Introduction
Diabetes is a disease characterized by a chronic high levels of blood glucose linked to defects in insulin secretion or action pathways, often associated with a loss of a functional beta-cell mass (BCM). Currently, neither type 1 diabetes (T1D) nor type 2 diabetes (T2D) can be cured by pharmaceutical interventions. Beta-cell replacement methods have been suggested as a curative treatment for T1D. For several decades, beta-cell replacement strategies in the context of T1D have been restricted solely to transplantation of the whole pancreas or isolated human islets from brain dead donors [
1]. A renewable cell source such as islets generated from human pluripotent stem cells (e.g., human embryonic stem cells (hESC) or induced pluripotent stem cells (iPSC)) has been proposed and explored to recover a functional BCM. This approach could help bypass the main challenges associated with human islet transplantation, such as frequent shortage of suitable donors and the need for extensive immunosuppressive treatments [
2,
3]. Over the past decades, physiologically relevant stem-cell-derived islets (SC-islets) with a cytoarchitecture and a functionality similar to those of the adult primary islets have been successfully generated [
4,
5]. Those SC-islets have demonstrated the capability to cure diabetic mice in vivo [
6,
7], making them a promising alternative to human islet transplantation. Recently, in an ongoing clinical phase I/II study, three patients with T1D have been transplanted with SC-islets through an ordinary hepatic vein infusion, resulting in glucose responsive insulin secretion, reduced usage of exogenous insulin and improvement in the HbA1c levels (ClinicalTrials.gov Identifier: NCT04786262). Despite the successful generation of functional SC-islets, monitoring the transplanted SC-islets remains a challenge, especially during the critical early post-transplantation period that dictates the survivability of the grafted cells as they are prone to immune rejections [
8]. Molecular imaging techniques and particularly positron emission tomography (PET) have been suggested as potential tools for monitoring the fate of islets after clinical transplantation [
9]. Therefore, the identification of molecular targets restricted to newly grafted SC-islets would greatly help us monitor their fate after transplantation. The DiGeorge syndrome critical region gene 2 (DGCR2) protein has not only been identified as biomarker of mature human pancreatic beta cells [
10], but also shown to be strongly expressed on insulin-positive hESC-derived pancreatic progenitor cells from immunostaining results [
11].
In this study, we describe the labeling of a novel Affibody molecule ZDGCR2:AM106 with fluorine-18 using an innovative click chemistry approach. The resulting PET tracer [18F]ZDGCR2:AM106 was then evaluated for molecular imaging of SC-islets grafts in vitro and in vivo, as well as for biodistribution in various healthy animal species.
Research design and methods
Radiochemistry
The labeling process comprised three reaction steps, adapted from a previously developed method for
18F-radiolabeling of antibody constructs [
12].
Synthesis of [18F]MeTz
[18F]fluoride (16–20 GBq) in aqueous solution was concentrated and trapped on a Chromabond PS-HCO3 Shorty/45 mg cartridge (Macherey–Nagel). The [18F]fluoride was then reacted on the solid support with a solution of the precursor N,N,N-trimethyl-5-((2,3,5,6-tetrafluorophenoxy)carbonyl) pyridin-2-aminium trifluoromethanesulfonate 1 (10 mg) in acetonitrile (0.8 mL) which was passed over the Chromabond cartridge followed by neat acetonitrile (0.7 mL). Both solutions were passed over the cartridge at a flow rate of 0.4 mL/min. The reaction formed [18F]F-Py-TFP instantly at room temperature, which was then released with the flow into a Teflon tubing acting as a reservoir. After the elution was completed, the collected solution in the Teflon tubing was pushed with air over an Oasis MCX Plus Short cartridge (Waters), preconditioned with acetonitrile (3 mL), to remove unreacted cationic precursor 1. The purified solution containing [18F]F-Py-TFP was then delivered to a septum-equipped vial (5 mL) containing (4-(6-methyl-1,2,4,5-tetrazin-3-yl)phenyl)methanamine hydrochloride 2 (1.0 mg) in DMSO (0.1 mL) to improve its solubility.
