Design, Synthesis, In Vitro, and Initial In Vivo Evaluation of Heterobivalent Peptidic Ligands Targeting Both NPY(Y1)- and GRP-Receptors—An Improvement for Breast Cancer Imaging?

Heterobivalent peptidic ligands (HBPLs), designed to address two different receptors independently, are highly promising tumor imaging agents. For example, breast cancer has been shown to concomitantly and complementarily overexpress the neuropeptide Y receptor subtype 1 (NPY(Y1)R) as well as the gastrin-releasing peptide receptor (GRPR). Thus, radiolabeled HBPLs being able to bind these two receptors should exhibit an improved tumor targeting efficiency compared to monospecific ligands. We developed here such bispecific HBPLs and radiolabeled them with 68Ga, achieving high radiochemical yields, purities, and molar activities. We evaluated the HBPLs and their monospecific reference peptides in vitro regarding stability and uptake into different breast cancer cell lines and found that the 68Ga-HBPLs were efficiently taken up via the GRPR. We also performed in vivo PET/CT imaging and ex vivo biodistribution studies in T-47D tumor-bearing mice for the most promising 68Ga-HBPL and compared the results to those obtained for its scrambled analogs. The tumors could easily be visualized by the newly developed 68Ga-HBPL and considerably higher tumor uptakes and tumor-to-background ratios were obtained compared to the scrambled analogs in and ex vivo. These results demonstrate the general feasibility of the approach to use bispecific radioligands for in vivo imaging of breast cancer.


Introduction
Radiolabeled peptides, being able to specifically bind certain receptors overexpressed on many malignancies, have become standard radiotracers for tumor-specific imaging by positron emission tomography (PET) in a clinical routine. However, these radiolabeled peptides are able to address only one target receptor type, being thus only able to visualize tumors expressing this particular receptor.
As different tumor lesions can however overexpress different receptor types and receptor expression can change upon metastasis and disease progression, a monovalent peptidic radioligand is often not able to visualize all tumor cells with the same efficiency and some lesions might be completely missed by the applied radiopeptide.
Radiolabeled heterobivalent peptide ligands (HBPLs) on the other hand-by their ability to specifically target more than one receptor type-have been proposed to be better-suited agents for tumor imaging as they enable the visualization of the tumor by different receptor types potentially overexpressed by the target cells [1,2].
HBPLs furthermore can have favorable effects for in vivo tumor imaging compared to monovalent ligands such as improved in vivo biodistribution and enhanced avidity caused by simultaneous binding if both receptors are present on the tumor cell surface. In this case, also a higher probability of rebinding is achieved for a heterobivalent binder in case of dissociation from the receptor compared to a monovalent peptide due to "forced proximity" of the second potential binder to the second target receptor.
Furthermore, by being able to target different receptor types on the tumor cell surface, an overall higher number of receptors can be addressed by a heterobivalent ligand, increasing the probability of binding and thus tumor visualization. Thus, HBPLs are also of special interest not only for the imaging of such tumors that express both receptor types concomitantly, but also for tumors exhibiting a heterogeneous target expression with varying receptor densities between different individuals or lesions and for differential target expression during disease progression. A non-uniform distribution of target receptors between lesions results-if using monomeric peptidic radioligands-in the visualization of only some of the lesions whereas others are not depicted. Applying a HBPL for imaging, such lesions can nevertheless be addressed as long as one of the target receptors is present ( Figure 1). Pharmaceuticals 2018, 11,x FOR PEER REVIEW 2 of 20 one target receptor type, being thus only able to visualize tumors expressing this particular receptor. As different tumor lesions can however overexpress different receptor types and receptor expression can change upon metastasis and disease progression, a monovalent peptidic radioligand is often not able to visualize all tumor cells with the same efficiency and some lesions might be completely missed by the applied radiopeptide. Radiolabeled heterobivalent peptide ligands (HBPLs) on the other hand-by their ability to specifically target more than one receptor type-have been proposed to be better-suited agents for tumor imaging as they enable the visualization of the tumor by different receptor types potentially overexpressed by the target cells [1,2].
HBPLs furthermore can have favorable effects for in vivo tumor imaging compared to monovalent ligands such as improved in vivo biodistribution and enhanced avidity caused by simultaneous binding if both receptors are present on the tumor cell surface. In this case, also a higher probability of rebinding is achieved for a heterobivalent binder in case of dissociation from the receptor compared to a monovalent peptide due to "forced proximity" of the second potential binder to the second target receptor.
Furthermore, by being able to target different receptor types on the tumor cell surface, an overall higher number of receptors can be addressed by a heterobivalent ligand, increasing the probability of binding and thus tumor visualization. Thus, HBPLs are also of special interest not only for the imaging of such tumors that express both receptor types concomitantly, but also for tumors exhibiting a heterogeneous target expression with varying receptor densities between different individuals or lesions and for differential target expression during disease progression. A nonuniform distribution of target receptors between lesions results-if using monomeric peptidic radioligands-in the visualization of only some of the lesions whereas others are not depicted. Applying a HBPL for imaging, such lesions can nevertheless be addressed as long as one of the target receptors is present ( Figure 1).

Figure 1.
Schematic depiction of the functional principle of radiolabeled NPY(Y1)R-and gastrinreleasing peptide receptor (GRPR)-binding heterobivalent peptidic ligands (HBPLs) for tumor imaging: Monomeric peptide radiotracers can only bind to one receptor type and thus miss tumor tissues or lesions that do not express the respective receptor due to tumor heterogeneity or disease progression whereas the use of radiolabeled HBPLs, which can bind to more than one target receptor type, results in a higher probability of target visualization.
