Development of a Neurotensin-Derived 68Ga-Labeled PET Ligand with High In Vivo Stability for Imaging of NTS1 Receptor-Expressing Tumors

Simple Summary Cancer diagnostics based on molecular imaging techniques such as positron emission tomography (PET) requires radiolabeled tracers, which are taken up by tumors. As the neurotensin receptor type 1 (NTS1R) is present in certain malignant tumors, radiolabeled NTS1R ligands can serve as molecular tools for tumor imaging. A straightforward approach for developing NTS1R PET ligands would be the preparation of fluorine-18 or gallium-68 labeled analogs of the peptide neurotensin. However, as neurotensin derivatives are prone to enzymatic cleavage, structural modifications are needed to prevent peptide degradation while retaining NTS1R affinity. Applying a new strategy for peptide stabilization, it is possible to develop a peptidic gallium-68 labeled NTS1R PET ligand with high in vivo stability and high NTS1R affinity. Investigations of the PET ligand in mice with subcutaneous NTS1R-positive tumors revealed the NTS1R-mediated visualization of the tumor. Future developments, such as NTS1R PET ligands with improved biodistribution, will benefit from these results. Abstract Overexpression of the neurotensin receptor type 1 (NTS1R), a peptide receptor located at the plasma membrane, has been reported for a variety of malignant tumors. Thus, targeting the NTS1R with 18F- or 68Ga-labeled ligands is considered a straightforward approach towards in vivo imaging of NTS1R-expressing tumors via positron emission tomography (PET). The development of suitable peptidic NTS1R PET ligands derived from neurotensin is challenging due to proteolytic degradation. In this study, we prepared a series of NTS1R PET ligands based on the C-terminal fragment of neurotensin (NT(8–13), Arg8-Arg9-Pro10-Tyr11-Ile12-Leu13) by attachment of the chelator 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) via an Nω-carbamoylated arginine side chain. Insertion of Ga3+ in the DOTA chelator gave potential PET ligands that were evaluated concerning NTS1R affinity (range of Ki values: 1.2–21 nM) and plasma stability. Four candidates were labeled with 68Ga3+ and used for biodistribution studies in HT-29 tumor-bearing mice. [68Ga]UR-LS130 ([68Ga]56), containing an N-terminal methyl group and a β,β-dimethylated tyrosine instead of Tyr11, showed the highest in vivo stability and afforded a tumor-to-muscle ratio of 16 at 45 min p.i. Likewise, dynamic PET scans enabled a clear tumor visualization. The accumulation of [68Ga]56 in the tumor was NTS1R-mediated, as proven by blocking studies.


Introduction
Neurotensin (NT), a linear 13 amino acid peptide, acts as a hormone in the gastrointestinal tract, regulating, inter alia, motility and mucosal regeneration [1], and as a neurotransmitter and neuromodulator in the central nervous system, where it is involved, inter alia, in the regulation of body temperature, food intake, blood pressure, nociception, memory, and hormone secretion [2][3][4][5][6][7]. The effects of neurotensin are mainly mediated by the neurotensin receptors 1 and 2 (NTS 1 R, NTS 2 R), members of the family of G-protein coupled receptors. The NTS 1 R has emerged as an interesting target for tumor visualization and therapy due to its overexpression in a variety of tumors such as breast cancer, colorectal carcinoma, and (the prognostically poor) pancreatic adenocarcinoma [8][9][10]. The carboxyterminal hexapeptide of NT (NT (8)(9)(10)(11)(12)(13), 1, Figure 1A) was identified as the biologically active fragment, exhibiting the potency of full-length NT [11][12][13]. Therefore, peptide 1 has previously served as a lead structure for the development of imaging agents addressing the NTS 1 R [14][15][16][17][18][19][20]. 68 Ga-and 18 F-labeled ligands of cell-surface receptors that are (over-)expressed in malignant tumors are considered useful tools for in vivo cancer imaging by positron emission tomography (PET) [21][22][23][24][25][26][27][28]. 18 F-labeled PET tracers are advantageous with respect to half-life (about 110 min) and achievable resolution, but require a cyclotron for radionuclide synthesis and usually two or more radiosynthetic reaction steps [29,30]. In contrast, the advantage of 68 Ga-labeled PET tracers lies in their fast one-step radiosynthesis (incorporation of 68 Ga in a chelator moiety) and convenient radionuclide accessibility ( 68 Ge/ 68 Ga-generator); however, these tracers result in lower resolution images and the short half-life (68 min) does not allow a transfer between clinics. The development of PET ligands with favorable properties (e.g., high receptor affinity, high in vivo stability and appropriate pharmacokinetics) is challenging. The development of peptidic PET tracers, often acting as receptor agonists, can be convenient with respect to high target affinity and attachment of the label [31][32][33], but high proteolytic stability in vivo might not be easily achieved [33,34]. With respect to NTS 1 R PET ligands, two main strategies have been pursued [25], i.e., investigations of peptidic agonists and of non-peptidic antagonists [35][36][37]. To date, reported 18 F-and 68 Ga-labeled NTS 1 R antagonists exhibit higher receptor affinities and higher in vivo stabilities compared to peptidic NTS 1 R PET ligands explored with regard to in vivo tumor imaging. However, the pharmacokinetic profile of the antagonists is not well-suited for PET imaging based on short-lived radionuclides [38]. Unlike antagonists, agonist binding induces receptor internalization; thus, peptidic PET ligands potentially allow for a higher tracer uptake in the tumor.  [39]) which were applied in SPPS for the preparation of aminofunctionalized precursor peptides. a Ki value; Schindler et al. [40], b IC50 value; Alshoukr et al. [41], c Ki value; Maschauer et al. [42].
In the present study, we aimed at the development of a peptidic NTS 1 R PET ligand matching up with reported receptor antagonists in terms of NTS 1 R affinity and in vivo stability. For this purpose, N ω -carbamoylated arginines derived from 6 or 7 were incorporated into 3 or slightly modified analogs of 3, optimized with respect to plasma stability, followed by the attachment of a DOTA chelator to the modified arginine side chain and insertion of stable ("cold") Ga 3+ or radioactive 68 Ga 3+ . The potential NTS 1 R PET ligands ("cold" compounds) were characterized with respect to NTS 1 R and NTS 2 R affinity, and plasma stability. For selected peptides, the 68 Ga-labeled analogs were prepared and studied in vivo in tumor-bearing mice.
Annotation concerning the 1 H-NMR spectra (solvent: DMSO-d 6 ): in order to allow an integration of the signals interfering with the broad water signal at ca. 3.5 ppm, spectra were additionally recorded in DMSO-d 6

