/c mice injected with [177Lu]Lu-**1** were analyzed by radio-HPLC. In addition, the soluble phase extracted from kidney and liver homogenates was analyzed. [177Lu]Lu-**1** showed a high stability in blood against enzymatic degradation in vivo with a total amount of radiometabolites of 19.5%, 35.9% and 43.1% at 10, 30 and 60 min p.i., respectively. The radiometabolites observed could be matched to [177Lu]Lu-**M1**, [ <sup>177</sup>Lu]Lu-**M2**, [177Lu]Lu-**M3**, [177Lu]Lu-**M5**, with hydrolysis of the C-terminal amide, as well as cleavage of the peptide bonds of Asp-1Nal, (*N*-Me)Nle-Asp and Gly-Trp. The same metabolites were also observed in urine, whereas the presence of [177Lu]Lu-**M1** was negligible. However, much higher levels of the radiometabolites were observed in urine, which was to be expected, given the renal pathway as main route of excretion. A similar pattern was observed also for liver and kidneys, where only minor additional radiometabolites could be detected. [177Lu]Lu-**M6** with cleavage of the peptide bond of Tyr-Gly was confirmed for both, liver and kidneys, while [177Lu]Lu-**M4** cleaved between Trp and (*N*-Me)Nle was detectable only in liver and [177Lu]Lu-**M7** cleaved between Pro and Tyr only in kidneys. No cleavage between DGlu and Pro was observed, confirming a stabilizing effect for the tertiary amide bond introduced into the peptide backbone. Radiochromatograms of the different samples analyzed are displayed in Figure 8a–d. The percentage of the intact radiopeptide and of the different radiometabolites found for the different time points p.i. are summarized in Table 3.


**Table 3.** Quantification of the percentage of intact radiopeptide and of the radiometabolites as analyzed by radio-HPC of blood and urine, as well as liver and kidneys of BALB/c mice injected with [177Lu]Lu-**1** for different time points p.i.

**Figure 8.** Radiochromatograms obtained from (**a**) blood, (**c**) liver, (**d**) kidneys and (**b**) urine of BALB/c mice injected with [ <sup>177</sup>Lu]Lu-**1** for the time point of 60 min p.i.; radiochromatograms of <sup>177</sup>Lu-labeled peptides **1** and **M1**–**M8** are shown for comparison (\* indicating radiolabeled with lutetium-177).

#### **3. Discussion**

The strategy of CCK2R targeting with radiolabeled gastrin analogs for diagnostic and therapeutic application in patients with advanced tumors, in particular MTC, has been pursued for more than two decades. The successful clinical use of radiolabeled somatostatin analogs for targeting somatostatin receptors in patients with neuroendocrine tumors could not yet be translated to radiolabeled minigastrin analogs targeting CCK2R. In our recent studies, we could develop a new stabilization strategy leading to increased stability of the linear peptide sequence against enzymatic degradation in vivo and improving the targeting properties [28,29]. In this study, we have further explored our stabilization strategy by introducing an additional modification in the *N*-terminal region of the peptide backbone in our

lead compound DOTA-MGS5. Alanine in position 2 was replaced by proline leading to the new peptide analog **1**. Furthermore, different possible metabolites were synthesized and analyzed in comparative studies with the aim to further investigate the enzymatic degradation in vivo. Pro was selected as a promising candidate for substitution, as the cyclic structure of the side chain with its conformational rigidity may protect the peptide against enzymatic degradation. The insertion of Pro into proteins influences the formation of α-helices and ß-sheets in dependence of the molecular environment conferring specific features to protein structure and folding [31]. It has been shown that the single change from the l- to d-configuration of the Glu residues in MG analogs alters the secondary structure of the peptide leading to improved serum stability [32].

In this study, the preclinical properties of [177Lu]Lu-**1** were evaluated and the metabolites thereof formed during degradation in vivo were characterized to explore additional possible stabilization strategies in the development of CCK2R targeting radiopeptides. To enable the characterization of the radiometabolites, eight different metabolites of **1** were synthesized in moderate yields. For in vitro studies, **1** and the metabolites thereof were radiolabeled with [177Lu]LuCl<sup>3</sup> at low molar activity allowing quantitative radiolabeling at high radiochemical purity >95%. Receptor affinity assays with the non-labeled peptides as well as cell uptake studies with the peptide derivatives radiolabeled with lutetium-177 were performed using A431-CCK2R cells. The results confirmed loss of receptor affinity as well as cell uptake for different metabolites studied, while the intact peptide showed a retained high CCK2R affinity and improved receptor-mediated cell uptake of more than 60% of the total activity added. The specificity of the cell uptake was verified via lack of uptake in A431-mock cells used as a control cell line. Hydrolysis of the C-terminal amide was sufficient to cause complete loss of receptor affinity, proving that the interaction of the amidated C-terminus with the binding pocket is essential for maintaining receptor affinity [33]. When incubated in fresh human serum, used as a model in vitro to measure the resistance against enzymatic degradation, radiolabeled **1** as well as **M2** and **M4** showed the highest stability after 24 h incubation (>98%). Radiolabeled **M1** showed a slightly lower stability of 89%, whereas for **M3**, less than 30% intact radiopeptide was detectable after 24 h incubation, suggesting a high enzymatic susceptibility. This could be explained by the lower protein binding of [177Lu]Lu-**M3** leading to higher levels of free radiopeptide susceptible to proteases when compared to [177Lu]Lu-**1** and radiolabeled **M1**, **M2** and **M4**. Thus, increased protein binding may play an additive role in protecting [177Lu]Lu-**1** against enzymatic degradation. Highly improved stability of [177Lu]Lu-**1** was observed also in rat tissue homogenates in vitro. It has been shown that unsubstituted DOTA-MG11 labeled with indium-111 is highly susceptible to proteolytic digestion and rapidly and completely degraded within 30 min in liver homogenate and within 10 min in kidney homogenate [34]. For [177Lu]Lu-**1** still 84% and 44% intact radiopeptide could be observed in liver and kidney homogenate, respectively, after 30 min incubation. Only at the later time point of 2 h after incubation, an almost complete breakdown occurred in kidney homogenate, whereas in liver homogenate >50% intact radiopeptide was still present. The resistance against enzymatic degradation in rat liver homogenate was also improved when compared to DOTA-MGS5 labeled with different radiometals for which less than 30% intact radiopeptide was found for the same time point, suggesting a potential additive stabilizing effect of the insertion of Pro in position 2 [28].

Incubation in organ homogenates is connected with a higher breakdown of the radiopeptide due to the exposure to extracellular and intracellular proteases released after tissue homogenization, possibly leading to an underestimation of the stability in vivo [16]. Therefore, additional metabolic biodistribution studies in female BALB/c mice were performed to monitor the metabolites formed in vivo. After intravenous injection of [177Lu]Lu-**1** in BALB/c mice, a highly improved stability with more than 80% intact radiopeptide in blood was observed at 10 min after injection. The in vivo stability during circulation was tested for up to 1 h p.i. still finding 56.9% intact [177Lu]Lu-**1**. Much higher amounts of metabolites were present in the urine of the mice at the different studied time points, with the intact radiopeptide decreasing from 29% at 10 min to 18% at 1 h after injection. The metabolites found in liver homogenate resembled the metabolites observed in blood. The same correspondence

was found for the metabolites in urine and kidney homogenate. A more rapid metabolism of the radiopeptide was observed in kidneys and urine (<30%), whereas in blood and liver, a much higher stability was observed. With the optimized radio-HPLC gradient allowing for a better separation of the different radiometabolites, possibly additional radiometabolites could be monitored, which were not detectable in previous studies with <sup>177</sup>Lu-labeled DOTA-MGS5 [28]. The additional substitution with Pro did, however, not show a considerable effect on in vivo stability. Still, the in vivo stability of [ <sup>177</sup>Lu]Lu-**1** is highly improved when compared to other MG analogs which are currently investigated in clinical trials [22,24]. For PP-F11 labeled with indium-111, the metabolic stability in the blood of mice was tested for the time point of 5 min p.i. finding >70% intact radiopeptide [28,35]. However, when analyzing the in vivo stability for a later time point, we found that <sup>177</sup>Lu-labeled PP-F11 is almost completely degraded at 30 min p.i. [28].

To identify the observed degradation products, different metabolites were synthesized and radiolabeled with lutetium-177 for comparative radio-HPLC analysis. The retention times of the different radiometabolites were matched with the metabolites found in vivo. In the blood of mice at different time points of up to 1 h p.i., only minor amounts of radiolabeled **M1** (<10%), **M2** (<8%), **M3** (<21%) and **M5** (<10%) resulting from hydrolysis of the C-terminal amide, as well as cleavage of the peptide bonds of Asp-1Nal, (*N*-Me)Nle-Asp and Gly-Trp were found. The same metabolites, except **M1**, were detected in urine, with [177Lu]Lu-**M2** (~40%) and [177Lu]Lu-**M5** (~30%) being the most prominent. In liver, besides similar amounts of the metabolites found in blood, additionally, radiolabeled **M4** with cleavage of the peptide bond of Trp-(*N*-Me)Nle was detectable at very low concentration (<3%), as well as radiolabeled **M6** with cleavage between Tyr and Gly (~20%). This could reflect the higher enzymatic turnover in the liver. In kidneys, only 7% of intact radiopeptide was found at 1 h p.i. even though still 18% were present in the urine at the same time point, whereas much higher amounts of [177Lu]Lu-**M2** (12%), [177Lu]Lu-**M5** (24%) and [177Lu]Lu-**M6** (53%) were detected. Furthermore, minor amounts of [177Lu]Lu-**M7** (~4%) with cleavage of the peptide bond of Pro-Tyr were observed. Despite the different radiometabolites found, the new minigastrin analog, with 56.9% intact radiopeptide still present in the blood at 1 h after injection, shows a highly improved stability in vivo.

