Identification and stabilization of a highly selective gastrin‐releasing peptide receptor agonist

The gastrin‐releasing peptide receptor (GRPR) is part of the bombesin receptor family and a well‐known target in cancer diagnosis and therapy. In the last decade, promising results have been achieved by using peptide‐drug conjugates, which allow selective targeting of GRPR expressing tumor cells. Most ligands, however, have been antagonists even though agonists can lead to higher tumor uptake owing to their internalization. So far, only a few studies focused on the identification of small GRPR‐selective agonists that are metabolically stable. Here, we developed novel bombesin analogs with high selectivity for the GRPR and improved blood plasma stability. The most promising analog [d‐Phe6, β‐Ala11, NMe‐Ala13, Nle14]Bn(6‐14) displays an activity of 0.3nM at the GRPR, a more than 4000‐fold selectivity over the other two bombesin receptors and more than 75% stability in human blood plasma after 24 hours. This analog is proposed as a promising drug shuttle for the intracellular delivery of different payloads in targeted tumor therapy approaches.

like nanoparticles, 13 quantum dots, 14 cytotoxic moieties, 15 photosensitizers, 16 or siRNA 17 have been attached to various bombesin peptides and were successfully delivered to tumor cells. Further, promising results have been achieved with bombesin analogs for the imaging of prostate cancer. 18 The majority of the developed peptide carriers are have been antagonists because they can target the GRPR without inducing the biological downstream responses in healthy tissues. A recent example for this approach is the conjugate [D-Phe 6 , NMe-Gly 11 , Sta 13 , Leu 14 ]Bn (6)(7)(8)(9)(10)(11)(12)(13)(14), which showed great results in the selective delivery of doxorubicin-filled nanoparticles. 19,20 Even though antagonists are discussed to provide better pharmacodynamics, the cellular uptake mechanism of the cargo at the tumor site is highly dependent on the payload itself. In contrast, the use of agonists lead to a receptormediated internalization into intracellular vesicles and thus facilitates a cargoindependent cellular uptake. Since the GRPR is subsequently separated in the endosome and recycles back to the cell membrane, this process can be repeated and may lead to an active intracellular accumulation of the peptide-drug conjugate. 21 In the past, derivatives such as [Nle 14 ]BBN (7)(8)(9)(10)(11)(12)(13)(14) or Bn (6)(7)(8)(9)(10)(11)(12)(13)(14), modified with radiotracers, were proposed as suitable conjugates for the detection of GRPR expressing tumors. [22][23][24][25][26][27] Even though these conjugates showed great tumor uptake, they were found to be metabolically instable, which resulted in a fast wash out. The rapid metabolic degradation of bombesin-based conjugates and their low selectivity profile impact their drug delivery capacity and limit medical application. 28 Thus, we developed GRPRselective bombesin analogs with sufficient blood plasma stability and demonstrated the intracellular delivery of a fluorophore-payload. The derivatives with the highest activity, greatest selectivity, and optimal metabolic stability can be considered as a drug shuttle for targeted tumor treatment.

| Solid phase peptide synthesis
All bombesin derivatives were synthesized by solid phase peptide synthesis (SPPS) using fluorenylmethyloxycarbonyl (Fmoc)/tBu strategy at a 15 μmol scale. A NovaSynTGR R resin was used to generate an amidated C-terminus. Automated SPPS was performed on a SYRO I peptide synthesizer using eightfold molar excess of N-α-Fmocprotected amino acids, ethyl 2-cyano-2-(hydroxyimino)acetate (Oxyma), and N,N′-diisopropylcarbodiimide (DIC) in DMF. All automated coupling steps were carried out twice with a reaction time of 40 minutes. Fmoc protecting groups were cleaved using 40% piperidine in DMF (v/v) for 3 minutes and 20% piperidine in DMF (v/v) for 10 minutes.

| Stable transfection
The plasmids (pCMV6_NMBR-tGFP, pCMV6_GRPR-tGFP, and pCMV6_BRS-3-tGFP) were amplified using E. coli DH5α cells and purified with a PureYield Kit. Correct constructs have been confirmed by sequencing. HEK293 cells were stably transfected with the plasmids using Lipofectamine 2000 and selected with G418.

| Cell culture
All cell lines were maintained in T75 cell culture flasks at 37°C, 95% humidity, and 5% CO 2 (standard conditions  Subsequently, cells were washed twice with OptiMEM to remove excessive peptide. Image recording was conducted directly after washing while cells were maintained in OptiMEM, using an Axio Observer.

