Radioiodinated Exendin-4 Is Superior to the Radiometal-Labelled Glucagon-Like Peptide-1 Receptor Probes Overcoming Their High Kidney Uptake

GLP-1 receptors are ideal targets for preoperative imaging of benign insulinoma and for quantifying the beta cell mass. The existing clinical tracers targeting GLP-1R are all agonists with low specific activity and very high kidney uptake. In order to solve those issues we evaluated GLP-1R agonist Ex-4 and antagonist Ex(9–39) radioiodinated at Tyr40 side by side with [Nle14,Lys40(Ahx-DOTA-68Ga)NH2]Ex-4 (68Ga-Ex-4) used in the clinic. The Kd, Bmax, internalization and binding kinetics of [Nle14,125I-Tyr40-NH2]Ex-4 and [Nle14,125I-Tyr40-NH2]Ex(9–39) were studied in vitro using Ins-1E cells. Biodistribution and imaging studies were performed in nude mice bearing Ins-1E xenografts. In vitro evaluation demonstrated high affinity binding of the [Nle14,125I-Tyr40-NH2]Ex-4 agonist to the Ins-1E cells with fast internalization kinetics reaching a plateau after 30 min. The antagonist [Nle14,125I-Tyr40-NH2]Ex(9–39) did not internalize and had a 4–fold higher Kd value compared to the agonist. In contrast to [Nle14,125I-Tyr40-NH2]Ex(9–39), which showed low and transient tumor uptake, [Nle14,125I-Tyr40-NH2]Ex-4 demonstrated excellent in vivo binding properties with tumor uptake identical to that of 68Ga-Ex-4, but substantially lower kidney uptake resulting in a tumor-to-kidney ratio of 9.7 at 1 h compared to 0.3 with 68Ga-Ex-4. Accumulation of activity in thyroid and stomach for both peptides, which was effectively blocked by irenat, confirms that in vivo deiodination is the mechanism behind the low kidney retention of iodinated peptides. The 124I congener of [Nle14,125I-Tyr40-NH2]Ex-4 demonstrated a similar favourable biodistribution profile in the PET imaging studies in contrast to the typical biodistribution pattern of [Nle14,Lys40(Ahx-DOTA-68Ga)NH2]Ex-4. Our results demonstrate that iodinated Ex-4 is a very promising tracer for imaging of benign insulinomas. It solves the problem of high kidney uptake of the radiometal-labelled tracers by improving the tumor-to-kidney ratio measured for [Nle14,Lys40(Ahx-DOTA-68Ga)NH2]Ex-4 by 32 fold.


Introduction
Glucagon like peptide-1 receptors (GLP-1R) are ideal targets for the imaging of benign insulinoma due to their high density on the surface of more than 90% of these tumors [1].Physiological expression of GLP-1 receptors on the pancreatic beta cells is also being exploited preclinically and potentially clinically for imaging of the pancreatic beta cell mass to quantify the loss of islets in type 1 diabetes [2,3].Since islet transplantation may be a promising treatment option for patients with diabetes, the GLP-1 targeted imaging is being explored for visualization of transplanted islets and monitoring their survival [4,5].
In case of benign insulinoma, preoperative localization plays a crucial role in the successful outcome of the surgery and in sparing the healthy pancreatic tissue.However, due to the small size of insulinoma lesions (often less than 2 cm) the precise localization is very challenging using conventional imaging techniques having a sensitivity of only 47% [6].This prompted an intensive research into developing more sensitive and non-invasive imaging modalities to localize benign insulinoma in the pancreas.A variety of GLP-1R targeting peptide probes labelled with 111 In [7,8], 99m Tc [8], 123 I [9] for SPECT and with 68 Ga [3,8,[10][11][12][13], 64 Cu [3,14], 18 F [5,15,16], and 89 Zr [13] for PET imaging showed promising results in preclinical studies.Some of the tracers have been successfully translated into the clinic [17][18][19][20][21][22] and showed highest sensitivity among the available techniques in detecting ''hidden" insulinomas in patients.
