**3. Discussion**

In the present study, a nastorazepide derivative functionalized with a DOTA chelator was synthesized to provide a potential precursor for the diagnosis and systemic therapy of CCK-2R-expressing tumors. In fact, the presence of a chelator moiety allowed for the coordination of several metal radionuclides, thus triggering the possibility of a radiotheragnostic approach. Nastorazepide is a well-known 1,5-benzodiazepine derivative that acts as an antagonist ligand for CCK-2R, and it is currently under development as a therapeutic drug for pancreatic cancer, gastroesophageal reflux disease, and peptic ulcers. The state of the art in the field of imaging and treatment of CCK-2R-expressing tumors with radiopharmaceuticals is mainly focused on the labelling of the natural ligands of this receptor (i.e., CCK and gastrin) or those derived thereof. However, issues regarding in vivo stability, receptors affinity, and high kidney retention have hindered this approach so far. An alternative and less explored pathway is the labelling of small molecules such as 1,5-benzodiazepine derivatives, although, to the best of our knowledge, this has only been pursued with the diagnostic radionuclide technetium-99 and, despite promising preliminary results, only minor efforts have been devoted to this approach [16,17].

The precursor synthesized here is composed by a nastorazepide moiety connected to a DOTA chelator with a short alkyl chain followed by a PEGylated portion introduced to increase the hydrophilicity of the whole molecule. The linker is an important part of the molecules because it ensures a sufficient separation between the pharmacophore and the radiometal binding moiety, which also influences the affinity. Differently from previously reported works, a linker completely not based on an amino acid sequence was assembled to avoid any possible site of enzymatic cleavage. While a nastorazepide core was synthesized in a six-step reaction pathway, as reported in the literature, the linker was obtained by a condensation reaction between *N*-Boc-1,6-hexanediamine (**1**) and CDI-

activated *N19*-Fmoc-4-oxo-9,12,15-trioxa-5,19-diazanonadecanoic acid (**2**). Consequently, the Boc protecting group was eliminated, and the linker was again condensed to the CDIactivated nastorazepide core. Finally, the Fmoc moiety was cleaved, and the chelator was inserted. The CDI-activation of the succinyl moiety of compound **2** and, more generally, the activation of the carboxylic groups with CDI in the presence of an Fmoc protecting group (such as reaction steps a and c in Scheme 1) were the trickiest steps of the pathways, yielding 46% and 32%, respectively. The moderate yields were supposed to derive from two side-reactions due to the well-known mechanism of action of CDI that leads to the release of an imidazole group [20]. This group, imparting a basic character to the reaction mixture, could have been responsible for a first side-reaction with compounds **2** and **4** that led to the elimination of the Fmoc-protecting group with the consequent formation of a free amine. The suggested mechanism of this reaction occurring for the CDI-activated compound **2** is reported in Figure 7. Moreover, a second imidazole-mediated side-reaction was supposed to involve compound **2** during the first condensation step of the pathway, thus yielding the cyclization of the succinyl group instead of its activation, as reported in Figure 8. The formation of this compound was confirmed by <sup>1</sup>H-NMR (Figure S10). To minimize the formation of these by-products, the original work-up or the reactions was modified by adding a 1 M acetic acid solution to the reaction mixtures to fix the pH to around 5 (thus protonating the imidazole and reducing its catalytic behavior) before the evaporation step.

**Figure 7.** Elimination of the fluorenylmethoxycarbonyl (Fmoc) group from CDI-activated compound **2** mediated by imidazole.

**Figure 8.** By-product obtained from the cyclization of the CDI-activated succinyl group of compound **2**.

Another step of the synthesis worth attention is the passage from compound **6** to compound **7**, i.e., the elimination of the Fmoc group after the condensation of the linker with Z-360. Conventional elimination conditions suggest the use of a 20% piperidine/DMF solution [21], but, in our case, a very low yield was obtained, probably due to the instability of the linker in a basic environment, as reported for similar chemical structures [22]. For this reason, the use of a milder base, such as a 50% morpholine/DMF solution, was preferred to obtain almost quantitative results [23]. Finally, in an alternative synthetic approach, CDI-activated Z-360 (**5**) was firstly reacted with *N*-Boc-1,6-hexanediamine (**1**), and, after the elimination of the Boc group, the obtained intermediate was further reacted with CDIactivated *N<sup>19</sup> -*Fmoc-4-oxo-9,12,15-trioxa-5,19-diaza-onadecanoic acid (**2**). However, this method did not show significant advantages with respect to the former one, exhibiting the same side-reactions described above and, moreover, implying an elevate loss of Z-360 due to the low yield of the first reaction. For these reasons, it was not pursued further. The overall yield of the process to obtain IP-001 was around 5% (22% starting from compound **5**), which was in line with a multi-step synthetic pathway, with the yield-limiting step being the synthesis of compound **5**. In view of further optimization, attention should be paid to the synthesis optimization of the latter.