Z1 microscope equipped with an ApoTome Imaging System and a
Heating Insert P Lab-Tek S1 unit (Zeiss, Oberkochen, Germany). Image processing was performed with AxioVision 3.1.

| Blood plasma stability assay
Fluorescently labeled peptides were incubated in a concentration of

| Design and synthesis of GRPR-selective peptides
We aimed for the development of a GRPR-selective peptide ligand with sufficient metabolic stability for a drug shuttle system. All analogs were synthesized by a combination of automated and manual SPPS on a NovaSynTGR R resin generating a C-terminal amide. were reported to be selective for the GRPR and therefore synthesized as controls. 29 To further increase the selectivity, peptide chimeras were by Leu 13 (10). Additionally, we reduced the side chain at position 13 even further and introduced Ala (11). After the synthesis was completed, all analogs were analyzed by MALDI-TOF and ESI-MS to prove correct molecule identity. Homogeneity was measured by RP-HPLC and is summarized in Table 1.

| Receptor selectivity investigation
The synthesized analogs were characterized with respect to receptor activation and selectivity by a Ca 2+ -mobilization assay ( Table 2). To validate the assay setup, the native ligands NMB (1), GRP (2), a shortened bombesin analog Bn(6-14) (3), and the universal ligand (4) were tested. The NMB (1) exhibited equimolar potencies at the NMBR and the GRPR with an EC 50 value 0.3nM whereas the BRS-3 was activated with an EC 50 value of 215nM. The GRPR preferring peptide 2 activated the GRPR with an EC 50 value of 0.6nM whereas the NMBR was activated with a slightly lower EC 50 value of 15nM and the BRS-3 was hardly activated. Analogs 3 and 4 exhibited similar potencies with respect to NMBR and GRPR activation but only the universal ligand (4) activated the BRS-3 with an EC 50 value of 5.5nM. Therefore, the universal ligand was used for normalization at each receptor. Peptide 5 exhibited slight selectivity for the GRPR due to a tenfold reduced potency at the NMBR and BRS-3 while the EC 50 value at the GRPR was only slightly decreased compared with the universal ligand (4).
The exchange of D-Phe 6 by D-Ala in analog 6 had no effect on the activation properties at all three receptors. Substitution of D-Val 8 to D-Ala (7) had also no effect on the GRPR activation, whereas the potency at the NMBR and BRS-3 was slightly reduced. Derivative 8 bearing two modifications exhibited a tenfold loss of activity at the GRPR and an even stronger potency loss at the NMBR and BRS-3 resulting in a peptide with 150-fold selectivity towards the NMBR and more than 550fold to the BRS-3. However, the combination of peptide 6 and 7 resulting in analog 9 did not provoke any significant increase in selectivity.
Since the combined analogs showed only slight selectivity for the GRPR mainly caused by a reduced NMBR and BRS-3 activity and not by a potency increase at the GRPR, we also investigated position 8 by single amino substitutions. Because GRP (2) and Bn (6)(7)(8)(9)(10)(11)(12)(13)(14) (3) contain a Leu at this position and exhibit a strongly reduced potency at the BRS-3, we exchanged the L-Phe 13 in conjugate 4 by L-Leu (10).
Surprisingly, this substitution had just minor effects on the potency at all three receptors. The exchange to D-Ala (11), however, led to a Note. Data were obtained by Ca 2+ -mobilization assay using stably transfected HEK293_X-tGFP ( drastic loss of activity at the NMBR and BRS-3, while the potency at the GRPR was only slightly affected. This substitution led to a highly selective peptide with more than 4000-fold selectivity over the NMBR and more than 8000-fold towards the BRS-3 (Table 2 and Figure 1). Next

