A new 68Ga-labeled somatostatin analog containing two iodo-amino acids for dual somatostatin receptor subtype 2 and 5 targeting

Background Somatostatin receptor (SST) targeting, specifically of the subtype 2 (SST2), with radiolabeled somatostatin analogs, is established for imaging and treatment of neuroendocrine tumors. Owing to the concomitant and heterogeneous expression of several subtypes on the same tumor, analogs targeting more subtypes than SST2 potentially target a broader spectrum of tumors and/or increase the uptake of a given tumor. The analog ST8950 ((4-amino-3-iodo)-d-Phe-c[Cys-(3-iodo)-Tyr-d-Trp-Lys-Val-Cys]-Thr-NH2), bearing 2 iodo-amino acids, exhibits sub-nanomolar affinity to SST2 and SST5. We report herein the development and preclinical evaluation of DOTA-ST8950 labeled with 68Ga, for imaging SST2- and SST5-expressing tumors. Comparative in vitro and in vivo studies were performed with the de-iodinated DOTA-ST8951 ((4-amino)-d-Phe-c[Cys-Tyr-d-Trp-Lys-Val-Cys]-Thr-NH2) and with the reference compounds DOTA-TATE (SST2 selective) and DOTA-NOC (for SST2 and SST5). Results Compared with natGa-DOTA-NOC, natGa-DOTA-ST8950 exhibited higher affinity to SST2 and SST5 (IC50 (95%CI), nM = 0.32 (0.20–0.50) and 1.9 (1.1–3.1) vs 0.70 (0.50-0.96) and 3.4 (1.8-6.2), respectively), while natGa-DOTA-ST8951 lost affinity for both subtypes. natGa-DOTA-ST8950 had the same potency for inducing SST2-mediated cAMP accumulation as natGa-DOTA-TATE and slightly better than natGa-DOTA-NOC (EC50, nM = 0.46 (0.23–0.92) vs 0.47 (0.15–1.5) vs 0.59 (0.18–1.9), respectively). [67Ga]Ga-DOTA-ST8950 had a similar internalization rate as [67Ga]Ga-DOTA-NOC in SST2-expressing cells (12.4 ± 1.6% vs 16.6 ± 2.2%, at 4 h, p = 0.0586). In vivo, [68Ga]Ga-DOTA-ST8950 showed high and specific accumulation in SST2- and SST5-expressing tumors, comparable with [68Ga]Ga-DOTA-NOC (26 ± 8 vs 30 ± 8 %IA/g, p = 0.4630 for SST2 and 15 ± 6 vs 12 ± 5 %IA/g, p = 0.3282, for SST5, 1 h p.i.) and accumulation in the SST-positive tissues, the kidneys and the liver. PET/CT images of [68Ga]Ga-DOTA-ST8950, performed in a dual HEK-SST2 and HEK-SST5 tumor xenografted model, clearly visualized both tumors and illustrated high tumor-to-background contrast. Conclusions [68Ga]Ga-DOTA-ST8950 reveals its potential for PET imaging SST2- and SST5-expressing tumors. It compares favorably with the clinically used [68Ga]Ga-DOTA-NOC in terms of tumor uptake; however, its uptake in the liver remains a challenge for clinical translation. In addition, this study reveals the essential role of the iodo-substitutions in positions 1 and 3 of [68Ga]Ga-DOTA-ST8950 for maintaining affinity to SST2 and SST5, as the de-iodinated [68Ga]Ga-DOTA-ST8951 lost affinity for both receptor subtypes.

