18F-labelled triazolyl-linked argininamides targeting the neuropeptide Y Y1R for PET imaging of mammary carcinoma

Neuropeptide Y Y1 receptors (Y1R) have been found to be overexpressed in a number of different tumours, such as breast, ovarian or renal cell cancer. In mammary carcinoma the high Y1R density together with its high incidence of 85% in primary human breast cancers and 100% in breast cancer derived lymph node metastases attracted special attention. Therefore, the aim of this study was the development of radioligands for Y1R imaging by positron emission tomography (PET) with a special emphasis on imaging agents with reduced lipophilicity to provide a PET ligand with improved biodistribution in comparison with previously published tracers targeting the Y1R. Three new radioligands based on BIBP3226, bearing an 18F-fluoroethoxy linker (12), an 18F-PEG-linker (13) or an 18F-fluoroglycosyl moiety (11) were radiosynthesised in high radioactivity yields. The new radioligands displayed Y1R affinities of 2.8 nM (12), 29 nM (13) and 208 nM (11) and were characterised in vitro regarding binding to human breast cancer MCF-7-Y1 cells and slices of tumour xenografts. In vivo, small animal PET studies were conducted in nude mice bearing MCF-7-Y1 tumours. The binding to tumours, solid tumour slices and tumour cells correlated well with the Y1R affinities. Although 12 and 13 showed displaceable and specific binding to Y1R in vitro and in vivo, the radioligands still need to be optimised to achieve higher tumour-to-background ratios for Y1R imaging by PET. Yet the present study is another step towards an optimized PET radioligand for imaging of Y1R in vivo.

and Y 2 R revealed high specific binding (up to >12000 dpm/mg) due to extremely high density of the receptors in breast cancer 9 . Interestingly, the neoplastic transformation comes along with a switch in YR subtype expression: while healthy breast tissue only expresses the subtype Y 2 R, the Y 1 R subtype was found in tumour tissue only or at least predominantly 9 . Therefore, the Y 1 R is a promising target for tumour diagnosis and targeted tumour therapy, due to the high receptor expression and the predominance of the Y 1 R subtype in breast cancer tissue.
As subtype selective ligands are highly attractive for specific targeting of Y 1 R-positive tumours, many efforts have been made to design such subtype selective ligands. In 2001, the Beck-Sickinger group reported the first Y 1 R-preferring NPY analogues. [Phe 7 ,Pro 34 ]pNPY showed the highest selectivity for Y 1 R over Y 2 R and Y 5 R (>1:3000-fold) 10 . This compound was further developed towards a 99m Tc-labelled radioligand, [ 99m Tc]Tc(CO) 3 -N α His-Ac-[Phe 7 ,Pro 34 ]-NPY, which was also used in first human imaging studies 11 . Based on this work, we have reported the first 18 F-labelled analogue of NPY, [Pra 4 ([ 18 F]FGlc),Phe 7 ,Pro 34 ]NPY, which was synthesised by 18 F-fluoroglycosylation using the corresponding alkyne-functionalised peptide 12 .
