Prototypic ¹⁸F-Labeled Argininamide-Type Neuropeptide Y Y₁R Antagonists as Tracers for PET Imaging of Mammary Carcinoma

Abstract

The neuropeptide Y (NPY) Y₁ receptor (Y₁R) selective radioligand (R)-N^α-(2,2-diphenylacetyl)-N^ω-[4-(2-[¹⁸F]fluoropropanoylamino)butyl]aminocarbonyl-N-(4-hydroxybenzyl)argininamide ([¹⁸F]23), derived from the high-affinity Y₁R antagonist BIBP3226, was developed for imaging studies of Y₁R-positive tumors. Starting from the argininamide core bearing amine-functionalized spacer moieties, a series of fluoropropanoylated and fluorobenzoylated derivatives was synthesized and studied for Y₁R affinity. The fluoropropanoylated derivative 23 displayed high affinity (K_i = 1.3 nM) and selectivity toward Y₁R. Radiosynthesis was accomplished via ¹⁸F-fluoropropanoylation, yielding [¹⁸F]23 with excellent stability in mice; however, the biodistribution study revealed pronounced hepatobiliary clearance with high accumulation in the gall bladder (>100 %ID/g). Despite the unfavorable biodistribution, [¹⁸F]23 was successfully used for imaging of Y₁R positive MCF-7 tumors in nude mice. Therefore, we suggest [¹⁸F]23 as a lead for the design of PET ligands with optimized physicochemical properties resulting in more favorable biodistribution and higher Y₁R-dependent enrichment in mammary carcinoma.

Neuropeptide Y (NPY), a 36 amino acid peptide, and NPY receptors are widely distributed in the central nervous system and in the periphery. NPY receptor mediated signaling is involved in the regulation and the modulation of numerous physiological processes. In humans, four NPY receptor subtypes, designated Y₁R, Y₂R, Y₄R, and Y₅R, are functionally expressed, all belonging to the superfamily of seven-transmembrane domain (G-protein coupled) receptors. Y₁ and Y₂ receptors were detected in a variety of human cancers and were, therefore, proposed as potential molecular targets for tumor diagnosis and treatment (theranostics).¹⁻⁶ It was reported that 85% of mammary carcinoma are Y₁R positive, whereas Y₂R levels were found to be low.¹ More than two-thirds of breast cancers are classified as estrogen receptor-positive, and in several studies an estrogen-dependent Y₁R expression in breast cancer cells was described.^7,8

Positron emission tomography (PET) allows noninvasive in vivo imaging of receptors expressed on tumors with high sensitivity and quantification of receptor densities, making PET a powerful imaging technique in diagnostic nuclear medicine, provided that appropriate radiotracers are available for in vivo use.⁹ As yet, efforts to develop radiolabeled Y₁R ligands for tumor diagnosis were based on peptidic agonists.^3,4,10−15 However, diaminopyridine-type ¹⁸F-labeled Y₁R antagonists, developed as brain-penetrant tracers for Y₁R imaging in rhesus monkey brain,^16,17 could represent potential molecular tools for PET imaging of peripheral Y₁R-expressing tumors, too. Recently, in vivo imaging of mammary carcinoma using an ¹⁸F-labeled NPY peptide analogue in MCF-7 tumor-bearing mice,¹⁸ and whole body scintimammography using a Y₁R selective ^99mTc-labeled peptide in humans¹³ were reported. In general, the use of nonpeptidic receptor ligands (antagonists) for tumor imaging has been explored in a few cases only, such as imaging of neurotensin receptor-positive tumors by ¹⁸F-labeled¹⁹ or ¹¹¹In-labeled radioligands of the diarylpyrazole type.²⁰

In this study we report on the exploration of ¹⁸F-labeled PET tracers for the Y₁R derived from the argininamide BIBP3226 (1a, Figure 1, Table 1), a highly potent and selective Y₁R antagonist.²¹ Lately, the bioisosteric substitution of the guanidine group in the argininamide moiety by an acyl- or carbamoylguanidine group (e.g., compounds 1b, 1c, Figure 1) was shown to be an effective strategy to prepare highly potent and selective fluorescent ligands,^22,23 bivalent ligands,²⁴⁻²⁶ and tritiated radioligands for the human Y₁R.²⁷⁻²⁹ Likewise, the concept of guanidine-/acylguanidine nonclassical bioisosterism has been successfully applied to Y₂R selective argininamides^30,31 and arginine-containing peptidic GPCR ligands.³²

Figure 1.

Radiolabeled Y₁R ligands derived from argininamide-type antagonists such as 1c.

