PET imaging of glycogen synthase kinase-3 in pancreatic cancer xenograft mouse models

Abstract

Glycogen synthase kinase-3 (GSK-3) contributes to tumorigenesis in pancreatic cancer by modulating cell proliferation and survival. This study evaluated the lead GSK-3 targeted PET radiotracers for neuro-PET imaging, [¹¹C]PF-367 and [¹¹C]OCM-44, in pancreatic cancer xenograft mice. Immunohistochemistry showed that GSK-3α and GSK-3β were overexpressed in PANC-1 xenografts. In autoradiography studies, higher specific binding was observed for [³H]PF-367 compared to [³H]OCM-44 when co-incubated with unlabeled PF-367 (59.2±1.8% vs 22.6±3.75%, respectively). Co-incubation of [¹¹C]OCM-44 with OCM-44 did not improve the specific binding (25.5±30.2%). In dynamic PET imaging of PANC-1 xenograft mouse models, tumors were not visualized with [¹¹C]PF-367 but were well visualized with [¹¹C]OCM-44. Time-activity curves revealed no difference in accumulation in PANC-1 tumor tissue compared to muscle tissue in [¹¹C]PF-367 baseline studies, while a significant difference was observed for [¹¹C]OCM-44 with a tumor-to-muscle ratio of 1.6. Tumor radioactivity accumulation following injection with [¹¹C]OCM-44 was not displaced by pre-treatment with unlabeled PF-367. Radiometabolite analysis showed that intact [¹¹C]PF-367 accounted for 7.5% of tumor radioactivity, with >30% in plasma, at 40 min post-injection of the radiotracer, and that intact [¹¹C]OCM-44 accounted for 20% of tumor radioactivity, with >60% in plasma. [¹¹C]OCM-44 is superior to [¹¹C]PF-367 for detecting lesions in preclinical mouse models of pancreatic cancer, however, both radiotracers undergo rapid metabolism in vivo. GSK-3 PET radiotracers with improved in vivo stability are needed for clinical translation. To our knowledge this work represents the first PET imaging study of GSK-3 in oncology.

Introduction

Pancreatic cancer (PnCa) is among the most lethal cancers and holds the highest mortality rate among all major cancers. It is the fourth leading cause of cancer-related deaths with a five-year survival rate of 10%, and a 75% mortality rate within the first year after diagnosis [1]. Few early symptoms are associated with pancreatic cancer, and the symptoms remain vague such as back pain, lethargy, and new onset diabetes [2]. In addition to being nearly asymptomatic, pancreatic cancer is very aggressive, such that patients are often at an advanced stage in disease progression at the time of initial diagnosis [3]. Less than 20% of pancreatic cancer patients are considered candidates for surgical resection due to tumor invasion of major blood vessels or metastases to the liver and/or other organs [4]. The high mortality rate of PnCa is largely due to a lack of biomarkers to diagnose the disease in its early stages when surgical removal of the tumor is still possible. New strategies for detection, diagnosis, and disease monitoring of PnCa are urgently needed.