Besides the peptidic NPY analogues, a number of non-peptide Y 1 R ligands have been reported, such as BIBP3226 13 , BIBO3304 14 , LY357897 15 and Y1-973 16 . The latter was radiolabelled with fluorine-18, a positron emitting radionuclide with beneficial decay characteristics, i.e. a half-life of 110 min, a clean decay profile (97% positron emission, 3% electron capture) and a low positron energy (max. 0.635 MeV), resulting a low maximum positron range of 2.4 mm in water and therefore leading to high-resolution positron emission tomography (PET) images 17 . [ 18 F]Y1-973 was studied as the first non-peptide Y 1 R antagonist to be successfully applied for in vivo imaging of Y 1 R in the central nervous system (CNS) of monkeys 16 . However, this compound is most likely not suitable for peripheral imaging of breast cancer due to its high lipophilicity. The (R)-argininamide BIBP3226 was described as the first highly potent and selective Y 1 R antagonist in 1994 13 . Recently, we reported prototypic 18 F-labelled argininamide-type Y 1 R antagonists derived from BIBP3226 as candidate radioligands for PET (1-4, Table 1) 18 . The most favorable compound (1 18 , Table 1) was an amine-functionalised carbamoyl-derivative of BIBP3226 which was synthesised by 18 F-fluoroacylation using 4-nitrophenyl-2-[ 18 F]fluoropropionate ([ 18 F] NPFP). This ligand is highly potent (K i (Y 1 R) = 1.3 nM) and subtype selective (>1:3000 over Y 2 R and >1:10000 over Y 4 R and Y 5 R) and showed excellent in vivo-stability in mice. In nude mice bearing Y 1 R-positive MCF-7 tumour xenografts, the compound showed a rapid blood clearance, but extraordinarily high accumulation in the gall bladder (>200%ID/g at 30 min p.i.). With only 0.51%ID/g at 30 min p.i., the tumour uptake was low, but  12 , the uptake of the compound in the kidneys was reduced by a factor of 10, being the major advantage of the hydrophobic small molecule antagonist PET radioligand. Based on these findings, the aim of this study was to reduce the lipophilicity of a candidate ligand in order to achieve a more suitable biodistribution with reduced biliary excretion and thus a better visibility of the tumour in PET imaging studies. Therefore, we synthesised three BIBP3226-derivatives, two with 18 F-fluoroethoxy-linkers and one with an 18 F-fluoroglucosyl moiety, and compared their properties in vitro on MCF-7-Y1 cells and in vivo using a nude mouse tumour xenograft model.

Results and Discussion
chemistry and radiochemistry. The alkynylated labelling precursor 7 was prepared from amine 4 19 by guanidinylation of 4 with the isothiourea derivative 5 in the presence of mercury(II) chloride yielding intermediate 6, which was treated with trifluoroacetic acid (TFA) to obtain alkyne 7 (Fig. 1). The latter was subjected to copper(I)-catalysed cycloadditions with azides 8 20 , 9 and 10 yielding the potential Y 1 R ligands 11-13 in a purity of >95% (Fig. 1).
The radiosynthesis of [ 18 F]11 was performed according to a copper-catalysed azide-alkyne cycloaddition (CuAAC)-based 18 F-fluoroglycosylation method (Fig. 2)  In-vitro characterisation. For the determination of the Y 1 R affinities of 11, 12 and 13 (synthesis see Fig. 1), competition binding assays were carried out using the radioligand [ 3 H]UR-MK299 (K d = 44 pM) on SK-N-MC neuroblastoma cells as described previously 22 . K i values of 11-13 are provided in Table 1 together with reference values from literature for the previously published compounds 1-3 for comparison. The precursor alkyne 7 was found to bind to the Y 1 R with an affinity of 0.94 nM and thus exhibiting a similar receptor affinity as BIBP3226 and the fluoroacylated compound 1 previously published by Keller et al. 18 . The 6-deoxy-6-fluoroglycosyl derivative 11 bound with an affinity of 208 nM to the Y 1 receptor, which is a 10-fold higher Y 1 R affinity compared to the 2-deoxy-2-fluoroglycosyl derivative 3 (K i = 2000 nM, Table 1) which had a longer spacer moiety between the fluoroglycosyl moiety and the binding motif 18 . Although exhibiting a lower affinity to the Y 1 R by a factor of about 200 compared to the lead compound BIBP3226, this finding supported our hypothesis, that shortening the linker from ten to only three atoms (3 vs. 11, Table 1) would result in a less pronounced decrease in Y 1 R affinity. Compared to the glycosyl derivatives, the compounds bearing the less polar fluoroethoxy groups revealed higher Y 1 R affinities: The K i value of 13 was determined to be 29 nM, which means that it has a sevenfold higher affinity than the fluoroglycosylated compound 11, and the affinity of ligand 12 comprising the short fluoroethoxy chain increased even more by one order of magnitude (K i = 2.8 nM).
To assess the YR subtype selectivity profiles of the three potential Y 1 R ligands 11, 12 and 13, the affinities to the other subtypes of the neuropeptide Y receptor, Y 2 R, Y 4 R and Y 5 R, were determined (Table S1, Supplementary  Information). None of the three compounds showed considerable binding to one of the other subtypes within the experimental range of concentrations (maximum concentration of 10 µM), confirming the subtype selectivity for Y 1 R of the ligands under study (11)(12)(13).