| Blood plasma stability investigation
Metabolic stability is an essential criterion for the medical application of peptide-drug conjugates. Therefore, a fluorescently labeled version of [D-Phe 6 , βAla 11 , Ala 13 , Nle 14 ]Bn(6-14) (11) was synthesized by introducing a 6-carboxytetramethylrhodamine to an oligo ethylene glycol linker that was attached to the N-terminus of peptide 11 yielding 11T (Table 3). This conjugate was then tested in a blood plasma stability assay. We observed that the original peak with a retention time of 16.3 min decreased over time and was completely degraded after 1 h revealing a half-life of 14.2 min (Figure 2A,D). The arising   correlating to the cleavage between Gln 7 and Trp 8 (Figure 2).

| Identification of the responsible peptidase
Since the cleavage of the peptide bond between His 12 and Ala 13 in

| Rational stabilization of the most selective analog
After the predominant cleavage site and the responsible peptidase were identified, we used multiple rational approaches to improve blood plasma stability and support the translation into medical application. This was expected to be challenging because the selectivitygaining amino acid substitution in [D-Phe 6 , βAla 11 , Ala 13 , Nle 14 ]Bn(6-14) (11) is in close proximity to the main cleavage side and most likely part of the ACE recognition site. Key amino acids were substituted by unnatural amino acids, which were chosen to be related to the original residues to maintain high selectivity while reducing recognition by the ACE to increase the metabolic stability.
As first approach, the amino acids at positions 12 and 13 were exchanged by their corresponding D-isomers. This led to peptide 12 bearing D-Ala 13 and analog 13 with D-His 12 . Next, in peptide 14, Nle 14 was replaced by O-methyl-L-homoserine to increase the C-terminal hydrophilicity and potentially impair the recognition by the ACE, which needs primarily hydrophobic C-terminal residues.
Furthermore, β-(2-thienyl)-L-alanine was introduced instead of the histidine at position 12 (15), and Ala 13 was substituted by 2aminoisobutyric acid (16). Finally, the peptide bond between the His 12 and Ala 13 was N-methylated to prevent degradation and increase thereby metabolic stability. To determine the stability, TAMRA labeled conjugates (11T-17T) were synthesized, and unlabeled analogs (12)(13)(14)(15)(16)(17) were used for the receptor activation and selectivity experiments. All derivatives were analyzed with respect to their correct molecule identity and homogeneity were determined by mass spectrometry and RP-HPLC (Table 3).

| Increased metabolic stability while GRPR selectivity is retained
The rationally modified analogs were tested in Ca 2+ -mobilization assays to determine whether selectivity and GRPR activity were retained while the blood plasma stability has been increased. Peptides 12 and 13 containing the D-isomer substitutions at position 12 and 13 were found to be completely inactive at the NMBR and BRS-3. Additionally, analog 12 exhibited almost a complete loss of activity at the GRPR, while peptide 13 displayed a more than 100-fold reduced activity at the GRPR with an EC 50 value of 33nM (Table 4). Even though both peptides showed an unsatisfied receptor activation profile, a clear improvement in stability has been observed with half-lives of more than 100 hours for the respective conjugates 12T and 13T.  (Table 4 and Figure 4). The EC 50 value at GRPR was slightly reduced to 3.9nM, and the activity at the NMBR and BRS-3 was still detectable. The exchange of the selectivity determining Ala 13 by 2aminoisobutyric acid (16) revealed an even better selectivity profile for the GRPR than the peptide 11. Next, to the increased stability of to 4.6 hours, the activity at the NMBR and BRS-3 receptor was completely abolished, and the EC 50 value of 0.2nM was not altered.
In the last approach, N-methylation was introduced between position 12 and 13, leading to analog 17. This modification led to a highly selective GRPR preferring peptide with an EC 50 value of 0.3nM and no NMBR and BRS-3 activity. Additionally, the corresponding TAMRA derivative (17T) displayed a half-life of 88.6 hours and accordingly high metabolic stability in human blood plasma (Table 4 and Figure 4).