Table 1. Structures and Y₁R Affinities (K_i Values) of Parent Compound 1a, Amine Precursors 8, 14, 15, Alkyne 39, and Fluorinated Argininamides 4, 7, 17–26, 28, 33–36, 38, and 41; Y₁R Antagonism (K_b Values) of 1a and Fluorinated Argininamides with a K_i below 100 nM; clogP Values of the Potential PET Ligands 4, 7, 17–26, 28, 33–36, 38, and 41 (Calculated with ACD/Labs 12.0 Using the Depicted Guanidine Tautomeric Form).

^aDissociation constant determined from the displacement of [³H]1b (K_d = 1.2 nM, c = 1.5 nM) on SK-N-MC cells; mean values ± SEM from at least two independent experiments each performed in triplicate.

^bInhibition of 10 nM pNPY induced [Ca²⁺]_i mobilization in HEL cells; mean values ± SEM from at least two independent experiments.

^cReported by Keller et al. 2011.²³

^dReported by Keller et al. 2008.²⁷

^eReported by Keller et al. 2011.²⁸

^fReported by Keller et al. 2013.²⁶

^gclogP value was calculated for the drawn structure.

In the present study we demonstrate that this strategy of bioisosterism could be adapted to the design of fluorinated Y₁R selective PET ligands. For this purpose the argininamide 1a was 2-fluoropropanoylated or 4-fluorobenzoylated either directly at the guanidine entity (compounds 4 and 7) or at an amine-functionalized spacer attached to the guanidine group (17–26, 28, 32–36) (Scheme 1). One of the early compounds was N^ω-2-fluoropropionylated 1a (4), a congener of the high-affinity Y₁R radioligand [³H]1b.²⁷ In contrast to the latter,^27,28 compound 4 rapidly decomposed in buffer at pH 7.4 (cf. Supporting Information) and is therefore inappropriate as a radiotracer. Whereas argininamide 4 was synthesized through acylation of compound 2(27) using 4-nitrophenyl 2-fluoropropanoate (3), the 4-fluorobenzoylated analogue 7 was obtained via activation of precursor 5 by treatment with trifluoromethanesulfonic acid anhydride and subsequent reaction with the reported amine 6 (Scheme 1).²⁷ The majority of the series of potential PET ligands (compounds 17–26 and 28) were prepared through convenient acylation of the amine-functionalized precursor molecules 8–15^23,26 and 27 with active ester 3 or 16, affording the products 17–26 and 28 with an average yield of 68% (Scheme 1). Argininamides 33–36 were obtained through guanidinylation of amine 6 with S-methylisothiourea derived guanidinylating reagents (29–32), which contained the complete N^ω-substituent of the resulting argininamide (Scheme 1).

Scheme 1. Synthesis of the Fluorinated Argininamides 4, 7, 17–26, 28, 33–36, 38, and 41.

Reagents: (a) (1) triethylamine, acetonitrile; (2) TFA, acetonitrile, 84%; (b) (1) trifluoromethanesulfonic anhydride, triethylamine, CH₂Cl₂; (2) amine 6, CH₂Cl₂, 40%; (c) triethylamine, acetonitrile or DMF, 48-91%; (d) (1) triethylamine, CH₂Cl₂; (2) TFA, CH₂Cl₂, 50%; (e) (1) HgCl₂, triethylamine, DMF; (2) TFA, CH₂Cl₂, 54-85%; (f) (1) triethylamine, CH₂Cl₂; (2) TFA, CH₂Cl₂, 71%; (h) CuSO₄·5H₂O, sodium ascorbate, EtOH/H₂O 1:1, 37%.

Compounds 38 and 41 represent potential PET ligands, which were inaccessible by 2-fluoropropionylation or 4-fluorobenzoylation. However, 38 was obtained through treatment of 2 with succinimidyl ester 37 and subsequent deprotection (Scheme 1). The triazole derivative 41 was prepared from the alkyne functionalized precursor 39 and 2-deoxy-2-fluoro-β-d-glucosyl azide (40)³³ by Cu(I)-promoted “click” reaction.

The Y₁R affinities of the series of argininamides were determined by displacement of [³H]1b on SK-N-MC neuroblastoma cells (Table 1). Most of the derivatives of 1a proved to be high-affinity Y₁R ligands with K_i values between 1 and 65 nM. For the potential PET ligands with K_i values <100 nM (compounds 17–26, 28, 33, 34, 38) Y₁R antagonism was proven in a spectrofluorimetric Fura-2 Ca²⁺ assay on human erythroleukemia (HEL) cells³⁴ (Table 1). As described recently, acylation of amine precursor 14, which contains a carbamoylguanidine moiety, with tritiated propionic acid afforded a Y₁R radioligand ([³H]1c, K_d = 2.0 nM).²⁸ Likewise, acylation of 14 with 2-fluoropropionic acid gave 23, the compound with the highest Y₁R affinity (K_i = 1.3 nM) among the fluorinated ligands.