Recently, a radiolabeled HBPL, being able to address the gastrin-releasing peptide receptor as well as integrin αvβ3 was successfully translated into the clinics for imaging of prostate cancer with PET/CT, showing a much higher tumor visualization sensitivity compared to the respective GRPRtargeting peptide monomer. This proves the clinical relevance of the heterobivalent peptide targeting concept [3].
To be able to develop a HBPL being able to more efficiently target tumor tissues than the respective peptide monomers, it is necessary to know the receptor expression profile on the target NPY(Y 1 )R and NPY(Y 1 )R-specific radioligand GRPR and GRPR-specific radioligand NPY(Y 1 )-and GRPR-specific radioligand Figure 1. Schematic depiction of the functional principle of radiolabeled NPY(Y 1 )R-and gastrin-releasing peptide receptor (GRPR)-binding heterobivalent peptidic ligands (HBPLs) for tumor imaging: Monomeric peptide radiotracers can only bind to one receptor type and thus miss tumor tissues or lesions that do not express the respective receptor due to tumor heterogeneity or disease progression whereas the use of radiolabeled HBPLs, which can bind to more than one target receptor type, results in a higher probability of target visualization.
Recently, a radiolabeled HBPL, being able to address the gastrin-releasing peptide receptor as well as integrin α v β 3 was successfully translated into the clinics for imaging of prostate cancer with PET/CT, showing a much higher tumor visualization sensitivity compared to the respective GRPR-targeting peptide monomer. This proves the clinical relevance of the heterobivalent peptide targeting concept [3].
To be able to develop a HBPL being able to more efficiently target tumor tissues than the respective peptide monomers, it is necessary to know the receptor expression profile on the target tumor type.
Regarding this point, some excellent systematical work has been carried out, determining the presence of certain receptor types and their densities on different human malignancies [2,[4][5][6].
Human breast cancer, for example, overexpresses in about 75% of all cases the gastrin-releasing peptide receptor (GRPR) [7] and in 66-85% the neuropeptide Y receptor subtype 1 (NPY(Y 1 )R) [8]. Both receptors are expressed to an insignificant amount on healthy breast tissue, thus rendering both receptor types well-suited target structures for sensitive and specific breast cancer imaging. Furthermore, Reubi and co-workers could show on 68 human breast cancer samples that 63/68 (93%) overexpressed one or both receptor types. Of these, 32 (51%) expressed both receptors concomitantly, whereas 18 (29%) expressed only the GRPR, and a further 13 (21%) expressed only the NPY(Y 1 )R [5]. Thus, the combination of two peptidic ligands being able to specifically bind the NPY(Y 1 )R and the GRPR to one radioligand should enable a considerably higher breast cancer visualization efficiency and sensitivity and thus give less false-negative results compared to the respective monovalent radioligands ( Figure 1).
To obtain a highly potent bispecific HBPL, it is mandatory that both peptide parts of the construct are still able to bind to their respective target receptor type despite the considerable chemical modifications necessary for peptide heterodimerization. Thus, a suitable molecular design has to be found that enables the binding of both peptides to their respective target receptor.
So far, only one example of a heterobivalent NPY(Y 1 )R-and GRPR-targeting ligand has been described [9,10] and for this substance, no descriptions of radiolabeling with a PET isotope, in vitro cell uptake or in vivo imaging data are available. Thus, the general feasibility of the approach has not been demonstrated so far.
Thus, the aims of this study were to: (i) Develop a synthesis strategy yielding different HBPLs varying in molecular design and consisting of a NPY(Y 1 )-and a GRPR-affine peptide as well as a chelating agent (for radiolabeling with the positron-emitting radiometal nuclide 68 Ga); (ii) Establish the 68 Ga-radiolabeling and determine the log D and in vitro stability of the resulting 68 Ga-HBPLs in human serum; (iii) Evaluate the uptake of the 68 Ga-HBPLs into human breast cancer cell lines in vitro to determine if the substances can still interact with both target receptors and are taken up by the tumor cells; (iv) Determine if the particular molecular design used has a measurable influence on tumor cell uptake; and (v) Show the general feasibility of the approach by investigating the tumor uptake of the most potent HBPL in vivo by PET/CT imaging and ex vivo biodistribution in a proof-of-concept study and determine if the tumor uptake profits from the heterodimerization of the receptor-specific peptides.

Results and Discussion
2.1. Synthesis of GRPR-and NPY(Y 1 )R-Binding HBPLs 22-26, Scrambled HBPL Analogs 24a-c, Blocking Agents 3 and 4 as well as Monomeric Reference Peptides 27 and 28 At first, a suitable synthesis strategy towards the GRPR-and NPY(Y 1 )R-binding HBPLs was developed. The target molecular design of the substances is depicted in Figure 2 and was based on the following considerations: (i) The structure was to be based on a symmetrically branched scaffold to obtain homogenous products and the scaffold should comprise the chelator NODA-GA ( (1,4,7-triazacyclononane-4,7-diyl)diacetic acid-1-glutaric acid), which is able to stably and efficiently complex 68 Ga [11]; (ii) The chelator should be spatially separated from the receptor-affine peptides by a short PEG linker to prevent an interference of the radiometal complex with receptor binding [12]; (iii) As we and others were able to show before for peptide di-and multimers, the distance between the peptides within the same molecule can have a significant influence on the achievable receptor interaction [13][14][15][16], thus different distances between both peptidic receptor ligands should be investigated for the target HBPLs by introducing linkers of different length; (iv) Furthermore, as it was proposed that also the rigidity of the molecules might influence cellular uptakes by a conformational stabilization of the spatial orientation of the receptor ligands, the used linker structures did not only differ in length, but also in rigidity. Schematic depiction of the general molecular design of the target HBPLs consisting of a chelating agent (NODA-GA), a short PEG4-linker between radiometal complex and peptides, the symmetrical branching unit, the linkers of different length and rigidity (green) and the GRPR-and NPY(Y1)R-binding peptides BBN7-14 (cyan) and [Lys 4 ,Trp 5 ,Nle 7 ]BVD15 (magenta).