| Intracellular payload delivery
The intention of this study was the identification of a possible drug shuttle system for the intracellular delivery of different payloads.
Therefore, the most promising analogs were investigated for their ability to deliver TAMRA molecules as payloads into endogenously GRPR expressing PC3 tumor cells and stably GRPR-tGFP transfected HEK293 cells ( Figure 5). After 1 hour of stimulation with 1μM 11T, internalization of the GRPR (green) was observed in HEK293 cells as already found for the unlabeled peptide 11 (Figure 1). In addition to the internalized GRPR, TAMRA-fluorescence was detected in small intracellular vesicles, which were found to be partially co-localized with the receptor ( Figure 5). Intracellular TAMRA fluorescence was also detected in PC3 cells after stimulation with 11T. Stimulation with the stabilized conjugates 14T, 16T, and 17T provided similar results for HEK293_GRPR cells as well as the for PC3 cells. All conjugates demonstrated uptake into intracellular vesicles supporting the concept of GRPR mediated delivery of certain payloads. Thus, these stabilized conjugates can be considered as targeting vectors for a GRPR based drug shuttle system.

| DISCUSSION
In the last decade, the bombesin receptor family evolved as a promising target for cancer diagnosis and therapy. 1 Its potential is promoted by their strong overexpression on various tumor types. Many different payloads have been attached to a number of bombesin analogs and were successfully delivered to GRPR expressing cells. 30 The majority of studies however was conducted with antagonistic bombesin derivatives like the GRPR-selective RM26 peptide, which was developed as a very promising candidate in clinical applications. 31 The N-terminal modification with DOTA allowed the incorporation of different radionucleotides and thereby promoted its path towards clinical phase I for early detection and imaging of prostate cancer. 32 Due to these promising results and their better pharmacodynamics, antagonists are   have been used as peptide-drug conjugates before but they possessed low selectivity for the GRPR or insufficient metabolic stability. Therefore, tumor targeting capacity and medical application are limited. 28 In the past, further studies based on these peptides were carried out, which led to improved GRPR selectivity and metabolic stability and thereby increased tumor-targeting properties. 38,39 We aimed for the development of a highly GRPR-selective peptide agonist with excellent metabolic stability. First, we synthesized and tested based three known ligands (5-7) as starting point for our study. 29 As all three control peptides were only slightly more selective than the starting peptide (4) Since the newly developed peptides activated the NMBR and BRS-3 still in a two-and three-digit nanomolar range, we investigated position 13, which was also identified to be important for the biological activity, uptake, and stability. 22,24,42 Since NMB (1) features a Phe at this position and activates the BRS-3 receptor with an EC 50 value of ca. 200nM, while the GRP (2) and Bn (6)(7)(8)(9)(10)(11)(12)(13)(14) (3) contain a Leu at this position and hardly activate the BRS-3, we expected a favored selectivity by an exchange of Phe 13 to Leu 13 (10). Surprisingly, this substitution led only to minor changes in the activation profile whereas the exchange by Ala led to a highly GRPR-selective analog Internalization study of stabilized conjugates. HEK293 cells stably expressing the tGFP tagged GRPR (green) and PC3 cells endogenously expressing the GRPR were stimulated with 1μM fluorescently labeled ligand (red) for 1 h at 37°C. Cell nuclei were stained with Hoechst 33342 (blue). Scale bar = 10 μm (HEK293_GRPR); 20 μm (PC3) (11). This derivative features an excellent selectivity for the GRPR with more than 4000-fold over NMBR and more than 8000-fold over BRS-3. The almost completely abolished activity at the NMBR and BRS-3 and the only slightly altered potency at the GRPR were surprising because this position was found to be equally important in Ac-Bn (7)(8)(9)(10)(11)(12)(13)(14) for NMBR and the GRPR activation. 43 With this very selective peptide in hand, we investigated its blood plasma stability to improve its potential even further.  (7)(8)(9)(10)(11)(12)(13)(14), we speculated that one of these two enzymes might also be involved in the cleavage of the newly developed peptide. 