The radiolabeling of IP-001 with indium-111 was performed in the standard conditions involving the labelling of a DOTA-conjugated small molecule with trivalent radiometals [24], obtaining an incorporation higher than 95% in a quite straightforward way. Stability in all tested conditions, and particularly in human serum and blood, was high and generally comparable at short incubation times (<24 h) to the best peptide-based candidates for the treatment of CCK-2R-expressing tumors, but it was much superior at longer times (intact amount of [111In]In-IP-001 >93% after 120 h in HB). On the other hand, the amount of [111In]In-IP-001 bound to the serum protein was almost comparable to other indium-111 labelled minigastrin analogues [13]. A direct comparison with other non-peptide-based radiotracers was not possible since, to the best of our knowledge, the high stability in HB at a late time found here was the first reported experimental confirmation of previous theoretic assumptions for this kind of molecular structure.

CCK-2R expression has been reported in cancers of different origins by immunohistochemistry on tumor specimens [25]. However, CCK-2R expression in human cancer cell lines has largely not been investigated. CCK-2R expression at the mRNA level has been reported in human PC3 prostate carcinoma, U373 glioma, U2OS osteosarcoma, Colo205 colon carcinoma [26], HepG2 hepatoma [27], gastric and colorectal cell lines, and MCF-7 breast cancer and Molt4 lymphoblastic leukemia cell lines [28]. To date, studies on CCK-2Rtargeting probes for cancer imaging have only been performed by exploiting transfected cell lines (the human A431 cancer cell line, HEK293 human embryonic kidney cells, NIH 3T3 mouse embryonic fibroblast cells, and Chinese hamster ovary cells) or rat pancreatic tumor cells [16,24,29–33]. However, the demonstration of CCK-2R overexpression by transfected versus parental cells has generally been lacking.

Herein, we demonstrated that the CCK-2R protein is indeed expressed by various human cancer cell lines, and we decided to perform experiments in vitro and in vivo on A549 cells as representative (Figure 1). We acknowledge the fact that our results are only partially comparable to other studies reported thus far, but we would like to emphasize that non-transfected tumor cells represent a closer model to the real physio-pathological condition than transfected cancer cell lines. With this in mind, we ascertained that [111In]In-IP-001 cell uptake in vitro was two-fold lower than what reported previously in AR4-2J rat pancreatic tumor cells for indium-111-labelled minigastrin derivatives [24] and fourfold lower than the total amount of technectium-99m nastorazepide-based derivatives accumulated by CCK-2R-transfected HEK293 cells [17].

In BALB/c nude mice developing A549 cell-derived tumors, [111In]In-IP-001 exhibited a quite rapid absorption rate in all the main organs (liver: 8.25 ± 2.21%; kidneys: 6.99 ± 0.97%; spleen: 3.88 ± 0.36%) and the tumor (2.36 ± 0.26% IA/g) within 4 h. A high amount of injected radioactivity was also found in the circulation, suggesting a strong interaction with serum albumin (Figures 3 and 4). Clearance was slow and performed by both the renal and hepatobiliary pathways, probably due to the relatively high lipophilicity of the radiotracer. After 24 h, the kidneys and liver (3.11 ± 0.52% IA/g and 3.55 ± 1.17% IA/g, respectively) remained the organs with the highest uptake, followed by the tumor (1.52 ± 0.71% IA/g), which was clearly delineated in the CT/SPECT images (Figure 3). Moreover, at 4 h post-injection, a certain level of radioactivity was also detected in the intestines, lungs, and pancreas (no statistically significant difference with the tumor), with a decrease of the signal of more than 73% over 24 h, indicating a transient accumulation. On the contrary, the tumor tissue still retained 65% of the accumulated [111In]In-IP-001 at 24 h post-injection. The lungs, pancreas, and intestines exhibited a more rapid wash-out with tumor/organ ratios of 2.02, 5.85, and 2.76, respectively (Figure 4). The slow kinetics of radiotracers based on the nastorazepide core was already reported, although a significantly more promising biodistribution was obtained in mice bearing tumors derived from transfected HEK293 cells [16,17], in which kidneys were the only organ with a comparable uptake with respect to the tumor at delayed times. As reported in Table 1, the addition of a PEGylated spacer and a strongly hydrophilic chelator like DOTA to the hydrophobic Z-360 core imparted a quite hydrophilic nature to IP-001. Indeed, at pH = 7.2, the chelator was completely dissociated, and the precursor was in the anionic form (i.e., IP-0013−). However, it is reasonable that the formation of a neutral complex with [111In]In3+ increased the lipophilicity of the final radiotracer (experimental value of 0.45), thus giving reason of its unspecific accumulation in the liver and spleen. Hence, further chemical optimizations are required, with a particular focus on the hydrophilicity of the linker, to obtain a novel ligand with a logP remarkably lower than <sup>−</sup>3.22. In general, [111In]In-IP-001 biodistribution reflected the behavior of a radiotracer with a moderate receptor affinity, as attested by the blocking studies performed with an excess of unlabeled ligand in which, despite the results showing a trend of lower [111In]In-IP-001 accumulation in the blocking setting (45% lower uptake in the tumor tissue), no statistically significant differences between the two groups of injected mice were found. Definitely, although a high stability was achieved by avoiding the use of peptides or peptide-mimetic moieties on the linker, affinity and hydrophilicity could likely be further tailored and enhanced by acting on this portion of the structure.