Despite the great clinical potential, several aspects of the existing GLP-1R specific tracers would greatly benefit from further improvements.Firstly, the high and persistent kidney uptake of radiometal-labelled tracers not only leads to an unnecessary high radiation dose to the kidney but also can complicate the preoperative and intraoperative insulinoma detection in the head and tail of the pancreas [17,20,23].Secondly, the specific activity of the existing tracers is relatively low requiring higher peptide mass for adequate activity to achieve a good image quality.The agonist peptide dose is however limited due to the risk of insulin-related side effects.GLP-1 receptor agonists belong to the potent incretin hormones; after binding to the receptor they internalize and activate GLP-1 receptor signalling leading to the glucose-dependent insulin secretion.
For this reason, antagonist probes would be much more attractive because they would not activate the downstream signalling.Moreover, antagonist ligands for other G-protein coupled receptors (GPCRs), including gastrin-releasing peptide receptor and somatostatin receptors have been shown to target more binding sites and as a result to have higher tumor uptake than agonists [24,25].Exendin(9-39) isolated from Heloderma suspectum venom was identified as GLP-1 receptor antagonist binding with high affinity [26].There are only few reports on radiolabeled Ex(9-39)-based tracers showing that Ex(9-39) derivatives radiolabeled with 68 Ga [27] or 18 F [28] are not suitable for imaging due to the low tumor uptake.
Our recent data also demonstrate that in contrast to Ex-4-based tracers, which tolerated various modifications at the C-terminal very well, the Ex(9-39) antagonist was very sensitive to conjugation of the 68 Ga-DOTA or 68 Ga-NODAGA complexes at the positions Lys 27 or Lys 40 [29].The resulting probes had low affinity and were not suitable for GLP-1 R imaging [29].However, Ex(9-39) radioiodinated with 125 I at Lys 27 using the Bolton Hunter (BH) method showed the same binding properties for the human GLP-1 receptor-expressing tissues in vitro as endogenous GLP-1 [30].We demonstrated that in vivo 125 I-BH-Ex(9-39) accumulated in the Ins-1E xenografts with an uptake similar to that of 68 Ga-Ex-4, the GLP-1R tracer currently used in the clinic [22].Most importantly, however radioiodination resulted in a significant reduction of the radioligand's kidney uptake [29].In comparison to 68 Ga-Ex-4, the tumor-to-kidney ratio of 125 I-BH-Ex-(9-39) improved 20-fold.
However, the Bolton-Hunter method is not so trivial to translate into the clinic.In the current study we selected the direct iodination approach, because it is a one-step process and the in vivo deiodination efficiency of 125 I-Tyr is even higher than that for 125 I-BH [31].As a consequence the kidney uptake can be expected to be even lower for the directly iodinated peptides.Since radioiodination of Ex(9-39) with 125 I-BH at position Lys 27 was tolerated better than conjugation of 68 Ga-DOTA or 68 Ga-NODAGA in the same position, we hypothesized that radioiodination at the C-terminal Tyr 40 may also potentially lead to a GLP-1R antagonist tracer with high affinity and favourable pharmacokinetics.In this study, the Ex-4 agonist and Ex(9-39) antagonist radioiodinated at the Tyr 40 residue were evaluated in vitro and in vivo and compared side by side to the reference compound, [Nle 14 ,Lys 40 (Ahx-DOTA-68 Ga)NH 2 ]Ex-4.

Reagents and instruments
All reagents and solvents were purchased from Sigma-Aldrich (Taufkirchen, Germany) and Carl Roth GmbH & Co. KG (Karlsruhe, Germany).The details for the analytical and semi-preparative HPLC set up are described in the S1 File.