Cell Culture and Preparation of HEK293T Cells Stably Expressing the Human NTS 2 R
All cells were cultured in 75 or 175 cm 2 flasks (Sarstedt, Nuümbrecht, Germany) in a humidified atmosphere (95% air, 5% CO 2 ) at 37 • C. HT-29 colon carcinoma cells (DSMZ-no. ACC 299) were maintained in antibiotic-free RPMI medium (Sigma-Aldrich) supplemented with 7.5% fetal bovine serum (FBS) (Sigma-Aldrich). HEK293T cells stably expressing the human NTS 2 R (HEK293T-hNTS 2 R cells) were essentially generated following a previously described procedure [61]. In brief, HEK293T cells (kind gift from Prof. Dr. Wulf Schneider, Institute for Medical Microbiology and Hygiene, University of Regensburg, Germany) were seeded on a 6-well plate (Sarstedt, Nümbrecht, Germany) in Dulbecco's modified Eagle's medium (Sigma-Aldrich) supplemented with 10% FBS, L-glutamine (2 mM) (Sigma-Aldrich) and Penicillin-Streptomycin (100 IU/mL and 0.1 mg/mL, respectively) (Sigma-Aldrich) at a density of 6 × 10 5 cells/well. On the next day, cells were transfected with 2 µg of cDNA encoding the hNTS 2 R (cDNA Resource Center, Rolla, MO, USA, catalog no. NTSR200000) using X-tremeGENE TM HP (Roche Diagnostics, Mannheim, Germany) as transfection reagent according to the manufacturer's protocol. After two days of transfection, cells were detached with trypsin-ethylenediamine-tetraacetic acid (EDTA, Biochrom, Berlin, Germany) and transferred to a 15-cm dish (Sarstedt, Nümbrecht, Germany). After the cells had attached to the dish, G418 (Biochrom, Berlin, Germany) was added at a final concentration of 1 mg/mL. Selection was achieved by exchanging the medium every two to three days for two weeks. Subsequently, a clone with high NTS 2 R-expression, which was assessed radiochemically after addition of 10 nM of [ 3 H]UR-MK300, was isolated. Cultivation was then continued with a reduced G418 concentration in the culture medium of 600 µg/mL.  [39] or 65.0 Ci/mmol; for structure see Figure S1, Supplementary Materials) at hNTS 1 R-expressing intact human HT-29 colon carcinoma cells were performed at 23 ± 1 • C as described previously [39]. Two different batches of the radioligand [ 3 H]UR-MK300 were used. The K d values of [ 3 H]UR-MK300 amounted to 0.55 nM (mean value from two independent saturation binding experiments, each performed in triplicate) [40] and 0.41 ± 0.12 nM (mean value ± SD from two independent saturation binding experiments, each performed in triplicate). Specific binding data (obtained by subtracting unspecific binding from total binding) were normalized (100% = specifically bound radioligand in the absence of competitor) and plotted over log(concentration of competitor) followed by a four-parameter logistic fit (SigmaPlot 12.5, Systat Software, San José, CA, USA) (note: in the case of 40, the lower curve plateau of the sigmoidal fit was constrained to >0). Resulting pIC 50  were treated with poly-D-lysine hydrobromide (Sigma-Aldrich) for 10 min. The wells were washed with H 2 O and the plates were dried on air at rt overnight. Alternatively, plates were treated with a sterile solution of 5% (w/v) gelatin (Sigma-Aldrich) in H 2 O (50 µL) at rt for 1.5-2 h. The gelatin solution was removed, followed by the addition of a solution of 2.5% (v/v) of glutaraldehyde (Sigma-Aldrich) in H 2 O (50 µL) at rt for 10 min. After removal of the glutaraldehyde solution, the wells were washed twelve times with H 2 O and two times with culture medium (150-300 µL). One day before the experiment, cells were seeded in the treated plates at a density of 9 × 10 4 cells/well. On the day of the experiment, the culture medium was carefully removed using a multi-channel pipette (Transferpette S-12, Brand, Wertheim, Germany) and the cells were washed once with Dulbecco's phosphate-buffered saline (D-PBS) containing Ca 2+ and Mg 2+ (1.8 mM CaCl 2 , 2.68 mM KCl, 1.47 mM KH 2 PO 4 , 3.98 mM MgSO 4 , 136.9 mM NaCl and 8.06 mM Na 2 HPO 4 ) (200 µL, rt) followed by the careful pre-filling of the wells with 180 µL (total binding) or 160 µL (unspecific and competition binding) of D-PBS, supplemented with 1% BSA and 100 µg/mL bacitracin (Serva, Heidelberg, Germany) (in the following referred to as binding buffer). To determine total binding, 20 µL of a solution of the radioligand in binding buffer (10-fold concentrated compared to the final concentration) were added. For the determination of unspecific binding, 20 µL of a solution of 1 in binding buffer (10-fold concentrated, used in 500-fold excess compared to the radioligand) and 20 µL of a 10-fold concentrated solution of the radioligand in binding buffer were added. To determine the displacing effect of a compound of interest, 20 µL of a solution of the respective compound in binding buffer (10-fold concentrated) and 20 µL of a 10-fold concentrated solution of the radioligand in binding buffer were added. During the incubation period of 2 h at 23 • C, the plates were gently shaken. After incubation, the liquid was carefully removed using a multi-channel pipette and the cells were carefully washed twice with ice-cold D-PBS (200 µL). 25 µL of lysis solution (8 M urea, 3 M acetic acid, and 1% Triton-X-100 in H 2 O) were added to each well and the plates were shaken at rt for 25 min, followed by the addition of liquid scintillator (Ultima Gold, PerkinElmer, Waltham, MA, USA) (200 µL). The plates were sealed with a transparent sealing tape (permanent seal for microplates, PerkinElmer, product no. 1450-461) and turned upside down several times to achieve complete mixing. Prior to the measurement of the radioactivity with a MicroBeta2 plate counter (PerkinElmer), the plates were kept in the dark for at least 1 h. All experiments were performed in triplicate. The K d values of [ 3 H]UR-MK300, determined for the different batches of radioligand by saturation binding experiments, amounted to 6.9 ± 1.8 nM (mean value ± SD from six independent determinations, each performed in triplicate) and 4.0 ± 1.5 nM (mean value ± SD from three independent determinations, each performed in triplicate) (for representative saturation binding curves see Figure S2, Supplementary Materials). Data from competition binding experiments were analyzed as described for NTS 1 R binding using a K d value of 6.9 nM (1, 8, 9, 14, 15, and 19)

