Advancing 89Zr-immuno-PET in neuroscience with a bispecific anti-amyloid-beta monoclonal antibody - The choice of chelator is essential

The accelerated approval of the monoclonal antibody (mAb) aducanumab as a treatment option for Alzheimer's Disease and the continued discussions about its efficacy have shown that a better understanding of immunotherapy for the treatment of neurodegenerative diseases is needed. 89Zr-immuno-PET could be a suitable tool to open new avenues for the diagnosis of CNS disorders, monitoring disease progression, and assessment of novel therapeutics. Herein, three different 89Zr-labeling strategies and direct radioiodination with 125I of a bispecific anti-amyloid-beta aducanumab derivate, consisting of aducanumab with a C-terminal fused anti-transferrin receptor binding single chain Fab fragment derived from 8D3 (Adu-8D3), were compared ex vivo and in vivo with regard to brain uptake and target engagement in an APP/PS1 Alzheimer's disease mouse model and wild type animals. Methods: Adu-8D3 and a negative control antibody, based on the HIV specific B12 antibody also carrying C-terminal fused 8D3 scFab (B12-8D3), were each conjugated with NCS-DFO, NCS-DFO*, or TFP-N-suc-DFO-Fe-ester, followed by radiolabeling with 89Zr. 125I was used as a substitute for 124I for labeling of both antibodies. 30 µg of radiolabeled mAb, corresponding to approximately 6 MBq 89Zr or 2.5 MBq 125I, were injected per mouse. PET imaging was performed 1, 3 and 7 days post injection (p.i.). All mice were sacrificed on day 7 p.i. and subjected to ex vivo biodistribution and brain autoradiography. Immunostaining on brain tissue was performed after autoradiography for further validation. Results: Ex vivo biodistribution revealed that the brain uptake of [89Zr]Zr-DFO*-NCS-Adu-8D3 (2.19 ±0.12 %ID/g) was as high as for its 125I-analog (2.21 ±0.15 %ID/g). [89Zr]Zr-DFO-NCS-Adu-8D3 and [89Zr]Zr-DFO-N-suc-Adu-8D3 showed significantly lower uptake (< 0.65 %ID/g), being in the same range as for the 89Zr-labeled controls (B12-8D3). Autoradiography of [89Zr]Zr-DFO*-NCS-Adu-8D3 and [125I]I-Adu-8D3 showed an amyloid-beta related granular uptake pattern of radioactivity. In contrast, the [89Zr]Zr-DFO-conjugates and the control antibody groups did not show any amyloid-beta related uptake pattern, indicating that DFO is inferior for 89Zr-immuno-PET imaging of the brain in comparison to DFO* for Adu-8D3. This was confirmed by day 7 PET images showing only amyloid-beta related brain uptake for [89Zr]Zr-DFO*-NCS-Adu-8D3. In wild type animals, such an uptake was not observed. Immunostaining showed a co-localization of all administered Adu-8D3 conjugates with amyloid-beta plaques. Conclusion: We successfully demonstrated that 89Zr-immuno-PET is suitable for imaging and quantifying amyloid-beta specific brain uptake using a bispecific aducanumab brain shuttling antibody, Adu-8D3, but only when using the novel chelator DFO*, and not DFO, for labeling with 89Zr.


Biodistribution
: Summary of all in vivo study groups; including number, sex, the weight of the mice, and corresponding ex vivo brain uptake day 7 p.i. by biodistribution. Results are expressed as mean (%ID/g) ±standard deviation (sd).    Results are expressed as mean (%ID/g) ±sd (n=4-5 mice per group). Significant differences between the groups are marked with asterisks (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001) or marked as non-significant (ns).   Figure S3: Corresponding biodistribution to Table S4 in APP/PS1 TG or WT mice, 7 days after injection of 30 µg radiolabeled protein. Results are expressed as mean (%ID/g) ±sd (n=4-5 mice per group).   Figure S5: Immunofluorescence analysis of brain sections of APP/PS1 mice. 11-13 month old APP/PS1 TG mice were injected with a fixed dose of 30 µg of radiolabeled protein. The same 20 µm cryo-sections used for autoradiography ( Figure 3) were stained with 0.125% Thioflavin S (yellow) and AF647-goat anti-human IgG (1:1000, purple) to detect injected antibody. Shown are the images of each separate and the merged channels, overlay of the two signals appears in white.

Quality controls -Size-exclusion high-performance liquid chromatography
Protein concentration was determined by size-exclusion high-performance liquid chromatography (SE-HPLC). In short, a Jasco or Shimadzu HPLC system was equipped with a Superdex® 200 Increase 10/300 GL (30 cm × 10 mm, 8.6 μm) size exclusion column (GE Healthcare Life Sciences) and a guard column using 0.05 M phosphate buffer/0.15 M NaCl/0.01 M NaN3 (pH = 6.7) as mobile phase with a run time of 40 min at 0.75 mL/min. Antibody concentration was assessed using the areas under the curve on the UV channel at 280 nm. The concentration was determined against a calibration curve of the unlabeled compound. All three used unmodified antibodies are shown in Figure S6-S8 and [ 89 Zr]Zr-DFO*-NCS-Adu-8D3 as an example in Figure S9.

Determination of the chelator-to-mAb ratio
To ensure high specific activity, we adapted our standard conjugation procedures by using higher equivalents of DFO-NCS, DFO*-NCS, or TFP-N-suc-DFO-Fe-ester during mAb conjugation, with longer reaction times and smaller reaction volumes to achieve a chelator-to-mAb ratio of around 2.5:1.
For the TFP-N-suc-DFO-Fe-ester modification of Adu-8D3, a sample was taken before the Fe-removal step and a SE-HPLC analysis was performed to determine the chelator-to-mAb ratio [3]. The ratio of the two signals from Fe-N-suc-DFO-Adu-8D3 and unreacted TFP-N-suc-DFO-Fe-ester at 430 nm was used to determine a chelator-to-mAb ratio of 2.85.
An SE-MS approach to analyse the chelator-to-mAb ratio, as reported by Sijbrandi et al. [4], for the analysis of antibody-drug conjugates did not provide reliable results for the DFO*-NCS and DFO-NCS modified mAb samples due to the thiourea bond in those conjugates as previously reported [5]. Furthermore, to evaluate the chelator-to-mAb ratio, we also performed a previously reported method by Chomet et al. [5] in which the chelator-mAb conjugate was not purified from unreacted chelator and radiolabeled with 89 Zr (1 MBq). This approach neither led to plausible results, probably because of the considerably higher equivalents (10 eq.) of DFO*-NCS or DFO-NCS used in our labelings. Finally, we considered titration methods as used by Vugts et al. [6]. However, this method could not be performed due to the large amounts of antibody needed, which makes this method more suitable for commercially available antibodies. Because of this, we have to rely on calculation for an estimated chelator-to-mAb ratio: with the molar activity of the used 89 Zr (20-30 MBq/nmol) and the achieved yields for both NCS modifications with 150-250 MBq 89 Zr and 500 µg (2.5 nmol) protein, it can be assumed that a chelator-to-mAb ratio of between 2:1 to 5:1 was reached.
Modification procedures and achieved radiolabelling yields for B12-8D3 and Adu-8D3 were similar, assuming a similar chelator-to-mAb ratio for the respective modification. Whole body PET images -day 7 p.i. Figure S11: Representative whole body PET/CT MIP images for all in vivo study groups day 7 p.i., injected with 30 µg of the radioimmunoconjugate.