Glycogen synthase kinase-3 (GSK-3) is a serine/threonine kinase that is responsible for the phosphorylation of many proteins [5]. GSK-3 is involved in multiple cell-signaling pathways with roles that can lead to pathogenesis of different diseases such as Alzheimer’s disease and several cancers [6]. GSK-3 has many substrates and is therefore involved in numerous signal transduction pathways including NF-κB signaling [7], WNT/β-catenin signaling [8] and EGFR/RAS/PI3K/PTEN/AKT/GSK-3/mTORC1 pathway [9]. The role of GSK-3 in cancer can be ambiguous such as both promoting and repressing cell proliferation, as well as acting as both a tumor suppressor and a tumor promoter, depending on the substrate. There are two isoforms of GSK-3, GSK-3α and GSK-3β, which have distinctive, non-redundant functions and both play a role in cancer cell survival, chemoresistance, and cancer progression in PnCa [9,10]. Preclinical and clinical studies have investigated the use of GSK-3 inhibitors in PnCa as therapeutic agents, or as sensitizing agents to improve the efficacy of standard chemotherapies. In preclinical studies, the GSK-3 inhibitors AR-A014418 and SB-216763 were found to suppress PnCa growth in vitro and in vivo [11,12] and the effect was increased when combined with gemcitabine [13,14]; treatment with the GSK-3 inhibitor 9-ING-41 alone induced cell killing in PnCa cell lines and the cytotoxicity was significantly increased when combined with gemcitabine [15]. The GSK-3β inhibitor, tideglusib, was found to suppress proliferation of PnCa cells [16]. In vivo studies showed that treatment with the GSK-3α/β inhibitor LY2090314 in combination with the chemotherapy Nab-pablitaxal (Abraxane^®) improved survival in mice bearing AsPC-1 PnCa orthotopic xenografts [17]. A clinical trial was initiated to administer LY2090314 in patients with metastatic cancer, including advanced PnCa (ClinicalTrials.gov identifier: NCT01632306) and a trial using 9-ING-41 for advanced cancers including PnCa is underway (ClinicalTrials.gov identifier: NCT03678883). A positron emission tomography (PET) radiotracer capable of quantifying GSK-3 density and distribution in vivo could provide new strategies for detection, diagnosis, and disease monitoring of PnCa.

Our laboratories reported the first radiotracer for imaging GSK-3, [¹¹C]AR-A014418 [18]. Several PET radiotracers targeting GSK-3 have been subsequently reported (Figure 1) and evaluated in preclinical brain-PET imaging studies [19-26]. The most promising radiotracers for imaging GSK-3α/β are oxazole-4-carboxamides, [¹¹C]PF-04802367 ([¹¹C]PF-367) and its derivative [¹¹C]OCM-44, that we developed with high affinity and selectivity for GSK-3α/β [27-29]. Despite extensive efforts to image GSK-3 in the brain, to date a PET radiotracer targeting GSK-3 has not yet been explored to visualize cancer tumors in vivo. This study evaluates [¹¹C]PF-367 and [¹¹C]OCM-44 in PnCa xenograft mouse models by immunohistochemistry (IHC), autoradiography, dynamic PET/MR imaging including blocking studies, ex vivo biodistribution, and radiometabolite analysis.

Figure 1.

PET radiotracers for imaging GSK-3. See [25,26] and references cited therein.

Materials and methods

General

Tritium labeled PF-04802367 ([³H]PF-367; molar activity (A_m) =3000 GBq/mmol (81 Ci/mmol), 37 MBq/mL (1 mCi/mL), radiochemical purity (RCP) =87.5%) and 5-(3-fluoro-4-methoxyphenyl)-N-(3-(pyridin-3-yl)propyl)oxazole-4-carboxamide ([³H]OCM-44; A_m=3000 GBq/mmol (81 Ci/mmol), 37 MBq/mL (1 mCi/mL), RCP=87.4%) were prepared by tritiomethylation from the respective desmethyl precursors at Novandi Chemistry AB (Södertälje, Sweden). PF-04802367 was purchased commercially (Millipore Sigma). Radiolabeling precursors and unlabeled OCM-44 were prepared by Sai Life Sciences, Ltd (Hyderabad, India).