Circular Dichroism (CD) Analysis
CD spectra of 100 µM aqueous solutions of 48 and 49 and a reference compound with the amino acid sequence Me-Arg-Arg-Pro-Tyr-Ile-Leu [40] were recorded in a 1 cm path length cuvette at 20 • C with a Jasco J-810 spectropolarimeter (Jasco, Tokyo, Japan) equipped with a PTC-423S Peltier temperature controller (Jasco). Instrumental parameters: spectral range, 180−300 nm; bandwidth, 1 nm; scanning speed, 500 nm/min. Each spectrum represents the average of three spectra, each recorded with 20 accumulations, after solvent subtraction. An "economy-size" singular-value decomposition (SVD) on a set of nine spectra (matrix A consisting of three spectra for each of 48, 49 and the reference compound) was calculated in MATLAB (MathWorks, Natick, MA) as A = U·S·V T . Here, U is a matrix whose columns contain the linearly independent spectral components, S is a diagonal matrix containing the singular values, and V T is the transpose of matrix V, which contains the linear coefficients associated with the spectral components in U. , the product was eluted from the cartridge with EtOH (effective volume ca. 1 mL) and the eluate was transferred into a 2-mL reaction vessel. The solvent was evaporated in a Savant Speed-Vac SVC100H vacuum concentrator (Savant Instruments, Farmingdale, NY, USA) equipped with pre-heated (100 • C) rotor inserts (aluminum blocks with bores for 1.5-and 2-mL reaction vessels) for approximately 50 min (note: a complete evaporation to dryness was avoided; the residual volume was approximately 10-30 µL), followed by uptake in 0.1% aq HCOOH (80-100 µL). The solution was subjected to preparative work-up using an HPLC system composed of a P4000 pump (Thermo Separation Products), a Degassex DG-4400 degasser (Phenomenex), a 2487 UV/visible detector (Waters) and a Rheodyne manual injector equipped with a 200 µL loop (note: the pump was directly controlled via the front panel and the UV/Vis-detectorwas remote-controlled using Waters Millennium Software). For the detection of γ-radiation, a B20/G-10 RADEye (Thermo Scientific, Erlangen, Germany) was placed close to the outlet tubing of the UV/Vis-detector (note: the vessel used for waste collection was shielded from the RADEye by 2 cm of lead). The stationary phase, a Luna C18(2), 3 µm, 150 × 4.6 mm (Phenomenex), was placed in a box of lead (wall thickness: 2 cm). The flow rate was 0.5 mL/min. Mixtures of A5 and B2 were used as mobile phase. The following linear gradients were applied: [ 68 Figure S7, Supplementary Materials). The eluate, containing the PET ligand, was collected in a 2-mL reaction vessel immediately followed by removal of the solvent in a vacuum concentrator equipped with pre-heated (100 • C) aluminum blocks (35-40 min). However, a complete evaporation of the solvent was avoided. The aqueous residue (20-40 µL) was taken up in PBS (ABX Kit) (80-500 µL) yielding a solution referred to as "tracer stock" in the following (final activity 0.157-1.69 GBq/mL). Quality controls of the PET ligands were performed by HPLC analysis using a system from Agilent Technologies (Waldbronn, Germany) consisting of a 1100 Series quaternary pump equipped with a 1260 Infinity degasser, a 1100 Series Autosampler, a 1100 Series Thermostated Column Compartment, a 1100 Series Diode Array Detector, and a GABI Star radiometric detector (Raytest Isotopenmessgeräte GmbH, Straubenhardt, Germany). A Luna C18(2), 3 µm, 100 × 4.6 mm (Phenomenex) served as stationary phase. The flow rate was 0.95 mL/min and the temperature of the Column Compartment was set to 25 • C. The following linear gradients were applied: [ 68 Table 1.  56 were determined by adding a solution of the tracer in PBS (100 µL, ca. 0.20-0.34 MBq) to a mixture of n-octanol (500 µL) and PBS (pH 7.4) (400 µL) in a 2-mL HPLC vial (Agilent, article number 5182-0714) with a screw cap equipped with a septum (Agilent, article number 5182-0717). After vortexing the mixture for 2 min, two 100 µL aliquots of the upper phase (n-octanol) were taken. To obtain a sample of the lower phase (PBS), the HPLC vial was held upside down and approximately 100-150 µL of the aqueous phase were removed using a syringe equipped with a canula and collected in a reaction vessel. Two 10 µL aliquots of the aqueous phase were subjected to measurement. The activity of the aliquots was measured with a Canberra Genie 2000 system (Canberra, Rüsselsheim, Germany) using the Gamma Acquisition & Analysis Software Genie 2000 3.4.1. Experiments were performed in triplicate. The decay-corrected counts per minute (cpm) values were averaged (n = 3) and transformed to a distribution coefficient logD 7.4 according to logD 7.4 = log(A octanol /A aqueous ), where A octanol is the mean of the decay-corrected cpm values obtained for samples of the n-octanol phase, and A aqueous is the mean of the decaycorrected cpm values obtained for samples of the aqueous phase.

Mouse Xenograft Model
8-12 weeks old female NMRI nude (nu/nu) mice (body weight 22-32 g) (Charles River, Sulzfeld, Germany) were kept under specified pathogen free (SPF) conditions at 23 • C, 55% relative humidity and a 12 h light/dark cycle in the central animal facility of the University of Regensburg using type III cages from Tecniplast (Hohenpeißenberg, Germany). The animals took food (Ssniff, Soest, Germay) and autoclaved tap water ad libitum. For tumor cell implantation, the culture medium of HT-29 cells was removed, the cells were detached from the culture flask by incubation for 2 min in trypsin-EDTA and the suspension was centrifuged (5 min, rt, 164 g). The supernatant was removed, and the cell pellet was washed twice with sterile PBS or serum-free medium (5-6 mL). The cells were resuspended in sterile PBS to a final density of 1 × 10 7 cells/mL. Mice were injected subcutaneously into the right flank with the HT-29 cell suspension (100 µL). After 2-4 weeks, when the tumors had reached a size of 50-500 mm 3 , animals were used for biodistribution and PET/computed tomography (CT) imaging studies (note: for PET/CT studies, the tumor size was at least 200 mm 3 ).

Biodistribution Studies
Aliquots of the tracer stock were diluted in PBS to give a volume of 200 µL. 80-100 µL (0.9-5.9 MBq) of this solution were injected into anesthetized HT-29 tumor-bearing nude mice via the tail vein. Animals were kept on a heating plate (set to 40 • C) or a pre-heated (ca. 45 • C) gel cushion and were killed by cardiac puncture 10 min, 25 min or 45 min p.i. immediately followed by taking blood, urine and tissue (i.e., tumor, kidney, liver, gall bladder, spleen, small intestine, heart, lung, brain, pancreas, femur and muscle) samples. Radioactivity measurement of the samples was performed with the Canberra Genie 2000 system described for the determination of the distribution coefficient logD 7.4 . Decay-corrected measured activities (cpm) were converted into activities (MBq) on the basis of an activity measurement with a 68 Ge-calibration standard source (Eckert&Ziegler, Berlin, Germany). Sample activities were converted to percentage of injected dose per gram tissue (%ID/g). Blocking experiments were performed by co-injection of the tracer with NTS 1 R ligand 48 (ca. 700 nmol per mouse). Mice were killed 45 min p.i. and analyzed as described above. The animals used for the 45- The analysis was performed with the urine obtained from biodistribution and PET imaging studies, respectively (t = 10 or 45 min). The urine was diluted with H 2 O (1:2-1:100 depending on the activity in the urine) and 10-80 µL of this solution were subjected to analysis by analytical HPLC using the same HPLC system and conditions as described for the quality controls of the PET ligands. To confirm the identity of [ 68 Ga]56 in the 45-min urine samples, an additional analysis was performed, using the aforementioned injection solution spiked with 100 µM 56. Representative chromatograms of the HPLC analysis of the urine samples are shown in Figures S9-S11 This analysis was performed with the blood obtained from biodistribution studies with [ 68 Ga]56 (t = 10 min). Blood (ca. 200 µL) was taken from the heart using a syringe that was rinsed with sodium heparin (25000 I.E., Ratiopharm, Ulm, Germany). The heparinized blood was transferred into a 1.5-mL reaction vessel immediately followed by centrifugation (5 min, 4 • C, 1200 g) using a Biofuge fresco centrifuge (Heraeus, Hanau, Germany). The supernatant (ca. 100 µL) was treated with the same volume of 10% aq TFA (precipitation of proteins) and the mixture was centrifuged (10 min, 4 • C, 16,100 g). The supernatant (70-100 µL) was subjected to analysis by analytical HPLC using the same HPLC system and conditions as described for the quality control of the PET ligand. To confirm the identity of [ 68 Ga]56 in the plasma samples, an additional analysis was performed, using the aforementioned injection solution spiked with 100 µM 56. Representative chromatograms of the HPLC analysis of the plasma samples are shown in Figure S11C-E, Supplementary Materials.

