Characterization of 111In-labeled Glucose-Dependent Insulinotropic Polypeptide as a Radiotracer for Neuroendocrine Tumors

Somatostatin receptor targeting is considered the standard nuclear medicine technique for visualization of neuroendocrine tumors (NET). Since not all NETs over-express somatostatin receptors, the search for novel targets, visualizing these NETs, is ongoing. Many NETs, expressing low somatostatin receptor levels, express glucose-dependent insulinotropic polypeptide (GIP) receptors (GIPR). Here, we evaluated the performance of [Lys37(DTPA)]N-acetyl-GIP1-42, a newly synthesized GIP analogue to investigate whether NET imaging via GIPR targeting is feasible. Therefore, [Lys37(DTPA)]N-acetyl-GIP1-42 was radiolabeled with 111In with specific activity up to 1.2 TBq/µmol and both in vitro and in vivo receptor targeting properties were examined. In vitro, [Lys37(111In-DTPA)]N-acetyl-GIP1-42 showed receptor-mediated binding to BHK-GIPR positive cells, NES2Y cells and isolated islets. In vivo, both NES2Y and GIPR-transfected BHK tumors were visualized on SPECT/CT. Furthermore, co-administration of an excess unlabeled GIP1-42 lowered tracer uptake from 0.7 ± 0.2%ID/g to 0.6 ± 0.01%ID/g (p = 0.78) in NES2Y tumors and significantly lowered tracer uptake from 3.3 ± 0.8 to 0.8 ± 0.2%ID/g (p = 0.0001) in GIPR-transfected BHK tumors. In conclusion, [Lys37(111In-DTPA)]N-acetyl-GIP1-42 shows receptor-mediated binding in various models. Furthermore, both GIPR-transfected BHK tumors and NES2Y tumors were visible on SPECT/CT using this tracer. Therefore, [Lys37(111In-DTPA)]N-acetyl-GIP1-42 SPECT seems promising for visualization of somatostatin receptor negative NETs.

Since insulinomas are NETs derived from pancreatic beta cells, incretin receptors are expected to be ideal candidate receptors for insulinoma imaging. Indeed, GLP-1R targeting tracers, such as [ 68 Ga]-or [ 111 In]-labeled exendin, can be applied for non-invasive in vivo insulinoma detection [22][23][24] . However, since malignant insulinomas display differential GLP-1R and somatostatin receptor expression patterns 25 , detection rates of these tumors remain limited to 50% by scintigraphy. Interestingly, GLP-1R negative malignant insulinomas and a majority of somatostatin negative NETs express enhanced GIPR levels 16 , rendering this receptor an interesting target for NET and insulinoma imaging.
Since it was described that GIP 1-30 exhibits reduced receptor binding affinity when compared to the full length peptide GIP 1-42 26,27 , we have investigated the potential of a newly synthesized GIP  analogue, [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP   (Fig. 1) as a radiotracer for NET imaging, starting from the initial hypothesis that a full-length peptide-based tracer might show improved characteristics for in vivo NET imaging when compared to GIP  . Therefore, we optimized the radiolabeling procedure and investigated its binding and internalization kinetics using GIPR-positive tumor cells (BHK-GIPR). Furthermore, we have also explored the tracers binding characteristics to NES2Y cells (a human beta cell-derived cell line, representing a more realistic model in terms of receptor expression) 28 and isolated islets of Langerhans. Finally, subcutaneous BHK-GIPR and NES2Y tumors were visualized by SPECT after injection of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP  .

Results
Radiolabeling and Serum Stability. [Lys 37 (DTPA)]N-acetyl-GIP 1-42 could be labeled with 111 In with a specific activity (now referred to as molar activity 29 ) of up to 1.2 TBq/µmol. Radiochemical purity exceeded 95% as determined by RP-HPLC and ITLC, resulting in a final molar activity exceeding 142.5 MBq/µg or 712.5MBq/ nmol when starting with 150 MBq [ 111 In]Cl 3 . Figure 2a shows the HPLC analysis of the labeling mixture. 111 In-EDTA eluted with a retention time of 3 minutes, whereas 111 In labeled [Lys 37 (DTPA)]N-acetyl-GIP 1-42 had a retention time of 14 minutes. After 12 minutes, a very small impurity (<2%) eluted from the column. Since GIP is known to be prone to inactivation by dipeptidyl peptidase IV (DPP IV), the stability of [Lys 37 ( 111 In-DTPA)] N-acetyl-GIP 1-42 was analyzed in human serum. The results of this stability analysis are shown in Fig. 2b and c. Up to 4 hours of incubation in human serum, the [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 remained intact. After 24 hours of incubation in human serum, 73% of the activity was still found as intact radiolabeled peptide, as determined by HPLC.
