89Zr-labeled ImmunoPET targeting the cancer stem cell antigen CD133 using fully-human antibody constructs

Background Cancer stem cells play an important role in driving tumor growth and treatment resistance, which makes them a promising therapeutic target to prevent cancer recurrence. Emerging cancer stem cell-targeted therapies would benefit from companion diagnostic imaging probes to aid in patient selection and monitoring response to therapy. To this end, zirconium-89-radiolabeled immunoPET probes that target the cancer stem cell-antigen CD133 were developed using fully human antibody and antibody scFv-Fc scaffolds. Results ImmunoPET probes [89Zr]-DFO-RW03IgG (CA = 0.7 ± 0.1), [89Zr]-DFO-RW03IgG (CA = 3.0 ± 0.3), and [89Zr]-DFO-RW03scFv − Fc (CA = 2.9 ± 0.3) were radiolabeled with zirconium-89 (radiochemical yield 42 ± 5%, 97 ± 2%, 86 ± 12%, respectively) and each was isolated in > 97% radiochemical purity with specific activities of 120 ± 30, 270 ± 90, and 200 ± 60 MBq/mg, respectively. In vitro binding assays showed a low-nanomolar binding affinity of 0.6 to 1.1 nM (95% CI) for DFO-RW03IgG (CA = 0.7 ± 0.1), 0.3 to 1.9 nM (95% CI) for DFO-RW03IgG (CA = 3.0 ± 0.3), and 1.5 to 3.3 nM (95% CI) for DFO-RW03scFv − Fc (C/A = 0.3). Biodistribution studies found that [89Zr]-DFO-RW03scFv − Fc (CA = 2.9 ± 0.3) exhibited the highest tumor uptake (23 ± 4, 21 ± 2, and 23 ± 4%ID/g at 24, 48, and 72 h, respectively) and showed low uptake (< 6%ID/g) in all off-target organs at each timepoint (24, 48, and 72 h). Comparatively, [89Zr]-DFO-RW03IgG (CA = 0.7 ± 0.1) and [89Zr]-DFO-RW03IgG (CA = 3.0 ± 0.3) both reached maximum tumor uptake (16 ± 3%ID/g and 16 ± 2%ID/g, respectively) at 96 h p.i. and showed higher liver uptake (10.2 ± 3%ID/g and 15 ± 3%ID/g, respectively) at that timepoint. Region of interest analysis to assess PET images of mice administered [89Zr]-DFO-RW03scFv − Fc (CA = 2.9 ± 0.3) showed that this probe reached a maximum tumor uptake of 22 ± 1%ID/cc at 96 h, providing a tumor-to-liver ratio that exceeded 1:1 at 48 h p.i. Antibody-antigen mediated tumor uptake was demonstrated through biodistribution and PET imaging studies, where for each probe, co-injection of excess unlabeled RW03IgG resulted in > 60% reduced tumor uptake. Conclusions Fully human CD133-targeted immunoPET probes [89Zr]-DFO-RW03IgG and [89Zr]-DFO-RW03scFv − Fc accumulate in CD133-expressing tumors to enable their delineation through PET imaging. Having identified [89Zr]-DFO-RW03scFv − Fc (CA = 2.9 ± 0.3) as the most attractive construct for CD133-expressing tumor delineation, the next step is to evaluate this probe using patient-derived tumor models to test its detection limit prior to clinical translation. Supplementary Information The online version contains supplementary material available at 10.1186/s13550-024-01091-9.