Synthesis of the Peptide Monomers 1−6
Aiming at the synthesis of GRPR-and NPY(Y1)R-binding HBPLs and their following in vitro and in vivo evaluation in human breast cancer tumor cells, we first synthesized the respective peptide monomers for subsequent heterodimerization on symmetrically branched scaffolds. As GRPR-affine peptide monomer, we chose the receptor agonist PESIN (PEG4-BBN7-14), which exhibits a favorably high stability, significant tumor uptake, as well as high tumor to background ratios in vivo [17,18], and can be modified at its N-terminal end without considerably changing its receptor binding affinity [13]. As NPY(Y1)R-affine peptide monomer, we chose [Lys 4 ,Trp 5 ,Nle 7 ]BVD15 as this peptide was shown to exhibit good affinities to the NPY(Y1)R even when further modified in position four [9,19,20].
The peptides were to be conjugated to the symmetrically branched scaffolds by a click chemistry approach to be able to obtain the desired rather complex target HBPLs efficiently. Furthermore, the coupling products have to be stable under physiological conditions. Different click chemistry reactions fulfill these requirements; of these, we chose the oxime formation between aminooxy functionalities and aldehydes.
Both peptides were synthesized by standard solid phase peptide synthesis (SPPS) methods [13,21] by successive conjugation of the respective Nα-Fmoc-amino acids after HBTU activation to the respective rink amide resin and finally modified on resin with bis-Boc-aminooxy acetic acid, giving aminooxy-PESIN (1) and [Lys 4 (aminooxy),Trp 5 ,Nle 7 ]BVD15 (2) (Figure 3). In case of PESIN, the aminooxy functionality was introduced at the N-terminal end as the peptide can be modified in this position without considerable alterations in binding affinity. In case of [Lys 4 ,Trp 5 ,Nle 7 ]BVD15, the aminooxy functionality was introduced in position 4 (Nε amine of lysine) as modifications in this position interfere least with receptor binding.  Aiming at the synthesis of GRPR-and NPY(Y 1 )R-binding HBPLs and their following in vitro and in vivo evaluation in human breast cancer tumor cells, we first synthesized the respective peptide monomers for subsequent heterodimerization on symmetrically branched scaffolds. As GRPR-affine peptide monomer, we chose the receptor agonist PESIN (PEG 4 -BBN [7][8][9][10][11][12][13][14], which exhibits a favorably high stability, significant tumor uptake, as well as high tumor to background ratios in vivo [17,18], and can be modified at its N-terminal end without considerably changing its receptor binding affinity [13]. As NPY(Y 1 )R-affine peptide monomer, we chose [Lys 4 ,Trp 5 ,Nle 7 ]BVD 15 as this peptide was shown to exhibit good affinities to the NPY(Y 1 )R even when further modified in position four [9,19,20].
The peptides were to be conjugated to the symmetrically branched scaffolds by a click chemistry approach to be able to obtain the desired rather complex target HBPLs efficiently. Furthermore, the coupling products have to be stable under physiological conditions. Different click chemistry reactions fulfill these requirements; of these, we chose the oxime formation between aminooxy functionalities and aldehydes.
Both peptides were synthesized by standard solid phase peptide synthesis (SPPS) methods [13,21] by successive conjugation of the respective N α -Fmoc-amino acids after HBTU activation to the respective rink amide resin and finally modified on resin with bis-Boc-aminooxy acetic acid, giving aminooxy-PESIN (1) and [Lys 4 (aminooxy),Trp 5 ,Nle 7 ]BVD 15 (2) (Figure 3). In case of PESIN, the aminooxy functionality was introduced at the N-terminal end as the peptide can be modified in this position without considerable alterations in binding affinity. In case of [Lys 4 ,Trp 5 ,Nle 7 ]BVD 15 , the aminooxy functionality was introduced in position 4 (N ε amine of lysine) as modifications in this position interfere least with receptor binding. Aiming at the synthesis of GRPR-and NPY(Y1)R-binding HBPLs and their following in vitro and in vivo evaluation in human breast cancer tumor cells, we first synthesized the respective peptide monomers for subsequent heterodimerization on symmetrically branched scaffolds. As GRPR-affine peptide monomer, we chose the receptor agonist PESIN (PEG4-BBN7-14), which exhibits a favorably high stability, significant tumor uptake, as well as high tumor to background ratios in vivo [17,18], and can be modified at its N-terminal end without considerably changing its receptor binding affinity [13]. As NPY(Y1)R-affine peptide monomer, we chose [Lys 4 ,Trp 5 ,Nle 7 ]BVD15 as this peptide was shown to exhibit good affinities to the NPY(Y1)R even when further modified in position four [9,19,20].