The octanol-water distribution coefficients (logD 7.4 ) of the radioligands were determined by the "shake flask" method and revealed a weak lipophilicity for all three compounds, being in good accordance to the calculated  www.nature.com/scientificreports www.nature.com/scientificreports/ values ( Table 2) The binding of the radioligands to human plasma proteins was determined by gel filtration and revealed a high fraction of protein bound radioligand: less than half of the amount of radioactivity in plasma was unbound and thus freely available in the blood (42% of [ 18 F]11, 38% of [ 18 F]12 and 37% of [ 18 F]13, Table 2).
The stability of the radiotracers was evaluated in vitro by radio-HPLC: None of the three radioligands showed any radioactive degradation products within 3 h of incubation in human serum at 37 °C (see Supplementary  Fig. S3 and Table 2).
To determine cellular accumulation in vitro, assays with the 18 F-labelled ligands were performed using the human breast cancer cells MCF-7-Y1. Cells were incubated either with the radioligand alone (total binding) or with the radioligand in the presence of 10 µM BIBP3226 as the blocking substance for determination of nonspecific binding. The total radioactivity in each well was defined as 100%. The highest specific accumulation was observed for the fluoroethoxy radioligand To substantiate the specific binding of the radioligands to Y 1 R, autoradiography experiments were performed in vitro using slices of MCF-7-Y1 tumour xenografts. The tumour slices were incubated with each of the three radioligands in the presence or absence of BIBP3226 (1 µM and 10 µM). As expected, fluoroglycosylated ligand

Biodistribution and small animal pet. The biodistribution of [ 18 F]11, [ 18 F]12 and [ 18 F]13 was evaluated
in vivo in healthy mice; the determined uptake values are depicted in Fig. 5. In general, the biodistribution of all three radioligands was very similar to that of [ 18 F]1 18 . However, a detailed HPLC analysis of blood samples taken early after radiotracer injection revealed that the 18 11 were determined in the blood samples. The radioligands and radiometabolites showed fast clearance from the blood, as there was no detectable radioactivity in the blood at 90 min p.i. Moderate amounts of radioactivity were detected in the kidneys and intestines, and exceptionally high radioactivity was observed in the gall bladder (up to 600%ID/g after 90 min for [ 18 F]11). All other organs did not show any significant accumulation of the radioligands. The uptake in the liver as the main organ for metabolism of xenobiotics was below 5%ID/g at 30 min and below 1.5%ID/g at 90 min p.i. The three 18 F-radioligands revealed a more differentiated result regarding the uptake in the kidney, which can be ascribed to the formation of hydrophilic radiometabolites in the blood (Supplemantary Fig. S4 18 , the majority of the injected radioactivity was found in the gall bladder (>100%ID/g) and in the intestines. For the glycosylated ligand [ 18 F]11 the excretion seems to occur more slowly as the accumulation in these organs increased from 30 to 90 min p.i., while it stayed constant or decreased in case of the fluoroethoxy ligands [ 18 F]12 and [ 18 F]13. The high accumulation in the bile might hamper tumour imaging by PET, as it was previously described for [ 18 F]1 18 . The aim of reducing the uptake in the gall bladder by using more hydrophilic radioligand analogues, was obviously not reached, although the lipophilicity of the radiotracers was in fact reduced. Since only very little radioactivity was detected in the bones, it is most likely that there was no cleavage of fluoride from the molecules in terms of metabolism.  Supplementary Information Fig. S4).    13 were more hydrophilic and demonstrated enhanced and specific tumour uptake leading to improved tumour-to-background-ratios, such that Y 1 R-positive tumours could clearly be visualised by PET. Nevertheless, there is still room for improvement for the design of an optimal radioligand for Y 1 R imaging by PET.
conclusion Three BIBP3226-derivatives, two with 18 F-fluoroethoxy-linkers and one with a 18 F-fluoroglucosyl moiety, were radiosynthesised in sufficient radioactivity yields and molar activities. Dependent on the size of the carbamoyl residues attached to the guanidine group the three radioligands showed receptor affinities for Y 1 R from 2.8-208 nM. The radioligand with the highest affinity ([ 18 F]12) revealed the highest specific binding to Y 1 R-positive cells and to tumour slices in vitro. Despite their different hydrophilicity (logD 7.4 values ranging from 0.43 to 2.03), the biodistribution of the three radioligands in healthy mice was very similar. In PET scans of tumour-bearing mice, [ 18 F]12 showed the highest specific binding to the Y 1 R-positive tumour in vivo, corresponding to the highest in-vitro affinity of the 18 F-labelled ligands under study. However, the PET imaging results suffered from high background levels, because of fast degradation of the radioligands in the blood and marked binding to plasma proteins. Therefore, the present study has to be regarded as another step towards the development of an optimal PET radioligand for Y 1 R imaging in vivo.