Background
Tumorigenic cancer stem cells (CSCs) demonstrate increased resistance to chemo-and radiotherapy treatments and these properties suggest a major role played by CSCs in driving cancer recurrence [1,2].Elevated CSC biomarkers can be found in aggressive, treatmentrefractory tumors including glioblastoma, breast, lung, and pancreatic cancer, and tumors rich in CSCs are more likely to recur and lead to worse patient outcomes [3][4][5][6][7].For these reasons, significant interest has been garnered towards the development of molecularly targeted therapies that can eliminate CSCs in tumors, such as immunotherapies, antibody-drug conjugates, and targeted radionuclide therapy [8][9][10].In contrast to CSC targeted molecular therapies, there has been comparatively less work done around the creation and evaluation of CSC targeted molecular imaging agents, either as companion diagnostics or standalone imaging agents.Imaging CSCs in tumors, which make up less than 1% of the tumor cell population, is challenging and may be best addressed through the use of high sensitivity imaging modalities like positron emission tomography (PET) using probes with potent (low-nanomolar) binding affinity for CSCselective antigens [11].
CD133 is a cell-surface glycoprotein overexpressed by CSCs in a variety of tumors making it a promising molecular target for CSC-targeted therapies and molecular imaging [12].Previous reports on CD133 imaging have labeled a humanized anti-CD133 antibody, AC133.1, with positron emitting radioisotopes copper-64 and zirconium-89 to successfully image CD133 expressing human colorectal and brain tumor xenograft mouse models [13,14].Other reports have used a zirconium-89 labeled CD133-targeted murine antibody derived from hybridoma clone B7 to image cell-line derived and orthotopic patient derived models of lung cancer [15,16].While these reports have established the feasibility of CD133 imaging using antibody constructs and advanced mouse models, further translation of these probes to the clinical setting may require further humanization due to potential immunogenicity of non-human antibody scaffolds [17].Furthermore, glycosylated Fc-regions present on these previously reported anti-CD133 antibody PET probes could induce Fc-effector immune responses that are undesired for molecular imaging applications [18].Minimally immunogenic, fully human antibody constructs with aglycosylated Fc-regions would better facilitate clinical translation of CD133 targeted companion diagnostic imaging probes.
Here we report the preparation and preclinical evaluation of a CD133-targeted imaging probe based on a fully human anti-CD133 antibody and an aglycosylated scFv-Fc antibody fragment with the same binding domain, both radiolabeled with zirconium-89 for PET.The immu-noPET probes were characterized for radiochemical yield and purity, immunoreactive fraction, and binding affinity prior to in vivo experiments.Tumor uptake and in vivo specificity of the probe were established through biodistribution and imaging studies using a CD133-expressing tumor xenograft mouse model.
[ 89 Zr]-oxalate was purchased from 3D Imaging (Little Rock, AR).The fully human anti-CD133 monoclonal antibody (RW03 IgG ) and scFv-Fc (RW03 scFv-Fc ) were provided by Dr. Jason Moffat (University of Toronto) under a research agreement with Century Therapeutics Canada.The Fab used for RW03 IgG and the scFv used for RW03 scFv-Fc were synthesized via the Cellectseq method from phage-display libraries, using CD133-transfected HEK293 cells and the parental HEK293 cells for positive and negative selection, respectively [19].

MALDI-MS
Matrix-Assisted Laser Desorption Ionization Mass Spectrometry (MALDI-MS) (Bruker UltrafleXtreme TOF/ TOF) was performed in positive mode to determine the average chelator-to-antibody scFv-Fc ratio (C/A).Samples were concentrated by centrifugal filtration to 5 mg/ mL and then diluted to 1 mg/mL with phosphate buffer (0.02 M, pH 7.4) and diluted 5-fold with a saturated solution of sinapinic acid in TA30 solvent (30:70 v/v acetonitrile: 0.1% TFA).The chelator-to-antibody ratio was calculated by subtracting the molecular weight of the parent IgG or scFv-Fc (MW S ) from the conjugate (MW C ) and dividing by the molecular weight of DFO (753 g/mol) (Eq.1).