In the second step, triethylamine (5 µL) in acetonitrile (100 µL) was added to the mixture, and the mixture was heated for 15 min at 55 °C. After diluting the mixture with aqueous trifluoroacetic acid (3 mL, 0.1%), the labeled tetrazine was isolated by semi-preparative HPLC. The column used was an ACE C18 5 µm (150 × 10 mm) column, and the eluent was a mixture of water, ethanol and TFA in a 62:38:0.01 ratio, with a flow rate of 5 mL/min. The column eluate was monitored using a UV detector (254 nm) and a radiodetector. The retention time for [18F]MeTz was 8.5 min. The collected HPLC fraction containing [18F]MeTz was reformulated in ethanol (0.5 mL) using the conventional procedure, which included dilution with water (20 mL), fixation on a Sep-Pak tC18 Plus Light Cartridge, a rinse with water (20 mL) and air prior to elution with the ethanol (0.5 mL) into a septum-equipped vial (1 ml). The ethanol was the then removed by evaporation under a stream of nitrogen gas and heat at 65 °C for 7 min, leaving the [18F]MeTz with only a few water droplets in the vial.
The radiochemical purity of [18F]MeTz was assessed by analytical HPLC using a Kinetex C18 column (2.6 µm 100 Å, 100 × 3.0 mm, Phenomenex). The eluent was a mixture of water and acetonitrile in a 75:25 ratio, with a flow rate of 0.7 mL/min. The retention time for [18F]MeTz was 6.6 min. The labeled tetrazine was identified by co-injection and matching retention times with an authentic reference standard (Norsk medisinsk syklotronsenter AS). The molar activity of the [18F]MeTz was assessed by determining the tracer concentration using the same analytical HPLC-procedure and measuring the activity in a dose calibrator.
Synthesis of [18F]ZDGCR2:AM106
ZDGCR2:AM106 (containing 59 residues with the following amino acid sequence: H2N-VDNKFNKEQTHARNEILQLPNLNRMQKSAFIRSLIDDPSQSANLLAEAKKLNDAQAPKC) was manufactured by custom solid-phase peptide synthesis by New England Peptide under our request. The unique cysteine residue at position 59 at the C-terminal was added albeit not initially being part of the original Affibody molecule sequence, to enable site-specific functionalization of the peptide at this position, a trans-cyclooctene (TCO) group was conjugated to the C-terminal cysteine via maleimide chemistry, via a polyethylene glycol (PEG3) linker (TCO–PEG3–maleimide). The resulting construct TCO–ZDGCR2:AM106 was purified, freeze-dried and aliquoted into vials of 100 µg. The purity was > 95% as assessed by HPLC. Freeze-dried TCO–ZDGCR2:AM106 was first put into solution (100 µg, 14 nmol in 300 µL PBS) before adding it to a vial containing [18F]MeTz (2.0 ± 0.6 GBq). The mixture was then allowed to react for 7–12 min at room temperature to form [18F]ZDGCR2:AM106. The labeled Affibody molecule was then purified from unreacted [18F]MeTz using size exclusion chromatography with a NAP-5 column (Cytiva). After preconditioning the column with PBS and 10% ethanol (two column volumes), the product solution was loaded and then eluted with the PBS and 10% ethanol (1 mL). The ready for use [18F]ZDGCR2:AM106 was collected in the eluate (approximately 1 mL), while the [18F]MeTz was retained on the column.
The radiochemical purity of [18F]ZDGCR2:AM106 was assessed by analytical HPLC using a Vydac 214MS C4 column (5 µm, 300 Å, 50 × 4.6 mm, Phenomenex). The eluent was a mixture of water + 0.1% TFA and acetonitrile in a gradient running from 5 to 80% acetonitrile over 10 min, with a flow rate of 4 mL/min. The retention time for [18F]ZDGCR2:AM106 was 3.5 min. [18F]ZDGCR2:AM106 Affibody molecule was identified by co-injection and co-elution with the non-labeled TCO-conjugated Affibody molecule ZDGCR2:AM106.
ELISA-based binding
Purified recombinant human DGCR2 His-tag protein (Bio-techne; #10,161-DG) was purchased for the binding studies. The freeze-dried protein was dissolved in PBS prior to coating on Pierce Copper-Coated High-Capacity Plates (Thermo Scientific) at a concentration of 10 µg/mL per well according to the manufacturer´s instruction, supplemented by a 15-min blocking step with PBS containing 1% bovine serum albumin (BSA). Wells containing PBS only were used as negative control. [18F]ZDGCR2:AM106 was incubated in each well at 5 MBq/mL for 60 min at RT with shaking. Three successive washes using PBS containing 0.05% Tween-20 were performed prior to measurement in a NaI well counter (Uppsala Imanet AB, Uppsala, Sweden).