In vitro binding and internalization kinetics. The results of the apparent IC 50 determination are shown in Fig. 3a. The apparent IC 50 of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 is 4.8 µM (95% confidence interval: 0.7-32.8 µM). However, the low internalization rate described below, render the observed value more likely to be a true IC 50 rather than an apparent IC 50 . Figure 3b summarizes the binding and internalization kinetics of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 by BHK-GIPR transfected cells and NES2Y cells as determined in vitro. Both cell lines displayed similar binding and internalization characteristics. After 1 hour, 0.6 ± 0.1% and 0.6 ± 0.02% of the added [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 was bound to BHK-GIPR transfected cells and NES2Y cells, respectively. At this time point, the internalized fractions were 1.1 ± 0.1% and 1.5 ± 0.3%, respectively. After 24 hours, the binding values had increased to 13.8 ± 1.7% and 8.6 ± 1.5%, respectively and the internalization values to 4.2 ± 0.6% and 4.7 ± 3.6% for BHK-GIPR and NES2Y cells, respectively. All values were corrected for non-specific binding, as determined by the addition of 25 µg unlabeled GIP 1-42 which significantly reduced the binding and internalization at all time points indicating GIPR-mediated binding in both cell lines (p < 0.05).  N-acetyl-GIP 1-42 dose had a pronounced effect on tumor uptake, which decreased significantly at peptide doses exceeding 0.2 µg ( Fig. 5a and b). Highest tumor uptake of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 was observed at peptide doses ≤ 0.2 µg (0.04 nmol) per mouse: 5.2 ± 1.4%ID/g at 4 hours post-injection. Administration of 0.5 µg (0.1 nmol) as a peptide dose resulted in a significantly lower tumor uptake: 2.4 ± 0.7%ID/g (p = 0.008) (Fig. 5b). Four hours post-injection, [Lys 37 ( 111 In-DTPA)]N-acetyl)GIP 1-42 also showed uptake in various organs such as the lung, spleen, pancreas, liver, stomach and duodenum. The high uptake in the kidneys has been described as being the result of tubular reabsorption through the scavenger receptors cubilin and megalin, which is the case for many peptides 30 . As the kidney uptake of radiolabeled Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 cannot be inhibited by an excess of unlabeled peptide, tubular reabsorption indeed appears to be the cause of the high kidney uptake. Raw data of this biodistribution study are shown in supplementary Table 1. Figure 6a summarizes the biodistribution of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP  in BALB/c nude mice bearing a subcutaneous BHK-GIPR transfected tumor. One hour after injection, highest tumor uptake was observed: 4.7 ± 0.8%ID/g. After 4 and 24 hours, tumor uptake decreased to 3.3 ± 0.8%ID/g and 2.0 ± 0.2%ID/g, respectively (Fig. 6a). Co-administration of an excess unlabeled GIP 1-42 significantly lowered tumor uptake to 0.8 ± 0.2%ID/g (p = 0.0001) 4 hours after injection, indicating GIPR-mediated tumor uptake. One hour after injection, tumor uptake in NES2Y tumors was 0.7 ± 0.2%ID/g (Fig. 6b). Co-administration of an excess unlabeled GIP 1-42 lowered tumor uptake to 0.6 ± 0.01%ID/g (p = 0.78) Raw data of these biodistribution studies are shown in supplementary Tables 2 and 3. SPECT/CT imaging. One and four hours after injection of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 , BHK-GIPR transfected tumors were clearly visible on SPECT/CT ( Fig. 7a and b). Co-injection of an excess unlabeled GIP  completely blocked tracer uptake in the BHK-GIPR transfected tumor (Fig. 7c). Also in mice bearing NES2Y tumors, clear tumor uptake was observed 1 hour after tracer injection ( Fig. 8a and b). Furthermore, co-injection of an excess unlabeled GIP 1-42 prevented tracer uptake in NES2Y tumors, suggesting GIPR-mediated tracer uptake in these tumors.