We are interested in developing somatostatin analogs with high affinity to SST2 and SST5 for targeting a broader spectrum of tumors and/or increasing the tumor uptake, when both receptor subtypes are concomitantly present. A library of disulfide-bridged octapeptides based on RC-121 (D-Phe-c(Cys-Tyr-D-Trp-Lys-Val-Cys)-Thr-NH 2 ) [23] that contains synthetic amino acids and modifications at positions 1, 3, and 8 was developed by Moore et al [24]. Out of this library, ST8950 (Fig. 1, peptide #9 in reference [24]) bearing the 2 iodoamino acids 4-amino-3-iodo-phenylalanin in position 1 and 3-iodo-tyrosine in position 3 exhibited subnanomolar affinity to both SST2 and SST5 and showed to be as potent as the natural SS-14 in the inhibition of growth hormone and prolactin release. We previously reported that ST8950 (AP102 in references [25,26]) has an intermediate agonistic potency between octreotide and pasireotide at SST2 and SST5 level [25] and reduces growth hormone secretion without causing hyperglycemia (a known undesirable effect of pasireotide) in a healthy rat model [26]. We aimed to develop and evaluate 68 Ga-labeled ST8950 for PET imaging of SST2-and SST5-expressing tumors and we used DOTA as a chelator. Knowing that modifications such as chelator conjugation and (radio)metallation impact on affinity and whole-body distribution of radiolabeled somatostatin analogs [10,27], with sometimes unexpected outcome, we decided to include in our study a second analog as an alternative. We chose the de-iodinated ST8951 (Fig. 1, peptide #2 in reference [24]) that exhibits also good affinity to SST2 and SST5 in an attempt to assess, additionally, the influence of the iodo-substitution on the Ga-DOTA conjugates.

Methods
(Radio)metallated peptide conjugates DOTA-ST8950 and DOTA-ST8951 were custom-made by PolyPeptide (San Diego, USA). The reference conjugates DOTA-TATE and DOTA-NOC were synthesized following Fmoc-solid-phase peptide synthesis, purified by preparative reverse phase high-performance liquid chromatography (RP-HPLC) and characterized by electrospray ionization mass spectroscopy (ESI-MS). The structural formulae of all four conjugates are presented in Fig. 1.
The nat Ga complexes of the four conjugates were prepared using 2.5-fold excess of nat Ga(NO 3 ) 3 × 9H 2 O in ammonium acetate buffer, 0.2 M, pH 4 at 95°C for 30 min. Free metal ions were eliminated via SepPak C-18 cartridge (Waters), pre-conditioned with methanol and water. The reaction mixture was loaded and the free nat Ga was eluted with water while the metallo-peptides were eluted with ethanol, evaporated to dryness, redissolved in water and lyophilized. 67 Ga-labeled conjugates were prepared by reacting 6 nmol of the corresponding conjugate in 250 μL Na-acetate buffer (0.2 M, pH 4.1) with [ 67 Ga]GaCl 3 (30-50 MBq, Mallinckrodt) at 95°C for 30 min. DOTA-ST8950 and DOTA-NOC were labeled with 68 Ga in an automatic Modular-Lab Pharm Tracer module (Eckert & Ziegler). Briefly, the 68 Ge/ 68 Ga-generator (IGG100, Eckert & Ziegler) was eluted with 7 mL HCl 0.1 N and the eluate (8 00 MBq) was loaded onto a cation exchange column (Strata-XC, Phenomenex). 68 Ga 3+ was eluted with 800 μL of acetone/HCl (97.6%/0.02 N) in a vial containing 2 mL Na-acetate buffer (0.2 M, pH 4.0) and 10 nmol of the conjugate. The stability of [ 68 Ga]Ga-DOTA-ST8950 was evaluated for 1 h after production at room temperature (RT), without any formulation of the product.
The quality control and the stability study were performed by analytical RP-HPLC on Phenomenex Jupiter Proteo 90 Å C12 (250 × 4.  Unit, Lausanne University Hospital, Switzerland) were stably transfected with plasmids encoding the human SST2 and SST5 (HEK-SST2 and HEK-SST5) and cultivated as previously described [25]. Nontransfected HEK cells were used as a negative control.
The binding affinities of nat Ga-DOTA-ST8950 and nat Ga-DOTA-ST8951, in comparison to nat Ga-DOTA-TATE and nat Ga-DOTA-NOC, were determined on HEK-SST2 and HEK-SST5 cells. SS-14, octreotide, and pasireotide were used as reference compounds. 125 I-labeled SS-14 (81.4 TBq/mmol, Perkin Elmer) was used as a radioligand for the competition binding assays. Binding assays were performed as described previously [25].

Log D measurement
Log D (pH = 7.4) was determined by the "shake-flask" method. To a pre-saturated mixture of 500 μL n-octanol and 500 μL of phosphate-buffered saline (PBS) at pH 7.4, 10 μL of 1 μM of 67 Ga-labeled conjugates was added. The solutions were vortexed for 1 h to reach equilibrium and then centrifuged (3000 rpm) for 10 min. From each phase, 100 μL was removed and measured in a γcounter. The partition coefficient was calculated as the average of the logarithmic values (n = 3) of the ratio between the radioactivity in the organic and the PBS phase.