It is well known from the literature that different enzymes are involved in the metabolism of members of the gastrin/CCK family such as the angiotensin converting enzyme (ACE), endopeptidase (NEP), aminopeptidase A (APA) and cathepsins. ACE is a zinc- and chloride-dependent peptidyl dipeptidase, widely distributed throughout the body including the lungs, gastrointestinal tract, vascular endothelium and blood, with broad specificity and besides inactivating vasoactive peptides also acts on other bioactive peptides [16,36]. In the degradation process of CCK and gastrin analogs with eight or less amino acids, ACE initially cleaves the amidated C-terminal dipeptide Asp-Phe-NH<sup>2</sup> and releases a further C-terminal di- or tripeptide in a secondary step [13]. Thirteen amino acid long MG analogs containing the penta-Glu motif seem to be ACE-resistant [14]. It has however been reported that radiolabeled MG analogs derived from MG11 and MG0 are not cleaved by ACE and ACE inhibitors cannot prevent the degradation in vivo [14,15]. NEP is a zinc-dependent cell-surface enzyme with wide distribution in the body, including the presence on granulocytes and endothelial cells of the vasculature compartment, as well as in major organs such as liver, kidneys and gastrointestinal tract, and is involved in the degradation of many bioactive peptides [16,37]. In the degradation process of CCK and gastrin analogs, NEP cleaves the peptide at Asp-Phe, Trp-Met, Gly-Trp, as well as Ala-Tyr in the case of gastrin [38,39]. The fact that co-injection of <sup>111</sup>In-labeled DOTA-MG11 together with a NEP inhibitor clearly increased the amount of intact radiopeptide in the blood of mice and led to a significant improvement of tumor uptake, points out the importance of NEP in the metabolism of radiolabeled MG analogs [15]. Incubation in vitro with neprylisin-1 and neprylisin-2 confirmed cleavage at Asp-Phe and Gly-Trp for DOTA-MG11, whereas for DOTA-PP-F11, only cleavage at Asp-Phe was observed and DOTA-PP-F11N with Met replaced by Nle seemed to be resistant against neprylisins. Interestingly, NEP inhibition did not result in improved tumor uptake of DOTA-PP-F11 and DOTA-PP-F11N [14]. APA is a membrane-bound type II zinc metalloprotease with broad tissue distribution [16,40]. It cleaves *N*-terminal glutamyl and aspartyl residues and is known to be involved in the degradation of CCK-8. *N*-terminal modification of CCK8 analogs led to resistance against APA, suggesting that also radiolabeled peptide analogs with *N*-terminal conjugation of a bifunctional chelator are APA-resistant [41]. During intracellular trafficking also other proteases may be involved in the degradation process. Cathepsins are a group of enzymes whose primarily function is to act as intralysosomal enzymes and in addition to that are involved in cancer development and progression [17,42]. When analyzing the stability against different cathepsins in vitro, cleavage sites at Asp-Phe and Met/Nle-Asp have been confirmed for different DOTA-conjugated MG analogs [14].

[ <sup>177</sup>Lu]Lu-**M2** and [177Lu]Lu-**M5** with cleavage at Asp-1Nal and Gly-Trp, respectively, were present in all samples studies, indicating a major involvement of NEP in the metabolization of [177Lu]Lu-**1** in vivo. The presence of [177Lu]Lu-**M3** in almost all samples studies, as well as of [177Lu]Lu-**M6** in the soluble phase of the homogenates from liver and kidney, suggests that also an ACE-dependent metabolism could occur. Kolenc-Peitl et al. have also suggested two different degradation pathways for MG analogs, one pathway via ACE-like enzyme activity with cleavage at Met-Asp, Gly-Trp, Tyr-Gly and Ala-Tyr, and another pathway directly releasing gastrin-6 with cleavage at Ala-Tyr [43]. Interestingly, hydrolysis of the C-terminal amide resulting in [177Lu]Lu-**M1**, especially in blood and liver, was additionally observed, suggesting that other enzymes are also involved in the metabolism of [ <sup>177</sup>Lu]Lu-**1**, which have not yet been characterized. Our results are in contrast to Sauter et al. who investigated the susceptibility of different DOTA-conjugated MG analogs against various proteases in vitro [14]. Based on the radiometabolites found in the different tissue samples obtained from BALB/c mice injected with [177Lu]Lu-**1,** we could not confirm ACE resistance in vivo, and also replacement of Met by Nle did not result in liability against neprylisins. However, the fact that the radiometabolites [ <sup>177</sup>Lu]Lu-**M2** and [177Lu]Lu-**M3** were found in all investigated tissues supports a possible involvement of cathepsins in the degradation process. For [177Lu]Lu-**1**, only minor amounts of [177Lu]Lu-**M4** cleaved at (*N*-Me)Nle-Trp were observed in liver homogenate, indicating a stabilizing effect of the introduced *N*-methylated peptide bond. In a previous study, we could already show that single substitution with 1Nal in position 8 did not improve in vivo stability and only additional substitution with (N-Me)Nle in position 6 allowed to stabilize the linear peptide against enzymatic degradation [28,29]. When introducing 1,4-disubstituted 1,2,3-triazoles as metabolically stable bioisosteres in replacement of the amide bonds in Nle-substituted DOTA-MG11, Grob et al. also found the highest impact on the stability against proteases in human blood for the triazole insertion at Trp-Nle. The stability was not considerably further improved by additional triazole insertion at Ala-Tyr or Tyr-Gly. On the other hand, the insertion of triazoles at Tyr-Gly, Ala-Tyr, DGlu-Ala had a remarkable effect on tumor uptake, which was however clearly inferior to DOTA-MGS5 [28,30,44]. The introduction of an *N*-methylated peptide bond at Nle-Trp showed both effects of increased stability and improved receptor interaction. DOTA-MGS5 radiolabeled with different radiometals, showing the combined substitution of (*N*-Me)Nle and 1Nal in position 6 and 8, respectively, displayed a considerably higher cell uptake of >50%, when compared to ~25% observed for <sup>111</sup>In-labeled DOTA-MGS1 with single replacement of Phe by 1Nal [29]. The combination of increased stability and improved receptor interaction led to a considerable improvement in tumor uptake. In mice xenografted with A431-CCK2R cells a very high tumor uptake of more than 20%IA/g in combination with improved tumor-to-kidney ratio (4–6) was observed for DOTA-MGS5 labeled with indium-111, lutetium-177 or gallium-68 [28]. In the present study, very high receptor-mediated cell uptake in A431-CCK2R cells of >60% was also found for [177Lu]Lu-**1** showing additional substitution with Pro in position 2. Interestingly, the radiometabolite [177Lu]Lu-**M8** with cleavage at DGlu-Pro was not found in any of the tissues examined, indicating a possible additional stabilizing effect of the introduction of Pro. In a previous study analyzing the blood of patients injected with <sup>111</sup>In-labeled DOTA-MG11, high amounts of the short chain radiometabolites of DOTA-DGlu, DOTA-DGlu-Ala and DOTA-DGlu-Ala-Tyr were confirmed at 10 min after injection [12].

The preliminary biodistribution profile obtained for [177Lu]Lu-**1** from the metabolic studies in mice confirmed a rapid clearance from blood and low unspecific uptake in most tissues, together with predominant renal excretion. The prolonged retention of radioactivity in CCK2R-expressing stomach and pancreas indicates that [177Lu]Lu-**1** also shows high potential for targeting CCK2R-expressing tumors [45].

#### **4. Materials and Methods**

#### *4.1. Materials*

All commercially obtained chemicals were of analytical grade and used without further purification. No-carrier-added [177Lu]LuCl<sup>3</sup> produced from highly enriched <sup>176</sup>Yb was purchased from Isotope Technologies (Garching, Germany). Dr. Luigi Aloj kindly provided the A431 human epidermoid carcinoma cell line stably transfected with the plasmid pCR3.1 containing the full coding sequence for the human CCK2R, as well as the same cell line transfected with the empty vector alone [46]. A431-CCK2R and A431-mock were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% (*v*/*v*) fetal bovine serum and 5 mL of a 100× penicillin-streptomycin-glutamine mixture at 37 ◦C in a humidified 95% air/5% CO<sup>2</sup> atmosphere. Media and supplements were purchased from Invitrogen Corporation (Lofer, Austria). **1** was purchased from piCHEM (Raaba-Grambach, Austria).

#### *4.2. Peptide Synthesis*

The different metabolites of **1**, namely **M1**–**M6** shown in Table 1, were synthesized using 9-fluorenylmethoxycarbonyl (Fmoc) chemistry. The peptides were assembled on 2-chlorotritylchloride (2-CTC) resin with capacity 1.6 mmol/g (Iris Biotech GmbH, Marktredwitz, Germany). The reactive side chains of the amino acids were masked with the following protection groups: tert-butyl ester for Asp and DGlu, tert-butyl ether for Tyr, and tertbutyloxycarbonyl (BOC) for Trp. All coupling reactions were performed using a 5-fold excess of Fmoc-protected amino acids, 1-hydroxy-7-aza-benzotriazole (HOAt) and *O*-(7-Azabenzotriazole-1-yl)-*N*,*N*,*N*′ ,*N*′ -tetramethyluronium hexa-fluorophosphate (HATU) in dimethtylformamide (DMF) and pH adjusted to 8 with *N*,*N*′ -diisopropylethylamine (DIPEA). The resin was loaded with 30% of total capacity and the remaining binding sites were capped with methanol/DIPEA/dichloromethane (DCM) in a ratio of 200 µL/100 µL/2 mL for 30 min at room temperature. Between every conjugation step, the product was washed 6 times with DMF for 1 min. Removal of the Fmoc protecting groups was obtained by two consecutive treatments with 5 mL of 20% piperidin in DMF for 5 and 15 min each. For the coupling of DOTA, a 3-fold molar excess of DOTA-tris(tert-butyl ester), HOAt and HATU was used. Cleavage of the peptides from the resin with concomitant removal of acid-labile protecting groups was achieved by treatment with a mixture of trifluoroacetic acid (TFA), triisopropylsilane, and water in a ratio of 95/2.5/2.5 (*v*/*v*/*v*). The crude peptides were precipitated in ice-cold ether before HPLC purification and characterized by analytical HPLC (Dionex, Germering, Germany and matrix-assisted laser desorption/ionization time of flight mass spectroscopy (MALDI-TOF MS) (Bruker Daltonics, Bremen, Germany). Purification was performed by RP-HPLC on a GILSON 322 chromatography system with a GILSON UV/VIS-155D multi-wavelength UV detector, equipped with an Eurospher II 100-5 C18 A column, 250 × 8 mm (Knauer, Berlin, Germany) or an Eurosil Bioselect 300-5 C18 A column, Vertex Plus, 300 × 8 mm, combined with an Eurosil Bioselect 300-5 C18 precolumn, Vertex Plus A, 30 × 8 mm (Knauer, Berlin, Germany), using a water/ACN/0.1% TFA gradient.