Radiochemical synthesis of [¹¹C]PF-367 and [¹¹C]OCM-44

Automated radiosynthesis of [¹¹C]PF-367 was carried out on a GE Tracerlab FX2 C-Pro™ synthesis module as previously described [28] with modifications to adapt a “loop method” [30]. Briefly, the HPLC loop was pre-loaded with desmethyl-PF-367 (0.5±0.1 mg) dissolved in 100 µL anhydrous MEK and 2.5 µL 1.0 M methanolic TBAOH. [¹¹C]CH₃OTf was flowed through the HPLC loop for 3 min at room temperature (RT) prior to injection onto a semi-preparative Luna^® C18(2) HPLC column (10 µm, 10×250 mm; Phenomenex, Torrance, USA). Isocratic semi-preparative HPLC purification was conducted at 5.0 mL/min with 40/60 CH₃CN/100 mM NH₄HCO₂ (v/v) as the mobile phase. The peak corresponding to [¹¹C]PF-367 was collected and diluted with a sterile mixture of 20 mL H₂O and 2 mL 1 M NaHCO₃. The diluted solution was passed through a pre-conditioned solid-phase extraction cartridge (SepPak^® tC18 Plus, Waters; Milford, USA) and the cartridge was rinsed with 10 mL of sterile water. The radiotracer was eluted with 1 mL EtOH, followed by 9 mL 0.9% saline. Analytical HPLC was performed at 1.0 mL/min with 40/60 CH₃CN/100 mM NH₄HCO₂ (v/v) mobile phase on a Luna C18 HPLC column (10 μm, 4.6×250 mm; Phenomenex).

Automated radiosynthesis of [¹¹C]OCM-44 was also carried out on a GE Tracerlab FX2 C-Pro™ synthesis module as previously described [28] with minor modifications to adapt a “loop method” [30]. Briefly, 1.0±0.1 mg of desmethyl-OCM-44 was dissolved in 80 µL anhydrous DMF, then 1.5 µL potassium tert-butoxide (1 M in THF) was added. This solution was pre-loaded on a 5 mL stainless-steel loop, 5 min prior to radiosynthesis. [¹¹C]CH₃I was flowed through the stainless-steel loop and held at RT for 5 min prior to being loaded onto a semi-preparative Luna C18 HPLC column (10 µm, 10×250 mm; Phenomenex) for isocratic separation with 40/60 CH₃CN/H₂0 + 0.1 N ammonium formate as the mobile phase. The peak corresponding to [¹¹C]OCM-44 was collected and diluted with a sterile mixture of 20 mL H₂O and 2 mL 1 M NaHCO₃. The diluted solution was passed through a pre-conditioned solid-phase extraction cartridge (SepPak^® tC18 Plus, Waters) and the cartridge was rinsed with 10 mL of sterile water. The radiotracer was eluted with 1 mL EtOH, followed by 9 mL 0.9% saline. Analytical HPLC was performed at 3.0 mL/min with 40/60 CH₃CN/100 mM NH₄HCO₂ (v/v) mobile phase on a Luna C18 HPLC column (10 μm, 4.6×250 mm; Phenomenex).

Tumor xenograft mouse models

PANC-1 human PnCa cells, with high GSK-3α and GSK-3β expression [12,31], were purchased from the American Type Culture Collection (Manassas, VA, USA). PANC-1 cells were cultured in McCoy’s 5A Modified medium (Gibco, Life Technologies; Burlington, ON, Canada) supplemented with 10% fetal bovine serum (Gibco), and 1% penicillin/streptomycin (Sigma-Aldrich, St. Louis, MO, USA). Cells were cultured in an atmosphere of 5% CO₂ at 37°C. Female ICRscid mice (Taconic Biosciences; Rensselaer, NY, USA), were inoculated subcutaneously (s.c.) on the right flank and imaged when tumors reached 0.75-1.0 cm diameter. Animal studies were conducted under a protocol (#817) approved by the Animal Care Committee at the Centre for Addition and Mental Health, following Canadian Council on Animal Care guidelines.