Human tissue and cell culture
hESC (H1 cell line, Wicell®) were differentiated into SC-islets using a previously published differentiation protocol of seven stages[
4] with the following differences: H1 cells were propagated on treated human recombinant Laminin 521 (BioLamina; LN 521) in mTeSR-Plus medium (STEMCELL Technologies; #100-0274/100-275). To prepare for the differentiation step, cells were seeded at a density of 16 million cells/10 cm dish or two million cells/3.5 cm dish on plates coated with Laminin 521 in mTeSR-Plus medium with the addition of 10 µM of ROCK inhibitor Y-27632 (STEMCELL Technologies; 72,304). Differentiation was then triggered 24 h after the seeding process. During the first four stages (S1-4), cells were differentiated on adherent culture and a single cell suspension was prepared during S4:2 to generate uniform cluster formation. At stage 5:2, all SC-islets were moved to an ultra-low attachment 6-wells plate, on a rotating platform at 95 rpm. The last stages (S5-7) of the differentiation were carried out in suspension culture. During the stages S1-5, the medium was changed every day, while it was changed every second day over S6 and S7.
Pancreatic endocrine islet fractions of 47% purity were obtained from non-diabetic deceased donors within the Nordic Network for the Clinical Islet Transplantation Laboratory (Uppsala University Hospital, Uppsala Sweden) through isolation methods described by Goto et al. [
13]. None of the tissues were procured from prisoners, and all the donors have provided written consent that their donated tissues may be entered into a biobank and used in medical research. The use of human tissues from Uppsala Biobank (registration #827) was approved by the Regional Ethics Board, Uppsala, Sweden (now the Swedish Ethical Review Authority) (2011/473, Ups 02–577, 2015/401), and they were anonymized, collected and treated according to local institutional and Swedish national rules and regulations. The need for informed consent was renounced by the Regional Ethics Board in Uppsala, and the study has been conducted in accordance with the Declaration of Helsinki.
The Hek293 cell line used as negative control was kindly provided by professor Di Yu (Uppsala University) and cultured in DMEM (Gibco) supplemented with 10% FBS (Gibco) and 1% PEST (Gibco). The cells were passaged using SemPro Accutase (Gibco) according to the manufacturer´s instructions.
Frozen section autoradiography
Hek293 cells were collected using Accutase and washed with PBS before centrifugation for 5 min at 300 × g to form a pellet. Pellets of pancreatic endocrine islet fractions of 47% purity and SC-islets were obtained through sedimentation. All the obtained pellets (Hek293, human pancreatic endocrine fraction, SC-islets) were then embedded into optimal cutting temperature (O.C.T) compound (Q Path mounting media, VWR) and frozen at -80 °C. Slices of 10 µm thickness were prepared on cryotome (Cryostat NX70, ThermoFisher) before mounting on SuperFrost Plus slides (ThermoFisher) and stored at − 80 °C prior to the experiment.
Frozen sections of Hek293 cells, endocrine pancreatic islet fractions and SC-islets were incubated at RT for 60 min with PBS and 0.35 MBq/mL of [18F]ZDGCR2:AM106. A series of two 3-min cold washes in PBS with 1% BSA and one 1-min wash in deionized water were performed at the end of the incubation time. Samples were air-dried and left for exposure on the BAS-IP storage phosphor screen (Cytiva) for 120 min. 20 µL droplets of reference radioactive solution on absorbent paper, cross-measured in a NaI well counter (Uppsala Imanet AB, Uppsala, Sweden) were also included to enable quantification. The resulting digital image readout was obtained using an Amersham Typhoon storage phosphor imager (GE healthcare). The resulting autoradiography images were analyzed using ImageJ with manual delineation (details of the analysis are given in supplementary methods).