Acylation of the guanidine group in 1a with 4-fluorobenzoic acid, affording 7, and the introduction of N^ω-alkoxycarbonyl substituents (35, 36) resulted in compounds with considerably decreased Y₁R affinity (Table 1). Even though alkyne derivative 39 showed high Y₁R binding (K_i = 1.3 nM), the glucosylated derivative 41 exhibited significantly decreased Y₁R affinity (K_i = 2.0 μM). Therefore, the synthesis of such fluoroglucosylated Y₁R ligands as potential PET tracers was discontinued.

In addition to Y₁R binding constants determined on SK-N-MC neuroblastoma cells, the Y₁R affinity of compounds 18, 23, and 24 was investigated at MCF-7-Y₁²⁸ breast cancer cells. The K_i values amounted to 31 ± 11, 3.3 ± 0.1, and 6.8 ± 0.5 nM, respectively, and were in good agreement with the K_i values provided in Table 1 (SK-N-MC cells). The Y₁R selectivity was confirmed for argininamides 23 and 24 in flow cytometric binding studies using cyanine labeled peptides (Cy5-pNPY, Dy-635-pNPY, Cy5-[K⁴]hPP) and cells transfected with the human Y₂, Y₄, and Y₅ receptor (Table S1).

According to the in vitro Y₁R binding data, the 2-fluoropropionylated derivative 23 exhibited the most promising Y₁R affinity among the potential PET ligands under study. In addition to receptor binding characteristics, the hydrophilicity of a PET tracer for imaging of peripheral tumors is also a critical issue with regard to the ratio between compound accumulation in the tissue of interest and elimination/excretion. The acyl- or carbamoylguanidine entity of the herein reported argininamide-type Y₁R ligands exhibit a considerably lower basicity (pK_a ≈ 7–8) than alkylguanidines (pK_a ≈ 12–13), though high enough to account for a partial protonation of the compounds at pH 7.4 and thus sufficient water solubility (clogD(pH 7.4) < clogP). Nevertheless, the 4-fluorobenzoylated derivatives showed clogP values as high as 4.1–5.7 (Table 1), predicting unfavorable biodistribution due to high lipophilicity. The most promising derivative 23 demonstrated lower lipophilicity with a clogP value of 3.4 when compared to the series of compounds under study (Table 1).

In a first approach toward a suitable PET ligand for Y₁R, [¹⁸F]17 was prepared via acylation of amine precursor 8 using 4-nitrophenyl-2-[¹⁸F]fluoropropionate ([¹⁸F]NPFP, [¹⁸F]3)³⁵ as shown in Scheme 2, and the biodistribution was studied in nude mice bearing subcutaneous SK-N-MC xenografts (Figure 2A). The radiosynthesis afforded [¹⁸F]17 in radiochemical yield (RCY) of 60% for the final reaction step, when the preliminary biodistribution study showed high uptake of [¹⁸F]17 in the gall bladder and only marginal uptake by the tumor (Figure 2A). Taking into account the highly laborious radiosynthesis of [¹⁸F]3,³⁵ together with difficulties in the detection of the SK-N-MC tumors with [¹⁸F]17 in PET imaging experiments (results not shown), we envisaged an improved PET ligand and imaging of Y₁R-positive breast cancer. The in vitro data, presented in Tables 1 and 1S, prompted us to synthesize the ¹⁸F-labeled form of 23, which has a 20-fold higher Y₁R affinity compared to compound 17 (cf. Table 1). Moreover, the in vivo studies were performed with nude mice bearing MCF-7 tumors, for which we proved the expression of Y₁R by immunofluorescence analysis (Figure S4, Supporting Information), confirming our previous studies.^8,27,28

Scheme 2. Synthesis of the PET ligands [¹⁸F]17 and [¹⁸F]23.

Reagents and conditions: [¹⁸F]17: 1.22 μmol of 8, triethylamine, acetonitrile/DMSO 10:1, 50 °C, 15 min, RCY = 60%. [¹⁸F]23: 1.47 μmol of 14, triethylamine, acetonitrile/DMSO 11:1, 50°C, 5 min, RCY = 94%.

Figure 2.