Determination of the Internalization of [ 68 Ga]56 in HT-29 Tumor Cells
HT-29 cells were seeded in 24-well TC plates (Sarstedt, catalog no. 83.3922) one day prior to the experiment at a density of 4.5 × 10 5 cells/well. Shortly before the experiment, the culture medium was removed and the cells were washed once with PBS (500 µL, rt) followed by pre-filling of the wells with 250 µL of binding buffer (see procedure for NTS 2 R binding studies). For the determination of unspecific binding, the binding buffer was supplemented with 1 (2 µM). 50 µL aliquots of the [ 68 Ga]56 stock in PBS (1.5-2.8 MBq) were added to the wells, and the cells were incubated at 37 • C for 5, 10, 15, 25, 35, 55, or 75 min under gentle shaking. After completed incubation, the plates were immediately placed on ice, the liquid was removed by suction and the cells were washed twice with ice-cold PBS (500 µL). The cells were washed twice with ice-cold acid strip buffer (50 mM glycine and 125 mM NaCl in H 2 O, pH 3.0) (300 µL) for 5 min each, and the washings were combined in 5-mL polypropylene tubes. The cells were then lysed by the addition of lysis solution (see procedure for NTS 2 R binding studies) (250 µL) and shaken at 37 • C for 15 min. The lysates were transferred into 5-mL polypropylene tubes, the wells were washed with lysis solution (250 µL) and the washings were combined with the lysates. The activities were measured with the Canberra Genie 2000 system as described for the determination of the distribution coefficient logD 7.4 . The decay-corrected cpm values of unspecific binding were subtracted from the decay-corrected cpm values of total binding to obtain specific binding data for both surface-bound [ 68 Ga]56 (acid strip) and internalized [ 68 Ga]56 (cell lysate). The amount of internalized specific binding was normalized based on total specific binding, and the mean values from two independent experiments (each performed in triplicate) were plotted against the incubation time.

Imaging Analysis
For dynamic analysis of the PET images, list mode data were replayed according to the following frame scheme: 6 × 10 s, 4 × 30 s, 1 × 2 min, 4 × 5 min, 2 × 10 min. All PET scans were corrected for normalization, detector dead time, attenuation, scatter, decay, random coincidences and prompt gamma coincidences. Attenuation corrected PET images (512 × 512 pixels, pixel size 0.40 mm, slice thickness 2.03 mm) were reconstructed by iterative reconstruction (8 iterations, 24 subsets, point spread function modelling) with a Gaussian post reconstruction filter with 1.0 mm full width at half maximum (FWHM). Additionally, static frames from 5 min to 10 min and for the entire acquisition of 45 min were reconstructed (512 × 512 pixels, pixel size 0.40 mm, slice thickness 0.6 mm) from the list mode data. By means of static recording over the complete acquisition period, possible movements of the mouse could be excluded. For determination of the tracer uptake in tumors and kidneys corresponding regions of interest (ROIs) were generated using a fixed threshold of 45% of the maximum tracer accumulation for tumors and 40% of the maximum tracer accumulation for kidneys. The resulting delineations were inspected visually and corrected manually if necessary. Manual correction was performed for all tumors in mice used for blocking experiments which exhibited only low diffuse tracer accumulation. For determination of the tracer uptake in the muscles, a ROI was delineated manually in the left femoral muscles observing a volume of approximately 0.05 mL. The ROIs were transferred to all time frames of the respective dynamic study. Time-activity-curves (TACs) were generated by computing the mean standardized uptake value normalized to body weight (SUV mean ) in each frame. ROI definition and ROI analyses were performed using the ROVER software, version 3.0.64 (ABX GmbH, Radeberg, Germany).