Analytical HPLC was performed using an UltiMate 3000 chromatography system consisting of a variable UV-detector (UV-VIS at λ = 220 nm), a HPLC pump, an autosampler, a radiodetector (GabiStar, Raytest, Straubenhardt, Germany), equipped with a Phenomenex Jupiter 4 µm Proteo 90 Å C12 column, 250 × 4.6 mm (Phenomenex Ltd., Aschaffenburg, Germany) and analyzed with Chromeleon Dionex Software (Version 7.2.9.11323). The radiodetector was equipped with two different loops, a low-sensitivity loop of 5 µL and a high-sensitivity loop of 250 µL. Mass spectrometry was performed using a Bruker microflex benchtop MALDI-TOF MS with 200 shots per spot in reflector acquisition mode with a positive ion source. For mass determination, samples were prepared on α-cyano-4-hydroxycinnamic acid (HCCA) matrix using dried droplet procedure. Flex Analysis 2.4 software was used to analyze the recorded data. HPLC chromatograms and MS spectra are presented in the supporting information (Figures S1 and S2). The lyophilized peptide derivatives were stored at −20 ◦C.

**M7** and **M8** were obtained using other strategies. For the synthesis of **M7**, the dipeptide DGlu-Pro was synthesized on 2-CTC resin following the synthesis protocol described above defining the amino acid sequence. After cleavage from the resin, precipitation in ice-cold ether, purification by HPLC and lyophilization, DGlu-Pro (0.015 g, 0.061 mmol) was transferred to a round-bottomed flask, dissolved in 1 mL ACN and pH adjusted to 8 with 25 µL DIPEA. DOTA-*N*-hydroxysuccinimid ester (0.023 g, 0.030 mmol) was added to the solution and the mixture was stirred at room temperature overnight. After evaporation of the solvent, the product was dissolved in 2 mL H2O, purified by HPLC, lyophilized and characterized by HPLC and MALDI-TOF MS. For the synthesis of **M8**, to a solution of d-glutamic acid (0.30 g, 2.04 mmol) in 6.25 mL dry methanol, 0.89 mL distilled thionyl chloride (1.46 g, 12.23 mmol) was added at 0 ◦C over 30 min. Then, the reaction mixture was stirred for 12 h at room temperature. The solvent was evaporated under reduced pressure, diluted with saturated NaHCO3, and extracted with dicloromethane (3 × 200 mL). The organic layer was washed with H2O, followed by brine, dried over Na2SO<sup>4</sup> and filtered. After evaporation of the solvent, the desired d-glutamic acid dimethyl ester was obtained as a yellowish oil (0.315 g, 1.79 mmol, 87% yield) [47]. NMR spectrum was measured on a 400 MHz Avance 4 Neo (Bruker) spectrometer. As solvents for NMR deuterated chloroform (CDCl3) was used (Euriso-top®). The chemical shifts (δ) were referenced to tetramethylsilane or the solvent peak and were given in parts per million (ppm). Coupling constants (J) were reported in Hertz (Hz). The following descriptors for signals were used: s = singlet, t = triplet, q = quartet, m = multiplet. <sup>1</sup>H-NMR (400 MHz, CDCL<sup>3</sup> δ ppm): 3.66 (s, 3H, CH3), 3.61 (s, 3H, CH3), 3.43–3.39 (q, 1H, CH), 2.43–2.39 (t, 2H, CH2), 2.04– 1.75 (m, 2H, CH2) (Figure S3). To d-glutamic dimethylester (0.030 mg, 0.174 mmol) in a round-bottomed flask a mixture of DOTA-tris(tert-butylester) (0.05 g, 0.087 mmol), HATU (0.066 g, 0.174 mmol), HOAt (0.024 g, 0.174 mmol) in 3 mL DCM, adjusted to pH 8 with 35 µL DIPEA, was added and the solution was stirred overnight at room temperature. The reaction solution was evaporated and the protecting groups were removed by adding 3 mL 50% TFA in DCM at 60 ◦C for 24 h. After evaporation, the crude product was dissolved in 2 mL 50% ACN. After HPLC purification and lyophilization, the final product was characterized by HPLC and MALDI-TOF MS.

#### *4.3. Radiolabeling and Characterization In Vitro*

For labeling with lutetium-177, the DOTA-peptides (8–12 µg in 10 µL) were incubated with 10–30 µL [177Lu]LuCl<sup>3</sup> solution (60–300 MBq in 0.05 M HCl) and a >1.2-fold volume of 0.4 M sodium acetate/0.24 M gentisic acid solution pH adjusted to 5 at 90 ◦C for 20 min. Radiochemical purity of the radiopeptides was analyzed using the analytical HPLC system described above using a flow rate of 1 mL/min together with the following water/ACN/0.1% TFA gradient: 0–3 min 10% ACN, 3–18 min 10–55% ACN, 18–20 min 55–80% ACN, 20–21 min 80–10% ACN, 21–25 min 10% ACN. [177Lu]Lu-**1** used in animal studies was purified by solid phase extraction (SPE). For this purpose, a C18 SepPak tLight cartridge (Waters, Milford, MA) was pretreated with 5 mL 99% ethanol followed by 5 mL 0.9% saline. The radiolabeling solution was passed through the cartridge and washed with 5 mL 0.9% saline to elute hydrophilic impurities. The radiolabeled peptide was eluted with 50% ethanol from the cartridge and diluted with 0.9% saline.

For the determination of the distribution coefficient (LogD) in octanol/PBS, the radiolabeled DOTA-peptides (100 pmol) in 500 µL PBS (pH 7.4) were added to 500 µL octanol in an Eppendorf microcentrifuge tube (*n* = 8). The mixture was vigorously vortexed at room temperature over a period of 15 min using a small shaker (MS3 Basic, IKA, Staufen, Germany) with speed of 1500 rpm. After a waiting time of 10 min sufficient for the separation of the two phases, 100 µL aliquots of both layers were measured in a gamma counter (2480 Wizard2 3", PerkinElmer Life Sciences and Analytical Sciences, formerly Wallac Oy, Turku, Finland) and the logD value was calculated.

For protein binding assessment, the radiolabeled DOTA-peptides were incubated in fresh human serum at 37 ◦C (500 pmol/mL, *n* = 2). After 1, 4 and 24 h of incubation, two samples were taken for each time point and analyzed by Sephadex G-50 size-exclusion chromatography (GE Healthcare Illustra, Little Chalfont, UK). The percentage of protein binding was determined by measuring the column and the eluate with the gamma-counter.

In vitro stability studies for the characterization of the metabolic stability of the radiolabeled peptide analogs in human serum were carried out with [177Lu]Lu-**1**, [177Lu]Lu-**M1**, [177Lu]Lu-**M2**, [ <sup>177</sup>Lu]Lu-**M3** and [177Lu]Lu-**M4** (*n* = 2). Additional stability studies in rat liver and kidney homogenates were performed for [177Lu]Lu-**1** (*n* = 2). Tissue homogenates were prepared from dissected organs by homogenization for 1 min at RT (IKA-Werke, Staufen, Germany) in 20 mM HEPES buffer pH 7.3 (30% *v*/*v*). The radioligands were incubated in the different media at a concentration of 500 pmol/mL (corresponding to an activity of 6–11 MBq). After incubation at 37 ◦C at different time points for up to 24 h, a 100 µl sample was taken in duplicates and analyzed by HPLC. Samples obtained from human serum and rat homogenates were treated with ACN at a ratio of 1:1.5 (*v*/*v*) to precipitate proteins, centrifuged (14,000 rpm, 2 min, centrifuge 5424, Eppendorf AG, Germany) and diluted with water at a ratio of 1:1 (*v*/*v*). A 100 µl aliquot of this solution was injected into the radio HPLC system. The degradation of the radioligand was evaluated based on the radiochemical purity after radiolabeling and the percentage of intact radiopeptide observed during incubation in the different media.

#### *4.4. Receptor Binding and Cell Uptake Studies*

The binding affinity of **1,** pentagastrin and of the different metabolites for the CCK2R was tested in a competition assay against [125I][3-iodo-Tyr12,Leu15]gastrin-I. Radioiodination of gastrin-I was carried out using the chloramine-T method. [125I][3-iodo-Tyr12,Leu15]gastrin-I at high molar activity was obtained by HPLC purification and stored in aliquots at −25 ◦C. Binding assays were carried out using 96-well filter plates (MultiScreenHTS-FB, Merck Group, Darmstadt, Germany) pre-treated with 10 mM TRIS/139 mM NaCl pH 7.4 (2 × 250 µL). For the assay, 200,000–400,000 A431-CCK2R cells per well were prepared in 20 mM HEPES buffer pH 7.4 containing 10 mM MgCl2, 14 µM bacitracin and 0.5% BSA. The cells (*n* = 3) were incubated with increasing concentrations of the peptide conjugates (0.0003–30,000 nM) and [125I][3-iodo-Tyr12,Leu15]gastrin-I (50,000 cpm) for 1 h at RT. Incubation was interrupted by filtration of the medium and rapid rinsing with ice-cold 10 mM TRIS/139 mM NaCl pH 7.4 (2 × 200 µL) and the filters were measured in the gamma-counter. Half maximal inhibitory concentration (IC50) values were calculated following nonlinear regression with Origin software (Microcal Origin 6.1, Northampton, MA, USA).