In vitro binding assays
Ins-1E cells (a gift from Dr. Gu ¨nter Pa ¨th) were maintained in RPMI medium, containing 10% FBS, 1mM sodium pyruvate, 50 μM ß-mercaptoethanol, 10 mM HEPES (pH 7.2), Penicillin G (100 IA/ml) and Streptomycin (10 mg/ml).One day prior to the experiments, cells (1.8 × 10 6 ) were plated in triplicates in the 6-well plates.For the saturation binding experiment, cells were incubated with 3 kBq of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 and increasing amounts of [Nle 14 , 127 I-Tyr 40 -NH 2 ]Ex-4 peptide (1-100 nM) in the growth medium, containing 1%FBS for 2 hours at +4˚C.To determine the non-specific binding, three additional wells were blocked with 1 μM of [Nle 14 , 127 I-Tyr 40 -NH 2 ]Ex-4.After washing with cold PBS cell bound activity was collected with 1M NaOH (3x1 ml).For internalization experiments, Ins-1E cells were incubated with 5 kBq of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4, containing 1 nM of [Nle 14 , 127 I-Tyr 40 -NH 2 ]Ex-4 at 37˚C.At specific time points the internalization was stopped by washing the cells with ice-cold PBS and the membrane bound and internalized activity were collected as described previously [33].125 I-radioactivity in samples was measured using a gamma counter (Cobra 5003; Packard Instruments).The percentage of specifically bound activity per 1x10 6 cells was assessed by comparison with standards and non-specific controls.
All animal experiments were conducted in accordance with the German animal protection law (TierSchG).The protocol was approved by the Animal Welfare Ethics committees of the University of Freiburg (Regierungspra ¨sidium Freiburg Az G-12/21).Female Balb/c Nude mice (18-20 g, 6-8 weeks old) were obtained from Janvier Labs (Saint-Berthevin Cedex, France) and were housed and handled in accordance with the good animal practice as defined by   were coinjected together with the radioiodinated peptide.For the blocking experiment, 80 nmol of Ex(9-39) was injected 5 minutes before the administration of iodinated peptide.The sodium iodide symporters were blocked by Irenat (Bayer) (120 mg/kg, i.v.).Animals (n = 3-4, per group) were euthanized by asphyxiation with excess isoflurane at 1, 4 and 24 h post-radiotracer administration, and tissues were removed, rinsed in water, dried in air, weighed and counted on a calibrated and normalized gamma-counter.

Small-animal PET/CT imaging
Mice were injected iv with [Nle 14 , 124 I-Tyr 40 -NH 2 ]Ex-4 (2.5-3.0MBq in 100 μL sterile saline).At 1, 2 and 4 hour post injection 20-40 minute static scans were acquired using microPET Focus 120 scanner (Concorde Microsystems), followed by 2 minute CT scans on micro-CT-Tomoscope Synergy (CT Imaging GmbH).PET sinograms were reconstructed using a 2dimensional ordered-subset expectation maximization (2D-OSEM) algorithm.Image counts per pixel per second were calibrated to activity concentrations (Bq/mL) by measuring a 3.5 cm cylinder phantom filled with a known concentration of radioactivity.The PET and CT images were co-registered using the Rover software (ABX, Radeberg, Germany).PET images were analyzed using Rover software (ABX, Radeberg, Germany).The regions of interest were drawn based on the CT to include the entire tumor or the kidney and the mean %IA/g was calculated.For the time activity curves the PET images recorded at 1, 2 and 4 h p.i. were merged together and the mean %IA/g were calculated from the VOI placed in the same position in the tumor.

Statistical analysis
A statistical analysis was performed using GraphPad Prism 5.01 (GraphPad Software, Inc) and Microsoft Excel.Data were analyzed by using the unpaired, two-tailed Student's t-test.