Chemistry
Standard Fmoc strategy solid-phase peptide synthesis (SPPS) was used for the preparation of the NT(8-13)-derived peptides 8, 9, 10 [60], 11, and 12, containing an aminofunctionalized arginine (position 8 or 9) derived from the reported Fmoc-and Boc-protected N ω -carbamoylated arginine building blocks 6 or 7 [39] (structures shown in Figure 1B) (Scheme 1). Peptides 8-12 were N α -methylated in position 8 (8-10 and 12) or 9 (11), and α-tert-butyl-Gly (Tle) was incorporated in position 12 in peptides 8 and 10-12 instead of Ile 12 . Details on the coupling conditions are provided in Table S1, Supplementary Materials. A reported procedure for the on-resin N α -methylation of peptides [66], which was recently used for N α -methylation of the carbamoylated arginine in 10 [60], was also successfully applied for N α -methylation of the carbamoylated arginines in 11 and 12. Coupling of Fmoc amino acids to an N α -methylated N-terminal amino acid using the standard coupling reagents HOBt, HBTU and DIPEA proved to be unfeasible; therefore, 11 was prepared by applying a combination of oxyma and DIC as activation reagents. Detailed information on the synthesis procedures and the applied coupling conditions are given in the Supplementary Materials.
Compounds 8-12 served as starting materials for the syntheses of the chelator-conjugated peptides 14-18 using the tris-tBu-protected DOTA reagent 13 for coupling to the aminofunctionality of the N ω -carbamoylated arginine in 8-12. Attempts to introduce the DOTA moiety using a non-protected DOTA succinimidyl ester caused severe separation problems due to nearly identical HPLC retention times of precursor peptide and product. However, the tBu-protected intermediates could easily be separated from the remaining starting material, followed by deprotection with acid overnight and purification, yielding 14-18 with HPLC purities of >99%.
The "cold" PET ligands 19-23 were prepared by incubation of 14-18 with natural 69 Ga 3+ in a HEPES buffer pH 4.2 at 100 • C (Scheme 1). Complete conversion of the starting material was achieved after only 10 min. Under these conditions, the peptides proved to be stable. It should be noted that the potential PET ligands 19-23 could not be separated from the remaining respective precursor peptide (C18 RP-HPLC) when using acetonitrile and 0.04% aqueous TFA as eluent. However, baseline separation was achieved using MeOH and 0.05% formic acid as mobile phase.
For the purpose of the preparation of less polar PET ligands, peptides 10 and 12, containing a tetramethylene and a dioxaoctamethylene linker, respectively, in the aminofunctionalized N ω -carbamoylated arginine, were conjugated to the fluorinated biphenyl-Ala spacer 24, yielding 25 and 26 (after subsequent Boc-deprotection), or to the fluorinated aminomethyl-benzoyl spacer 27, affording 28 and 29 (after subsequent Bocdeprotection), using HOBt, HBTU and DIPEA as coupling reagents. The side chain of the carbamoylated arginine in 26 was further elongated by treatment with succinimidyl ester 30, yielding a terminal aminohexanoyl moiety in the arginine side chain after subsequent Boc-deprotection (31).
Compounds 31, 25, and 26 were treated with 13 as described above, giving the DOTAconjugated compounds 32, 34 and 35 after removal of protecting groups (note: 28 and 29 were not further processed by coupling to DOTA as they proved to be more polar (shorter RP-HPLC retention times) than 25, 31 and 26). Finally, 32, 34, and 35 were converted into the potential PET ligands 33, 36, and 37 by insertion of Ga 3+ . Unlike the synthesis of 19-23, complete conversion of the starting material was only achieved after incubation at 100 • C for 30 min (as verified by analytical HPLC).
Aiming at a PET ligand with high in vivo stability, a series of N-terminally methylated NT (8)(9)(10)(11)(12)(13) derivatives was synthesized by SPPS containing various commercially available unnatural amino acids in position 12 (38)(39)(40)(41)(42)(43)(44)(45), 13 (46 and 47), or 11 (48 and 49) (Figure 2). Whereas the unnatural amino acids incorporated in peptides 38-47 represent enantiomerically pure derivatives of Ile and Leu, the incorporation of racemic β,β-dimethyl-tyrosine (β,β-diMe-Tyr) in position 11 yielded the epimers 48 and 49 (see Figure 2). In the case of 46 and 47, a 2-ClTrt-Cl resin had to be used instead of a H-Leu-2-ClTrt resin. For the coupling of the carboxy-terminal amino acid, the 2-ClTrt-Cl resin was treated with the respective Fmoc amino acid and DIPEA in CH 2 Cl 2 overnight. After quenching of unreacted starting material with MeOH, the loading of the resin with β-cyclopropyl-Ala (46) or α-methyl-Leu (47) was estimated to amount to 50% compared to the original loading of the resin with chloride. After side chain deprotection and cleavage from the resin, the overall-yields of 38-49 amounted to 15-74%. The amino-functionalized peptides 50 and 52 were treated with 13 as described above for the synthesis of 8-12 (cf. Scheme 1) to give the DOTA-conjugated compounds 54 and 55, respectively, in high yields (76% and 79%) after subsequent Boc-deprotection (Scheme 2). Insertion of Ga 3+ into the chelator moiety resulted in the PET ligand candidates 56 (UR-LS130) and 57 in high yields of 97% and 91%, respectively. Replacement of the N α -methylated Arg 8 in 48 and 49 by an N α -methylated, N ωcarbamoylated arginine derived from 6, led to peptides 50 and 51, and the additional replacement of Ile by Tle yielded 52 and 53 ( Figure 2). Isolation of the epimers 50/51 and 52/53, respectively, from one batch was necessary due to the usage of the same racemic β,β-diMe-Tyr building block as described for 48 and 49.
The amino-functionalized peptides 50 and 52 were treated with 13 as described above for the synthesis of 8-12 (cf. Scheme 1) to give the DOTA-conjugated compounds 54 and 55, respectively, in high yields (76% and 79%) after subsequent Boc-deprotection (Scheme 2). Insertion of Ga 3+ into the chelator moiety resulted in the PET ligand candidates 56 (UR-LS130) and 57 in high yields of 97% and 91%, respectively.

Circular Dichoism (CD) Analysis
To determine the configuration at the α-carbon of the β,β-dimethylated tyrosine at position 11, we measured CD spectra of 48 and 49 and compared them to the CD spectrum of the peptide Me-Arg-Arg-Pro-Tyr-Ile-Leu [40], representing an all-L-configured reference compound with tyrosine instead of β,β-dimethylated tyrosine in position 11 as the only difference to 48 and 49 ( Figure 3). To facilitate the assignment, we factorized the CD spectra into linearly independent spectral components by singular value decomposition (SVD), as described elsewhere [67]. As the three compounds differ in two properties, i.e., configuration at the α-carbon, β,β-dimethylation, or both, we expected three linear components, each contributing with a certain linear coefficient with either a positive or a negative sign. Indeed, performing the SVD on a set of nine spectra (three spectra each) resulted in three components being different from noise (cf. Figure 3B), whose reconstruction (under omission of linear components supposedly containing noise contribution only) resulted in nearly identical spectra as in Figure 3A (Figure 3C). Despite factorizing the SVD according to numerical variance and not to structural origin of spectral features, a rough assignment of the linear components was possible. Reconstruction of the spectra exclusively from spectral component 1 and the corresponding linear coefficients resulted in the spectra shown in Figure 3D. As this component represents the features with highest agreement between the three species, they are presumably associated with the backbone conformation of the peptides. In agreement with previous NMR and CD data on neurotensin in water [68], the maximum at 220 nm and the minimum at 190 nm indicate a lack of consecutive order in these peptides. After reconstruction of spectra with linear component 2, the resulting spectra for the reference compound and peptide 48 were nearly identical, whereas the spectrum reconstructed for 49 had the opposite sign. Therefore, component 2 is the one that indicates the configuration of the α-carbon, which allows assignment of the L-configuration to the β,β-dimethylated tyrosine in 48 and the D-configuration to the β,β-dimethylated tyrosine in 49. Finally, component 3 accounts for all the remaining spectral differences between the three species such as contributions from presence or absence of β,β-dimethylation. The configuration (R or S) of the 17emethylated tyrosine in 50-53 was assigned based on the comparison of elution orders in RP-HPLC (48, 50 and 52 elute before 49, 51 and 53, respectively).