Radiochemistry
Zirconium-89 radiolabeling was performed according to previously established procedures [20] (Scheme 1).In a typical labeling, 0.35 ± 0.15 mg conjugate was exchanged into HEPES buffer (1 M, pH 7.4) using a PD-10 column and concentrated to 1.25 mg/mL using centrifugal filtration.The [ 89 Zr]-oxalate stock containing 185 MBq was diluted to 250 µL with oxalic acid (1 M).The [ 89 Zr]-oxalate solution was slowly adjusted to pH 7 using a solution of Na 2 CO 3 (1 M), monitored using pH strips.A 100 µL aliquot containing 74 MBq of the neutralized zirconium-89 solution was added to 0.35 ± 0.15 mg (280 ± 120 µL) of conjugate and incubated for 1 h at 37 o C while shaking.The crude radiochemical yield was monitored by radio-iTLC at 0, 20, 40, and 60 min into the reaction.After 60 min, the reaction was quenched with 1 µl of EDTA (5 mM) for 10 min and purified using a PD-10 column equilibrated with ABST buffer.Fractions were characterized for radiochemical purity (RCP) by radio-iTLC, and fractions with > 97% RCP were pooled.
The lower limit for specific activity was calculated using the activity of the final (pooled) product and the mass of IgG or scFv-Fc added to the reaction.After purification the product was filtered (0.22 µM) and activity losses to the filter were recorded.

Stability analysis
The radiochemical stability of [ 89 Zr]-DFO-RW03 scFv − Fc was evaluated over 6 days in either saline or mouse serum.From the 37 MBq/mL stock solution of radioligand (in ABST), an aliquot (100 µL) was added to a 500 µL volume of either saline or mouse serum.Free zirconium-89 and soluble Zr 4+ complex presence was evaluated by radio-iTLC up to 1-week post-synthesis using EDTA (0.1 M, pH 7) as the eluent.

Flow cytometry
To assess the effect of conjugation on antibody binding, HT-29 cells (3 × 10 6 cells/mL) were incubated with native RW03 IgG , native RW03 scFv − Fc , or the DFO-RW03 conjugates to determine the half-maximal effective concentration (EC 50 ) [14].After incubation with the antibody or DFO-conjugate for 1 h on ice, cells were pelleted and washed to removed unbound antibody.Cells were then incubated with a secondary antibody (1:100 dilution) (Goat anti-Human IgG-FITC, BioRad, 204,002) for 30 min on ice.To stain dead cells, 7AAD (1:100) (Miltenyi 130-111-568) was added and incubated for at least 5 min.The median fluorescence intensity (MFI) for each measurement was plotted against antibody concentration and data was normalized to the highest MFI.The curves were fitted with a one-site binding sigmoidal curve using GraphPad Prism 8 to determine the EC 50 values.

Statistical analysis
Unless otherwise noted, data is presented as the mean ± SEM.All tests for statistical significance were performed using GraphPad Prism 8. Asymmetrical confidence intervals (95% CI) were used to statistically describe results for in vitro cell binding assays.Student's t-test was used to describe statistically significant differences (p < 0.05) in radioimmunoconjugate organ uptake between cohorts.
To further demonstrate the capability for [ 89 Zr]-DFO(2.9)-RW03scFv − Fc to accumulate in CD133 expressing tumors, a PET-imaging study was conducted where mice were administered [ 89 Zr]-DFO(2.9)-RW03scFv − Fc in the presence and absence of unlabeled RW03 (Fig. 3, Tables S4-3 & S4-4).Images of both cohorts at 1 h p.i. showed high uptake of [ 89 Zr]-DFO(2.9)-RW03scFv − Fc in blood pool (heart) (no RW03 IgG : 33 ± 5%ID/cc, with RW03 IgG : 38 ± 4%ID/cc) and low tumor uptake (< 3%ID/cc).The tumor uptake of [ 89 Zr]-DFO(2.9)-RW03scFv − Fc increased over the duration of the study, with tumor uptake at 72 h (20.3 ± 0.5%ID/cc) in good agreement with tumor uptake observed at 72 h in the biodistribution study (23 ± 4%ID/g) (see Fig. 1).As early as 24 h p.i. the tumor-to-heart ratio equalized in the non-blocked cohort and at this timepoint a significant reduction (p = 0.006) in tumor uptake was observed in the blocked cohort (10 ± 1%ID/cc) relative to the non-blocked cohort (15.6 ± 0.5%ID/cc).The tumor-to-blood ratio and tumor-to-liver ratio increased in the non-blocked cohort over the duration of the study, and both reached a maximum of 3:1 at 168 h p.i.The tumor-to-blood ratio and tumor-to-liver ratio in the blocked cohort equalized at 72 h and 24 h p.i., respectively, and both ratios stayed at 1:1 over the duration over the study (168 h p.i.).Uptake in the knee joint contralateral to the tumor was highest at the last imaging timepoint (168 h p.i.) in both cohorts (5.8 ± 0.6%ID/cc and 4.5 ± 0.2%ID/cc for the non-blocked and blocked cohorts, respectively).The ROI analysis at 168 h p.i. is in good agreement with ex vivo biodistribution performed on mice from the imaging study, which found tumor uptake of 18.2 ± 0.6%ID/g and 5 ± 1%ID/g in the non-blocked and blocked cohorts at 168 h, respectively (Fig. S4-1, Tables S4-5 & S4-6).