PET/MRI/CT of mice with intramuscular SC-islet grafts
A total of n = 4 NMRI nu/nu mice (Taconic M&B) with weight of 20–30 g were transplanted with 200–800 SC-islets in the abdominal and/or neck muscle. Mice were anesthetized with isoflurane before exposition of the muscle. SC-islets were packed in a butterfly needle and implanted into the muscle layer. Subcutaneous carprofen 5 mg kg-1 (Rimadyl Vet) was used as analgesic. Approximately four weeks after transplantation, for each mouse, a magnetic resonance imaging (MRI) (517 ms TR, 12.6 ms TE, 17 slices, 1 Th, 0.1 Gap, 80 × 60 mm FOV, 0.313 mm resolution, 4 NEX, 6 min 30 s acquisition time) or computer tomography (CT) (50 kVp, 610 µA, 480 proj, 1:4 binning, 251 × 251 × 251 µm voxel size) acquisition for attenuation and anatomical localization was first performed. 400 kBq/g of [18F]ZDGCR2:AM106 was subsequently administered through the lateral tail vein. A dynamic PET scan of 60 min was then acquired and reconstructed with the Tera-Tomo 3D method (6 × 10’’; 4 × 1’; 1 × 5’; 5 × 10’ frame, 0.4 × 0.4x0.4 mm voxels, 212 × 212y 235z matrix size, 4 iterations). All the sequences described previously were performed on a nanoScan PET/MRI (for the MRI and PET sequences) or nanoScan SPECT/CT scanner (for the CT sequences) both from Mediso. Euthanasia of the mice was conducted 60 min post-injection by cervical dislocation. The intramuscular grafts were then collected alongside a piece of reference muscle for measurement in a NaI well counter (Uppsala Imanet AB, Uppsala, Sweden).
The time-activity curve (TAC) derived from each frames of the PET scan were used as inputs to calculate the volume of distribution (Vt) via the Logan plot model using the kinetic modeling tool PKIN (PMOD technologies LLC). The left ventricle of the heart was used as the blood input function and corrected for the plasma to blood ratio as well as the metabolites parent fraction.
Immunofluorescent staining
SC-islets were fixed in 4% paraformaldehyde (PFA) at RT for 20 min and frozen. Muscle grafts were fixed overnight in 4% PFA at 4˚C, followed by incubation in 15% sucrose in PBS during 2–3 h and then 30% sucrose in PBS overnight at 4˚C and frozen. Cryosections of SC-islets or grafts were prepared with a thickness of 10 µm. Sections were blocked with 3% donkey serum (Jackson Immunoresearch Laboratories) in PBS for 30 min followed by staining with the primary polyclonal rabbit antibodies against DGCR2 (1:400, Thermo Fisher Scientific, #PA5-51,474), and polyclonal guinea pig antibodies against human insulin (1:400, Fitzgerald, Acton, #20-IP30) at 4 °C overnight. The next day, the following secondary antibodies diluted 1:300: Alexa Fluor 488-conjugated donkey anti-rabbit (Jackson Immunoresearch Laboratories) and Alexa Fluor 594-conjugated donkey anti-guinea pig (Jackson Immunoresearch Laboratories) were incubated for 1 h at RT. The nuclei were stained with DAPI. All images were obtained using a laser scanning confocal microscope Zeiss LSM 780.
Biodistribution of [18F]ZDGCR2:AM106 in rats and pigs
A MRI acquisition for attenuation and localization was first performed on n = 2 rats (Sprague Dawley, males, weight = 250–300 g) using the following parameters: 438.2 ms TR, 3.9 ms TE, 235 slices, 5 Th, 0 Gap, 80 mm FOV, 0.5 mm resolution, 2 NEX, 20 min acquisition time. Subsequently, 50 kBq/g of [18F]ZDGCR2:AM106 (equivalent to a peptide dose of 10 µg) dissolved in PBS and maximum 10% ethanol was injected intravenously through the tail vein of each rat. A dynamic PET imaging of 90 min was then acquired using the following parameters: 3 bed positions of 2 × 1.4’, 2 × 3.2’, 1 or 2 × 10’, 0.4 × 0.4 × 0.4 mm voxels, 212 × 21 2 × 583 matrix size. All of the sequences described previously were performed using a nanoScan PET/MRI scanner (Mediso).