Circular Dichoism (CD) Analysis
To determine the configuration at the α-carbon of the β,β-dimethylated tyrosine at position 11, we measured CD spectra of 48 and 49 and compared them to the CD spectrum of the peptide Me-Arg-Arg-Pro-Tyr-Ile-Leu [40], representing an all-L-configured reference compound with tyrosine instead of β,β-dimethylated tyrosine in position 11 as the only difference to 48 and 49 ( Figure 3). To facilitate the assignment, we factorized the CD spectra into linearly independent spectral components by singular value decomposition (SVD), as described elsewhere [67]. As the three compounds differ in two properties, i.e., configuration at the α-carbon, β,β-dimethylation, or both, we expected three linear components, each contributing with a certain linear coefficient with either a positive or a negative sign. Indeed, performing the SVD on a set of nine spectra (three spectra each) resulted in three components being different from noise (cf. Figure 3B), whose reconstruction (under omission of linear components supposedly containing noise contribution only) resulted in nearly identical spectra as in Figure 3A (Figure 3C). Despite factorizing the SVD according to numerical variance and not to structural origin of spectral features, a rough assignment of the linear components was possible. Reconstruction of the spectra exclusively from spectral component 1 and the corresponding linear coefficients resulted in the spectra shown in Figure 3D. As this component represents the features with highest agreement between the three species, they are presumably associated with the backbone conformation of the peptides. In agreement with previous NMR and CD data on neurotensin in water [68], the maximum at 220 nm and the minimum at 190 nm indicate a lack of consecutive order in these peptides. After reconstruction of spectra with linear component 2, the resulting spectra for the reference compound and peptide 48 were nearly identical, whereas the spectrum reconstructed for 49 had the opposite sign. Therefore, component 2 is the one that indicates the configuration of the α-carbon, which allows assignment of the L-configuration to the β,β-dimethylated tyrosine in 48 and the D-configuration to the β,β-dimethylated tyrosine in 49. Finally, component 3 accounts for all the remaining spectral differences between the three species such as contributions from presence or absence of β,β-dimethylation.

see Supplementary Materials).
The N α -unmethylated peptides 1 and 2 were reported to undergo very rapid degradation in plasma [40,43]. Therefore, N-terminal methylation or methylation of Arg 9 , both impairing NTS1R binding only to a minor extent [40], was applied to the synthesized peptides throughout.
The N α -unmethylated peptides 1 and 2 were reported to undergo very rapid degradation in plasma [40,43]. Therefore, N-terminal methylation or methylation of Arg 9 , both impairing NTS 1 R binding only to a minor extent [40], was applied to the synthesized peptides throughout.
In the initial set of prepared peptides (8,9,11,12, and 14-23) low stability in plasma (≤15% intact peptide after 24 h) was found for compounds containing Ile in position 12 (9, 15 and 20), confirming the importance of Tle 12 for the stabilization of the C-terminus against proteolytic degradation. Compounds with Tle 12 , devoid of a Ga 3+ -occupied chelator (8, 10-12, 14, and 16-18), showed a high stability towards proteolytic degradation (≥93% intact peptide after 24 h). However, insertion of Ga 3+ (19 and 21-23) led to a considerable decrease in stability (≤46% intact peptide after 24 h, Table 2). This observation can be explained by changes in compound structure, hydrophilicity and charge distribution upon insertion of the gallium cation and rearrangement of the carboxylic arms of the chelators, thereby facilitating the recognition by proteases.
The difference with respect to the linker in the side chain of the N ω -carbamoylated arginines of compounds 10 and 12 did not affect NTS 1 R receptor affinity (K i = 2.8 and 2.7 nM, respectively), but the change in the position of the N α -methylated and N ω -carbamoylated arginine (position 8 in 10, position 9 in 11) led to a slight decrease in NTS 1 R affinity from K i = 2.8 nM (10) to 13 nM (11). The latter finding is in agreement with reports on the higher importance of Arg 9 for NTS 1 R binding compared to Arg 8 [41,72,73].
The initially prepared set of PET ligand candidates with Tle in position 12 (19 and 21-23) exhibited K i values (NTS 1 R) in the range of 7.5-20 nM. The second set of potential PET ligands (33, 36, and 37), containing a lipophilic fluorinated biphenyl moiety, showed higher NTS 1 R affinities with K i values of 2.5-4.2 nM. The position and the type of carbamoylated arginine played only a minor role in terms of NTS 1 R binding for the potential PET ligands and their precursors. Within the final set of potential PET ligands (56 and 57), containing S-configured β,β-diMe-Tyr in position 11, the peptide with Ile in position 12 (56) displayed excellent NTS 1 R binding (K i = 1.2 nM), while its congener 57, containing Tle in position 12, showed 18-fold lower NTS 1 R affinity (K i = 21 nM).
The series of NT(8-13) derivatives containing various unnatural amino acids in position 11, 12 or 13 (38)(39)(40)(41)(42)(43)(44)(45)(46)(47)(48)(49) was prepared to develop a PET ligand with improved in vivo stability. As the side chains of Ile 12 and Leu 13 in 1 were hypothesized to contribute to receptor binding via hydrophobic interactions with aliphatic residues in the binding pocket of the NTS 1 R [73,74], hydrophobic unnatural amino acids structurally related to Ile and Leu, were incorporated in position 12 or 13 (38-47, cf. Figure 2). The artificial amino acids were quite well tolerated with respect to NTS 1 R binding, provided that they contained no additional alkyl substituent at the α-carbon (38, 40-43, and 46). The incorporation of amino acids with an additional alkyl group (methyl, ethyl) at the α-carbon resulted in a loss of NTS 1 R binding (Table S3, Supplementary Materials). Strikingly, sub-nanomolar NTS 1 R affinity (K i = 0.14 nM) was achieved with compound 48, containing S-configured β,β-diMe-Tyr 11 instead of Tyr 11 . As 48 showed also excellent in vitro plasma stability (see Table S3, Supplementary Materials), it served as a lead structure for the synthesis of the PET ligand candidates 56 and 57. The epimer of 48 (peptide 49), containing R-configured β,β-diMe-Tyr 11 , displayed considerably lower NTS 1 R binding compared to 48. This was in agreement with the results of variations in position 11 of 1, including the incorporation of D-configured tyrosine derivatives, revealing that D-configured tyrosine analogs caused a decrease in NTS 1 R binding [40,71,[75][76][77].
All compounds that were investigated at the NTS 2 R showed lower K i values at the NTS 1 R than at the NTS 2 R (difference most pronounced for peptide 15: K i values of 2.4 and 55 nM, respectively) revealing moderate NTS 1 R selectivity (Table 2). However, it is unlikely that low or missing NTS 1 R selectivity hampers the imaging of NTS 1 R-expressing tumors in the periphery, as the NTS 2 R is primarily expressed in the central nervous system [78][79][80].
In addition to the investigation of NTS 1 R and NTS 2 R binding, the agonistic activities of 21 and 56 at the G q -coupled NTS 1 R were determined in a Fura-2 Ca 2+ -assay using HT-29 colon carcinoma cells. The potential PET ligands 21 and 56 proved to be full agonists with maximal responses comparable to that of 1 (see Figure S6, Supplementary Materials). As also found for 1, the NTS 1 R agonistic potencies of 21 and 56 were lower compared to their NTS 1 R binding affinities (cf. Table 2 and Figure S6). A plausible explanation for this observation is in the non-equilibrium conditions in the case of the functional Ca 2+ -assay precluding a complete association of the agonist to the receptor (signal is recorded within 3 min after agonist addition), which must be compensated by higher agonist concentrations.