The peptides were to be conjugated to the symmetrically branched scaffolds by a click chemistry approach to be able to obtain the desired rather complex target HBPLs efficiently. Furthermore, the coupling products have to be stable under physiological conditions. Different click chemistry reactions fulfill these requirements; of these, we chose the oxime formation between aminooxy functionalities and aldehydes.
Both peptides were synthesized by standard solid phase peptide synthesis (SPPS) methods [13,21] by successive conjugation of the respective Nα-Fmoc-amino acids after HBTU activation to the respective rink amide resin and finally modified on resin with bis-Boc-aminooxy acetic acid, giving aminooxy-PESIN (1) and [Lys 4 (aminooxy),Trp 5 ,Nle 7 ]BVD15 (2) ( Figure 3). In case of PESIN, the aminooxy functionality was introduced at the N-terminal end as the peptide can be modified in this position without considerable alterations in binding affinity. In case of [Lys 4 ,Trp 5 ,Nle 7 ]BVD15, the aminooxy functionality was introduced in position 4 (Nε amine of lysine) as modifications in this position interfere least with receptor binding.  As the target HBPLs should be evaluated in vitro regarding their ability to be taken up by human breast cancer cell lines and the contribution of both parts of the HBPLs on tumor cell uptake should be assessed, we further synthesized the peptide monomers bombesin (3) and [Lys 4 ,Trp 5 ,Nle 7 ]BVD 15 (4) as blocking substances for the GRPR and the NPY(Y 1 )R during these experiments ( Figure 4).
Pharmaceuticals 2018, 11, x FOR PEER REVIEW 5 of 20 As the target HBPLs should be evaluated in vitro regarding their ability to be taken up by human breast cancer cell lines and the contribution of both parts of the HBPLs on tumor cell uptake should be assessed, we further synthesized the peptide monomers bombesin (3) and [Lys 4 ,Trp 5 ,Nle 7 ]BVD15 (4) as blocking substances for the GRPR and the NPY(Y1)R during these experiments ( Figure 4). Regarding the in vivo evaluation of the HBPLs in tumor-bearing animals and the verification of the receptor specificity of the observed tumor uptakes and the contribution of both peptides of the HBPLs to overall tumor uptakes, two different approaches can be followed. The first one is to block the respective target receptor analogous to the in vitro assays by adding blocking substances 3 or 4 and the other one is to use scrambled HBPL analogs. To compare the in vivo tumor uptake of an HBPL to that of its scrambled analogs instead of performing blocking studies however eliminates possible difficulties that might arise from the low stability of the monomeric receptor ligands.
Thus, three different scrambled HBPL analogs were synthesized: PESIN combined with scrambled [Lys 4 (aminooxy),Trp 5 ,Nle 7 ]BVD15, scrambled PESIN combined with [Lys 4 (aminooxy), Trp 5 ,Nle 7 ]BVD15 and both peptides of the HBPL scrambled. For this purpose, the two scrambled aminooxy-modified peptide monomers aminooxy-PESINscrambled (5) and [Lys 4 (aminooxy),Trp 5 ,Nle 7 ] BVD15,scrambled (6) ( Figure 5) were synthesized and analogously to 1 and 2 used during the following HBPL syntheses.  Regarding the in vivo evaluation of the HBPLs in tumor-bearing animals and the verification of the receptor specificity of the observed tumor uptakes and the contribution of both peptides of the HBPLs to overall tumor uptakes, two different approaches can be followed. The first one is to block the respective target receptor analogous to the in vitro assays by adding blocking substances 3 or 4 and the other one is to use scrambled HBPL analogs. To compare the in vivo tumor uptake of an HBPL to that of its scrambled analogs instead of performing blocking studies however eliminates possible difficulties that might arise from the low stability of the monomeric receptor ligands.
Pharmaceuticals 2018, 11, x FOR PEER REVIEW 5 of 20 As the target HBPLs should be evaluated in vitro regarding their ability to be taken up by human breast cancer cell lines and the contribution of both parts of the HBPLs on tumor cell uptake should be assessed, we further synthesized the peptide monomers bombesin (3) and [Lys 4 ,Trp 5 ,Nle 7 ]BVD15 (4) as blocking substances for the GRPR and the NPY(Y1)R during these experiments ( Figure 4). Regarding the in vivo evaluation of the HBPLs in tumor-bearing animals and the verification of the receptor specificity of the observed tumor uptakes and the contribution of both peptides of the HBPLs to overall tumor uptakes, two different approaches can be followed. The first one is to block the respective target receptor analogous to the in vitro assays by adding blocking substances 3 or 4 and the other one is to use scrambled HBPL analogs. To compare the in vivo tumor uptake of an HBPL to that of its scrambled analogs instead of performing blocking studies however eliminates possible difficulties that might arise from the low stability of the monomeric receptor ligands.