For internalization experiments with [177Lu]Lu-**1** and its metabolites, A431-CCK2R and A431-mock cells were seeded at a density of 1.0×10<sup>6</sup> per well in 6-well plates (Greiner Labortechnik, Kremsmuenster, Austria)and grown for 48 h until reaching almost confluence. On the day of the experiment, the medium was replaced by 1.2 mL of fresh DMEM medium supplied with 1% (*v*/*v*) fetal bovine serum. The cells (*n* = 3) were incubated with ~25,000 cpm of the radioligand in 300 µL PBS/0.5% BSA in a total volume of 1.5 mL, corresponding to a final concentration of 0.4 nM of total peptide. After 1 h and 2 h incubation, the cell uptake was interrupted by removal of the medium and rapid rinsing with 2 × 1 mL PBS/0.5% BSA. Thereafter, the cells were incubated twice at ambient temperature with acid wash buffer (50 mM glycine buffer pH 2.8, 0.1 M NaCl) for 5 min, to remove the membrane-bound radioligand. Finally, the cells were lysed by treatment in 1 N NaOH and collected (internalized radioligand fraction). All fractions were counted in the gamma-counter and mean values were calculated. The internalized fraction was expressed in relation to the total radioactivity added to the cells. In an additional experiment, the cell uptake of [177Lu]Lu-**M1**, [177Lu]Lu-**M2**, [177Lu]Lu-**M3** and [177Lu]Lu-**M4** was investigated in A431-CCK2R cells for the time point of 1 and 2 h incubation.

#### *4.5. Metabolic Biodistribution Studies and Characterization of the Radiometabolites In Vivo*

Metabolic biodistribution studies were performed in accordance with the ethical standards of the institution and approved by the Austrian Ministry of Science (BMWFW-66.011/0072-V/3b/2019). These studies were carried out in 5- to 7-week-old female BALB/c mice (*n* = 3) injected with [177Lu]Lu-**1** via a lateral tail vein. To allow monitoring of the metabolites by radio-HPLC, mice were injected with 30 MBq corresponding to 0.8 nmol total peptide. The mice were euthanized by cervical dislocation after different time points of 10 min, 30 min and 1 h p.i. and the urine and a venous blood sample were collected at the time of sacrifice. Liver and kidneys were dissected and homogenized in 20 mM HEPES buffer pH 7.3 at a ratio of 1:1 (*v*/*v*) with an Ultra-Turrax T8 homogenizer (IKA-Werke, Staufen, Germany) for 1 min at RT. Before radio-HPLC analysis, the samples (except urine) were treated with ACN at a ratio of 1:1 (*v*/*v*) to precipitate proteins, centrifuged at 2000× *g* for 2 min and diluted with water at a ratio of 1:1 (*v*/*v*). For HPLC analysis of the samples, the analytical HPLC system described above with a flow rate of 1 mL/min was used together with the following optimized water/ACN/0.1% TFA gradient to allow a better separation of [177Lu]Lu-**1** and the different radiometabolites: 0–7 min 1–7% ACN, 7–8 min 7–10% ACN, 8–18 min 10–18% ACN, 18–20 min 18–32% ACN, 20–27 min 32% ACN, 27–37 min 32–0% ACN, 37–40 min 80% ACN, 40–40.1 min 80–1% ACN, 40.1–45 min 1% ACN. Urine was measured with the low-sensitivity loop of the radiodetector. The remaining samples were analyzed with the high-sensitivity loop due to the lower radioactivity present in blood, liver and kidney homogenate. To quantify the percentage of intact [177Lu]Lu-**1** and [177Lu]Lu-**M1** showing some overlap in the radiochromatogram when using the high-sensitivity loop of the radiodetector, the two radiopeptides were co-injected at different known ratios to enable a more accurate separation and integration of the two peaks (see supporting Figure S4). Mice from the metabolic biodistribution studies were subjected to a further dissection of all remaining tissues (blood, lung, heart, femur, spleen, muscle, intestine, pancreas and stomach). All organs, also including liver and kidneys, were weighed, and their radioactivity was measured in the gamma counter together with a standard and the residual body. To quantify the uptake of radioactivity in liver and kidneys, a part of liver and kidney homogenate was measured in the gamma counter to extrapolate the radioactivity for the whole organ.

#### **5. Conclusions**

Radiolabeled MG analogs with the modified receptor-specific C-terminal sequence Trp-(*N*-Me)Nle -Asp-1Nal-NH<sup>2</sup> are promising new candidates for diagnostic and therapeutic use in patients with advanced MTC and other CCK2R-expressing malignancies. [177Lu]Lu-**1** with introduction of an additional tertiary peptide trough substitution with Pro in position 2, shows a highly improved stability against enzymatic degradation in vivo. From the radiometabolites identified in the blood of mice injected with [177Lu]Lu-**1** hydrolysis of the C-terminal amide and cleavage of the peptide bonds of Asp-1Nal, (*N*-Me)Nle-Asp and Gly-Trp were found to occur in vivo. The high receptor-mediated cell uptake and favorable biodistribution profile in normal BALB/c mice support further studies evaluating the tumor targeting potential of [177Lu]Lu-**1** and other alternative derivatives thereof.

**Supplementary Materials:** The following are available online, Figure S1: Representative UV-chromatogram of the metabolites **M1**–**M8**; Figure S2: MALDI-TOF-MS of the different synthesized metabolites **M1**–**M8**; Figure S3: 400 MHz 1H NMR of D-glutamic acid dimethyl ester; Figure S4. Radiochromatograms of [177Lu]Lu-**1** and [ <sup>177</sup>Lu]Lu-**M1** co-analyzed in different ratios of approximately (a) 1:1, (b) 2:1 and (c) 10:1 (*v*/*v*) using the radiodetector equipped with the high sensitivity loop (250 µL)

**Author Contributions:** Conceptualization, Supervision, Project Administration and Funding Acquisition, E.v.G. and S.S.; Methodology and Investigation, A.A.H., M.K., M.R., R.G., N.H. and E.v.G.; Writing—Original Draft Preparation: A.A.H.; Writing—Review & Editing, E.v.G. and M.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was financially supported by the Austrian Science Fund (FWF), project P 27844.

**Acknowledgments:** Christine Rangger and Joachim Pfister are greatly acknowledged for technical assistance in cell culture and animal studies. Open Access Funding is supported by the Austrian Science Fund (FWF).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **PET Radiotracers for CNS-Adrenergic Receptors: Developments and Perspectives**

#### **Santosh Reddy Alluri <sup>1</sup> , Sung Won Kim <sup>2</sup> , Nora D. Volkow 2,3,\* and Kun-Eek Kil 1,4,\***


#### Academic Editor: Krishan Kumar

Received: 31 July 2020; Accepted: 1 September 2020; Published: 3 September 2020

**Abstract:** Epinephrine (E) and norepinephrine (NE) play diverse roles in our body's physiology. In addition to their role in the peripheral nervous system (PNS), E/NE systems including their receptors are critical to the central nervous system (CNS) and to mental health. Various antipsychotics, antidepressants, and psychostimulants exert their influence partially through different subtypes of adrenergic receptors (ARs). Despite the potential of pharmacological applications and long history of research related to E/NE systems, research efforts to identify the roles of ARs in the human brain taking advantage of imaging have been limited by the lack of subtype specific ligands for ARs and brain penetrability issues. This review provides an overview of the development of positron emission tomography (PET) radiotracers for in vivo imaging of AR system in the brain.

**Keywords:** adrenergic receptor; positron emission tomography; radiotracer

#### **1. Introduction**

Positron emission tomography (PET) is a noninvasive and highly sensitive in vivo imaging technique that uses small amounts of radiotracers to detect the concentration of relevant biomarkers in tissues such as receptors, enzymes, and transporters. These radiotracers can be used to characterize neurochemical changes in neuropsychiatric diseases and also to measure drug pharmacokinetics and pharmacodynamics directly in the human body including the brain [1,2]. PET technology requires positron-emitting radioisotopes, a radiotracer synthesis unit, a PET scanner, and data acquisition components. Among the PET isotopes, cyclotron produced carbon-11 (C-11, *t*1/<sup>2</sup> = 20.34 min), nitrogen-13 (N-13, *t*1/<sup>2</sup> = 9.96 min), and fluorine-18 (F-18, *t*1/<sup>2</sup> = 109.77 min) are frequently used for PET-neuroimaging. PET imaging is particularly valuable to characterize investigational drugs and their target proteins, providing a valuable tool for clinical neuroscience [3].