At one hour p.i. of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 1.4±0.1%IA/organwas accumulated in the thyroid, confirming the dehalogenation of the tracer in vivo.However, blocking the sodium iodide symporter with irenat reduced the thyroid uptake by 94% (Table 2).Among the three studied peptide doses of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 (0.5 pmol, 10 pmol and 100 pmol) the lowest peptide mass of 0.5 pmol corresponding to the highest specific activity showed the most favourable biodistribution profile with highest tumor uptake and highest tumor-to-normal organ ratios (Table 2).
Biodistribution of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex  Corresponding to the lower affinity of the antagonist, also the tumor uptake of [Nle 14 , 125 I--Tyr 40 -NH 2 ]Ex(9-39) was lower, 12.7±4.1%IA/gat 1 h p.i. and it decreased after 4 h to 1.9±0.5% IA/g, similar to the activity in the blood pool (Table 3).The uptake in other receptor-expressing organs was also lower (Table 3).Blocking with an excess of cold antagonist reduced the uptake in the tumor, pancreas and lung by 80%, 93% and 88% respectively (Table 3).The uptake in the kidney at 1 h p.i. was also very low (7.6±1.2%IA/g) and identical to the agonist kidney uptake.However, due to the lower tumor uptake, the tumor-to-kidney ratio dropped to 1.7.The tumor-to-blood and tumor-to muscle ratios of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex(9-39) were substantially lower than the ratios for the agonist.The uptake in the thyroid was 1.3±0.2%IA/organat 1 h p.i., indicating tracer deiodination in vivo.In terms of the peptide dose, the lowest dose of 0.5 pmol yielded the highest tumor uptake and best tumor-to-normal organ ratios (Table 3).kidney had the highest accumulation of activity (201.3±30.6%IA/g)followed by the Ins-1E tumor (58.3±15.6%IA/g), the pancreas (25.5±5.2%IA/g), and the lung (14.3±1.2%IA/g).The tumor-to-blood and tumor-to-muscle ratios were 48.5 and 215.9, respectively (Table 4).The tumor-to-kidney ratio was very low, only 0.3.2). a Thyroid uptake is reported as %IA/organ.Reason for this is the complicated delineation of the thyroid, which resulted in excision of variable amounts of adjacent tissue upon dissection and concomitantly incorrect values when uptake is reported as %IA/g.The MIP PET image of the reference GLP-1 receptor imaging agent [Nle 14 ,Lys 40 (Ahx-DOTA-68 Ga)NH 2 ]Ex-4 at 1 h p.i (Fig 5B ) demonstrates a typical biodistribution profile for the radiometal labelled Ex-4 analogue with a high tumor uptake but also very high uptake in the kidney.

Discussion
GLP-1 receptor targeting peptides are important imaging tools for preoperative localization of benign insulinoma.Currently available tracers show a high and persistent renal uptake and low specific activity, resulting in side effects such as hypoglycaemia due to insulin secretion  after activation of the GLP-1 receptor.We attempted to solve these issues by using radioiodinated analogues of the agonist Ex-4 and the antagonist Ex(9-39).
[Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 showed high affinity binding to Ins-1E cells in vitro with a K d of 4.1±1.1 nM.The K d value for [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex(9-39), however was 4.4 fold higher, demonstrating a significantly lower affinity towards GLP-1R.In contrast to the radiometallabelled GLP-1R tracers, the fraction of internalized [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 did not increase with time after reaching the maximum of 3% at 0.5 h.This phenomenon has been previously described for radioiodinated tracers [9] and is related to the intracellular degradation and release of the main catabolite 125 I-Tyr, which is not trapped in the lysosomal compartment as do the radiometal-containing catabolites [34].As a result, some steady state between the amount of internalized radioactivity and the amount of released radioactivity is reached.
[Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex(9-39) behaved in in vitro binding assays as expected from an antagonist.In comparison to the agonist [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4, it showed higher cell membrane binding but a 5.5-fold-lower internalization into Ins-1E cells.Interestingly, the Bmax value for the iodinated antagonist was higher (0.070±0.007 versus 0.045±0.003,p<0.05), potentially indicating that it recognizes more binding sites than the iodinated agonist.This feature of the antagonist tracer has been described for the SST-2 receptors, where antagonists recognized 10-15 fold more binding sites, compared to the agonists [24].However, in spite of the higher number of binding sites, the lower affinity and lower internalization resulted in a lower total cellular accumulation of the GLP-1R antagonist (2.53%) in comparison to the GLP-1R agonist (4.43%).
In vivo, [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 demonstrated high and specific tumor uptake, which was equal to the tumor uptake of 68 Ga-Ex-4.But most importantly the kidney uptake was very low resulting in the high tumor-to-kidney ratio of 9.7.This ratio was even better than the ratio of 3.5, which we reported for the antagonist 125 I-BH-Ex(9-39) [29], and was the highest among the published GLP-1R tracers.The tumor-to-pancreas ratios were also very similar for [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 and 68 Ga-Ex-4, which would translate into similar imaging contrast for insulinomas against the normal pancreas.Somewhat surprisingly, the tumor uptake of the antagonist [Nle 14 , 125 I-Tyr 40-NH 2 ]Ex(9-39) was significantly lower, only 12.7±4.05%IA/gat 1 h p.i. and the activity was completely washed out after 4 h p.i.This may be explained by the higher K d value of the compound.The kidney uptake of the iodinated antagonist was identical to the agonist (7.63±1.21%IA/gand 7.51 ±0.73%IA/g, respectively), but the tumor-to-kidney ratio and other tumor-to-normal organ ratios were much lower.
The mechanism behind the low kidney retention of 125 I-Ex-4 is related to the non-residualizing nature of the 125 I label [34].After reabsorption and lysosomal degradation in the proximal tubular cells of the kidneys, the main catabolite 125 I-Tyr, is not retained in the lysosomes but instead freely diffuses out of the cells, and the iodide released upon deiodination rapidly accumulates in the thyroid tissue and stomach via the sodium iodide symporter.Indeed, the radioactivity in the thyroid and stomach for both tracers, was effectively blocked by inhibiting the sodium iodide symporter with irenat.
As we hypothesized, the kidney uptake of the directly iodinated [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 and [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex(9-39) (7.5%IA/g) at 1 h p.i. was even lower than that of the Bolton-Hunter-labelled analogue 125 I-BH-Ex(9-39) (12.1±1.4%IA/g)[29].Higher accumulation of radioactivity in the stomach and the thyroid after injection of [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex-4 and [Nle 14 , 125 I-Tyr 40 -NH 2 ]Ex(9-39) in comparison to 125 I-BH-Ex(9-39) [29] confirms more efficient dehalogenation for the directly iodinated peptides.The relatively fast washout of activity from the tumor within 4 hours for the directly iodinated agonist can also be explained by in vivo dehalogenation.Since fast washout of activity from normal organs will result in a lower radiation exposure, the 124 I congener of the directly iodinated Ex-4 agonist may prove a valuable candidate for clinical PET imaging, provided that an early time-point is chosen for imaging.However, relatively fast reduction in the tumor uptake of iodinated Ex-4 would limit the time window for intraoperative localization of insulinoma using a surgical probe, compared to the In-111-labeled Ex-4 analogs used in clinic.Based on the tumor uptake and tumor-to-normal organ ratios, the optimal imaging time point for the iodinated Ex-4 tracer would be at 1 hour p.i., which is similar to 68 Ga-Ex-4.