Immunohistochemistry

IHC was performed to determine GSK-3α and GSK-3β density in PANC-1 tumors. Tumors were fixed in 10% formalin for 48 h then embedded in paraffin and prepared in 4 µm sections onto microscope slides. Slides were dewaxed through changes of xylene, followed by hydration through decreasing grades of alcohol in water (100%, 95%, 70%). Slides were blocked with 3% hydrogen peroxide then antigen retrieval was performed with pepsin for slides being stained for CD68. Serum block was applied as per directed by the MACH-4 Universal HRP-Polymer kit (Inter Medico, Markham, ON, CA) followed by incubation with primary antibodies mouse anti-CD68 (M0876, 1:400; Agilent Dako, Santa Clara, CA, USA), rabbit polyclonal anti-GSK-3α (Ab62331, 1:200; Abcam, Cambridge, UK), or rabbit recombinant anti-GSK-3β (phosphor S9) antibody (Ab75814, 1:100; Abcam) at RT for 1 h. Color was developed using DAB (K3468, Agilent Dako) and counter stained with hematoxylin. Slides were dehydrated by reversing the rehydration procedure and sections were mounted with mounting medium (3801120, Leica, Buffalo Grove, IL, USA). Slides were scanned with a slide scanner (VS200 Slideview, Olympus, Tokyo, Japan).

Autoradiography

Tumors were excised and immediately stored at -80°C until cryostat sectioning. Frozen tumors were coated with Tissue-Tek (Sakura Finetek, Tokyo, Japan) embedding medium for frozen tissue specimens to ensure the optimal cutting temperature (O.C.T.), a serial of 10 µm thick cryosections was generated using a Cryostar NX50 cryostat (Thermo Fisher Scientific, Runcorn, UK). Slides were stored at -80°C until receptor binding assays were performed. Prior to autoradiography studies, tissue sections were removed from freezing conditions and acclimated to RT for 30 min.

For total binding, slides were incubated for 60 min at RT with 8 nM [³H]PF-367 or [³H]OCM-44 in HEPES-buffered Krebs-Ringer Solution, pH 7.2 containing DMSO (Invitrogen, Oregon, USA). For non-specific binding, slides were incubated for 60 min at RT with 8 nM [³H]PF-367 or [³H]OCM-44 in HEPES-buffered Krebs-Ringer Solution, pH 7.2, adding 10 µM unlabeled DMSO-dissolved PF-367 or OCM-44. Slides were washed with ice-cold 0.9% saline once for 60 seconds followed by a 10-second rinse with ice-cold distilled water. Slides were dried under airflow.

After incubating with the tritium labeled radiotracers, washing, and drying, the tissue sections and tritium standard (ART 0123A, American Radiolabeled Chemicals, Missouri, USA) were placed into a Halsey Rigidform Cassette (Halsey X-Ray Products, New York, USA) and exposed to a TR2040 phosphor screen (FUJIFILM, Tokyo, Japan) for five days at RT, then scanned using an Amersham Typhoon™ Biomolecular Imager (GE Healthcare, Massachusetts, USA). Receptor binding was quantified (nCi/mg) using a Microcomputer Imaging Device (MCID) computer-based imaging system 7.1 (Imaging Research Inc., Ontario, CA) based on the tritium standard exposed with the slides. Non-specific binding was subtracted from total binding to calculate specific binding.

Dynamic PET/MR imaging and biodistribution studies

PET/magnetic resonance (MR) image acquisition and analysis were performed as previously described [32]. Tumor-bearing mice were anesthetized by isoflurane in O₂ (4%, 2 L/min induction; 1-2%, 1 L/min maintenance) for lateral tail-vein catheterization then transferred to a nanoScan™ PET/MRI 3T scanner (Mediso, Budapest, Hungary). Anesthesia was maintained throughout PET/MR scanning while body temperature and respiration parameters were monitored. Mice bearing PANC-1 xenografts were injected through the tail-vein catheter with [¹¹C]PF-367 (2.1-9.18 MBq, 5.47-44.15 nmol/kg, 3.72-24.93 GBq/µmol) or [¹¹C]OCM-44 (6.17-12.77 MBq, 0.53-22.45 nmol/kg, 63.36-368.94 GBq/µmol). For blocking studies, mice bearing PANC-1 xenografts were scanned at baseline (n=2 for [¹¹C]PF-367; n=5 for [¹¹C]OCM-44) or under blocking conditions (n=2), in which mice were pre-treated with an i.p. injection of 2 mg unlabeled PF-367 in 20 µL dimethylsulfoxide (Sigma) 60 min prior to bolus injection of [¹¹C]PF-367 or [¹¹C]OCM-44. Following the 60 min PET scans, mice were sacrificed by cervical dislocation and tissue samples were collected, weighed, and transferred to γ-counting tubes for biodistribution analysis. Tissue radioactivity was measured with a γ-counter (2480 Wizard^2TM, PerkinElmer, MA, USA) and expressed as % ID/g. Image analyses and extraction of time-activity curves (TACs) from regions of interest (ROIs) were performed in Amide v1.0.4.