PET radiotracers for the central nervous system (CNS) should have proper pharmacokinetic profile in the brain and to achieve this property, they were designed to meet five molecular properties: (a) molecular weight <500 kDa, (b) Log D7.4 between ~1 to 3 (lipophilicity factor), (c) number of hydrogen bond donors <5, (d) number of hydrogen bond acceptors <10, and (e) topological polar surface area <90 Å<sup>2</sup> . Otherwise, they may either not cross the blood-brain barrier (BBB) or show lack of specific signal due to high nonspecific binding [3–5]. In addition, an ideal PET radiotracer should have, though unachievable, (a) high affinity (preferably subnanomolar range) for its target, (b) high selectivity between subtypes or void on off-targets, (c) high dynamic range in specific

binding, (d) appropriate metabolic profile, (e) no adverse toxicology effects, and (f) kinetics suitable for mathematical modelling, which requires fast transfer to BBB and clearance in non-target tissues [3]. In general, a radiotracer with binding potential [ratio of target density (*B*max) to ligand's affinity (*K*<sup>d</sup> or *K*<sup>i</sup> )] value of at least 10 is expected to provide a reliable specific signal in vivo [6,7]. Furthermore, there are some technical challenges associated with CNS PET radiotracer development, which include: limited synthesis time and/or complex syntheses to prepare a radiotracer, attainment of high molar activity, accurate radiometric metabolite analysis. In addition, the radiotracer ought to ideally address targets that are relevant to brain function and for the diagnosis and therapy of a disease [3,8]. Epinephrine (E) and norepinephrine (NE) were discovered in 1894 and 1907, respectively. They are both neurotransmitters and hormones that belong to the group of catecholamines crucial to the function of the body and brain [9–11]. Neither E nor NE crosses the BBB but they are synthetized in the brain. NE is synthesized in synapse from *L*-phenylalanine in four steps enzymatically through phenylalanine hydroxylase (PAH), tyrosine hydroxylase, dopa decarboxylase (DDC), and dopamine beta-hydroxylase (DBH), respectively. Further, NE is methylated to convert into E with phenylethanolamine-*N*-methyl transferase (PNMT) [12–14]. NE in synaptic cleft is removed either by reuptake via NE transporter (NET) or via metabolism by monoamine oxidase A (MAO-A) or catechol-*O*-methyl transferase (COMT) into various transitional metabolites. Though the noradrenergic neurons are confined to a few relatively small brain areas such as midbrain, pons, locus coeruleus, caudal ventrolateral nucleus, and medulla, they send extensive projections to most brain regions [15]. Table 1 summarizes the brain distribution (reported in mice) and function of various AR subtypes and their involvement in some brain disorders for CNS-ARs.

The effects of NE and E in the CNS and PNS is mediated mainly through two main classes of adrenergic receptors (ARs): alpha-ARs (α-ARs) and beta-ARs (β-ARs) [16]. The ARs were first identified in 1948 and pharmacological and molecular cloning techniques since then have identified various subclasses of ARs [17,18]. α-ARs are divided into α1 and α2 subclasses, wherein, each of these has three subtypes: α1A, α1B, α1D and α2A, α2B, α2C. The three subtypes of β-ARs are β1, β2, and β3 [19]. These receptors in the CNS are G-protein coupled receptors (GPCR) and are implicated in pathophysiology of various diseases, biochemical pathways, and biological functions [18].


**Table 1.** Brain distribution of AR subtypes and their associated brain disorders.

Given its biological significance, the adrenergic system has emerged as an important target for PET studies. Although ARs play major roles in the brain, most PET studies of ARs have focused on cardiac imaging [49,50] and PET studies of CNS-ARs are very limited. The other components of adrenergic system such as NET and MAO-A were studied using PET. Various C-11 and F-18 radiotracers of NET-selective anti-depressants (e.g., reboxetine) were examined using PET to monitor the function of noradrenergic system in CNS. Radiotracers, for example, (*S,S*)-[11C]*O*-methyl-reboxetine ([11C]MRB) and [18F]FMeNER-D2, were shown to exhibit desirable in vivo properties and their regional distribution in the brain is consistent with known distribution of NET in preclinical/clinical settings [51–54]. These radiotracers have been employed to monitor NET availability in different related diseases, including obesity, major depressive disorder, and Parkinson's disease. Likewise, radiotracers, for instance, [ <sup>11</sup>C]Harmine demonstrated clinical success for in vivo brain imaging of MAO-A, respectively [55–57]. [ <sup>11</sup>C]Harmine has been applied in several-PET neuroimaging studies to study the role of MAO-A in different pathological conditions, including nicotine/alcohol dependence, and Alzheimer's disease.

However, the metabolic functions of MAO-A do not necessarily reflect the activities of adrenergic system as they also metabolize other neurotransmitters, such as, dopamine and serotonin (5-HT). Therefore, the in vivo activities of adrenergic neurons, thus far, were exclusively examined by PET imaging based on NET. Since the activities of NET reflect only on presynaptic systems of adrenergic neurons, novel PET radiotracers that can observe the features of postsynaptic adrenergic system are still required. PET-neuroimaging of AR subtypes combined with NET can scrutinize the unique implications of adrenergic system in various pathophysiological conditions of the brain. Subtype specific PET radiotracers for CNS-ARs have the potential to help clarify the roles of ARs in brain pathophysiology and provide suggestions towards the diagnosis and treatment for diseases such as depression, attention deficit hyperactivity disorder (ADHD), substance use disorders, schizophrenia, and neurodegenerative diseases that involve ARs. This review, based on reports published through 2019, summarizes the development of PET radioligands for CNS-ARs in animal models and human subjects and presents suggestions for further development for CNS-AR radiotracers.

#### **2.** α**1-AR PET Radiotracers**

Three highly homologous subunits of α-1 ARs, α1A, α1B, and α1D, have been shown to have different amino acid sequence, pharmacological properties, and tissue distributions [58–60]. A detailed review of α1-AR pharmacology is given by Michael and Perez [58]. Binding of NE/E to any of the α1-AR subtypes is stimulatory and activates Gq/11-signalling pathway, which involves phospholipase C activation, generation of secondary messengers, inositol triphosphate and diacyl glycerol and intracellular calcium mobilization. In PNS, as well as in the brain's vascular system it results in smooth muscle contraction and vasoconstriction. While the signaling effects of α1-ARs in the cardiovascular system are well studied [61,62], the role of α1-ARs in CNS is complex and not completely understood. The Bmax values of α-1 ARs were measured from saturation assays using [3H]prazosin (a selective α-1 blocker) with tissue homogenates from rats and the observed binding capacities (fmol/mg tissue) of prazosin in cortex, hippocampus, and cerebellum were 14.49 ± 0.38, 11.03 ± 0.39, and 7.72 ± 0.11, respectively [23,63]. The α1-ARs are postsynaptic receptors and can also modulate release of NE. In the human brain, α1-AR subtypes are localized in amygdala, cerebellum, thalamus, hippocampus, and to some extent in striatum [20,22,64].

The anti-depressant effects of noradrenergic enhancing drugs as well as their effects on anxiety and stress reactivity points to their relevance of α1A, α1D to these behaviors [21,28,65]. Furthermore, α1A-ARs regulate GABAergic and NMDAergic neurons [20]. A decrease in α1A-AR mRNA expression was observed in prefrontal cortex of subjects with dementia [66,67]. The α1B-ARs has also been shown to be crucial to brain function and disease [18]. For example, α1B-knockout (KO) mice study revealed that α1B-AR modulates behavior, showing increased reaction to novel situations [24,27]. In addition, the locomotor and rewarding effects of psychostimulants and opiates were decreased in α1B-KO mice, highlighting their role in the pharmacological effects of these drugs [24]. On the other

α

hand, studies using the α1B-overexpression model suggest their involvement in neurodegenerative diseases [25]. α

α

Several α1-AR agonists and antagonists (α1-blockers) are available in the market as drugs to treat various heart and brain disorders [68,69]. Pharmacology of most of these drugs is complicated by the fact that they have strong affinities for other receptor systems, such as serotonin and dopamine receptors. The need to develop α1-AR selective/α1-AR subtype specific drugs is demanding. Undoubtedly, PET radiotracers selective for α1-AR are valuable to assess α1-ARs contribution to brain function and disease. In the late 1980s, prazosin, used for the treatment of hypertension, was labelled with carbon-11 to image α1-ARs with PET [49]. Following this, [11C]prazosin analogous, [11C]bunazosin and [11C]GB67 (Figure 1A) were developed as PET radiotracers to image α1-ARs in the cardiovascular system [49,70]. These tracers were shown to have limited BBB permeability and were deemed to be not suitable for PET-neuroimaging. Efforts to develop PET radiotracers to image α1-ARs in the CNS was mainly based on antipsychotic drugs such as clozapine, sertindole, olanzapine, and risperidone that have mixed binding affinities for D2, 5HT2A receptors and α1-ARs in the nanomolar range. Their affinity for α1-ARs has been shown to contribute to antipsychotic efficacy uncovering their role in psychoses [71,72]. α α α α α α α α α α α

α α **Figure 1.** (**A**) Earlier PET radiotracers, [11C]Prazosin, [11C]Bunazosin, and [11C]GB67 for cardiac α1-AR imaging. (**B**) Antagonist PET radiotracers based on sertindole. (**C**) Antagonist PET radiotracers based on octoclothepin for brain α1-AR imaging.

α α μ μ α Two C-11 labelled analogs of the atypical antipsychotic drug sertindole were first reported by Balle et al. In the early 2000s (Figure 1B) [73,74]. Several other analogs with good affinities (*K*<sup>i</sup> < 10 nM) for α1-ARs were subsequently reported by the same group. Two analogs, in which chlorine in sertindole (**1**) is replaced by a 2-methyl-tetrazol-5-yl (**2**) and a 1-methyl-1,2,3-triazol-4-yl (**3**), were labelled with C-11 [73,74]. The in vitro affinities (*K*<sup>i</sup> ) of **1**, **2** and **3** for α1-ARs are 1.4, 1.8 and 9.5 nM, respectively. Both [11C]**2** and [11C]**3** were prepared using <sup>11</sup>C-methylation with [11C]methyl triflate from their corresponding *N*-desmethyl precursors. The molar activity of [11C]**2** and [11C]**3** was reported at 70 GBq/µmol and 15 GBq/µmol, respectively. Their brain distribution examined with PET in Cynomolgus monkeys showed that brain uptake of [11C]**2** and [11C]**3** was slow and low, with [11C]**3** showing somewhat higher brain uptake than [11C]**2**. Their brain distribution was homogenous and specific binding to α1-ARs could not be demonstrated. It was also concluded that these two radiotracers were not suitable to image α1-ARs in brain owing to rapid metabolism, substantial distribution to other organs, and substrates for active efflux mechanism.