In vitro characterization
For cell experiments, stably SST2-and SST5-expressing cells were seeded in 6-well plates (10 6 cells/well) and incubated overnight with Dulbecco's modified Eagle's medium (DMEM) with 1% fetal bovine serum (FBS, Biochrom GmbH, Merck Millipore) to obtain a good cell adherence. The plates were pre-treated with a solution of 10% poly-lysine to promote cell adherence.

Internalization assays
The cells were washed with PBS and incubated with fresh medium (DMEM with 1% FBS) for 1 h at 37°C/5% CO 2 .
[ 67 Ga]Ga-DOTA-ST8950, [ 67 Ga]Ga-DOTA-ST8951, [ 67 Ga]Ga-DOTA-TATE, or [ 67 Ga]Ga-DOTA-NOC (2.5 nM) were added to the medium, and the cells were incubated for 0.5, 1, 2, and 4 h at 37°C/5% CO 2 (in triplicates). The internalization process was stopped by removing the medium and washing the cells with icecold PBS, followed by 2 × 5 min treatment with ice-cold glycine solution (0.05 M, pH 2.8), to distinguish between cell surface-bound (acid releasable) and internalized (acid resistant) radio-conjugate. Finally, the cells were detached with 1 M NaOH at 37°C. To determine nonspecific uptake, selected wells were incubated with the radio-conjugate in the presence of 1000-fold excess of SS-14. Internalization and bound rate are expressed as a percentage of the applied radioactivity.

Off-rate experiments
HEK-SST2 cells were incubated with [ 67 Ga]Ga-DOTA-ST8950, [ 67 Ga]Ga-DOTA-TATE, or [ 67 Ga]Ga-DOTA-NOC (2.5 nM) for 2 h. The medium was removed and the wells were washed with ice-cold PBS. The surfacebound radio-conjugate was removed with a glycine solution (pH 2.8) on ice, as described above. Cells were then incubated again at 37°C with a fresh medium. At 10, 20, 30, 60, 120, and 240 min, the medium was removed for quantification of radioactivity and replaced with a fresh 37°C medium. At the end of the experiment, the cells were detached with 1 M NaOH and collected for quantification of the radioactivity.

In vivo evaluation
The Veterinary Office (Department of Health) of the Cantonal Basel-Stadt approved the animal experiments (approval no. 2799) in accordance with the Swiss regulations for animal treatment. Female athymic Nude-Foxn1 nu /Foxn1 + mice (Envigo, The Netherlands), 4-6 weeks old, were inoculated subcutaneously with 10 7 HEK-SST2 cells on the right shoulder and 10 7 HEK-SST5 cells on the left shoulder, suspended in 100 μL sterile PBS. The tumors were allowed to grow for 2-3 weeks until reach an average volume of 100 mm 3 , considering both tumor types. The average tumor mass was 0.22 g (0.14-0.31 g) for SST2 tumors and 0.13 g (range 0.08-0.21) for SST5 tumors. For the biodistribution (cohorts of n = 3-8 mice) and imaging studies, the mice were euthanized by keeping them in a CO 2 chamber 2 min, followed by a slow increase of the concentration of CO 2 gas. The mice with the largest tumors were used for PET imaging. (100 μL/100 pmol/5 MBq) was administered to mice bearing dual HEK-SST2 and HEK-SST5 tumors. Onehour p.i. the mice were euthanized, and the bladder was emptied by gently pressing with hands around the bladder area. The excess urine was soaked by cotton, followed by repetitive cleaning of the area with ethanol. The mice were scanned for 60 min using a human PET/CT scanner (Discovery STE, GE Medical Systems). A scout scan (180°, 10 mA, 120 kV) was performed to establish a protocol for all other scans. CT scans were acquired with a minimum slice thickness of 0.625 mm, pitch 1.375:1) and the highest possible tube current for these settings (320 mA @ 120 keV). PET emission events were collected in 3D scanning mode (septa out) over 60 min. Images were corrected for the decay of 68 Ga and random events and reconstructed using the manufacturer's 3D OSEM algorithm to 47 slices (display FOV = 6.4 cm, 128 × 128 matrix, resulting pixel size = 0.5 mm), once for each mouse separately in the center of the reconstruction cylinder. The in vivo images are presented as fused images of PET maximum intensity projection (MIP) and CT.