Synthesis of the Heterobivalent Ligands 22-26, 24a-c and Monomeric Reference Peptides 27 and 28
The branched bis-amines 7−11 and bis-aldehydes 12−16 (Scheme 1) were synthesized following a published procedure [22] with minor modifications (see Supplementary Materials for detailed description). In the following, these NODA-GA-modified branched bis-aldehyde scaffolds 12−16 were efficiently reacted with aminooxy-PESIN (1) and aminooxy-PESIN scrambled (5) to the monovalent intermediates 17−21 and 19a. These were further reacted with [Lys 4 (aminooxy),Trp 5 ,Nle 7 ]BVD 15 (2) and its scrambled analog 6 to the final heterobivalent peptidic target structures 22−26 and their partly or fully scrambled analogs 24a-c (Scheme 1). The branched bis-amines 7−11 and bis-aldehydes 12−16 (Scheme 1) were synthesized following a published procedure [22] with minor modifications (see Supplementary Materials for detailed description). In the following, these NODA-GA-modified branched bis-aldehyde scaffolds 12−16 were efficiently reacted with aminooxy-PESIN (1) and aminooxy-PESINscrambled (5)   1 as well as its scrambled analog 5 reacted efficiently within minutes with the branched bisaldehydes 12−16, giving the respective monovalent conjugation products 17−21 and 19a in satisfactory yields of 47% to 58%. Higher yields could not be obtained as 1 and 5 had to be applied in  1 as well as its scrambled analog 5 reacted efficiently within minutes with the branched bis-aldehydes 12−16, giving the respective monovalent conjugation products 17−21 and 19a in satisfactory yields of 47% to 58%. Higher yields could not be obtained as 1 and 5 had to be applied in a lower amount than the bis-aldehydes 12−16 to minimize the formation of the respective homobivalent PESIN-dimers, being the only observed side products in this reaction. These monovalent intermediates were in the following reacted with 2 or 6, proceeding equally efficient than the first reaction step within minutes, giving the target HBPLs 22−26 as well as the scrambled analogs 24a-c in good yields of 49% to 82%.
Pharmaceuticals 2018, 11, x FOR PEER REVIEW 7 of 20 a lower amount than the bis-aldehydes 12−16 to minimize the formation of the respective homobivalent PESIN-dimers, being the only observed side products in this reaction. These monovalent intermediates were in the following reacted with 2 or 6, proceeding equally efficient than the first reaction step within minutes, giving the target HBPLs 22−26 as well as the scrambled analogs 24a-c in good yields of 49% to 82%. The HBPLs exhibited-depending on the linker structure used-distances between both peptidic receptor ligands of 46 (no additional linker units used), 64 (PEG2 linkers), 78 (PEG4 linkers), 60 (ACMP linkers), and 74 (two successive ACMP linkers) bond lengths.

68 Ga-Radiolabeling, logD and Stability Determination of Peptide Heterodimers
Regarding a favorable in vivo biodistribution of the radioligands, the logD of the HBPLs should be in a comparable range as that of the lead peptide monomers as we and others were able to show before that a high lipophilicity negatively influences tumor uptake, organ distribution, and unspecific background accumulation, resulting in a limited usefulness of the radiopeptides for tumor visualization  (28), serving as mono-specific reference substances for the HBPLs in the following in vitro tumor cell uptake studies.  Figure S1A) as well as non-optimized molar activities of 10−15 GBq/µmol (used for in vitro assays and obtained by using an itG generator system) or 40−46 GBq/µmol (used for in vivo evaluations, obtained by using an Eckert & Ziegler IGG100 generator system), starting from 110−150 or 420-460 MBq of 68 Ga 3+ , respectively.
Regarding a favorable in vivo biodistribution of the radioligands, the log D of the HBPLs should be in a comparable range as that of the lead peptide monomers as we and others were able to show before that a high lipophilicity negatively influences tumor uptake, organ distribution, and unspecific background accumulation, resulting in a limited usefulness of the radiopeptides for tumor visualization [23][24][25] In the following, we intended to determine if we could observe an independent binding of both peptide parts of the HBPLs to both target receptor types, being the prerequisite for improved/more likely tumor uptake (→ Figure 1). This can be achieved by tumor cell uptake studies of the radiotracers as it was shown before for radiolabeled somatostatin analogs that the in vitro cell uptake directly correlates to in vivo tumor uptakes [27], demonstrating the relevance of such in vitro tumor cell uptake studies.
The human breast cancer cell line T-47D was described to express both the GRPR [9,28] as well as the NPY(Y 1 )R [29,30] (where β-estradiol in the medium increases NPY(Y 1 )R-expression) and thus should be the ideal cell line to determine if both parts of the developed HBPLs bind to their respective receptor and if a synergistic effect of peptide heterodimerization on tumor cell uptake can be achieved. Of course, it would also have been feasible to use different cells lines expressing either the GRPR or the NPY(Y 1 )R to demonstrate that both peptides of the HBPLs are still able to address their respective target receptor, but a cell line expressing both receptors concomitantly is far more advantageous to showcase the potential beneficial effects of heterodimerization and to determine the part each of the peptides contributes to tumor cell uptake in case of a concomitant receptor expression.
Thus, we first determined the uptake of the HBPLs  These results indicate that the heterodimerization and the resulting significant chemical modification of the peptides as well as the compounds' complexity and size do not affect the GRPR-specific tumor cell uptake of the HBPLs compared to the monomeric reference [ 68 Ga] . Also, the molecular design of the heterobivalent ligands comprising linkers of different length and rigidity does not seem to have a significant effect on the GRPR-mediated cellular uptake of the respective HBPL radiotracer.
In the following, we tested the uptake of the peptide monomers [ 68 Ga] and [ 68 Ga] in three further standard cell lines of breast cancer: BT-474, MCF-7 and MDA-MB-231 cells. Of these, the BT-474 cells were also described to express both the GRPR [31] and the NPY(Y 1 )R [31], whereas MCF-7 cells were described to be NPY(Y 1 )R positive [30,32] but expressing the GRPR only to a low extent [9] and MDA-MB-231 cells were described to be GRPR positive [33] but expressing the NPY(Y 1 )R to a low extent [29,30] [ 68 Ga] , is not able to efficiently address the NPY(Y 1 )R; (ii) The NPY(Y 1 )R is expressed in such low amounts that the tested radioligands cannot be efficiently taken up by the cells by this receptor; or (iii) The molar activities of the tested  These results indicate that the heterodimerization and the resulting significant chemical modification of the peptides as well as the compounds' complexity and size do not affect the GRPR-specific tumor cell uptake of the HBPLs compared to the monomeric reference [ 68 Ga]27. Also, the molecular design of the heterobivalent ligands comprising linkers of different length and rigidity does not seem to have a significant effect on the GRPR-mediated cellular uptake of the respective HBPL radiotracer.