In the early 2010s, further optimization of structure-activity relationship (SAR) studies led to the identification of two more sertindole analogs, **4** and **5** (Lu AA27122), with higher α1-AR selectivity over D<sup>2</sup> receptors (Figure 1B). Compared to the in vitro affinity for α1A-AR subtype (*K*<sup>i</sup> = 0.16, 0.52 nM for **4** and **5**), the affinity for the α1B and α1D subtypes is 4 to 15 times less potent (Table 2). The LogD7.4 values for **4** (2.7) and **5** (1.9) were in the optimal range for in vivo brain imaging [75–77]. Both [11C]**4** and [ <sup>11</sup>C]**5** were prepared in a similar manner like above with >370 GBq/µmol molar activity for non-human primate studies [75]. Interestingly, while [11C]**4** showed very poor brain uptake, [11C]**5** had suitable brain uptake (4.6% ID/cc at 36 min). However, its binding was not blocked with a pharmacological dose of prazosin pretreatment, indicating lack of α1-AR specificity.


**Table 2.** In vitro affinities of compounds **1** to **7** for α1-ARs, D<sup>2</sup> and 5HT2C receptors [73,75,76].

At the same time, another C-11 labelled radiotracer, [11C]**7** ([11C]Lu AE43936, Figure 1C) for brain α1-ARs was developed and evaluated by Risgaard et al. [76]. This radiotracer was based on the antipsychotic octoclothepin (**6**, Figure 1C), which belongs to the tricylic dibenzotheiepin group and has inverse agonist effects at dopamine, serotonin, and α1 receptor sites [78]. Two enantiomers of **7** (*R*/*S*) were radiolabelled on the basis of their varying selectivity and specificity for α1-AR subtypes (Table 2) and imaged with PET in female Danish Landrace pigs. The baseline PET imaging results indicated that neither of the radiolabelled isomers entered the pig's brain. Pretreatment with cyclosporin A (CsA) [79] increased the brain uptake of (*R*)-**7** in α1-AR rich cortex, thalamus (above 2 SUV), suggesting that (*R*)-**7** was a substrate for active efflux transporters. Further cell studies specified that (*R*)-**7** is a substrate for p-glycoprotein (Pgp).

So far, none of the reported radiotracers showed promising results for in vivo brain imaging of α1-ARs for they were limited by poor BBB penetration, being substrates of Pgp or lack of binding specificity. Therefore, novel α1-AR radiotracers that overcome brain permeability and display good affinity and subtype-selectivity are required to evaluate the role of α1-AR subtypes in brain pathophysiology. Several α1-AR subtype specific (high affinity for one subtype over the other) and nonspecific compounds have been reported over the years [26,29,80,81]. Either direct radiolabeling (if possible) or chemical modification and then radiolabeling of the most specific compounds could be an optimal approach for developing new α1-subtype specific radiotracer, considering the aforementioned CNS-PET radiotracer criteria.

#### **3.** α**2-AR Subtype and Nonspecific PET Radiotracers**

Unlike α1-ARs, α2-ARs decrease adenyl cyclase activity in association with G<sup>i</sup> heterotrimeric G-protein and hence are inhibitory. They mediate many NE effects including cognition and readiness for action [11,30]. The brain distribution of the three subtypes, α2A, α2B, and α2C, was characterized by autoradiography and immunohistochemistry techniques. Among the three subtypes, α2A-ARs are the most abundant in the brain and localized in locus coeruleus, midbrain, hippocampus, hypothalamus, amygdala, cerebral cortex and brain stem. The α2B-ARs are located in thalamus and hypothalamus and α2C-ARs in cortex, hippocampus, olfactory tubercle and basal ganglia [51–53]. In the mouse brain, ~90% of α2-ARs are α2A-ARs and ~10% are α2C-ARs [30–32,82]. Notably, binding experiments

α

γ

α

μ

using [3H]2-methoxyidazoxan (a selective α-2 AR antagonist) with postmortem human brain detected 100% of α2A-ARs population in the hippocampus, cerebellum, and brainstem (*B*max = 34–90 fmol/mg protein). In addition to this, α2A-AR (*B*max = 53 fmol/mg of protein) and α2B/C-AR (*B*max = 8 fmol/mg of protein) were detected in the frontal cortex [30]. An extensive array of agonists and antagonists for α2-ARs have been developed. The main limitation of these ligands is lack of subtype selectivity for α2-ARs and off-target binding to other receptors [83,84]. α2A-ARs are mostly presynaptic and agonists inhibit NE release from the terminals and are used to treat hypertension, drug withdrawal, and ADHD whereas antagonists increase NE release and are used as antidepressants. Development of α2-AR subtype selective PET tracers would facilitate medication development and help gain further understanding of their role in brain diseases. α α

α α

α α

α

α

μ μ

α

α

During the 1980s, two research groups identified two potent α2-AR selective antagonists via radioligand binding assays. One WY-26703 belongs to the benzoquinolizine class, and the other MK-492 belongs to the benzo[b]furo-quinolizine class [85,86]. Based on these templates, Bylund's group developed two PET tracers: [11C]WY-26703 (**8**) in 1992 and [11C]MK-912 (**9**) in 1998 (Figure 2) [87,88]. Both radiotracers were prepared from their respective *N*-desmethyl precursors via <sup>11</sup>C-methylation with 30.71–34.41 GBq/µmol molar activity. The in vitro binding assays and ex vivo biodistribution studies (tissue dissection followed by γ-counting) in rodents indicated that both radiotracers crossed the BBB and **9** showed higher affinity and specific binding to α2-ARs than **8**. However, PET studies of **8** and **9** in Rhesus monkeys showed fast washout from brain and nonspecific binding; thus it was concluded that they were not appropriate for PET imaging of α2-ARs in brain. α μ μ

**Figure 2.** Various classes of αα2-ARs antagonist radiotracers.

In 1997, Pike's group developed [11C]RS-15385-197 (**10**) and [11C]79948-197 (**11**) as PET α2-ARs ligands (Figure 2) [89]. These radiotracers were prepared from their respective *O*-desmethyl precursors through the <sup>11</sup>C-methylation method with 61 ± 17 GBq/µmol (**10**) and 64 ± 3 GBq/µmol molar activity (**11**). Biodistribution, brain uptake, and metabolic profile studies were done in male Sprague-Dawley rats. They observed specific signals in brain (mainly cerebellum) at 30–90 min (70–95% radioactivity of parent radioligand), which was analogous to their results with [3H]**10**. Nonspecific binding in brain was **11** > **10**, which mostly likely reflected their differential metabolism (**11** > **10**). Thus, they chose **10** to quantify α2-ARs in the human brain using PET [90]. Studies in two volunteers with **10** revealed a low brain uptake index (BUI) due to high affinity to human plasma proteins. Consequently, **10** was not studied further.

In 2002, Crouzel's group chose atipamezole, an α2-AR selective antagonist, to develop [ <sup>11</sup>C]atipamezole, **12** (Figure 2) [91]. The radiotracer was prepared through an unique approach using 2-ethyl-2-oxoacetylindane, [11C]formaldehyde ([11C]HCHO) in the presence of zinc oxide and ammonium hydroxide (similar to Debus-Radziszewski imidazole synthesis) with an overall yield of 1.5%. However, no PET studies were reported with **12**.

Furthermore, in 2002, Smith's group developed two tetracyclic based anti-depressant radiotracers, mianserin (**13**) and mitrazepine (**14**), that have potent antagonist properties at α2-ARs and also at serotonin receptors (5HT2A, 5HT2C) and labelled them with C-11 to prepare **13** and **14** (Figure 2) [92,93]. The radiotracers **13** and **14** were prepared from their respective *N*-desmethyl precursors via <sup>11</sup>C-methylation with ~40 GBq/µmol and 5–7 GBq/µmol molar activity, respectively. PET studies in female pigs with **13** showed limited binding potential in brain whereas **14** showed more favorable

properties including slow metabolism, fast brain uptake and sufficient target-to-background ratio for pharmacokinetic parameters estimation.

Radiotracer **14** had higher binding in the frontal cortex, thalamus, and basal ganglia where pretreatment with unlabeled mitrazepine revealed that its binding was reversible, whereas, in the cerebellum and olfactory tubercle, it was not. Notably, using the α2-AR subtype KO mouse model they validated the receptor selectivity of **14**. In 2004 and 2009, this group conducted a clinical trial with volunteers using **14** to study its distribution, metabolism and pharmacokinetics [94,95]. The results revealed that **14** can serve as a PET radiotracer to image α2-ARs in the brain, though identification of its metabolites and its nonselective binding are limitations. α α

Two years later, Leysen's group developed a reversible, potent and selective α2-AR antagonist, viz. R107474 (**15**, Figure 3) [96]. They prepared [11C]**15** through Pictet-Spengler condensation method using [11C]HCHO and the respective secondary amine with 24–28 GBq/µmol molar activity at the end of bombardment (EOB). They carried out ex vivo autoradiography to measure in vivo α2A-ARs and α2C-ARs occupancy of **15** in rats. Biodistribution studies showed rapid uptake of **15** into brain and other tissues with the brain showing the highest uptake other than liver and kidneys. In the brain the highest uptake of **15** was in the septum (3.54 ± 0.52 ID/g) and entorhinal cortex (1.57 ± 0.50 ID/g) whereas the lowest was in the cerebellum, a region with very low density of α2-ARs. However, the potential of **15** was not investigated further. α μ α α α

α . **Figure 3.** Anti-depressive & antihypertensive based α2-AR PET radiotracers.

α α In 2006, Jacobsen's group developed C-11 labelled yohimbine (**16**, Figure 3) [97], an antihypertensive agent. Yohimbine has potent antagonist properties at α2-ARs, but also interacts with α1 and 5-hydroxy tryptamine 1A receptors (5-HT1A). The radiotracer **16** was prepared through <sup>11</sup>C-methylation of yohimbinic acid using C-11 methyl iodide ([11C]CH3I) and obtained with 40 GBq/µmol molar activity.

