The influence of different metal-chelate conjugates of pentixafor on the CXCR4 affinity

Background The overexpression of the chemokine receptor 4 (CXCR4) in different epithelial, mesenchymal, and hematopoietic cancers makes CXCR4 an attractive diagnostic and therapeutic target. However, targeting the CXCR4 receptor with small cyclic pentapeptide-based radiopharmaceuticals remains challenging because minor structural modifications within the ligand-linker-chelate structure often significantly affect the receptor affinity. Based on the excellent in vivo properties of CXCR4-directed pentapeptide [68Ga]pentixafor (cyclo(-d-Tyr-N-Me-d-Orn(AMB-DOTA)-l-Arg-l-2-Nal-Gly-)), this study aims to broaden the spectrum of applicable (radio)metal-labeled pentixafor analogs. Methods Cyclic pentapeptides, based on the pentixafor scaffold, were synthesized by a combined solid- and solution-phase peptide synthesis. The CXCR4 receptor affinities of the cold reference compounds were determined in competitive binding assays using CXCR4-expressing Jurkat T - cell leukemia cells and [125I]FC131 as the radioligand. Results Metalated pentixafor derivatives with cyclic and acyclic chelators were synthesized by solid-phase peptide synthesis and evaluated in vitro. The resulting CXCR4 affinities (IC50) were highly dependent on the chelator and metal used. Two pentapeptides, Ga-NOTA and Bi-DOTA conjugates, offer an improved affinity compared to [68Ga]pentixafor. Conclusions Based on the pentapeptide [68Ga]pentixafor, a broad range of metal-labeled analogs were investigated. The affinities of the new compounds were found to be strongly dependent on both the chelator and the metal used. Bi-labeled pentixafor showed high receptor affinity and seems to be a promising ligand for further preclinical evaluation and future α-emitter-based endoradiotherapy.


Background
The chemokine receptor 4 (CXCR4) and its sole known natural ligand stromal cell derived factor-1 (SDF-1, CXCL12) are physiologically involved in leukocyte recruitment, homing, and retention of hematopoietic stem and progenitor cells [1,2]. CXCR4 also holds a substantial role in various pathological conditions and represents a highly attractive therapeutic target. Overexpression of CXCR4 has been linked to cancer proliferation, cell migration, and tissue-specific homing of cancer cells as well as resistance to conventional and targeted therapies [3][4][5].
Based on the success of [ 68 Ga]pentixafor, this study aims to broaden the potential spectrum of radiometallabeled CXCR4 ligands, both for imaging and therapy, by development and in vitro evaluation of a range of tracers labeled with alternative radionuclides, such as 111 In 3+ (for single photon emission computed tomography (SPECT)); 18 F − and 89 Zr 4+ (for PET); or 90 Y 3+ , 177 Lu 3+ , and 213 Bi 3+ (for endoradiotherapy).
However, the experiences gained during the development of pentixafor have shown that, compared with [ 68 Ga]pentixafor, unlabeled pentixafor and other radiometalated pentixafor derivatives exhibit significantly lower CXCR4 receptor affinities. Thus, in contrast to other peptides, such as somatostatin receptor (SSTR), gastrin-releasing peptide receptor (GRPR), or α v β 3 binding peptides, the affinity of [ 68 Ga]pentixafor towards CXCR4 is determined by the entire ligand-spacerchelator-radiometal construct. Consequently, a more or less independent "bioactive substructure" or "pharmacophor" (e.g., the pentapeptide core A depicted in Fig. 1) cannot be identified. In this study, we investigated pentixafor derivatives with alternative cyclic and acyclic chelators and evaluated these ligands in vitro. With regard to the utilized chelators, the following nuclides relevant for medical purposes have been investigated: Ga 3+ , AlF 2+ , Zr 4+ , Cu 2+ , In 3+ , Lu 3+ , Y 3+ , and Bi 3+ (Fig. 1).