As a proof-of-concept, the 124 I analogue of the most promising tracer candidate, the agonist [Nle 14 , 124 I-Tyr 40 -NH 2 ]Ex-4, was evaluated in PET-studies.Although absolute tumor uptake levels were somewhat lower than expected, the PET-images confirmed the very favourable biodistribution profile of the 124 I congener in comparison to 68 Ga-Ex-4, with a pronounced and specific tumor uptake and low kidney retention.We strongly believe in the potential of 124 Iand 123 I-labeled Ex-4 derivatives for GLP-1 targeted imaging and currently develop automated labelling strategies.The goal is production of these tracers in high radiochemical purity, specific activity and improved radiochemical yields, which is an important factor given the high cost of 123 I and 124 I in particular, and the activities of these tracers needed for comprehensive imaging studies in animals and finally for clinical translation.

Conclusions
Even though the antagonist [Nle 14 , 125 I-Tyr 40 NH 2 ]Ex(9-39) recognized more binding sites on GLP-1R, it had a lower affinity than the agonist and it may not be suitable for clinical translation.The [Nle 14 , 125 I-Tyr 40 NH 2 ]Ex-4 agonist, in contrast, showed excellent tumor uptake and at the same time exhibited pharmacokinetics superior to all other GLP-1R tracers presently available, with particularly high tumor-to-kidney ratio and good contrast to normal organs.Preclinical PET imaging data strongly suggest that Ex-4 radioiodinated with 124 I may be a promising alternative to the radio-metal labelled derivatives for imaging of GLP-1 receptor positive insulinoma.It can further improve the sensitivity of the preoperative localization of benign insulinoma.

doi: 10 . 4
1371/journal.pone.0170435.t002PET imaging of GLP-1 receptors in mice using [Nle 14 , 124 I-Tyr 40 -NH 2 ] Ex-The in vivo properties of [Nle 14 , 124 I-Tyr 40 -NH 2 ]Ex-4 were evaluated using PET imaging.Maximum-Intensity-Projection (MIP) PET images (Fig 5A, 1 st panel) demonstrate that at 1 hour p.i. of [Nle 14 , 124 I-Tyr 40 -NH 2 ]Ex-4, the highest specific uptake was detected in Ins-1E tumor (10.8%IA/g based on image quantification).In the blocked mouse (Fig 5A, 2 nd panel) the tumor uptake was inhibited by 82% (1.9%IA/g left in the tumor).Apart from the bladder, high accumulation of radioactivity was also found in the thyroid (19.7%IA/g) and the stomach (28.3%IA/g) confirming the tracer radiodeiodination in vivo.The stomach uptake was partially inhibited in the blocked mouse, suggesting some GLP-1 receptor mediated uptake in the stomach, which was also described by others in the field.Blocking of the sodium iodide symporter with irenat did not affect the tumor uptake (11.4%IA/g) but completely inhibited the thyroid uptake and most of the stomach uptake (Fig 5A, 3 rd panel).The high amount of radioactivity in the bladder already after 1 hour p.i. suggests that the tracer or its radiometabolites are cleared through the kidney.Finally, HPLC analysis of the urine taken at 1 and 4 h p.i. of [Nle 14 , 124 I-Tyr 40 -NH 2 ]Ex-4 revealed that there was only free 124 I-iodide and no intact peptide present in the mouse urine (S4 Fig).The tumor time-activity curve derived from quantification of the PET images recorded at 1, 2, and 4 hours p.i. of irenat and [(Nle 14 , 124 I-Tyr 40 )NH 2 ]Ex-4 shows a decrease in the tumor radioactivity from 11.4%IA/g at 1 h p.i. to 5.4%IA/g at 4h p.i. (Fig 6).

Fig 3 .
Fig 3. Radio-HPLC showing the radiochemical purity of formulated [Nle 14 , 124 I-Tyr 40 -NH 2 ]Ex-4 without (A) and with co-injection of cold reference [Nle 14 , 127 I-Tyr 40 -NH 2 ]Ex-4 (B).UV-and radio-detectors were in series, resulting in a lag time of about 15 sec for the radiotrace.Numbers in the chromatograms refer to the peak retention time in minutes.doi:10.1371/journal.pone.0170435.g003