In the following, we tested the uptake of the peptide monomers [ 68 Ga]27 and [ 68 Ga]28 in three further standard cell lines of breast cancer: BT-474, MCF-7 and MDA-MB-231 cells. Of these, the BT-474 cells were also described to express both the GRPR [31] and the NPY(Y 1 )R [31], whereas MCF-7 cells were described to be NPY(Y 1 )R positive [30,32] but expressing the GRPR only to a low extent [9] and MDA-MB-231 cells were described to be GRPR positive [33] but expressing the NPY(Y 1 )R to a low extent [29,30] 26, is not able to efficiently address the NPY(Y 1 )R; (ii) The NPY(Y 1 )R is expressed in such low amounts that the tested radioligands cannot be efficiently taken up by the cells by this receptor; or (iii) The molar activities of the tested radioligands were too low and a significant self-blocking of the NPY(Y 1 )R-mediated uptake took place, preventing the cell uptake.
In order to determine if the NPY(Y 1 )R is actually present on T- In all of the three tested cell lines, the specific uptake of 125 I-PYY showed to be negligibly low (between 0.2 and 2.5% of the total activity applied), confirming that the target NPY(Y 1 ) receptor might be present on the cells but if so, then only to a very low amount, being too low to give useful results in the cell uptake studies.
Furthermore, this low receptor density prevents a successful determination of the receptor affinity of the developed radioligands. This is in contrast to other studies, using identically treated MCF-7 cells to determine NPY(Y 1 )R affinities of newly developed NPY(Y 1 )R-affine radioligands [9,19]. We could however not reproduce these results using MCF-7 cells of different suppliers due to the observed low expression of NPY(Y 1 )R on the cells.
As it was described that MCF-7 cells can express the target NPY(Y 1 )R to a much higher extent in vivo than under in vitro conditions [34], this might also be the case for the T-47D cell line, expressing besides the NPY(Y 1 )R also the for our scientific question important GRPR. Thus, we in the following performed initial in vivo evaluations of HBPL [ 68 Ga]24, showing highly promising results in the preceding evaluations. Of the developed HBPLs, [ 68 Ga]24 showed the highest stability and hydrophilicity as well as a slightly higher tumor cell uptake than the other analogs in vitro and thus represents a potent representative of the developed HBPLs for a following proof-of-concept in vivo evaluation. By evaluating these radioligands under the same conditions, the proportion of both peptide binders on the uptake of [ 68 Ga]24 and also the receptor-specificity of the uptake of the radioligand via both receptors should be determined. The evaluation of the partly scrambled monovalent analogs instead of the monomeric peptides, which would also have been possible, exhibits the advantage that all evaluated radioligands show a similar pharmacokinetic distribution and also a similar possible degradation pattern. Using the monovalent peptides bombesin or NPY/BVD for receptor blocking (to show the receptor specificity of both peptide parts of [ 68 Ga]24 and to determine the contribution of both peptide parts of the HBPL to overall tumor uptake) might have resulted in an incomplete blocking of the respective receptor as the GRPR-affine bombesin as well as the NPY(Y 1 )R-affine peptides NPY and BVD are known for their limited in vivo stability.

Proof-of-
For the PET/CT imaging studies, 5.5-8.0 MBq of the respective 68 Ga-radioligand were administered via the lateral tail vein under isoflurane anesthesia to the tumor-bearing animals. Directly after completion of the diagnostic scans, the animals were sacrificed, the organs were collected and measured in a gamma-counter. The results of these in vivo PET/CT imaging studies and the ex vivo biodistribution data are given in Figure 8 and Table S1. This indicates that both parts of HBPL [ 68 Ga]24 contributed to in vivo tumor uptake and that its uptake into the tumor was GRPR-and also NPY(Y1)R-specific.   As can easily be seen from the in vivo PET/CT imaging as well as the ex vivo biodistribution data, all developed ligands showed a rather high kidney and liver uptake. Thus, further improvements in ligand design must be carried out to result in clinically relevant imaging agents for improved breast cancer visualization with PET/CT.  This indicates that both parts of HBPL [ 68 Ga]24 contributed to in vivo tumor uptake and that its uptake into the tumor was GRPR-and also NPY(Y 1 )R-specific.
As can be observed from the time-activity-curves obtained by PET imaging (Figure 9a), those ligands comprising an intact variant of the GRPR-targeting ligand BBN 7-14 showed a stable plateau phase in tumor accumulation ([ 68 Ga]24b) or only a slight decrease ([ 68 Ga]24) whereas [ 68 Ga]24a, comprising a scrambled variant of BBN 7-14 , shows a surge in tumor uptake followed by a decline of activity in the tumor. This might be attributable to the limited stability of BVD analogs under in vivo conditions, which has been described before [35] and might result in radioligand metabolization during imaging. If then the GRPR-binding peptide is still able to bind to its target receptor, the tumor uptake nevertheless remains largely stable due to a GRPR-mediated uptake. However, if only a scrambled variant of BBN 7-14 is present, no receptor-affine peptide binder remains for continuous radiotracer uptake into the tumor, resulting in an overall decrease in tumor accumulation.