Synthesis of A
The peptide was synthesized according to a previously published procedure [12]. In short, synthesis was carried out using a standard Fmoc strategy using a TCP resin as solid support and HOBt/TBTU as coupling reagents. After selective N-methylation of D-Orn, D-Tyr was coupled to the peptide with HATU/HOAt, cleaved from the resin, and finally cyclized.

Coupling of chelators and metal complexation
A molar excess of the activated chelator was added to a free amino group of the peptide analog. Subsequent to successful coupling, the chelator-conjugated peptide was deprotected and purified. Metal complexation was performed in the presence of weakly chelating acetate buffers to reduce the likelihood of hydrolysis. Solutions for metal labeling comprised LuCl 3 (20 mM), pH = 6.0; InCl 3 (20 mM), pH = 4.5; and YCl 3 (20 mM), pH = 5.9, each in ammonium acetate (0.1 M) and Ga(NO 3 ) 3 (2 mM) pH = 3.0; Cu(OAc) 2 pH = 6.0; and ZrCl 4 (20 mM) pH = 1.3, each in water. The chelator-conjugated peptide (250 μL, 2 mM, 1 equiv) was dissolved in H 2 O and DMSO up to 50 % (v/v), if necessary, and the metal (1-10 equiv) was added, pH adjusted, and heated for 30 min. Final metalated peptides were obtained in a purity ≥ 95 %, and used for in vitro studies without further purification, unless stated otherwise.
CXCR4 affinities were determined in competitive binding assays using Jurkat cells with [ 125 I]FC131 as the radioligand according to a protocol similar to previously published [7]. FC131 (cyclo(-D-Tyr-L-Arg-L-Arg-L-Nal-Gly-)) [14] was synthesized and iodinated as described previously [7]. Jurkat cells (4 × 10 5 cells per vial) were incubated with the respective peptide of interest at the final concentrations ranging from 10 −11 to 10 −5 M and app. 0.1 nM of [ 125 I]FC131. The total sample volume was 250 μL. After an incubation time of 120 min, the vials were centrifuged at 1300 rpm (Heraeus Megafuge, Thermo) for 3 min and the supernatant was removed. The cells were washed twice with 200 μL ice-cold Hank's balanced salt solution (HBSS). After each washing step, the samples were centrifuged and the supernatant removed. Finally, the amount of displaced and bound radioligand in the combined fractions of the supernatant and the cell pellet was quantified. The half maximal inhibitory concentration (IC 50 ) values were determined using GraphPad Prism software.
When switching to the corresponding DOTAGA derivatives, the IC 50 values indicate a similar order within the series of investigated metal complexes with generally lower affinities (Table 1), except the Bi 3+ -complex, which, again surprisingly, showed only a small decrease of CXCR4 affinity. Thus, with respect to receptor affinity, DOTAGApentixafor was found to offer no advantage over DOTApentixafor analogs.
A corresponding evaluation of the NOTA and NODAGA derivatives identified [ nat Ga 3+ ]3 as the ligand with the highest affinity in this study (IC 50 : 17.8 ± 7.7 nM). All other complexes, including the [ nat F]AlF-NOTA derivative, seem to be unsuitable for further preclinical evaluation or potential clinical application. Similar disappointing results were obtained for NODA-Bn-SCN, DTPA, DTPA-Bn-SCN, and DFO-BN-SCN derivatives.