Biodistribution data of [¹⁸F]17 (A) and [¹⁸F]23 (B) in SK-N-MC or MCF-7 tumor-bearing mice. Each bar represents the mean ± SEM of 3–4 mice (except for SK-N-MC tumor, n = 2).

Apart from higher Y₁R affinity, 23 showed slightly higher hydrophilicity over 17 (clogP = 3.4 vs 3.9), and most importantly, 23 contains a carbamoylguanidine structure, which was recently reported to be highly stable at pH 7.4, contrary to the acylguanidine moiety of 17.^23,28,32

[¹⁸F]23 was prepared by treatment of an excess of amine precursor 14 with [¹⁸F]3 (Scheme 2). [¹⁸F]3 was prepared following a previously established protocol,³⁵ with some important modifications to optimize the overall synthesis time and RCY (Supporting Information): The ¹⁸F-fluorination of the 9′-anthrylmethyl-2-bromopropionate precursor was performed under mild reaction conditions in the presence of KH₂PO₄,³³ resulting in a reduced precursor concentration needed for radiolabeling (30 mM instead of 116 mM³⁵). This allowed us to omit the HPLC separation after the ¹⁸F-fluorination step and to perform the subsequent reaction steps of the radiosynthesis of [¹⁸F]3 by a simple one-pot procedure (Supporting Information). After isolation of [¹⁸F]3 by semipreparative HPLC and conjugation to amine 14, the radioligand [¹⁸F]23 was obtained in an overall RCY of 5–8% (uncorrected for decay) within an overall synthesis time of only 70–80 min and in a radiochemical purity of >99% and molar radioactivity of 12–21 GBq/μmol (n = 6).

Because [¹⁸F]23 revealed excellent in vitro Y₁R binding affinity and sufficient RCY from the optimized radiosynthesis, we proceeded with the evaluation of the radioligand in vitro and in vivo using tumor-bearing nude mice with Y₁R-positive MCF-7 tumors. The experimental logD_7.4 determination of [¹⁸F]23 revealed a value of 2.34 ± 0.03, which differed from the calculated clogD_7.4 value of 3.2 (ACD/Laboratories), possibly due to an underestimation of the protonated fraction of the ligand at physiological pH. Stability experiments in vivo displayed excellent stability of [¹⁸F]23, as no degradation products could be observed in the blood of mice at 90 min postinjection (p.i.) (Figure S2, Supporting Information).

Using MCF-7 tumor-bearing nude mice with a tumor mass of 80–240 mg, we studied whether [¹⁸F]23 could be suitable for the visualization of Y₁R positive tumors. As shown in Figure 2B, [¹⁸F]23 revealed fast blood clearance in nude mice bearing MCF-7 xenografts at 30 and 90 min p.i., low uptake in the liver, but exceptionally high uptake in the gall bladder (>100 %ID/g). The biodistribution of [¹⁸F]23 was surprisingly similar to that of [¹⁸F]17 (Figure 2A). In this context, it has to be noted that these data cannot be strictly compared due to the use of different animal models.

[¹⁸F]23 showed marginally higher uptake in the intestines and lower uptake in the femur. Since decomposition of ¹⁸F-fluoropropionylated radiotracers frequently leads to increased accumulation of [¹⁸F]fluoride in the bones,^36,37 the lower uptake of [¹⁸F]23 in the femur could be ascribed to the higher stability of the carbamoylguanidine as compared to the acylguanidine structure of [¹⁸F]17. The uptake of [¹⁸F]23 in MCF-7 tumors was 0.51 ± 0.04 %ID/g at 30 min p.i., and the tracer demonstrated high tumor retention with an uptake of 0.43 ± 0.08 %ID/g at 90 min p.i. (Figure 2B). Consequently, the tumor/blood ratio was about 2, also at the later time point at 90 min p.i., suggesting a suitable signal-to-noise ratio for PET imaging. In comparison with the MCF-7 tumor uptake of the previously described ¹⁸F-labeled NPY peptide analogue (0.7 %ID/g at 120 min p.i.),¹⁸ the tumor uptake of [¹⁸F]23 (0.4 %ID/g at 90 min p.i.) is lower, as the biodistribution of [¹⁸F]23 and its pronounced uptake in the gall bladder clearly hampers the suitability of [¹⁸F]23 for PET imaging purposes. However, [¹⁸F]23 revealed a significantly reduced kidney uptake by a factor of about 10 when compared with the peptide tracer, which is, as expected, the major advantage of the nonpeptide radioligand compared with a peptide ligand for PET imaging.