μ α α α PET studies were performed in pigs to obtain whole-body and dosimetry recordings and for dynamic brain imaging. Interestingly, no radioactive metabolites of **16** were reported in pig plasma and binding of **16** was observed in α2-AR-rich regions where it was displaceable by co-injection of pharmacological doses of yohimbine or selective α2-AR antagonist (Figure 4). Later, **16** was used to image α2-ARs in the human brain (*n* = 6) using PET [98]. Highest binding of **16** was observed in cortex and hippocampus and the lowest in corpus callosum, which was used as a reference region to estimate the average total distribution (*V*T) in other brain regions. The radiotracer **16** seems to be a suitable radiotracer to image α2-ARs but has similar issues as of **14,** which need to be addressed.

α α α α The concentration of α2-subtype receptors in the brain is low (5–90 fmol/mg range) increasing the challenge for their detection by PET [31]. Therefore, α2- subtype specific PET tracers (with subnanomolar affinities and >30-fold selectivity) still need to be developed for the quantification of α2-subtype receptors and to assess their role in brain diseases. A few research groups have developed α2-subtype specific PET radiotracers, but success has been limited.

<sup>−</sup> α **Figure 4.** Parametric maps of **16** in living porcine brain. (**A**) Baseline study using **16** showed regional differences in its distribution. (**B**) Blocking experiment (yohimbine at 0.07 mg/kg) reduced the scale of distribution volume (*V*d) to ~2 mL g−<sup>1</sup> in all the α2-AR bound regions. (**C**) Increased dose of yohimbine (1.6 mg/kg) had no further significant effect in comparison to the low dose (*n* = 3) Maps are superimposed on the MR image. Adapted from JNM publication by Jacobsen S, Pedersen, K.; Smith, D.F.; Jensen, S.B.; Munk, O.L.; Cumming P [97]. Permission obtained from SNMMI. <sup>−</sup> α *α*

#### *α 3.1.* α*2A-Specific PET Radiotracers*

α α α α α α α α α μ α Kumar's group in 2010 developed [11C]MPTQ (**17**, Figure 5) for the quantification of α2A-ARs in vivo [35]. Compound **17** was shown to have blocking effects on α2A-ARs in vivo in brain and has stronger affinities for α2A-AR (*K*<sup>i</sup> = 1.6 nM) than α2C-AR (*K*<sup>i</sup> = 4.5 nM) and 5-HTT (serotonin transporter, *K*<sup>i</sup> = 16 nM) [99]. They anticipated no binding of **17** to α2B and α2C-ARs since the densities of these receptors are lower than α2A-ARs. In addition, the 10-fold higher affinity of **17** α2A-ARs over 5-HTT is advantageous for α2A-ARs as both have similar *B*max values. The radiosynthesis of **17** was accomplished through <sup>11</sup>C-methylation of its respective *N*-desmethyl precursor with 74–88.8 GBq/µmol molar activity at the end of synthesis (EOS). PET studies in baboons with **17** showed that it penetrated the BBB and accumulated in α2A-AR-rich brain areas. They ruled out binding of **17** to 5-HTT due to its low uptake in the hippocampus, temporal cortex, and occipital cortex, which are the brain regions with the highest binding of 5-HTT radiotracers. No further studies were reported using **17**. α α α α α α α α μ α

α **Figure 5.** α2A-antagonist (**17**) and agonist (**18**) PET radiotracers.

α α α α α α μ α α α α α μ α In search of a selective agonist to α2-ARs, Lehmann's group identified 1-[(imidazolidin-2-yl) imino]indazole (marsanidine) [100] and later developed an α2A subtype specific ligand by introducing fluorine to marsanidine [33]. The reported binding affinity (*K*d) of 6-fluoromarsanidine for α2A (33 nM) is higher than for α2B (72 nM) and α2C (600 nM). Solin et al., in 2019, prepared 6-[18F]fluoromarsanidine (**18**, Figure 5) through electrophilic <sup>18</sup>F-radiofluorination using [18F]selectfluor bis(triflate) and a corresponding precursor with 3–26 GBq/µmol molar activity at the EOS [34]. In vivo PET was performed in rats and α2A-KO mice, but the radiotracer was not continued further because of its rapid metabolism and high nonspecific uptake in rat and mouse brain.

α

#### *3.2.* α*2C-Specific PET Radiotracers*

Animal models, such as the forced swimming test (FST) and the prepulse inhibition (PPI) are used to screen for anti-depressants and anti-psychotics, respectively. The use of α2C-KO and α2C-overexpression (α2C-OE) mouse models in FST and PPI paradigms suggested that α2C-specific compounds may have therapeutic benefits for depression and schizophrenia [32,38]. In 2007, Orion pharma from Finland identified an acridine-based compound, JP-1302 [38], and a research group from Japan identified a methyl benzofuran based compound, MBF [101], as selective α2C antagonists. Both these ligands have high affinities for α2C (JP-1302 *K*<sup>i</sup> = 28 nM, MBF *K<sup>i</sup>* = 20 nM) than for α2A (JP-1302 *K*<sup>i</sup> = 3500 nM, MBF *K*<sup>i</sup> = 17,000 nM) and α2B (JP-1302 *K*<sup>i</sup> = 1500 nM, MBF *K*<sup>i</sup> = 750 nM). Based on these findings, Zhang's group, in 2010, synthesized [11C]JP-1302 (**19**) and [11C]MBF (**20**)(Figure 6) as PET probes to evaluate their BBB penetration and α2C selective binding in the brain [37]. The radiotracers **19** (molar activity 95 ± 24 GBq/µmol) and **20** (molar activity 62 ± 15 GBq/µmol) were prepared using <sup>11</sup>C-methylation from *N*-desmethyl and *O*-desmethyl precursors, respectively. PET studies were conducted in WT and Pgp, breast cancer resistance protein (BCRP) KO mice using both radiotracers. α α α α α α α α α μ μ

α **Figure 6.** PET radiotracers for α2C-ARs.

α α This combined KO model is useful to evaluate whether brain penetration of PET probes is sensitive to Pgp and BCRP. After injection of the radiotracers, their levels in the brain were low in WT mice whereas they were higher in Pgp and BCRP KO mice. The regional binding of these radiotracers did not correspond with the regional brain distribution of α2C, so it was concluded that they were inadequate to evaluate α2C-ARs in brain with PET.

μ α α α α α In 2014, researchers from Turku PET center and Orion Pharma reported the radiosynthesis of [ <sup>11</sup>C]ORM13070 (**21**, Figure 6) with molar activity 690 ± 340 GBq/µmol and its evaluation in rats and in α2A and α2AC KO mice with PET [36,102]. The binding affinities of **21** for α2C (3.8 nM) is higher than for α2A (109 nM) and α2B (23 nM).

α α α α α α α α α The in vivo PET and ex vivo autoradiography of **21** in rat indicated that its brain distribution corresponds to the regional distribution of α2C in brain, with highest levels in striatum and olfactory tubercle. Pretreatment with atipamezole, a α2-sutype nonselective antagonist blocked the binding of **21** into these regions. Furthermore, by using α2A and α2AC KO model mice, they demonstrated α2C specificity of **21**. The brain uptake of **21** in α2A-KO and WT mice was similar whereas, negligible uptake occurred in α2AC KO (Figure 7, left). They represented time-activity curves for striatum and cerebellar cortex of three mice types (Figure 7, right) and the radioactivity ratios at 5–15 min for α2A, α2AC KO mice, and WT mice were 1.51–1.51, 1.06–1.09 and 1.51–1.57, respectively.

Accordingly, **21** was studied in healthy men to estimate its metabolism, pharmacokinetics, whole-body distribution and radiation dosimetry [39]. Good results were obtained in rodent and human PET studies with **21**, except for its fast washout from brain. Better pharmacokinetics, higher affinity, and specificity can potentially be enhanced by structural modifications to **21**. Given that the α2A-ARs are widely distributed in brain in contrast to α2C-ARs, a candidate with subnanomolar affinity for α2C-ARs (>50-fold affinity than α2A-ARs) is needed for a PET radiotracer. As α2C-ARs are of interest as therapeutic targets in brain diseases, the α2C-specifc PET radiotracers would facilitate their development as medications and help in investigations of α2C-ARs in the human brain.

α α α α α α **Figure 7.** PET/CT images and time-activity curves of **21** for striatum and cerebellar cortex of (**A**) α2A KO (**B**) α2AC KO and (**C**) WT mice. Brain uptake of **21** in α2AC KO is negligible and is similar in α2A KO and WT mice with 7.8–8.1% ID/g at 1 min and 1.2% ID/g at 30 min after **21** injection. The striatum to cerebellar cortex radioactivity ratios (at 5–15 min) for α2AC KO mice did not differ and for α2A KO and WT mice are alike. Adapted from JNM publication by Arponen E.; Helin, S.; Marjamäki, P.; Grönroos, T.; Holm, P.; Löyttyniemi, E.; Någren, K.; Scheinin, M.; Haaparanta-Solin, M.; Sallinen, J.; [36]. Permission obtained from SNMMI.

#### **4.** β**-ARs and Nonselective PET Radiotracers**

β

α α α α α α α **β** β β β β β β β β-ARs are associated with Gs-heterotrimeric G-protein and mediate intracellular signaling through adenyl cyclase activation and cyclic adenosine monophosphate (cAMP) production. β-ARs are classified into β1, β2, and β3 subtypes, in which, the former two have been much more explored [42,46]. Quite a lot of selective and nonselective β-AR agonists and antagonists (blockers) are available as drugs in the market to treat various cardiac and pulmonary disorders. In the brain, β-ARs are localized in the frontal cortex, striatum, thalamus, putamen, amygdala, cerebellum and hippocampus [48]. The density of β-ARs in brain is sensitive to brain pathophysiology. Notably, the density of β-ARs decrease with age [40]. Light microscopic autoradiography using [3H]dihydroaloprenolol (a selective β-blocker) with rat brain sections has shown a wide distribution of β-ARs in forebrain and cerebellum regions (*B*max = 23 fmol/mg tissue) [103]. Similarly, *B*max value of 18 fmol/mg protein was reported in pre-frontal cortex of subjects with Parkinson's disease [48,104]. By altering the Ca2<sup>+</sup> levels through *N-*methyl-*D-*aspartate (NMDA) receptors in hippocampus, β-ARs modulate synaptic plasticity, including that needed for memory [44,45]. The blockade of β-ARs is associated with a small increased risk for Alzheimer's and Parkinson's disease [43,105]. In addition, abnormal function and densities of β-ARs have been reported in mood disorders and schizophrenia [41,47,106].