Discussion
Experiences in the development of CXCR4-targeting peptides showed that affinities to the CXCR4 receptor can be significantly affected by even moderate structural modifications in the pentapeptide core [16][17][18][19], the linker unit [20], or the chelate [7,8]. This is in contrast to previous experiences of GPCR, glycoprotein, or enzymetargeting peptides, such as SSTRs, GRPRs (bombesin), and integrins (e.g., αvß 3 , RGD peptides), as well as the prostatespecific membrane antigen (PSMA), to mention only a few. A variety of peptides towards these targets have been developed and-unlike CXCR4-conjugated with a broad spectrum of linker/chelator moieties. For αvß 3binding RGD peptides, the Lys side chain of the typically used c(Arg-Gly-Asp-D-Phe-Lys) does not influence the binding of the peptide in the cleft between the two α v and β 3 subunits that forms the heterodimeric transmembrane glycoprotein [21]. Thus, these RGD peptides tolerate the introduction of spacers and chelator or the formation of multimers, such as dimers, tetramers, and octamers [22,23]. Similar freedom of variation, although not that multifarious, has been found for SST ligands. Tyr 3 -octreotate for instance was conjugated to both DTPA and DOTA and labeled with nat In, nat Ga, or nat Y. All conjugates, including metal-free octreotate bound hSST2 with high affinity (0.2 to 3.9 nM), regardless of the chelator and metal used [24]. This was confirmed by other SST2 binding peptides, coupled to NODAGA, CB-TE2A, or DOTA and labeled with nat Ga or nat Cu resulting in affinities in a range from 1.3 to 12.5 nM [25]. Moreover, modification of   [26]. Hence, compared to CXCR4-binding pentixafor derivatives, where discrepancies varied from 17.6 to 1165 nM (Ga-NOTA and Cu-DOTAGA conjugated to the same highly affine scaffold of CPCR4.2), the effect of chelator and metal exchange was much less pronounced. Profound investigations on bombesin-receptor mediated imaging agents have shown that bombesin analogs can be conjugated and labeled with a broad range of chelators and metals under retention of their receptor affinity [27]. Smith et al. listed 12 bombesin conjugates, labeled with a variety of metal chelation systems, all of them with an unchanged affinity in the range of 0.5 to 10.5 nM [28,29].
Regarding PSMA, 14 99m Tc-based imaging agents and five copper compounds were investigated with various common chelators of 99m Tc and 64 Cu, resulting in high affinity for every compound with K i values ranging from 0.03 to 16.3 nM [30] and 0.19 to 13.26 nM [31] for Tc and Cu compounds, respectively. Free, Ga 3+ -, and Lu 3 + -labeled PSMA DOTA and DOTAGA conjugates were shown to be highly specific as well, ranging from 10.2 to 54.7 nM [32]. This list can be prolonged with 18 F-tracers developed by Pomper et al. [33,34] and other further 68 Ga-tracers developed by Eder et al. [35,36].
Although different groups used different models for affinity evaluation, the relative trend shows that, in contrast to the examples above, the CXCR4 affinity is strongly influenced by the entire ligand-spacer-chelatorradiometal construct. During the development of linkerbridged dimers, Demmer et al. could demonstrate that even dimers, consisting of one high affinity and one "non-" CXCR4-binding peptide exhibit higher affinity when compared with the high affinity monomer conjugated with the used linker [20]. The authors conclude a subsite binding of the second peptide unit close to the main binding pocket. Based on the results of this study, we conclude that binding of the AMB-[M 3+ ]chelator moiety of [ nat Ga]pentixafor and [ nat Ga 3+ ]3 significantly contributes to and is a prerequisite for high affinity binding of the entire peptide ligand. Consequently, depending on the chelator, metalation can have a significant effect on the affinity towards CXCR4.

Conclusions
In summary, these studies demonstrated that pentixafor, consisting of the cyclic peptide cyclo(-D-Tyr-N-Me-D-Orn-L-Arg-L-2-Nal-Gly-) and conjugated at the Orn side chain with AMB-[ nat Ga]DOTA, represents a highly optimized ligand. As a result of this study, two further ligands, a Ga-NOTA ([ nat Ga 3+ ]3) and a Bi-DOTA ([ nat Bi 3+ ]1) derivative with slightly higher affinity to hCXCR4, have been developed. Whereas the Ga 3+ -ligand [ nat Ga 3+ ]3 suffers from a lower hydrophilicity and thus presumably inferior pharmacokinetics compared to [ nat Ga]pentixafor, the Bi 3+ -complex is expected to be a very promising new ligand for further studies towards α-emitter-based endoradiotherapeutic approaches, including multiple myeloma and other lymphoproliferative disorders.