Discussion
In this study, we investigated the potential of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 for NET detection. The tracer showed GIPR-mediated binding to BHK-GIPR positive cells and NES2Y cells, and to isolated islets in vitro. Optimal GIPR-mediated tumor targeting of BHK-GIPR positive tumors in vivo, was observed 1 hour post injection using a peptide dose of 0.2 µg (0.04 nmol). Furthermore, both BHK-GIPR transfected tumors and NES2Y tumors were visualized by SPECT.
It has been described that C-terminally truncated GIP analogues, such as GIP 1-30 , exhibit reduced receptor binding affinity when compared to intact GIP 1-42 , while retaining their full insulinotropic effect 26,27 . Since the goal of this study was to apply the analogue for in vivo imaging, high receptor affinity should be preserved rather than biological activity. Therefore, we selected GIP 1-42 as a basis for our radiotracer. Another point of consideration while designing a GIPR targeting radiotracer, is the susceptibility of the native peptide to DPP IV degradation 31 .
Since it was shown that several N-terminal modifications showed increased half-life in vitro 32,33 , we acetylated the N-terminal tyrosine residue to obtain a better DPP IV degradation-resistant tracer. Indeed, we observed that the tracer remained stable in serum for at least 4 hours after incubation and even after 24 hours of incubation 73% of the tracer remained intact.
In order to obtain high quality images, high target-to-background ratios are required. Tracer metabolites labeled with a radiometal via a chelator, can be trapped in the lysosomes upon tracer internalization and degradation, resulting in enhanced tracer accumulation in the target cells 23,34 . Therefore, we investigated tracer internalization rates in vitro. Internalization kinetics of the tracer were comparable and relatively slow: after 24 hours, both BHK-GIPR positive and NES2Y cell showed similar internalization rates of 4.2 ± 0.6%ID/g and 4.7 ± 3.6%ID/g,  respectively. Interestingly, the majority of the tracer was bound to the cell surface in both cell lines. These observations could be explained by very recent findings of Ismail et al. indicating that N-terminal acetylation of the peptide hampers receptor internalization 35 .
In BHK-GIPR positive tumors, GIP 1-42 doses lower than 0.5 µg (0.1 nmol) resulted in maximal tumor accumulation. This indicates that administration of higher peptide doses results in partial saturation of the GIPR in vivo. However, since the tracer could be labeled with a specific activity up to 1.2 TBq/µmol, administration of an activity dose for imaging dose is still feasible. For in vivo SPECT imaging, high tumor-to-blood ratios are required.  However, since tracer uptake in the blood was still very low at a peptide dose of 0.2 µg, we selected this dose as the optimal dose for in vivo imaging since it resulted in the highest absolute tumor uptake.
Although [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 showed high and specific tumor uptake in GIPR-transfected GIPR tumors, tumor uptake in NES2Y tumors was lower and co-injection of an excess unlabeled GIP 1-42 did not result in significantly lower tracer uptake. This clear difference in tracer uptake might be explained by the completely different origin and purpose of both cell lines. While the BHK-GIPR cell line was created to express artificially high receptor levels, the NES2Y cells are derived from patients with congenital hyperinsulinism of infancy (28). Since these cells do not overexpress the target receptor and it is known that the natural expression levels of the GIPR are much lower compared to its brother incretin peptide GLP-1 36 , the lower uptake is to be expected. As for the blocking study, the small size of the blocking group (n = 2) impairs generation of representative results. Therefore, the obtained significance level (p = 0.78) might be distorted. On the SPECT images however, tracer accumulation was observed in NES2Y tumors but was no longer observed after co-injection of an excess unlabeled GIP   (Fig. 8b), suggesting specific tumor targeting. Nevertheless, several tracer characteristics might be optimized to further improve tumor targeting. Firstly, other N-terminal modifications than acetylation, such as glucitol or pyroglutamil insertion, have been shown to improve the half-life of the peptide in the presence of DPP IV 32,33 and might also result in higher internalization rates. Secondly, receptor binding and internalization might increase with longer circulation times. As shown in our biodistribution studies, [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 is cleared from the blood very rapidly, hampering longer tracer circulation and thus higher tumor uptake. Several methods such as PEGylation, multimerization or introduction of free fatty acid tails might increase circulation time 37 and thereby tracer accumulation in the tumor. However, a good balance between tracer retention in blood and tracer targeting dynamics has to be found in order to obtain favorable target-to-background ratios for in vivo imaging.
[Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 was produced at a high molar activity and showed receptor mediated binding to various GIPR expressing cells in vitro. Subcutaneous BHK-GIPR positive tumors showed GIPR-mediated tracer accumulation with optimal in vivo targeting 1 hour post-injection, at a peptide dose of 0.2 µg (0.04 nmol). Furthermore, although very low, uptake of a GIP-based tracer was demonstrated for the first time in a human beta-cell derived tumor. Unfortunately, specificity of tracer binding could not be shown due to the small size of the blocking group. In conclusion, [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 shows promising results as a radiotracer for NET imaging. However, the initial working hypothesis that a full-length peptide-based tracer might show improved characteristics for in vivo NET imaging when compared to GIP 1-30 could not be confirmed.

Radiolabeling.
[Lys 37 (DTPA)]N-acetyl-GIP 1-42 (MW = 5422 g/mol) and GIP 1-42 (MW = 5046.7 g/mol) were purchased from Peptide Specialty Laboratories (PSL GmbH, Heidelberg, Germany). DTPA was conjugated to the ε-amino group of the Lysine (K37) and the N-terminal tyrosine was acetylated to reduce its susceptibility to  labeling mixture was incubated in human serum (1:10) at 37 °C. After 1, 2, 4 and 24 hours, a sample was taken and serum proteins were precipitated by adding acetonitrile. This mixture was centrifuged and after dilution, the radiochemical purity of the compound in the supernatant was analyzed using ITLC and RP-HPLC as described above.
Cell Culture. Baby hamster kidney cells, transfected with the human GIP (BHK-GIPR cells) were maintained in DMEM Glutamax (cat. Nr. 6195, GIBCO, BRL Life Sciences Technologies, Bleiswijk, The Netherlands) supplemented with 10% fetal bovine serum (FCS), (HyClone, Celbio, Logan, UT, USA) and 1 mg/ml G418, in a humidified atmosphere containing 5% CO 2 at 37 °C. The human beta cell line NES2Y was cultured as described previously (28). Briefly, cells were maintained in RPMI 1640 (R0883, GIBCO) supplemented with 10% FCS, (HyClone, Celbio), 2 mM L-glutamine (Sigma Aldrich), 100 U/ml penicillin (Sigma Aldrich) and 100 µg/ml streptomycin (Sigma Aldrich) (complete RPMI), in a humidified atmosphere containing 5% CO 2 at 37 °C. To determine the surface-bound fraction, ice-cold acid buffer (0.1 M acetic acid, 154 mM NaCl, pH 2.5) was added and cells were incubated for 10 minutes at 4 °C. After washing the cells twice with binding buffer to makes sure that all surface-bound tracer was removed from the cells, the cell-associated activity was measured in a well type gamma counter. After removal of the surface-bound activity, the internalized fraction was represented by the remaining cell-associated activity. The internalized fraction (remaining cell-associated activity after acid wash) and the receptor bound fraction (activity removed by acid wash) were determined in a well-type gamma counter.

Animals.
All animal experiments were approved by the Animal Welfare Body of the Radboud University, Nijmegen, The Netherlands and carried out in accordance with their guidelines. Six to eight weeks old, female BALB/c nude mice and C3H mice were purchased from Janvier labs (Le Genest Saint Isle, France). For all in vivo experiments, 6-8 weeks old female BALB/c nude mice were injected subcutaneously in the right shoulder with 0.2 ml of a 2.5 × 10 7 cells/ml suspension of BHK-GIPR transfected cells or with 0.2 ml of a 2.5 × 10 7 cells/ml suspension of NES2Y cells mixed with matrigel (2:1) in the right shoulder. When the tumor reached a diameter of 2-5 mm, mice were randomly divided in groups.