Radiosynthesis and Distribution Coefficients
The potential PET tracers 21, 33, 37 and 56 all showed high NTS 1 R affinity and high in vitro plasma stability. Thus, radiolabeling with 68 Ga 3+ was performed to prepare the respective PET tracers, i.e., [  Radiosynthesis was performed by incubation of the precursor compounds 16, 32, 35 or 54 (7.5-15 nmol) in HEPES buffer pH 5.5 with [ 68 Ga]GaCl 3 at 125 • C for 6 or 16 min (for details see Table 1). After the synthesis, the PET tracers were separated from the respective precursor using an analytical HPLC system in order to increase specific activity and to obtain a homogenous tracer preparation. The isolation from the precursors was feasible on a C18 reversed-phase material with mixtures of MeOH and 0.1% formic acid as mobile phase (comparable conditions as used for the purification of the "cold" PET ligands). For an exemplary chromatogram of the micro-preparative HPLC see Figure S7 (Supplementary Materials). Radiosynthesis and evaporation of the solvent was accomplished in approximately 90 min, and the separation of the PET tracer from the precursor including the second evaporation step took approximately another 50-60 min.  Biodistribution studies in HT-29 tumor-bearing nude mice were firstly performed with [ 68 Ga]21 and resulted in a tumor-to-muscle ratio of 9.5 at 45 min after injection of the tracer (cf. Figure S8 and  Figure 4A), tumor-to-muscle ratios 45 min p.i.  Figure S8 and  Figure S9, Supplementary Materials). Notably, for these tracers, only one main metabolite was found in the urine samples, being more polar than the respective intact tracer. Prompted by these findings, the focus was set on the development of NT(8-13) derived PET tracers with higher proteolytic stability, as a low in vivo stability of the tracers could also be a major reason (other than fast elimination) for the low tracer accumulation in the tumor. The preparation and in vitro characterization of peptides 38-49 brought forth the potential PET tracer 56 exhibiting high NTS1R affinity (Ki = 1.2 nM) and high in vitro plasma stability ( Table 2). The biodistribution of [ 68 Ga]56 in HT-29 tumor-bearing nude mice was investigated at 10, 25, and 45 min p.i. (Figure 5 and Table 3). Blocking experiments were performed for the time of highest tumor-uptake (45 min p.i.). At 10 min after injection of [ 68 Ga]56, the major fraction of activity was found in the kidneys (47% ID/g), remaining at this level over time (47% ID/g and 55% ID/g after 25 min and 45 min, respectively). The activity in the liver was much lower (7.2, 5.7 and 5.2% ID/g after 10, 25 and 45 min, respectively), indicating predominant renal excretion of [ 68 Ga]56, which was expected for this highly hydrophilic tracer (Table 3, Figure 4A). The activity in the blood dropped from 3.1% ID/g (10 min p.i.) to 1.3% ID/g (45 min p.i.), while it increased in the tumor from 3.6% ID/g (10 min) over 7.3% ID/g (25 min) to 8.4% ID/g (45 min). Tumor-tomuscle ratios increased over time (3.3 at 10 min p.i. and 13 at 25 min p.i.), reaching a value of 16 after 45 min, which was considerably higher than the tumor-to-muscle ratios of [ 68   Prompted by these findings, the focus was set on the development of NT(8-13) derived PET tracers with higher proteolytic stability, as a low in vivo stability of the tracers could also be a major reason (other than fast elimination) for the low tracer accumulation in the tumor. The preparation and in vitro characterization of peptides 38-49 brought forth the potential PET tracer 56 exhibiting high NTS 1 R affinity (K i = 1.2 nM) and high in vitro plasma stability ( Table 2). The biodistribution of [ 68 Ga]56 in HT-29 tumor-bearing nude mice was investigated at 10, 25, and 45 min p.i. (Figure 5 and Table 3). Blocking experiments were performed for the time of highest tumor-uptake (45 min p.i.). At 10 min after injection of [ 68 Ga]56, the major fraction of activity was found in the kidneys (47% ID/g), remaining at this level over time (47% ID/g and 55% ID/g after 25 min and 45 min, respectively). The activity in the liver was much lower (7.2, 5.7 and 5.2% ID/g after 10, 25 and 45 min, respectively), indicating predominant renal excretion of [ 68 Ga]56, which was expected for this highly hydrophilic tracer (Table 3, Figure 4A). The activity in the blood dropped from 3.1% ID/g (10 min p.i.) to 1.3% ID/g (45 min p.i.), while it increased in the tumor from 3.6% ID/g (10 min) over 7.3% ID/g (25 min) to 8.4% ID/g (45 min). Tumor-to-muscle ratios increased over time (3.3 at 10 min p.i. and 13 at 25 min p.i.), reaching a value of 16 after 45 min, which was considerably higher than the tumor-to-muscle ratios of [ 68 Figure S11B and Figure S10, Supplementary Materials). The high in vivo stability of [ 68 Ga]56, which obviously contributes to an increased tracer accumulation in the tumor, was confirmed by HPLC analysis of an ex vivo blood plasma sample at 10 min p.i. showing that intact [ 68 Ga]56 (identity confirmed by spiking of the plasma sample with 56) accounted for more than 80% of the radioactivity in the processed sample (cf. Figure S11C- Figure S11B and Figure S10, Supplementary Materials). The high in vivo stability of [ 68 Ga]56, which obviously contributes to an increased tracer accumulation in the tumor, was confirmed by HPLC analysis of an ex vivo blood plasma sample at 10 min p.i. showing that intact [ 68 Ga]56 (identity confirmed by spiking of the plasma sample with 56) accounted for more than 80% of the radioactivity in the processed sample (cf. Figure S11C-E, Supplementary Materials). Given are mean values ± SD (n = 4). Blocking data were obtained by co-injection of 48 (560-700 nmol/mouse). Statistical analysis of differences between tracer accumulation in the tumor of unblocked and blocked animals was performed by an unpaired two-tailed t-test (p <0.05 was considered statistically significant). ** p <0.01. *** p <0.001.  . Blocking data were obtained by co-injection of 48 (560-700 nmol/mouse). Statistical analysis of differences between tracer accumulation in the tumor of unblocked and blocked animals was performed by an unpaired two-tailed t-test (p <0.05 was considered statistically significant). ** p <0.01. *** p <0.001. In reported studies, the 111 In-labeled analog of compound 4 was investigated in terms of in vivo stability 15 min after injection in mice, revealing 22% of remaining intact tracer in the plasma sample [41]. The 68 Ga-labeled analog of 5, investigated in vitro in human serum, gave 93% intact tracer after 60 min of incubation [42]. In vivo studies in mice with a 68 Ga-labeled tracer structurally closely related to 5, showing low NTS 1 R affinity (K i = 180 nM), resulted in 90% intact tracer in blood plasma 10 min after tracer administration [16]. Noteworthy, a recently reported 18 F-labeled fluoroglycosylated NTS 1 R PET ligand derived from 10, containing the same peptide core structure as [ 68  C for up to 75 min followed by removal of extracellularly bound peptidic receptor ligand using the acid-trip method. This experiment showed that the fraction of internalized tracer was >80% after only 5 min of incubation, reaching a plateau of approximately 95% after 55 min ( Figure 6). This feature is considered favorable with respect to an accumulation of the tracer in the tumor in vivo, particularly when it comes to therapeutic applications using, e.g., alpha-emitting tracers such as 225 Ac-labeled radiopharmaceuticals. In reported studies, the 111 In-labeled analog of compound 4 was investigated in terms of in vivo stability 15 min after injection in mice, revealing 22% of remaining intact tracer in the plasma sample [41]. The 68 Ga-labeled analog of 5, investigated in vitro in human serum, gave 93% intact tracer after 60 min of incubation [42]. In vivo studies in mice with a 68 Ga-labeled tracer structurally closely related to 5, showing low NTS1R affinity (Ki = 180 nM), resulted in 90% intact tracer in blood plasma 10 min after tracer administration [16]. Noteworthy, a recently reported 18 F-labeled fluoroglycosylated NTS1R PET ligand derived from 10, containing the same peptide core structure as [ 68 Ga]21, [ 68 Ga]33 and [ 68 Ga]37, exhibited low in vivo stability in mice (30% of remaining intact tracer in plasma 10 min p.i., and no detectable tracer in plasma 20 min p.i.) [60]. Consequently, the high in vivo stability accomplished with [ 68 Ga]56 in combination with retained high NTS1R affinity, represents an important achievement in the field of peptidic NTS1R PET ligands.