Radiometabolite analysis

Tumor-bearing mice were sacrificed by cervical dislocation 40 min post-injection (p.i.) with [¹¹C]PF-367 (n=1, 10.4 MBq, 2.82 nmol/kg, 97.13 GBq/µmol) or [¹¹C]OCM-44 (n=1, 4.85 MBq, 412.46 nmol/kg, 0.32 GBq/µmol). Blood was collected by cardiac heart puncture then centrifuged for 5 min at 2,000× RCF. Tumors were excised and homogenized with a BeadBug™ (Benchmark Scientific; Sayreville, NJ, USA) in 1.6 mL acetonitrile with respective 50 ng of unlabeled parent compounds PF-367 or OCM-44, then 0.8 mL H₂O was added. The homogenate was centrifuged at 10,000× RCF for 5 min. Parent radiotracer was separated from plasma and tumor homogenates by column-switching HPLC in a mobile phase of 40/60 CH₃CN/100 mM NH₄HCO₂ (v/v), as previously described by our laboratories [32], and analyzed with PowerChrom 2.6.15 (eDAQ) [33].

Statistical analysis

Data are represented as the mean ± SD. Statistical comparisons were performed by an unpaired t-test (P<0.05) with GraphPad Prism Version 9.

Results

Radiosynthesis of [¹¹C]PF-367 and [¹¹C]OCM-44

[¹¹C]PF-367 was produced and had a retention time of approximately 9.5-10.5 min during the semi-preparative purification (Figure 2A and 2B). [¹¹C]PF-367 was synthesized with RCP>99%, as confirmed by co-elution of the formulated radiotracer with authentic PF-367 standard by analytical HPLC (Figure 2C and 2D). [¹¹C]PF-367 had an average A_m=41.4 GBq/μmol (n=5, 1119 mCi/μmol).

Figure 2.

HPLC chromatograms from the production of [¹¹C]PF-367. Semi-preparative HPLC chromatogram of (A) radioactive (gamma) signal and (B) UV signal. Analytical HPLC chromatograms showing (C) radioactive (gamma) signal from the formulated radiotracer and (D) UV signal from the authentic PF-367 reference standard.

[¹¹C]OCM-44 was separated by semi-preparative HPLC with a retention time of approximately 11.7-12.7 min (Figure 3A and 3B). [¹¹C]OCM-44 was synthesized with a average A_m=161.3 GBq/μmol (n=8, 4360 mCi/μmol) and RCP>99%, as confirmed by co-elution of authentic standard by analytical HPLC (Figure 3C and 3D).

Figure 3.

HPLC chromatograms from the production of [¹¹C]OCM-44. Semi-preparative HPLC chromatogram of (A) radioactive (gamma) signal and (B) UV signal. Analytical HPLC chromatograms showing (C) radioactive (gamma) signal from the formulated radiotracer and (D) UV signal from the authentic OCM-44 reference standard.

Immunohistochemistry

GSK-3α and GSK-3β density were examined by IHC staining in human PnCa PANC-1 xenografts (Figure 4). IHC staining confirmed that PANC-1 xenografts had high expression (brown staining) of GSK-3α and GSK-3β. Tissue sections of PANC-1 tumors were also analyzed for CD-68, a surface biomarker on cells of human macrophage/monocyte origin, as a negative control.

Figure 4.