Discussion
Cancer stem cell-targeted immunoPET probes could aid in the development and utilization of CSC-targeted therapies by identifying CSC-rich tumors for patient selection or by detecting residual CSCs in patients following treatment [22].CD133 antigen is selective for CSCs in a wide variety of tumors [3][4][5][6][7], which has driven development of CD133-targeted therapies aimed to eliminate CSCs for improved treatment outcomes [8][9][10].To aid in clinical translation of these therapies, previously reported molecular imaging tools have utilized CD133-targeted humanized or murine monoclonal antibodies, which while demonstrating promising preclinical data, have faced barriers to clinical translation due to immunogenicity concerns [14,23].Here, we report the synthesis of CD133-targeting immunoPET probes, [ 89 Zr]-DFO(0.7)-RW03IgG , [ 89 Zr]-DFO(3.0)-RW03IgG , and [ 89 Zr]-DFO(2.9)-RW03scFv − Fc using a fully human IgG and scFv-Fc fragment, and demonstrate the capability of these probes to visualize CD133-expressing tumors in mice through PET imaging.
Intact IgGs and scFv-Fc antibody fragments demonstrate similar pharmacokinetics owing to the fragment  , d, f) uptake is expressed as mean radioactivity per mL tissue volume (%ID/cc) ± SEM crystallizable (Fc)-region, which prolongs blood circulation through saturable binding with Fc-receptors present in the liver and other tissues [24].For both IgGs and scFv-Fcs, the long half-life (3.3 days) of zirconium-89 and ability to radiolabel proteins under non-denaturing conditions make it an attractive isotope for immunoPET probes.We found that conjugation of RW03 scFv − Fc with DFO did not reduce the CD133-binding affinity of the conjugates relative to native RW03 IgG or RW03 scFv − Fc for up to 3 DFO per antibody, in-line with previously reported results [25].
Previous reports have demonstrated decreased tumor uptake at higher injected antibody doses, with one report finding that increasing the injected antibody dose 10-fold (from 1 to 10 µg) led to about a 50% decrease in tumor uptake (from ∼ 35 to 17%ID/g at 72 h p.i. for 1 and 10 µg injected doses, respectively) [28].PET imaging studies on [ 89 Zr]-DFO-RW03 scFv − Fc used a 16-fold higher antibody dose than that used in biodistribution studies (16.0 ± 0.2 µg vs. 1.0 ± 0.2 µg for PET and biodistribution studies, respectively) and tumor uptake between the two studies did not significantly differ (e.g.20.3 ± 0.5%ID/cc and 23 ± 4%ID/g for PET and biodistribution at 24 h p.i., respectively).For clinical studies it may be necessary to optimize the amount of cold antibody used.For example, in the ZIPHIR Trial (NCT01565200) patients received 37 MBq of [ 89 Zr]-labeled trastuzumab and 50 mg cold trastuzumab for a final effective specific activity of 0.74 MBq/mg.The high effective SA achieved here for [ 89 Zr]-DFO-RW03 scFv − Fc (220 MBq/mg) is promising for imaging the rare CSC population in tumors and provides a large window for titrating the amount of cold RW03 IgG that can be co-administered to reduce off-target uptake in a clinical setting.