β β β β β β Several radioligands, mostly based on β-blockers, were validated for imaging of β-ARs in the heart [50]. The majority of β-blockers possess a hydroxyl propylamine moiety in their structures that is vital for binding to β-ARs and this moiety was maintained in most of these radioligands. PET radiotracers have succeeded in imaging and quantifying myocardial and pulmonary β-ARs in human [107,108], whereas, PET radiotracers for cerebral β-ARs have been more challenging. The clinical PET radiotracers for cardiac β-ARs have negative Log P values (<−2), which is not suitable for imaging the brain. Several lipophilic and high to moderate affinity β-AR nonselective antagonists were explored as PET radiotracers to image β-ARs in the brain.

β During the 1980s, propranolol, a β-blocker drug was labelled with C-11 (**22**, Figure 8) but was unsuitable as a PET ligand for β-ARs because of high nonspecific binding in vivo [109,110]. Subsequently, Berridge's group described the synthesis of two isomers (*R*/*S*) of [18F]fluorocarazolol

β β

β

β

β

β

(**23**, Figure 8) through reductive amination using [18F]fluoroacetone and desisopropylcarazolol with 18.5–37 GBq/µmol molar activity [111,112]. The radiotracer **23** has subnanomolar *K<sup>i</sup>* values for β-ARs (β1 0.4 nM, β2 0.1 nM) and Log P7.4 value of 2.19. The same group used *S*-**23** for PET imaging of the pig heart and lungs to validate the β-AR biding. In 1997, Waarde et al., employed *S*-**23** to image β-ARs in the human brain and obtained positive results [113]. They observed specific binding (blocked with pindolol) of *S*-**23** in β-AR rich areas, striatum and various cortical areas. However, the radiotracer was discontinued for further human studies as fluorocarazolol was positive for the Ames test i.e., mutagenic [114]. μ β β β β β β μ β β β β β β

β

β

β

β

β −

β −

β

β

β

β

β

β

β **Figure 8.** Early PET radiotracers for cerebral β-ARs. β

β β − β μ β β β Two research groups conducted biodistribution studies in rats using [18F]fluoropropranolol (**24**) and [11C]ICI 118,551 (**25**) (Figure 8), which failed because of their nonspecific binding [115,116]. In 2001, Fazio's group described two isomers (*R*/*S*) of C-11 labelled bisprolol (β1 *K*<sup>i</sup> 1.6 nM, β2 *K*<sup>i</sup> 100 nM and Log P7.4 = −0.2) (**26**, Figure 9) to image β1-ARs in the brain [117]. The radiosynthesis of **26** was accomplished via reductive amination using [11C]acetone and desisopropyl bisprolol precursor with 129.5 ± 37 GBq/µmol molar activity at the EOS. They observed little specific uptake of **26** in β1-AR rich regions in the rat's brain and also high nonspecific uptake in the pituitary (1.8 ± 0.3 ID at 30 min), a region with high β2-ARs levels. No further studies were reported using **26** to image β-ARs. β β − β μ β β β

β β **Figure 9.** Radiotracers based on various β-AR blockers.

β β β β In 2002, Elsinga's group reported five different potent and lipophilic β-AR antagonists (carvedilol, pindolol, toliprolol, bupranolol, and penbutolol) as PET probes to image β-ARs in rat brain [118]. C-11 labelled carvedilol (**27**, Figure 9; molar activity 12.97–25.9 GBq/µmol) was prepared through <sup>11</sup>C-methylation using [11C]CH3I and its respective *O-*desmethyl precursor, whereas, [11C]pindolol (**28**, molar activity 25.9–37 GBq/µmol) and [11C]toliprolol (**29**, molar activity 22.2–25.9 GBq/µmol) were prepared via reductive amination using [11C]acetone and the respective desisopropyl precursors. The F-18 tracers of bupranolol (**30**, molar activity 11.1–18.5 GBq/µmol) and penbutolol (**31**, 22.2–99.9 GBq/µmol) were also prepared by means of reductive amination but using [18F]fluoroacetone and the respective des-fluoro-isopropyl precursors.

μ

The five radiotracers had strong affinities (subnanomolar *K*d) for β1 and β2-ARs. Although these radiotracers had sufficient affinity and lipophilicity for in vivo imaging, none showed good brain uptake. This group also evaluated S-[18F]fluoroethylcarazolol (**32**, β1 *K*<sup>i</sup> = 0.5 nM, β2 *K*<sup>i</sup> = 0.4 nM and Log P7.4 = 1.94) for in vivo imaging of β-ARs in rat brain [119]. The radiotracer **32** (Figure 9) was prepared via an epoxide ring-opening using [18F]fluoroethylamine and the corresponding epoxide with >10 GBq/µmol molar activity. The radiotracer accumulated in brain with uptake reflecting cerebral β-ARs binding. However, no further PET imaging studies were reported using **32** probably because of its analogous nature to **23** which was shown to be positive Ames test [114,120]. β β β β β μ β

μ

μ

μ

μ

In 2008, Elsinga's and Vasdev's groups chose exaprolol (β-AR *K*<sup>d</sup> = 9–9.5 nM) and developed *S*-[11C]exaprolol (**33**) and *S*-[18F]fluoroexaprolol (**34**), respectively, to image β-ARs with PET (Figure 10) [120,121]. Radiotracer **33** was prepared via reductive amination using [11C]acetone and desisopropylexaprolol precursor with >10 GBq/µmol molar activity and the radiotracer **34** was prepared through a nucleophilic substitution reaction using [18F]fluoride and a corresponding tosylate precursor followed by reductive hydrolysis, with 34.29 GBq/µmol molar activity. Regardless of good binding and kinetic properties, both these radiotracers showed high nonspecific uptake in the brain and were found to be inadequate for PET imaging of β-ARs. β β μ μ β

β **Figure 10.** Another set of latest β-AR PET radiotracers.

β μ β Again in 2014, Elsinga's group developed [18F]FPTC (**35**, Figure 10) for PET imaging of β-ARs in brain [122]. The radiotracer **35** is a derivative of carazolol, wherein, isopropylamine group of carazolol was replaced by a PEGylated triazole group. It was prepared through Huisgen's 1,3-dipolar cycloaddition (click reaction) using F-18 labelled PEGylated alkyne and the corresponding azide with >120 GBq/µmol molar activity. Although **35** was shown to have appropriate LogP7.4 (2.48) and specific binding in in vitro assays, it could not visualize β-ARs in the brain, lung or heart using micro-PET.

β β β β β β Thus, the development of PET radiotracers for neuroimaging of β-ARs remains a challenge and as of now, there are no β-AR subtype specific PET radiotracers. Such radiotracers are important to expand our understanding of the role of β-ARs in aging and memory formation and also to assess their function in behavioral disorders. Future research, as suggested by Elsinga and Waarde [48], should consider modifying the imaging agents used for myocardial β-ARs rather than radiolabeling existing β-blocker drugs. Alterations should optimize Log P7.4 (2-3), high affinity and selectivity to β-ARs and no substrate affinity for Pgp.

#### **5. Conclusions**

Over the past four decades, significant efforts have been made to develop CNS-ARs PET ligands for brain imaging. Despite these efforts, very few PET radiotracers are available to selectively image AR subtypes in the brain. The development of specific radiotracers is hindered mainly by the low receptor densities of each AR subclass within the brain, which requires further optimization processes for highly potent and BBB permeable ligands. Though challenging, AR subtype specific agonist/antagonist PET radiotracers are needed to ascertain AR's role in brain pathophysiology and for medication development.

**Funding:** This research was funded by 'Tier 2 Research and Creative Works Strategic Investment Program Project of the University of Missouri (Internal Grant)' and the "Intramural Program of the National Institute on Alcohol Abuse and Alcoholism, grant number Y1AA-3009" and "The APC was funded by Y1AA-3009".

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


β

β

β

	- 99. Andrés, J.I.; Alcázar, J.; Alonso, J.M.; Alvarez, R.M.; Bakker, M.H.; Biesmans, I.; Cid, J.M.; De Lucas, A.I.; Fernández, J.; Font, L.M.; et al. Discovery of a new series of centrally active tricyclic isoxazoles combining serotonin (5-HT) reuptake inhibition with α2- adrenoceptor blocking activity. *J. Med. Chem.* **2005**, *48*, 2054–2071. [CrossRef] α
	- 100. S . a̧ a̧ α czewski, F.; Kornicka, A.; Rybczy ´nska, A.; Hudson, A.L.; Shu, S.M.; Gdaniec, M.; Boblewski, K.; Lehmann, A. 1-[(imidazolidin-2-yl)imino]indazole. Highly α2/I 1 selective agonist: Synthesis, X-ray structure, and biological activity. *J. Med. Chem.* **2008**, *51*, 3599–3608. [CrossRef]
	- 101. Hagihara, K.; Kashima, H.; Iida, K.; Enokizono, J.; Uchida, S.I.; Nonaka, H.; Kurokawa, M.; Shimada, J. Novel 4-(6,7-dimethoxy-1,2,3,4-tetrahydroisoquinolin-2-yl)methylbenzofuran derivatives as selective α2C-adrenergic receptor antagonists. *Bioorg. Med. Chem. Lett.* **2007**, *17*, 1616–1621. [CrossRef] [PubMed]

β

′ β


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*