Data analysis
Statistical analysis was performed by unpaired two-tailed t test using GraphPad Prism 7 software (GraphPad Software Inc.). P values of < 0.05 were considered significant.

Discussion
Various expression and co-expression patterns have been described for the 5 somatostatin receptor subtypes (SST1-5), depending on the tumor type and origin [5,6,28]. Interestingly, tumor area lacking expression of a given subtype may be populated by another one [4,5,8].
Hence, somatostatin analogs with affinity to more than one receptor subtypes are of great interest as they address receptor subtype co-expression and heterogeneous expression patterns. Two independent studies comparing the diagnostic efficacy [ 68 Ga]Ga-DOTA-NOC, a somatostatin analog with a high affinity for SST2 and SST5 and a lower affinity for SST3, with the SST2-selective [ 68 Ga]Ga-DOTA-TATE in NET patients, provided controversial results on the clinical outcome of multi-receptor subtype targeting. Kabasakal et al. [29], concluded that even though the images have comparable diagnostic accuracy, [ 68 Ga]Ga-DOTA-TATE detected more lesions. Contrary to this, Wild et al. [30] reported that [ 68 Ga]Ga-DOTA-NOC detected significantly more lesions than [ 68 Ga]Ga-DOTA-TATE (sensitivity: 93.5 vs 85.5%) and it changed the clinical management in 17% of the studied patients. Recently, Lamarca et al. [31] confirmed the role of [ 68 Ga]Ga-DOTA-NOC PET imaging for the optimization of the clinical management in lung carcinoid patients. Overall, the clinical data support that multi-receptor subtype targeting is relevant for improving the diagnostic accuracy and sensitivity of PET imaging of SST-expressing tumors. Therefore, effort needs to be made for developing new radiotracers in this direction.
With the aim to develop 68 Ga-tracers for combined SST2 and SST5 targeting, we focused on the series of compounds reported by Moore et al. [24], who used synthetic amino acids, among them iodo-substituted ones, to improve binding affinities. There are several cases in the literature where (radio)iodination of somatostatin analogs either did not affect or improve the binding affinity and/or potency [8,32,33]. In the series of Moore et al., iodination at position 3 (3-iodo-Tyr 3 , peptide #6 in reference [24]) showed improved affinity to SST5 by an order of magnitude, followed unluckily, by a 4-fold reduction in the affinity to SST2. Similar observations on SST5 were reported by Schotellius et al. [33]. Iodination at position 3 of DOTA-TATE (DOTA-3-iodo-Tyr 3octreotate: HA-TATE) enhanced the affinity of nat Ga-DOTA-HA-TATE to SST5, compared with nat Ga-DOTA-TATE (IC 50 = 102 ± 65 vs > 1000 nM, respectively), but did not affect the affinity to SST2, contrary to Moore et al. Taken together, the two studies indicate that iodo-substitution of Tyr 3 on the octreotide motif improves affinity to SST5.
Modification of ST8950 at the N-terminal by coupling of DOTA and complexation with Ga 3+ does not affect the affinity for SST2, while reduces the affinity for SST5 by a factor of approx. 2. Nevertheless, nat Ga-DOTA-ST8950 retains   [24]) after the coupling of the chelate nat Ga-DOTA losses massively its affinity for both, SST2 and SST5 (IC 50 = 7.5 and 24 nM, respectively). The effect is more prominent in the SST2, for which ST8951 shows higher selectively. In addition, we evaluated to which extend ST8950 and ST8951 retain their agonistic potencies after Ga-DOTA conjugation, in light of examples in the literature indicating that modifications like DOTA conjugation can change the function of a somatostatin analog from an antagonist to an agonist [34]. The agonistic potencies of nat Ga-DOTA-ST8950 and nat Ga-DOTA-ST8951 followed an identical trend as their affinities to both SST2 and for SST5. In vitro experiments on HEK-SST5 did not show any internalization (neither cell surface-bound) for any of the tested radiotracers. This may be due to the particular cellular distribution and trafficking of SST5 [35,36] and not due to the radiotracers. This phenomenon was also observed by others using somatostatin analogs with a high affinity for SST5 [10,35]. Cescato et al. [35] showed that SST5 internalization can be induced by natural somatostatin peptides but not by synthetic high-affinity SST5 agonists. Indeed, Maina et al showed in vitro internalization on HEK-SST5 of an 111 In-labeled modified analog of the natural SS-28 [22], even though very low (approx. 