Islet Isolation. Islets of Langerhans were isolated from donor C3H mice by collagenase digestion as follows.
Mice were euthanized by CO 2 /O 2 suffocation and 2 ml of ice cold RPMI 1640 containing 1 mg/ml collagenase type V (Sigma Aldrich) were infused via the pancreatic duct in situ. After dissection, the perfused pancreata were collected in serum free medium containing collagenase and kept on ice until digestion. Pancreata were digested for 12 min at 37 °C. Digestion was stopped by adding RPMI medium containing 10% FCS, 2 mM L-glutamine, 100 U/ml penicillin and 100 U/ml streptomycin (complete RPMI). After washing, digested pancreata were passed through a mesh. Afterwards, islets were purified on a discontinuous Ficoll gradient (densities: 1.108; 1.096; 1.037; Cellgro by Mediatech Inc, Manassas, VA, USA) by centrifugation at 625 × g for 16 min without brake. Islets were collected from the intersection of the second and third layer and remaining Ficoll was removed by washing with complete RPMI. Isolated islets were cultured overnight in complete RPMI medium in a humidified atmosphere containing 5% CO 2 at 37 °C.
In Vitro Binding to Isolated Islets. After overnight recovery from the isolation procedure, islets from the donor C3H mice were collected, counted and resuspended in complete RPMI. After transwell saturation using binding buffer, islets were transferred to 24 well transwell plates (200 islets per transwell) (Corning Inc. Tewksbury, MA, USA) and washed with binding buffer. Subsequently, approximately 8 kBq [Lys 37 ( 111 In-DTPA)] N-acetyl-GIP 1-42 (13.5 fmol) was added followed by incubation for 24 hours at 37 °C. To investigate the GIPR-mediated binding, 25 µg of unlabeled GIP 1-42 was added together with [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 in a separate set of wells. After incubation, islets were washed extensively with binding buffer and islet-associated radioactivity was measured in a well type gamma counter.
Peptide Dose Escalation Study. To  Biodistribution Studies. BALB/c nude mice bearing BHK-GIPR transfected tumors were injected intravenously with approximately 3 MBq (peptide dose 0.2 µg = 0.04 nmol) of [Lys 37 ( 111 In-DTPA)]N-acetyl-GIP 1-42 via the tail vein. At various time points after injection (1, 4 and 24 hours), mice were euthanized by CO 2 /O 2 suffocation and blood, tumor and other relevant tissues were dissected, weighed and measured in a well type gamma counter. The percentage injected dose per gram (%ID/g) was calculated for each tissue. To determine whether the GIP 1-42 uptake was GIPR mediated, 25 µg of unlabeled GIP 1-42 was co-injected in a separate group (n = 5) and mice were euthanized four hours post-injection. BALB/c nude mice bearing NES2Y tumors (n = 5) were injected intravenously with approximately 3 MBq (peptide dose 0.2 µg = 0.04 nmol) of [Lys 37 ( 111 In-DTPA)] N-acetyl-GIP 1-42 via the tail vein. One hour after injection, mice were euthanized by CO 2 /O 2 suffocation and blood, tumor and other relevant tissues were dissected, weighed and measured in a well type gamma counter. The percentage injected dose per gram (%ID/g) was determined for each tissue. To determine whether the GIP 1-42 uptake was GIPR mediated, 25 µg of unlabeled GIP 1-42 was co-injected in a separate group (n = 2) and mice were euthanized one hours post-injection.
SPECT/CT. One (BHK-GIPR tumors and NES2Y tumors) or four (BHK-GIPR tumors)hours (n = 2/group) prior to SPECT imaging, BALB/c nude mice were injected intravenously in the tail vein with 20.6 ± 0. Statistics. All values are expressed as mean ± standard deviation (SD). Statistical analysis was performed using unpaired two-tailed t-test using GraphPad Prism v5.03 (GraphPad Softwarem Inc., San Diego, CA, USA). The level of significance was set at p < 0.05.

Ethical approval.
All animal experiments were approved by the Animal Welfare Body of the Radboud University, Nijmegen, The Netherlands and carried out in accordance with their guidelines.
Data availability statement. all data are available at the department of Nuclear medicine of the Radboud University medical center, Nijmegen.