GSK-3α, GSK-3β, and CD-68 expression in PANC-1 PnCa xenograft tissue sections. IHC staining shows high expression of (A) GSK-3α and (B) GSK-3β, and (C) low expression of CD-68 in PANC-1 PnCa xenografts.

Autoradiography

Figure 5 depicts the results of autoradiography studies which assessed the specific binding of [³H]PF-367 and [³H]OCM-44 co-incubated with unlabeled PF-367 or OCM-44 in PANC-1 tumor tissue sections. Co-incubation of 8 nM [³H]PF-367 with 10 µM unlabeled PF-367 on PANC-1 tumor sections exhibited specific binding of 59.2±1.8%, which was reduced, though not significantly, to 48.9±12.0% when co-incubated with 10 µM unlabeled OCM-44. No difference in specific binding was observed when [³H]OCM-44 was co-incubated with 10 µM unlabeled PF-367 or OCM-44, 22.6±3.75% and 25.5±30.2%, respectively. The specific binding of [³H]PF-367 was significantly higher than [³H]OCM-44 in PANC-1 tumor tissues when co-incubated with PF-367 (P=0.0001). No significant difference was observed between [³H]PF-367 or [³H]OCM-44 in PANC-1 tumor tissues when co-incubated with OCM-44.

Figure 5.

Autoradiography with [³H]PF-367 and [³H]OCM-44 in PANC-1 PnCa tumor tissue sections. Representative images of autoradiography in PANC-1 tumor sections following incubation with (A) [³H]PF-367 alone (left) and co-incubated with PF-367 (right), (B) [³H]OCM-44 alone (left) and with OCM-44 (right), (C) [³H]PF-367 alone (left) and with OCM-44 (right), (D) [³H]OCM-44 alone (left) and with PF-367 (right), and (E) the associated quantification of the total binding (TB), non-specific binding (NSB), and specific binding (SB) shown as the % of total binding. Error bars represent the mean ± SD.

Dynamic PET/MR imaging of tumor-bearing mice

We evaluated [¹¹C]PF-367 and [¹¹C]OCM-44 in ICRscid mice bearing s.c. PANC-1 PnCa xenografts by dynamic PET/MR imaging and validated the PET scan data analysis by ex vivo biodistribution studies. Figure 6A shows a representative PET image of a mouse bearing a s.c. PANC-1 xenograft following injection with [¹¹C]PF-367 (0-60 min summed image) and did not reveal radioactivity accumulation in the tumor. Analysis of TACs support that the image acquired at baseline showed no difference between radioactivity accumulation in the tumor and in muscle tissue (Figure 6C). Blocking studies were performed by pre-treatment with unlabeled PF-367 prior to [¹¹C]PF-367 bolus injection and revealed no difference between the tumor/muscle (T/M) ratios under baseline and blocked conditions (Figure 6D). Figure 6E shows a representative PET image of a mouse bearing a s.c. PANC-1 xenograft following injection with [¹¹C]OCM-44 (0-60 min summed image) in which the tumor was well visualized. However, as shown in Figure 6F, the radioactivity accumulation in the tumor was not displaced by pre-treatment with unlabeled PF-367. Figure 6G shows that the TAC analysis of [¹¹C]OCM-44 at baseline confirms a significantly higher tumor tissue radioactivity accumulation compared to muscle tissue with an average uptake from 50-60 min of 1.41±0.16% ID/g vs 0.91±0.19% ID/g (P=0.0022) resulting in 1.6 T/M. Figure 6H shows the TAC analysis of blocking studies performed by pre-treatment with unlabeled PF-367 prior to bolus [¹¹C]OCM-44 injection and reveals a decrease between the T/M at baseline or blocked conditions, 1.60±0.30% ID/g vs 1.09±0.31% ID/g, albeit the result is not statistically different (P=0.1040).

Figure 6.