A total of n = 2 pigs (Yorkshire x Swedish Landrace x Hampshire, males, weight = 25–30 kg) were anesthetized and intubated according to a standardized procedure previously described [
14]. A CT acquisition for attenuation and localization was first performed (100 kV, 80–400 mA, noise index 10, rotation 0.50, full spiral, slice thickness 3.75 mm, pitch 0.98:1, recon diameter 50 mm) before injecting approximately 30 MBq/kg of [
18F]Z
DGCR2:AM106 (equivalent to a peptide dose of 50 µg) dissolved in PBS. A dynamic PET imaging of 90 min was then acquired using the following parameters: 33 frames of 12 × 10, 6 × 30, 5 × 120, 5 × 300 and 5 × 600 s, VPFX-S, 3 i/16 s, 256 × 256 × 89 pixels, 3 mm post-filter, 500-mm-diameter zoom. Once the PET/CT dynamic scan was finished, a pancreas contrast CT acquisition using 3 mL/kg Omnipaque (GE Healthcare), bolus tracking with 100 HU threshold and region of interest placed in aorta, 15-s monitor delay for arterial phase, and 60-s delay for venous phase with 3.5 mL/s flow was performed. All of the sequences described previously were performed using a Discovery MI PET/CT scanner (GE Healthcare).
The resulting PET data were analyzed via manual segmentation using the PBAS modeling tool (PMOD technologies LLC). The uptake in kBq/cc was converted to standardized uptake values (SUV) by correcting for the administered radioactivity and the weight of each animal.
Discussion
DGCR2 is a putative excellent target for molecular imaging of mature beta cells as highlighted by previous studies [
10,
11]. In the present study, we demonstrate that DGCR2 is expressed in SC-islets similar to adult primary islets, with significant co-localization with insulin, and with stable expression over several weeks following engraftment in either the abdominal or neck muscle. The Affibody molecule Z
DGCR2:AM106 was labeled with fluorine-18, recognized for its low energy range thus offering better resolution, especially crucial for smaller structures such as pancreatic islets. The labeling was performed using a novel TCO–tetrazine click chemistry approach. Importantly, the TCO–tetrazine reaction involving the TCO-conjugated peptide can be performed using mild conditions (i.e., neutral pH and room temperature), therefore minimally affecting the peptide tertiary structure or binding capability. Introducing covalently bonded fluorine-18 via Inverse Electron Demand Diels–Alder (IEDDA) reactions offers multiple potential advantages when labeling large molecules such as proteins and peptides. This method facilitates site-specific labeling of the molecule, as the label is attached exclusively to a specific group on the molecule eliminating non-specific labeling. Such precision can help maintain the biological activity and function of the labeled molecule. Covalent bonding can also improve the stability of the labeled molecule. Non-covalently bound labels can be more prone to dissociation, which can affect the specificity of the labeled compound. In contrast, covalently bound labels are more stable and are less likely to dissociate or degrade, which can make the labeled molecule more reliable for use in imaging. With this in mind the Affibody molecule DGCR2, functionalized with a TCO group, was covalently radiolabeled with fluorine-18 using [
18F]MeTz via a IEDDA reaction. Similar to radiolabeling of large molecules with radiometals using chelating chemistry the separation of the TCO– Z
DGCR2:AM106 starting material from the [
18F]Z
DGCR2:AM106 was not feasible. Considering that the Affibody molecule was site-specifically functionalized with only one TCO group per molecule, the ratio between the radiolabeled tetrazine compound and the number of available TCO groups was concluded to be near equimolar, with a ratio between 1:1 and 1.8:1 based on the estimated concentrations of the two reactants. The ability to perform the fluorine-18 labeling and form covalent bonds at near equimolar conditions demonstrates the extreme efficiency of the IDDEA reaction. In this study, the amount of precursor material used was reduced by a factor of 1,000 compared to the typical conditions required for nucleophilic substitution reactions with [
18F]fluoride, which usually require micromolar quantities, enabling the labeling of [
18F]Z
DGCR2:AM106 with high apparent molar activity without the need for separation from its precursor. Additionally, [
18F]Z
DGCR2:AM106 demonstrated robust stability in human plasma (parent fraction > 90%) up to 90 min and therefore excluding any risks of stability issues (Additional File
2: Figure S1).