Summary and conclusions
[ 89 Zr]-DFO-RW03 scFv − Fc , a CD133 targeting immu-noPET probe based on a fully human monoclonal antibody was developed and its ability to target CD133expressing tumors demonstrated through biodistribution and PET imaging studies.These showed favorable tumor uptake and low off-target uptake, including in the liver and bone.[ 89 Zr]-DFO-RW03 scFv − Fc is a promising immu-noPET tracer as a companion diagnostic and warrants further evaluation in PDX tumor models that better represent CSC populations in patient tumors.

Fig. 3 [
Fig. 3 [ 89 Zr]-DFO-RW03 scFv-Fc ImmunoPET imaging CD133 expressing tumors.[ 89 Zr]-DFO(2.9)-RW03scFv-Fc PET imaging study (a & c) and corresponding ROI analysis (b & d).Mice received (a & b) [ 89 Zr]-DFO(2.9)-RW03scFv-Fc (1.42 ± 0.07 MBq, 16.0 ± 0.2 ug) or (c & d) [ 89 Zr]-DFO(2.9)-RW03scFv-Fc (1.40 ± 0.09 MBq, 16.0 ± 0.3 ug) co-injected with RW03 IgG (0.5 mg).For PET images, the transverse and coronal planes are presented as maximum intensity projections (MIP).White arrow = tumor.For ROI analysis (b, d) uptake is expressed as mean radioactivity per mL tissue volume (%ID/cc) ± SEM experiments performed at McMaster University were approved by the Animal Research Ethics Board (AREB) at McMaster University.Animal experiments conducted with the Spatio-Temporal Targeting and Amplification of Radiation Response (STTARR) Innovation Centre (Toronto, Canada) were approved by the University Health Network (UHN).Animal experiments adhered to all applicable institutional and national guidelines, including guidelines of the Canadian Council on Animal Care and the Committee for Research and Ethi- cal Issues of the International Association for the Study of Pain.Female BALB/c nu/nu mice (Charles River Laboratories, St Constant, QC) at 4-6 weeks of age were sterile housed and maintained at 24 °C with a 12 h light/ dark cycle and were provided autoclaved food and water ad libitum.HT-29 xenografts were established in the hind limb by subcutaneous injection of 2 × 10 6 cells formulated in 1:1 sterile DPBS (ThermoFisher 14,287,080)/ Matrigel BD (Matrigel Matrix High Concentration, Phenol Red-Free, LDEV-Free, Corning 354,262).Radiotracer injections were performed 10 days post-tumor engraftment.All radiotracer injections were intravenous (i.v.) and administered through the lateral tail vein.Mice were humanely euthanized under anesthesia (5% isoflurane) by cervical dislocation.For biodistribution studies, blood, adipose tissue, bone, brain, heart, kidneys, large intestine and cecum (with contents), liver, lungs, pancreas, skeletal muscle, small intestine (with contents), spleen, stomach (with contents), bladder with urine and tail were collected at pre-determined timepoints, weighed, and counted in an automated gamma counter (Wallac Wizard 1470 gamma counter, PerkinElmer).Decay correction was used to normalize organ activity measurements to the time of dose preparation for data calculations with respect to the injected dose (reported as % injected dose per gram (%ID/g) and per organ (%ID/O)).