Determination of plasma protein binding. The binding of 18 F-labelled compounds to plasma proteins was determined using gel filtration columns(illustra ™ MicroSpin ™ G-50 Columns, GE Healthcare Life Sciences, Freiburg).
An aliquot of the radiotracer (approx. 100 kBq) was added to 200 µL of saline, and 100 µL of human plasma, respectively. Both samples were incubated at 37 °C for 10 min. MicroSpin ™ columns were prepared according to the user instruction. Then 25 µL of the incubated radiotracer were given onto the columns and the devices were centrifuged (2000 × g, 2 min). Eluate and solid phase were analysed in the γ-counter (Wallac Wizard). The percentage of unbound radioligand was calculated as free % For autoradiography studies slides were thawed and pre-incubated for 15 min in incubation buffer (50 mM TRIS HCl, pH 7.4, containing 120 mM NaCl, 5 mM MgCl 2 ,) at room temperature. Thereafter, 800 µL of incubation buffer, containing 0.1 MBq of the respective radiotracer were pipetted on the slide and incubated at room temperature for 60 min. For displacement studies BIBP3226 (1 µM or 10 µM) was added to the incubation buffer before pipetting on the slides. Afterwards slides were washed by placing in ice cold incubation buffer (3 × 2 min) followed by short dipping in ice cold distilled water. Slides were carefully dried in a stream of warm air and finally placed on an autoradiography film (Fuji Imaging Plate BAS-IP SR 2025 E, Fujifilm, Düsseldorf) overnight prior to readout (25 µm resolution) on the autoradiograph (HD-CR-35 Bio, Raytest, Straubenhardt) and analysis with the software AIDA (Raytest). Additional sections were stained with hematoxylin and eosin (H&E) for comparison with the autoradiography images. Biodistribution. Biodistribution studies were conducted using female NMRI outbred mice (HsdWin:NMRI) purchased from Envigo (Horst, The Netherlands). Mice were kept in groups of four to five animals in individually ventilated cages in a twelve hours dark/light cycle with unlimited access to water and standard chow. At the age of eight to nine weeks six animals per radiotracer were injected with 2-4 MBq of the respective radiotracer under isoflurane anesthesia. Mice were sacrificed by cervical dislocation 30 or 90 min p.i. of the radiotracer. Blood as well as the organs/tissues lung, liver, kidneys, heart, spleen, brain, muscle, intestines, gall bladder and bones were harvested and analysed in the γ-counter. Samples were weighed and radioactivity in different tissues was calculated as percentage of the total injected dose per gram tissue (%ID/g).

In-vivo
Small animal PET. For the NPY Y 1 R xenograft model, female NMRI nude mice (HsdCpb:NMRI-Foxn1 nu ) were purchased from Envigo (Horst, The Netherlands) at the age of three weeks. Mice were kept in groups of four to five animals in individually ventilated cages in a twelve hours dark/light cycle with unlimited access to water and standard chow. At the age of nine to ten weeks one 17β-estradiol pellet per animal (0.72 mg per pellet, 3 mm diameter) with a 60-day release time (Innovative Research of America, Sarasota, FL, USA) was subcutaneously implanted on the back under isoflurane anesthesia. After three days, approximately 10 6 MCF-7-Y1 tumour cells (in 50 µL PBS) were mixed with Matrigel (50 µL, BD Biosciences, Heidelberg) and then injected subcutaneously at the back. Tumour diameters and weight of the animals were recorded five times a week. Imaging studies were performed four weeks after inoculation of the cells.

Data Availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files). Parts of this study have been reported in the PhD thesis 'Selective neuropeptide and opioid receptor radioligands for imaging studies in vivo by positron emission tomography (PET)' by Julian J. Ott 26 .