In summary, we were able to show here for the first time that the general concept to assemble a GRPR-and a NPY(Y 1 )R-affine peptide to one combined radioligand is in general feasible regarding a contribution of both peptides of the HBPL to in vivo tumor uptake and thus is beneficial with regard to overall tumor uptake and tumor visualization probability compared to the monospecific agents.
Bis-amines 7-11 and bis-aldehydes 12-16 were synthesized according to published procedures [22] with minor modifications of the synthesis protocols. Details of these syntheses can be found in the supplementary information.
Unless otherwise stated, the coupling reactions during solid phase-based syntheses were usually carried out in DMF for 30 min using 4 eq. of acid, 3.9 eq. of HBTU as coupling reagent and 4 eq. of DIPEA (N,N-Diisopropylethylamine) as base. Fmoc protecting groups were removed using 50% (v/v) piperidine in DMF.
For the in vivo evaluations, five week old female Fox Chase SCID mice were obtained from Janvier and implanted with estradiol pellets (0.36 mg/60 days; obtained from Innovative Research of America) one week prior to tumor cell inoculation. Tumor cells were inoculated using Matrigel basal membrane matrix with reduced growth factor (obtained from VWR). For PET/CT measurements, a small animal Albira II PET/SPECT/CT system (Bruker, Eggenstein-Leopoldshafen, Germany) was used.
Synthesis of aminooxy-PESIN (1) and aminooxy-PESIN scrambled (5). The peptides were synthesized on solid support by standard Fmoc solid-phase peptide synthesis using a commercially available standard Rink amide MBHA resin, HBTU as coupling reagent, standard N α -Fmoc-amino acids, N ω -Fmoc-PEG 4 -OH and bis-Boc-aminooxy acetic acid. All amino acids (apart from bis-Boc-aminooxy acetic acid which was reacted for 60 min) were coupled within 30 min. The crude aminooxy-modified peptides were cleaved from the solid support using a mixture of TFA:TIS:H 2 O of 95 :2.5:2.5 (v/v)  6 was purified using a gradient of 10-50% MeCN + 0.1% TFA in 8 min (R t = 4.31 min) and isolated as white solid after lyophilization in yields of 20% (26.0 mg; 19.7 µmol). MALDI-MS (m/z) using Synthesis of DOTA-PESIN (27). The peptide was synthesized on solid support by standard Fmoc solid-phase peptide synthesis using a commercially available standard Rink amide MBHA resin, HBTU as coupling reagent, standard N α -Fmoc-amino acids, and N ω -Fmoc-PEG 4 -OH. After the conjugation of the PEG 4 linker to the peptide sequence, DOTA-(tBu) 3 was coupled within 120 min using an excess of the synthon of 2.7 eq. together with 2.6 eq. HBTU and 4 eq. DIPEA. The crude product was cleaved from the solid support using a mixture of TFA  (28). The peptide was synthesized on solid support by standard Fmoc solid-phase peptide synthesis using a commercially available standard Rink amide MBHA resin, HBTU as coupling reagent, standard N α -Fmoc-amino acids and Fmoc-Lys(Mtt)-OH. After conjugation of the last amino acid, the lysine side chain Mtt-protecting group was removed with diluted TFA (TFA:DCM 1:99 (v/v)) within 2 h and DOTA-(tBu) 3 was coupled in this position within 120 min using an excess of the synthon of 2.7 eq. together with 2.6 eq. HBTU and 4 eq. DIPEA. The crude DOTA-modified peptide was cleaved from the solid support using a mixture of TFA  (1.25 M, 90-95 µL). After reaction for 10 min at 45 • C, the reaction mixtures were analyzed by analytical radio-HPLC. The radiolabeled products were found to be 95-99% pure and obtained in molar activities of 40-46 GBq/µmol (non-optimized). The pH of the radiotracer solution was adjusted to 6.0-7.0 using HEPES buffer (2.0 M, pH 8.0, 200 µL) and used for the in vivo studies. Determination of radiotracer lipophilicity. The heterobivalent ligands (22)(23)(24)(25)(26) as well as the monomeric reference compounds (27 and 28) were radiolabeled with 68 Ga as described before and 2 µL of the product solution (~65 pmol of the respective radioligand) were added to a mixture of phosphate buffer (0.05 M, pH 7.4, 800 µL) and 1-octanol (800 µL) and incubated for 5 min at ambient temperature under vigorous shaking. Both phases were separated by centrifugation and 100 µL of each phase were measured for radioactivity in a gamma-counter. From these data, the distribution coefficient log D was calculated from the following equation: logD o/w = log(cpm o /cpm w ), where: cpm o = activity in the 1-octanol phase [cpm] (cpm = counts per minute), cpm w = activity in the aqueous phase [cpm]. These experiments were performed six times independently.
Determination of the stability of the ligands in human serum. The heterobivalent ligands (22)(23)(24)(25)(26) as well as the monomeric reference compounds (27 and 28) were radiolabeled with 68 Ga as described before and 125 µL of the product solution were added to 500 µL of human serum and incubated at 37 • C. At defined time-points of 5, 15, 30, 60 and 90 min, aliquots of 75 µL of the mixture were added to 75 µL of ethanol and the precipitation of serum proteins was enhanced by ice-cooling for 2 min. After centrifugation, supernatant and precipitate were measured for radioactivity and the supernatant was analyzed by analytical radio-HPLC. These experiments were performed thrice.