[¹⁸F]23 was injected into MCF-7 xenografted nude mice for PET imaging studies and demonstrated displaceable and specific Y₁R-mediated tumor uptake (Figure 3A,B), regardless of its biodistribution with predominant accumulation in the gall bladder, intestines, and kidneys (Figure 3C). The in vivo specificity of [¹⁸F]23 was proven by coinjection of 1a (40 μg/animal, bolus injection of 100 μL over a few seconds), demonstrating a significantly diminished tracer uptake in the tumors of coinjected animals (0.42 ± 0.13 %ID/g (n = 5, Figure 3A) vs 0.21 ± 0.07 %ID/g (n = 4, Figure 3B), P < 0.05, Mann–Whitney U test, two-tailed). In addition, to exclude that differences in tumor uptake could be a result of overall biodistribution changes when 1a is administered prior to tracer injection, we determined tumor-to-muscle ratios by PET image analysis, confirming significantly diminished tumor uptake of [¹⁸F]23 by coinjection of 1a (3.1 ± 0.9 (n = 5, control) vs 1.8 ± 0.7 (n = 4, coinjection), P < 0.05, paired t test). These results confirmed the visualization of Y₁R-positive MCF-7 tumors by the antagonist [¹⁸F]23 in vivo by PET.

Figure 3.

Representative coronal PET images of a MCF-7 tumor-bearing mouse injected with [¹⁸F]23 alone (A) or with [¹⁸F]23 and 1a (40 μg, B). The blue crosshair indicates the tumor. (C) PET image of the same mouse with adjusted scale bar showing the biodistribution of [¹⁸F]23 in the plane of the intestines (left) and in the kidneys (right): b, bladder; g, gall bladder; i, intestines; k, kidneys; l, liver.

In conclusion, we developed the antagonist radioligand [¹⁸F]23 as the first nonpeptide ¹⁸F-labeled tracer that has been evaluated for in vivo small animal PET imaging of Y₁R-positive mammary carcinoma. The biodistribution of [¹⁸F]23 is characterized by its hepatobiliary clearance with very slow clearance of the gall bladder. We suggest [¹⁸F]23 as a lead for the design of improved PET ligands with more favorable biodistribution and higher Y₁R-dependent tumor uptake to enable PET imaging studies. Further preclinical studies of derivatives of [¹⁸F]23 are currently underway.

Glossary

ABBREVIATIONS

brs

broad singlet

CH₂Cl₂

dichloromethane

DIPEA

diisopropylethylamine

EDC·HCl

1-(3-(dimethylamino)propyl)-3-ethylcarbodiimide hydrochloride

EtOAc

ethyl acetate

HEL cells

human erythroleukemia cells

HOBt

1-hydroxybenzotriazole

hPP

human pancreatic polypeptide

k

retention (or capacity) factor (HPLC)

K_d

dissociation (or binding) constant obtained from a saturation binding experiment

K_i

dissociation (or binding) constant obtained from a competition binding experiment

logD_7.4

logarithm of distribution coefficient (pH of the aqueous phase = 7.4)

MeCN

acetonitrile

NPY

neuropeptide Y

p.i.

postinjection

pNPY

porcine NPY

RCY

radiochemical yield

RP-HPLC

reversed-phase HPLC

SEM

standard error of the mean

TBTU

2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate

Y₁R

NPY Y₁ receptor

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.6b00467.

• Experimental description of the synthesis and analytical data of compounds 3–5, 7, 16–26, 28–39, 41 and further intermediates. RP-HPLC chromatograms of compounds 4, 7, 17–26, 28, 33–36, 38, 39 and 41. Investigations on the chemical stability of 4. Pharmacology-related experimental protocols. Data on Y₁R selectivity of 23 and 24. Experimental protocols for the synthesis of the PET ligands [¹⁸F]17 and [¹⁸F]23, for animal studies, PET imaging, Western blot, and immunofluorescence analysis (PDF)

Author Present Address

^∥ Ascendis Pharma GmbH, Im Neuenheimer Feld 584, D-69120 Heidelberg, Germany.

Author Present Address

^⊥ Piramal Imaging GmbH, Tegeler Str. 7, D-13353 Berlin, Germany.

Author Contributions

All authors have given approval to the final version of the manuscript.

This work was supported by the Deutsche Forschungsgemeinschaft (grant MA 4295/1–3 and Graduate Training Programs (Graduiertenkollegs) GRK 760 (“Medicinal Chemistry: Molecular Recognition–Ligand-Receptor Interactions”) and GRK 1910 (“Medicinal Chemistry of Selective GPCR Ligands”)).

The authors declare no competing financial interest.