2.5% after 1 h at 37°C) and unusually high (50%) nonspecific portion. Our data are in line with the published findings. Importantly, the lack of in vitro internalization in SST5-expressing cells does not exclude the accumulation of the radiotracer in SST5-expressing tumors in vivo. Our in vivo data prove this.
The in vivo distribution of [ 68 Ga]Ga-DOTA-ST8950 is representative of radiolabeled somatostatin analogs, regarding the accumulation in SST-positive tissues, such as the stomach, the pancreas, and the pituitary.  DOTA-NOC (p = 0.4630 for SST2 and p = 0.3282 for SST5), proven to be receptor subtype mediated by the 90% reduction found on the SST-negative tumor. The kidneys were the second tissue after the tumors accumulating radioactivity, which was expected due to the renal excretion of this class of radiotracers. Unluckily, the lipophilic character of [ 68 Ga]Ga-DOTA-ST8950 was reflected on its biodistribution profile, with rather high blood and liver values. Nevertheless, when we compared with the 68 Ga-labeled pasireotide which also targets SST2 and SST5 (referred as 68 Ga-DOTA-SOM230 in [15,16]) in the same animal model, [ 68 Ga]Ga-DOTA-ST8950 has advantages in terms of lower blood values (1.9 ± 0.6 vs 4.1 ± 0.9 %IA/g at 1 h p.i.) and liver uptake (6.4 ± 1.9 vs 12.9 ± 2.2 %IA/g at 1 h p.i.). Liu et al. confirmed similarly high blood values and even higher liver uptake for the 68 Ga-labeled pasireotide (referred as 68 Ga-DOTA-PA1 in [17] (Table 4, p = 0.0141 and p = 0.0005, respectively) and much higher when compared with our previous data on [ 68 Ga]Ga-DOTA-TATE [37] (blood: 1.9 ± 0.6 vs 0.4 ± 0.0 %IA/g and liver 6.4 ± 1.9 vs 0.4 ± 0.2 %IA/g, respectively). On the other hand, [ 68 Ga]Ga-DOTA-NOC demonstrated higher uptake in the SST-positive stomach and pancreas, while [ 68 Ga]Ga-DOTA-TATE cannot be used for imaging SST5-expressing tumors as it is unable to bind to this receptor subtype.
PET/CT imaging is reflecting the biodistribution data, with clear visualization of SST2-and SST5-expressing tumors and high image contrast for [ 68 Ga]Ga-DOTA-ST8950 and for [ 68 Ga]Ga-DOTA-NOC. The higher kidney uptake of [ 68 Ga]Ga-DOTA-ST8950, compared with [ 68 Ga]Ga-DOTA-NOC (p = 0.0399), is of no concern for a diagnostic tracer. However, the accumulation of [ 68 Ga]Ga-DOTA-ST8950 in the liver is a drawback. Especially when recognizing that the liver is the first site of metastasis of NETs, and therefore, low background activity is needed for a good image contrast and diagnostic accuracy. Two approaches are considered to circumvent this problem: (a) a chemical approach that involves modification of the structure by introducing hydrophilic spacers and/or amino acids and (b) a pharmacological approach by enhancing the tumor uptake via epigenetic receptor upregulation [38], improving tumor-to-liver ratio. The second approach is mainly considered for the therapeutic counterpart [ 177 Lu]Lu-DOTA-ST8950.

Conclusion
The preclinical evaluation of the 2-iodo-substituted somatostatin analog [ 68 Ga]Ga-DOTA-ST8950 reveals its potential as PET tracer for in vivo imaging of SST2-and SST5-expressing tumors, which may be of interest for gastroenteropancreatic neuroendocrine tumors, pituitary tumors, and gastric cancers. Its in vivo uptake in the tumors compares favorably with the uptake of the clinically used [ 68 Ga]Ga-DOTA-NOC, but its high accumulation in the liver remains a challenge for clinical translation. While iodination in positions 1 and 3 seemed not to be a prerequisite for a good binding affinity of ST8950 and of the de-iodinated ST8951 to SST2 and SST5, this is proven to be essential in their [ 68 Ga]Ga-DOTA-chelated versions. [ 68 Ga]Ga-DOTA-ST8951 lost its affinity and potency for both subtypes, and it is disqualified for usage as a PET tracer.