Cell culture. All cell lines were grown in suitable culture medium at 37 • C in a humidified CO 2 (5%) atmosphere. The human breast cancer cell lines T-47D, MDA-MB-231 and MCF-7 were grown in Dulbecco's Modified Eagle Medium (DMEM), supplemented with 10% (v/v) fetal calf serum (FCS) and 1% (v/v) L-Glutamine. For a high expression of the NPY(Y 1 ) receptor on T-47D cells, the medium for this cell line was further supplemented with 0.15% (w/v) β-estradiol. The BT-474 and PC-3 cell lines were grown in RPMI-1640 medium, also supplemented with 10% (v/v) fetal calf serum (FCS) and 1% (v/v) L-Glutamine.
Internalization studies. Cells (T-47D, MDA-MB-231, MCF-7, BT-474 and PC-3, 1.5 × 10 6 cells per well) were seeded into 6-well plates and incubated overnight at 37 • C in a humidified CO 2 (5%) atmosphere. The next day, the medium was removed and the cells were washed twice with the respective medium without supplements (ice-cold, 1 mL) and incubated with 3.7-4.0 kBq (0.37-4.0 pmol) of the respective 68 Ga-radiolabeled ligand [ 68 Ga]22−[ 68 Ga]26, [ 68 Ga]27 or [ 68 Ga]28 (in 1.5 mL medium, containing 0.5% (w/v) BSA) for defined time-points of 1, 2, 3 or 4 h at 37 • C in a humidified CO 2 (5%) atmosphere. A 1000-fold excess of the respective peptide (3 or 4) was used for blocking to determine the non-specific cell uptake. At each time point, the medium was removed and the cells were washed twice with the respective medium without supplements (ice-cold, 1 mL). Cells were treated twice with 1 mL glycine buffer (ice-cold, 50 mM glycine, 100 mM NaCl, pH 2.8) for 5 min at room temperature, followed by 2 mL NaOH solution (1 M) for 10 min at 37 • C. The supernatants were collected and the radioactivity measured in a gamma counter. The internalized and surface bound activity was expressed as percentage of measured to total added activity. Each data point was generated thrice in triplicates.
In vivo experiments. All animal experiments were performed in compliance with the German animal protection laws and protocols of the local committee (Regierungspräsidium Karlsruhe, approval number: 35-9185.81/G-206/15). 20, six week old female immunodeficient Fox Chase SCID (CB17/Icr-Prkdc scid /IcrIcoCrl) mice with an average weight of 20 g were subcutaneously implanted with 17β-estradiol pellets (0.36 mg/60 days). 4 days later, the tumors were induced by subcutaneous inoculation of 5 × 10 6 T-47D cells into the left flank of the approval number s. After induction, the tumors were allowed to grow for 8-10 weeks and reached a diameter of about 0.5 cm.
For imaging, the animals were anaesthetized with isoflurane and injected with 5.5-8.0 MBq of the respective radioligand ([ 68 Ga]24, [ 68 Ga]24a or [ 68 Ga]24b) into the lateral tail vein. Dynamic PET images were acquired over 90 min and CT images were obtained within further 30 min. After the end of the diagnostic scan, the animals were sacrificed, the organs were collected and measured in a gamma-counter.
The dynamic PET images were reconstructed using the Albira Suite Reconstructor (Bruker) with an iterative dynamic reconstruction with 12 iterations using an 2D-Maximum-Likelihood Expectation-Maximization (MLEM) algorithm and a cubic image voxel size of 0.5 mm after scatter and decay correction. Data were divided into time frames from 1 to 10 min (10 × 1 min, 10 × 2 min, 6 × 5 min and 3 × 10 min) for the assessment of temporal changes in regional tracer accumulation. The CT images were obtained at 45 kVp, with currents of 0.4 mA (high dose, good resolution). Acquisitions of 400 projections were taken and a 250 µm isotropic voxel size image was reconstructed via filtered back projection. The reconstructed PET data were manually fused with the CT images using PMOD 3.6.1.1. and analyzed. Volumes of interest (VOIs) were defined for the quantification of tracer accumulation in heart, liver, kidneys, tumor, and muscle. The results for each VOI were calculated as SUV (kBq/cm 3 ) averaged for each time frame.

Conclusions
We were able to show that is chemically and radiochemically feasible to synthesize radiolabeled heterobivalent peptides consisting of a GRPR-and a NPY(Y 1 )R-affine peptide on symmetrically branched scaffolds, resulting in bispecific heterobivalent peptidic PET radiotracers. The compounds demonstrated high stabilities in human serum, hydrophilicities comparable to the monomeric lead peptides and high GRPR-mediated tumor cell uptakes in vitro.
The performed in vivo imaging and ex vivo biodistribution studies indicated a contribution of both peptides of the evaluated HBPL to overall in vivo tumor uptake, showing the feasibility of the general concept to develop GRPR-and NPY(Y 1 )R-bispecific PET radiotracers with regard to an improved and more sensitive tumor visualization of human breast cancer.
Nevertheless, the results also show that further work is required to obtain GRPR-and NPY(Y 1 )R-bispecific imaging agents being useful for clinical application due to the high kidney and liver accumulation of the agents developed so far.
Supplementary Materials: Supporting material for this article, comprising the syntheses of compounds 7-11 and 12-16, the results of the log D determinations, radio-HPLC chromatograms and serum stability data, results of in vitro cell uptake studies and ex vivo biodistribution data is available online at http://www.mdpi.com/1424-8247/11/3/65/s1.