46 Therefore, 11T was incubated with the ACE-inhibitor enalaprilat and the NEP-inhibitor sacubitrilat. In contrast to the NEP-Inhibitor, the ACE-inhibitor led to an increased half-life indicating the relevance of ACE for the stability of this bombesin derivative. Since this enzyme is highly abundant in the human blood and leads to a rapid degradation of the selective peptide, analog 11 had to be stabilized. 47 Even though frequently used stabilization approaches as palmitoylation 48,49 and PEGylation 50 lead in general to very good results, we aimed to stabilize the actual cleavage site by amino acid substitution to preserve the small molecular size allowing a maximized payload loading. This approach turned out to be fairly challenging due to the close proximity of the cleavage site to the selectivity gaining substitution at position 13. First, we separately exchanged the amino acids L-His 12 and L-Ala 13 by their isomeric counterparts, yielding highly stable derivative with half-lives of more than 100 hours. Unfortunately, both analogs displayed a strong potency loss at the GRPR. While analog 13 displayed with an EC 50 value of 33nM a still reasonable potency, peptide 12 exhibited nearly no activity at the GRPR, which was not surprising because this position was shown before to be critical for GRPR activity and selectivity. Next, the C-terminal Nle was exchanged by the more hydrophilic Hse(Me) because the ACE-homolog ACE2 requires a hydrophobic or basic C-terminal amino acid for recognition. 51 We expected to achieve a positive result with respect to receptor activation as the exchange of the original Met 14 to Nle decreased the uptake into tumor cells in previous studies. 24,42 In fact, the reintroduction of the more hydrophilic Hse(Me) at this position not only led to a more stable analog with a half-life of 38.1 hours but also abolished the NMBR and BRS-3 receptor activity completely whereas the potency change at the GRPR was negligible. These findings are partially in agreement with a recently published study in which the same exchange was performed in a 177 Lu labeled Bn (6)(7)(8)(9)(10)(11)(12)(13)(14) conjugate. Even though no increase in blood plasma stability could be observed, the uptake into tumor xenografts was improved. 52 Additionally, His 12 was exchanged by Tha (15), which led only to a minor stabilization in blood plasma and resulted in a tenfold loss of activity at the GRPR. The introduced sulfur obviously hinders the correct conformation of the peptide and leads to a reduced potency.
However, the exchange of Ala 13 by Aib (16) resulted in a half-life of 4.6 hours, while selectivity for the GRPR as well as the potency was maintained. The mild improvement in stability and the conserved activation profile can be explained by the obvious similarities of Aib (16) to Ala (11) and Leu (10), which had no significant influence on the GRPR activation.
As a last modification, an N-methylated alanine was introduced to prevent peptide bond cleavage in blood plasma (17). This substitution yielded the most stable analog with a half-life of 88.9 hours, which also retained its GRPR activity and selectivity completely. This result is in accordance with the proposed binding mode, which suggested that the amide bond between His 12 and Leu 13 is not primarily important for formation of the internal hydrogen bonding network. 43 The relevance of this position for the metabolic stability was already investigated by various groups, which exchanged the Leu 13 in Bn(7-14) by the unnatural amino acid cyclohexylalanine (Cha) or by a statin group leading to a derivative with significantly improved stability in human blood plasma. However, these substitutions led also to reduced internalization rates into PC3 cells and binding affinities for the GRPR. 53,54 In contrast to previously developed conjugates, the herein designed analogs feature not only subnanomolar activity and great selectivity for the GRPR, they also facilitate the uptake into GRPR expressing cells. We nicely demonstrated the successful uptake of TAMRA molecules into tumor cells by fluorescence microscopy proving the intracellular delivery of conjugated payloads. These analogs can be considered as universal targeting vectors for the selective delivery of bioactive molecules to GRPR expressing tumor cells.

| CONCLUSION
Here, we present the identification of a highly GRPR-selective ago-