Biodistribution of PET
To estimate the internalization rate of NTS1 receptors occupied by [ 68 Ga]56, HT-29 tumor cells were incubated with [ 68 Ga]56 at 37 °C for up to 75 min followed by removal of extracellularly bound peptidic receptor ligand using the acid-trip method. This experiment showed that the fraction of internalized tracer was >80% after only 5 min of incubation, reaching a plateau of approximately 95% after 55 min ( Figure 6). This feature is considered favorable with respect to an accumulation of the tracer in the tumor in vivo, particularly when it comes to therapeutic applications using, e.g., alpha-emitting tracers such as 225 Ac-labeled radiopharmaceuticals.

PET/CT Imaging with [ 68 Ga]56
Dynamic PET scans of HT-29 tumor-bearing nude mice injected with [ 68 Ga]56 were performed for 45 min. Notably, a PET/CT scanner (Siemens Biograph mCT-S(40)) for clinical routine tumor diagnostics in patients was used for these studies. It was shown previously that this instrument is applicable for imaging of small animals with sufficiently large tumor implants [65,81]. Blocking experiments were carried out by co-injection of an excess of the NTS1R ligand 48. Time-activity-curves (TACs) for the tumor and muscle (from nonblocking and blocking experiments) as well as for the kidneys, generated from the SUVs acquired for the respective ROIs, are depicted in Figure 7A,B. These data confirmed the results from the biodistribution studies with [ 68 Ga]56: the tracer uptake in the tumor increased over time and the activity level in the kidneys reached a plateau after

PET/CT Imaging with [ 68 Ga]56
Dynamic PET scans of HT-29 tumor-bearing nude mice injected with [ 68 Ga]56 were performed for 45 min. Notably, a PET/CT scanner (Siemens Biograph mCT-S(40)) for clinical routine tumor diagnostics in patients was used for these studies. It was shown previously that this instrument is applicable for imaging of small animals with sufficiently large tumor implants [65,81]. Blocking experiments were carried out by co-injection of an excess of the NTS 1 R ligand 48. Time-activity-curves (TACs) for the tumor and muscle (from non-blocking and blocking experiments) as well as for the kidneys, generated from the SUVs acquired for the respective ROIs, are depicted in Figure 7A,B. These data confirmed the results from the biodistribution studies with [ 68 Ga]56: the tracer uptake in the tumor increased over time and the activity level in the kidneys reached a plateau after approximately 10 min. Thus, tumor-to-muscle ratios, determined from the SUVs (cf. inset table in Figure 7A), were comparable with the tumor-to-muscle ratios obtained from biodistribution studies. The blocking experiments confirmed the specific (NTS 1 R-mediated) uptake of [ 68 Ga]56 in the tumor ( Figure 7A). Representative PET images of tumor-bearing mice injected with [ 68 Ga]56 alone or with [ 68 Ga]56 and an excess of 48 are shown in Figure 7C. Accumulation of the tracer in the tumor was clearly visible.  Figure  7C. Accumulation of the tracer in the tumor was clearly visible.

Conclusions
We herein describe the preparation, analysis and biological characterization of a series of peptidic PET tracer candidates, which led to the discovery of the DOTA(Ga 3+ )-conjugated NTS1R ligand UR-LS130 (56) showing high stability in human plasma (t1/2 >24 h) and higher NTS1R affinity (Ki = 1.2 nM) compared to previously reported NTS1R PET ligands with high in vitro plasma stability. A novel feature of this Ga 3+ -containing peptidic PET ligand is the attachment of the chelator via the side chain of an arginine.

Conclusions
We herein describe the preparation, analysis and biological characterization of a series of peptidic PET tracer candidates, which led to the discovery of the DOTA(Ga 3+ )-conjugated NTS 1 R ligand UR-LS130 (56) showing high stability in human plasma (t 1/2 >24 h) and higher NTS 1 R affinity (K i = 1.2 nM) compared to previously reported NTS 1 R PET ligands with high in vitro plasma stability. A novel feature of this Ga 3+ -containing peptidic PET ligand is the attachment of the chelator via the side chain of an arginine. [ 68 Ga]56 displayed high in vivo stability and a clear accumulation in NTS 1 R-expressing HT-29 tumors. Notably, 56 contains no Tle, but Ile in position 12, like endogenous neurotensin. Instead, 56 contains a β,β-diMe-Tyr in position 11, leading to excellent stability in vitro and in vivo. To date, replacement of Ile 12 by Tle 12 is the state of the art to achieve proteolytic stabilization of the C-terminus of NT(8-13) derived PET tracers. However, in the present study, we show that the Ile 12 /Tle 12 exchange can be insufficient to achieve high in vivo stability. Unlike the Ile 12 /Tle 12 exchange, which affects NTS 1 R binding, the recently introduced alternative based on the exchange of Leu 13 by trimethylsilylalanine proved to be beneficial with respect to NTS 1 R binding, but turned out to be less favorable than the Tle 12 -approach regarding proteolytic stability [38]. In contrast, the new diMe-Tyr 11 -approach combines retained NTS 1 R affinity and high tracer stability. Taking into consideration the reported impact of the metal ion chelator on a tracer's biodistribution profile and accumulation in the tumor [37,82], further improvement of the tracer [ 68 Ga]56 could be undertaken by conjugation to a chelator different from DOTA. With [ 68 Ga]56 we present the first peptidic NTS 1 R PET ligand with high in vivo stability exhibiting comparable NTS 1 R affinity as reported 177 Lu-labeled NTS 1 R antagonists [83,84], which are favored over peptides for tumor endoradiotherapy due to higher in vivo stability [85]. Thus, the achievements of this work could promote the development of NT(8-13)-derived radiotherapeutics for cancer treatment, as an alternative to 177 Lu-labeled NTS 1 R antagonists.

Data Availability Statement:
The data presented in this study are available in this article.