Dynamic PET/MR imaging and TACs of GSK-3 targeted [¹¹C]PF-367 and [¹¹C]OCM-44 in PANC-1 PnCa xenograft mouse models. Representative iterative static PET/MR images (0-60 min summed image) of [¹¹C]PF-367 at (A) baseline, (B) with PF-367 blockade, and (C, D) respective TACs. Representative iterative static PET/MR images (0-60 min summed image) of [¹¹C]OCM-44 at (E) baseline, (F) with PF-367 blockade, and (G, H) respective TACs. Tumor sites are indicated by red arrows.

PET imaging analyses were substantiated by ex vivo biodistribution studies of selected tissues performed 60 min following radiotracer injection. No difference between radioactivity accumulation in tumor tissue and muscle tissue was observed for [¹¹C]PF-367 (1.39±0.17% ID/g and 1.31±0.36% ID/g, respectively, (P=0.785)). Radioactivity accumulation in tumor tissue was significantly higher than that in muscle tissue for [¹¹C]OCM-44 (1.11±0.08% ID/g and 0.59±0.09% ID/g, (P=0.0236)). The resultant tumor-to-muscle ratios for [¹¹C]PF-367 and [¹¹C]OCM-44 were 1.1 and 1.9, respectively. No significant difference was observed in tumor radioactivity accumulation of [¹¹C]PF-367 or [¹¹C]OCM-44, which revealed 1.39±0.17% ID/g and 1.11±0.08% ID/g, respectively (P=0.1634). Additionally, tumor radioactivity accumulation of [¹¹C]PF-367 and [¹¹C]OCM-44 when pre-treated with unlabeled PF-367 blockade were observed at 1.39±0.72% ID/g and 1.30±0.40% ID/g, respectively, and were not significantly different from their respective baselines (P=0.9924 and P=0.5865).

Radiometabolite analysis

Ex vivo composition of [¹¹C]PF-367 and [¹¹C]OCM-44 in plasma and tumor tissue were determined at 40 min p.i. in tumor-bearing mice by HPLC analysis of relative amounts of parent radioligand and radiometabolites. At 40 min p.i., intact [¹¹C]PF-367 (Figure 7A) represented 32.6% of the total radioactivity in the plasma and 7.5% in tumor tissue while intact [¹¹C]OCM-44 (Figure 7B) represented 63.1% of the total radioactivity in the plasma and 19.9% in tumor tissue. These results indicate that [¹¹C]PF-367 is metabolized at a faster rate than [¹¹C]OCM-44 yet both radiotracers exhibit rapid metabolism in mice.

Figure 7.

Radiometabolite analysis of [¹¹C]PF-367 and [¹¹C]OCM-44 in PANC-1 tumor-bearing mice at 40 min p.i. (A) Column-switching HPLC chromatogram of [¹¹C]PF-367 (i) product following radiosynthesis; HPLC chromatogram with normalized peak analysis from (ii) plasma and (iii) tumor homogenates. (B) Column-switching HPLC chromatogram of [¹¹C]OCM-44 (i) product following radiosynthesis; HPLC chromatogram with normalized peak analysis from (ii) plasma and (iii) tumor homogenates.

Discussion

Radiotracers with high selectivity and affinity for GSK-3 have been sought after and optimized for brain-PET imaging for nearly two decades [18]. The two lead GSK-3 PET radiotracers to date are [¹¹C]PF-367 and [¹¹C]OCM-44, which have high selectivity and affinity for GSK-3. In light of the increased applications of these PET radiotracers we simplified the automated radiosynthesis on a commercial platform, by adapting the “loop method” for ¹¹C-methylation reactions with and [¹¹C]CH₃OTf and [¹¹C]CH₃I, for [¹¹C]PF-367 and [¹¹C]OCM-44, respectively. Preliminary evaluation of [¹¹C]CH₃OTf, compared with [¹¹C]CH₃I, for alkylation of [¹¹C]PF-367 suppressed the formation of a volatile impurity and led to higher RCYs (not optimized). In the present study, we sought to evaluate [¹¹C]PF-367 and [¹¹C]OCM-44 as the first PET radiotracers targeting GSK-3 in oncology. We thereby directed our efforts to visualize PnCa tumors in xenograft mouse models.