The binding capability of [
18F]Z
DGCR2:AM106 was first evaluated in vitro toward the recombinant human DGCR2 using an ELISA-based method and further cross-checked via bio-layer interferometry (Additional File
3: Figure S2). The binding was found to be maintained, suggesting the labeling process did not affect the binding capability of [
18F]Z
DGCR2:AM106 to DGCR2. Additionally, the frozen section autoradiography showcased a significantly higher binding of [
18F]Z
DGCR2:AM106 to frozen sections of SC-islets compared with frozen sections of human pancreatic endocrine islet fractions of 90% purity and Hek293 cells, suggesting that DGCR2 could be a more suitable biomarker for stem-cell-derived human pancreatic islets compared to native pancreatic islets.
Interestingly, in vivo visualization using [18F]ZDGCR2:AM106 was demonstrated in the pilot study using mice grafted with SC-islets, with confirmed DGCR2 expression. However, the uptake signal of [18F]ZDGCR2:AM106 within the SC-islets grafts was only apparent during the early frames of the PET scan (≈ 10 min p.i.). Thus, this observation of signal in the grafts may be explained either by rapid, increased tracer delivery early during the PET examination, perhaps due to a perfusion effect. Secondly, it may also be that [18F]ZDGCR2:AM106 binds reversibly to DGCR2, with a modest dissociation rate causing a decrease in uptake signal early during the scan.
To assess the biodistribution of [18F]ZDGCR2:AM106 and define an imaging window with optimal contrast, PET imaging was performed in healthy rats, as well as larger animals such as pigs who usually exhibit strong translational value and high homology to humans. The biodistribution in rats highlighted the strong renal clearance of [18F]ZDGCR2:AM106, expected to a certain extent from a peptide compound with low lipophilic properties. Similarly, the biodistribution of [18F]ZDGCR2:AM106 in pigs showed excretion of the tracer mainly through the kidneys. Low level of hepatic uptake could be observed in both rats and pigs likely due to the presence of some tetrazine metabolites, but are quickly cleared out. The TAC highlighted a peak in the renal clearance at 30 min p.i., emphasizing the rapid washout property of the Affibody molecule which is a desirable feature for an imaging probe to obtain optimal contrast during scanning. The close proximity of the kidneys with the pancreas and spleen poses a high risk of spill-over signal and thus highly impairs the possibility of native beta-cell imaging, but should not be a limitation for imaging ectopic transplanted grafts. Lastly, no uptake of [18F]ZDGCR2:AM106 within the brain could be detected, suggesting that the tracer is unable to cross the blood–brain barrier.
The exact function of DGCR2 is unknown, but the deletion of the associated gene causes 22q11.2DS (previously known as DiGeorge syndrome), a developmental disorder. Interestingly, 22q11.2DS was recently shown to also be associated with both defects in the immune system, obesity and Type 2 Diabetes [
15,
16]. Furthermore, DGCR2 is localized in a region on chromosome 22, which is the home of several long non-coding RNA sequences that have been identified in pancreatic beta cells, as well as islet enhancer clusters [
17,
18]. DGCR2 expression has also been independently reported on extracellular vesicles released from beta cells [
19].
A final reflection is that the selection of lead candidate Z
DGCR2:AM106 could be improved, notably by modifying the selection criteria from the affinity maturation process of the Affibody molecules [
20] by selecting affinity binders primarily focusing on an optimal dissociation rate. We could also lower the temperature selection threshold for the identified Affibody molecules, as the TCO–tetrazine click chemistry radiolabeling method described in this paper can be performed in mild and non-denaturing conditions.
In summary, we have presented the radiolabeling method based on TCO–tetrazine click chemistry to produce the highly stable [18F]ZDGCR2:AM106 for PET imaging of human DGCR2. [18F]ZDGCR2:AM106 exhibited retained binding capacity to both recombinant human DGCR2 and SC-islets sections in addition to a suitable biodistribution profile in multiple animal species. Successfully transplanted SC-islets grafts in immunodeficient mice were confirmed to retain expression of DGCR2 for many weeks, showing significant overlap with insulin staining. Lastly, the SC-islets could be visualized in vivo using [18F]ZDGCR2:AM106, although the Affibody molecule demonstrated rapid washout from the target tissue.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.