In agreement with the literature, our IHC studies showed that both GSK-3α and GSK-3β are present and evenly distributed through PANC-1 PnCa xenografts [12]. In autoradiography studies using PANC-1 PnCa xenograft tissue, higher specific binding of [³H]PF-367 for GSK-3 was observed compared to [³H]OCM-44, with 2.6-fold greater specific binding when blocked with PF-367 and 2-fold greater specific binding when blocked with OCM-44. Despite high selectivity and specific binding observed ex vivo in PANC-1 xenografts incubated with [³H]PF-367, no radioactivity accumulation was observed in PANC-1 tumors in vivo by PET imaging with [¹¹C]PF-367. The lack of tumor uptake following administration of [¹¹C]PF-367 in PET imaging studies is rationalized by its extensive radiometabolism in mice despite reports of reasonable in vitro stability in human hepatic microsomes [27], with only 7.5% of the radioactivity accumulated in the tumor accounted for by intact parent [¹¹C]PF-367 after 40 min. PET imaging studies with [¹¹C]OCM-44 revealed a T/M of 1.6. However, the tumor uptake could not be displaced in blocking studies with pre-treatment of PF-367, which may be attributed to the rapid metabolism of both PF-367 and OCM-44 in mice. Radiometabolite analysis from arterial blood for both of these tracers is underway in non-human primates and will be published elsewhere with detailed kinetic analysis.

A limitation of this study is the lack of potent and selective GSK-3 inhibitors with reasonable metabolic stability available for blocking studies. Although excess unlabeled PF-367 was used to block the radioactivity accumulation of [¹¹C]PF-367 and [¹¹C]OCM-44, the use of unlabeled OCM-44 or other newer generation GSK-3 inhibitors with slower metabolism are likely needed to adequately confirm specificity of radiotracer binding. The rapid metabolism observed with [¹¹C]PF-367 and [¹¹C]OCM-44 may restrict their applications in PET imaging of oncology. A lack of suitable inhibitors for demonstrating in vivo selectivity of novel targets can present an obstacle to discovering and screening lead radiotracers for PET imaging novel biomarkers [34]. Furthermore, the roles of GSK-3α and GSK-3β in PnCa are not functionally redundant despite their structural similarity. In NF-κB cell signaling, GSK-3α inhibition leads to suppression of cell growth in PnCa [35], while GSK-3β up-regulates NF-κB activity which stimulates cell proliferation, pro-tumorigenic cytokine production, resistance to apoptosis, and chemoresistance in PnCa [12,36-38]. GSK-3β is also involved in the WNT/β-catenin pathway by modulating resistance to radiation therapy [39]. Future studies investigating GSK-3 in PnCa with a PET radiotracer that is selective for GSK-3α or -β would also be worthwhile to investigate in oncology as newer generations of GSK-3 PET radiotracers are discovered [28].

Conclusions

To our knowledge this works represents the first PET imaging study of GSK-3 in cancer. Herein we simplified the automated radiosynthesis of the two leading GSK-3 PET radiotracers, [¹¹C]PF-367 and [¹¹C]OCM-44 to adapt the “loop method” for ¹¹C-methylation. [¹¹C]OCM-44 was utilized to successfully image GSK-3 overexpressing PnCa tumors in PANC-1 xenograft mouse models. Although the rapid metabolism in the periphery and short half-life of the radionuclide may preclude the use of [¹¹C]PF-367 and [¹¹C]OCM-44 in oncology, next generation GSK-3 and related kinase inhibitors have potential for imaging signal transduction pathways.

Disclosure of conflict of interest

NV is a co-founder of MedChem Imaging, Inc. and serves as an Advisory Board Member for 4M Therapeutics, Inc. The above-mentioned interests had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. All other authors declare no potential conflict of interest.